in health
Transcripción
in health
Volume 14 Supplement 1 July/August 2011 ISSN 1098-3015 Supplement to IN HEALTH 1st Special Issue: Pharmacoeconomics and Outcomes Research in Latin America www.ispor.org VALUE IN HEALTH VOLUME 14 SUPPLEMENT 1 JULY/AUGUST 2011 PAGES S1 - S152 ELSEVIER EDITORIAL BOARD Co-Editor-in-Chief Michael Drummond, PhD University of York Heslington, York, UK [email protected] Co-Editor-in-Chief C. Daniel Mullins, PhD University of Maryland Baltimore, MD, USA [email protected] CO-EDITORS EDITORIAL ADVISORY BOARD Chris Bingefors, PhD, MSc Uppsala University Uppsala, Sweden [email protected] Alan Brennan, PhD University of Sheffield Sheffield, UK [email protected] Jalpa Doshi, PhD University of Pennsylvania Philadelphia, PA, USA [email protected] Sheri Fehnel, PhD, MA RTI Health Solutions Research Triangle Park, NC, USA [email protected] Benjamin P. Geisler, MD, MPH Wing Tech, Inc. Blue Hill, ME, USA [email protected] Dan Greenberg, PhD Ben-Gurion University of the Negev Beer-Sheva, Israel [email protected] John Hornberger, MD, MS Cedar Associates, LLC Menlo Park, CA, USA [email protected] Teresa Kauf, PhD University of Florida Gainesville, FL, USA [email protected] Gordon G. Liu, PhD Peking University, Guanghua School of Management PR China [email protected] Andrew Lloyd, DPhil, BSc Oxford Outcomes, Ltd. Oxford, UK [email protected] Andrea Manca, PhD, MSc University of York York, UK [email protected] Michelle Naughton, PhD Wake Forest University School of Medicine Winston-Salem, NC, USA [email protected] Paul Scuffham, PhD, BA Griffith University - School of Medicine Queensland, Australia [email protected] Johan L. (Hans) Severens, PhD Erasmus University Rotterdam, The Netherlands [email protected] Ya-Chen (Tina) Shih, PhD, MS University of Chicago Chicago, IL, USA [email protected] Ulla S. Skjoldborg, PhD, MA Eli Lilly Denmark A/S Copenhagen, Denmark [email protected] Marc L. Berger, MD Ingenix Life Sciences New York, NY, USA [email protected] Mark Nuijten, PhD, MD, MBA Ars Accessus Medica Dorpsstraat, The Netherlands [email protected] Andrew Briggs, DPhil University of Glasgow Glasgow, UK [email protected] Chris L. Pashos, PhD United BioSource Corporation Lexington, MA, USA [email protected] DS Pete Fullerton, PhD Strategic Pharmacy Innovations Seattle, WA, USA [email protected] Jean Paul Gagnon, PhD Pittstown, NJ, USA [email protected] Henry Glick, PhD University of Pennsylvania Philadelphia, PA, USA [email protected] Don Husereau, MSc, BSc University of Ottawa Ottawa, ON, Canada [email protected] Dennis Revicki, PhD UBC Bethesda, MD, USA [email protected] Paul Kind University of York Heslington, York, UK [email protected] Paul C. Langley, PhD University of Minnesota Woodbury, MN, USA [email protected] Pablo Lapuerta, MD Bristol-Myers Squibb Princeton, NJ, USA [email protected] Adrian Levy, PhD Dalhousie University Halifax, NS, Canada [email protected] Steven E. Marx, PharmD, MS Abbott Laboratories Abbott Park, IL, USA [email protected] Karl A. Matuszewski, MS, PharmD First DataBank, Inc. South San Francisco, CA, USA [email protected] William F. McGhan, PharmD, PhD University of the Sciences in Philadelphia Philadelphia, PA, USA [email protected] Richard J. Milne, PhD University of Auckland Auckland, New Zealand [email protected] Louis A. Morris, PhD Louis A. Morris & Associates Dix Hills, NY, USA [email protected] Peter J. Neumann, ScD Tufts—New England Medical Center Boston, MA, USA [email protected] Kevin A. Schulman, MD Duke Clinical Research Institute Durham, NC, USA [email protected] Uwe Seibert, MD, MPH, MSc, ScD University of Health Sciences, Medical Informatics & Technology Hall i.T., Austria [email protected] Sean Sullivan, PhD University of Washington Seattle, WA, USA [email protected] Milton C. Weinstein, PhD Harvard School of Public Health Boston, MA, USA [email protected] MANAGEMENT ADVISORY BOARD Bong-Min Yang, PhD (Chair) Seoul University Seoul, South Korea [email protected] Donald Patrick, PhD, MSPH University of Washington Seattle, WA, USA [email protected] Shelby Reed, PhD, RPh Duke University Durham, NC, USA [email protected] EDITORIAL OFFICE Managing Editor Stephen L. Priori ISPOR Lawrenceville, NJ, USA [email protected] Editorial Assistant Danielle Mroz ISPOR Lawrenceville, NJ, USA [email protected] Editorial Office. Value in Health, ISPOR, 3100 Princeton Pike, Suite 3E, Lawrenceville, NJ 08648. ISPOR Office. Marilyn Dix Smith, RPh, PhD, Executive Director, 3100 Princeton Pike, Suite 3E, Lawrenceville, NJ 08648. Tel: (609) 219-0773, Fax: (609) 2190774, E-mail: [email protected], Web site: http://www.ispor.org/valueinhealth_ index.asp. CUSTOMER SERVICE (orders, claims, online, change of address): Elsevier Health Sciences Division, Subscription Customer Service, 3251 Riverport Lane, Maryland Heights, MO 63043. Tel: (800) 654-2452 (U.S. and Canada); (314) 447-8871 (outside U.S. and Canada). Fax: (314) 447-8029. E-mail: [email protected] (for print support); [email protected] (for online support). Address changes must be submitted four weeks in advance. YEARLY SUBSCRIPTION RATES: United States and possessions: Individual $250.00; Institution $535.00. All other countries (prices include airspeed delivery): Individual $250.00; Institution $535.00. Single Issues $67.00. To receive student/resident rate, orders must be accompanied by name of affiliated institution, date of term and the signature of program/residency coordinator on institution letterhead. Orders will be billed at the individual rate until proof of status is received. Current prices are in effect for back volumes and back issues. Further information on this journal is available from the Publisher or from this journal’s website (http://www.elsevier.com/locate/jval). Information on other Elsevier products is available through Elsevier’s website (http://www. elsevier.com). Author inquiries For inquiries relating to the submission of articles (including electronic submission where available), visit http:/www.elsevier.com/authors. The site also provides the facility to track accepted articles and set up e-mail alerts to inform you of when an article’s status has changed, as well as detailed artwork guidelines, copyright information, frequently asked questions, and more. Please see Information for Authors for individual journal. Contact details for questions arising after acceptance of an article, especially those relating to proofs, are provided after registration of an article for publication. English language help service: Upon request, Elsevier will direct authors to an agent who can check and improve the English of their paper (before submission). Please contact [email protected] for further information. Reprints. For queries about author offprints, e-mail authorsupport@ elsevier.com. To order 100 or more reprints for educational, commercial, or promotional use, contact the Commercial Reprints Department, Elsevier Inc., 360 Park Avenue South, New York, NY 10010-1710. Fax: (212) 462-1935; email [email protected]. Reprints of single articles available online may be obtained by purchasing Pay-Per-View access for $36 per article on the journal website http://www.elsevier.com/locate/jval. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. This journal and the individual contributions contained in it are protected under copyright by International Society for Pharmacoeconomics and Outcomes Research, and the following terms and conditions apply to their use: Photocopying Single photocopies of single articles may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier’s Rights Department in Oxford, UK: phone (215) 238-7869 or ⫹44 (0) 1865 843830, fax ⫹44 (0) 1865 853333, e-mail [email protected]. Requests may also be completed online via the Elsevier homepage (http://www.elsevier.com/locate/ permissions). In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (978) 750-8400, fax: (978) 750-4744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (⫹44) 20 7631 5555; fax: (⫹44) 20 7631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Subscribers may reproduce tables of contents or prepare lists of articles including abstracts for internal circulation within their institutions. Permission of the Publisher is required for resale or distribution outside the institution. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this journal, including any article or part of an article. Except as outlined above, no part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier Rights Department, at the fax and e-mail addresses noted above. Abstracting and Indexing Value in Health is indexed in Index Medicus/MEDLINE, Current Contents/Social & Behavioral Sciences, SciSearch/SCI Expanded, Social Sciences Citation Index, International Pharmaceutical Abstracts, Embase/Excerpta Medica, and PsychInfo/Psychological Abstracts, and Journal Citation Reports/Science Edition (Thomson ISI). Notice No responsibility is assumed by the Publisher or the International Society for Pharmacoeconomics and Outcomes Research for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. Although all advertising material is expected to conform to ethical (medical) standards, inclusion in this publication does not constitute a guarantee or endorsement of the quality or value of such product or of the claims made of it by its manufacturer. Supplement to VOLUME 14 SUPPLEMENT 1 JULY/AUGUST 2011 TABLE OF CONTENTS S1 EDITORIAL Pharmacoeconomics and Outcomes Research in Latin America: A Promising and Developing Field Federico Augustovski and Marcos Bosi Ferraz S3 HEALTH POLICY ANALYSIS Implementing Pharmacoeconomic Guidelines in Latin America: Lessons Learned Federico Augustovski, Guillermo Melendez, Alexandre Lemgruber, and Michael Drummond S8 Is Equity of Access to Health Care Achievable in Latin America? Gabriela Tannus Branco Araújo, Joaquı́n E. Caporale, Stephen Stefani, Diana Pinto, and Antonio Caso S13 Insurance Status and Demographic and Clinical Factors Associated with Pharmacologic Treatment of Depression: Associations in a Cohort in Buenos Aires Gerardo Machnicki, Carol Dillon, and Ricardo F. Allegri S16 Precios de Referencia Internacional y Análisis de Costo Minimización para la Regulación de Precios de Medicamentos en Colombia Caludia Vacca, Angela Acosta, and Ivan Rodriguez S20 ECONOMIC ANALYSIS Hospitalization Costs for Heart Failure in People with Type 2 Diabetes: Cost-Effectiveness of its Prevention Measured by a Simulated Preventive Treatment Joaquı́n E. Caporale, Jorge Elgart, Guillermina Pfirter, Pablo Martı́nez, Gloria Viñes, Jorge T. Insúa, and Juan J. Gagliardino S24 Chronic Hepatitis B Treatment: The Cost-Effectiveness of Interferon Compared to Lamivudine Alessandra Maciel Almeida, Anderson Lourenço da Silva, Mariângela Leal Cherchiglia, Eli Iola Gurgel Andrade, Gustavo Laine Araújo de Oliveira, and Francisco de Assis Acurcio S29 Análise de Custo-Efetividade da Sinvastatina versus Atorvastatina na Prevenção Secundária de Eventos Cardiovasculares no Sistema Único de Saúde Brasileiro Denizar Vianna Araujo, Camila Pepe Ribeiro de Souza, Luciana Ribeiro Bahia, Helena Cramer Veiga Rey, Braulio dos Santos Junior, Bernardo Rangel Tura, Otavio Berwanger, Anna Maria Buehler, and Marcus Tolentino Silva S33 Costo-Efectividad del Cardiodesfibrilador Implantable en Pacientes con Factores de Riesgo de Muerte Súbita en Argentina Andrea Alcaraz, Jorge González-Zuelgaray, and Federico Augustovski S39 Costo Efectividad de Posaconazol versus Fluconazol/Itraconazol en el Tratamiento Profiláctico de las Infecciones Fúngicas Invasivas en México Kely Rely, Pierre K. Alexandre, and Guillermo Salinas Escudero S43 Costo-Efectividad del Tratamiento de Salmeterol/Fluticasona en Comparación con Leucotrieno Montelukast para el Control del Asma Infantil Kely Rely, Sebastián Emanuel González McQuire, Pierre K. Alexandre, and Guillermo Salinas Escudero S48 Costos de la Esclerosis Múltiple en Colombia Martin Romero, Carlos Arango, Nelson Alvis, Juan Camilo Suarez, and Aristides Duque TABLE OF CONTENTS - continued S51 Development and Validation of a Microsimulation Economic Model to Evaluate the Disease Burden Associated with Smoking and the Cost-Effectiveness of Tobacco Control Interventions in Latin America Andres Pichon-Riviere, Federico Augustovski, Ariel Bardach, and Lisandro Colantonio, for the Latinclen Tobacco Research Group S60 Estimation and Comparison of EQ-5D Health States’ Utility Weights for Pneumoccocal and Human Papillomavirus Diseases in Argentina, Chile, and the United Kingdom Julieta Galante, Federico Augustovski, Lisandro Colantonio, Ariel Bardach, Joaquin Caporale, Sebastian Garcia Marti, and Paul Kind S65 Evaluación Económica de un Programa de Inmunización Infantil en México Basado en la Vacuna Neumocócica Conjugada 13-Valente Emilio Muciño-Ortega, Joaquı́n Federico Mould-Quevedo, Raymond Farkouh, and David Strutton S71 Gastos do Ministério da Saúde do Brasil com Medicamentos de Alto Custo: Uma Análise Centrada no Paciente Cristina Mariano Ruas Brandão, Augusto Afonso Guerra Júnior, Mariângela Leal Cherchiglia, Eli Iola Gurgel Andrade, Alessandra Maciel Almeida, Grazielle Dias da Silva, Odilon Vanni de Queiroz, Daniel Resende Faleiros, and Francisco de Assis Acurcio S78 Health Care Resource Use and Costs in Opioid-Treated Patients with and without Constipation in Brazil Maira L. S. Takemoto, R. A. Fernandes, G. R. Almeida, R. D. C. Monteiro, M. Colombini-Neto, and A. Bertola-Neto S82 Ideal Vial Size for Bortezomib: Real-World Data on Waste and Cost Reduction in Treatment of Multiple Myeloma in Brazil Luciana Clark, Ana Paula Castro, Anna Flávia Fortes, Fábio Santos, Otávio Clark, Tobias Engel, Bruna Pegoretti, Vanessa Teich, Denizar Vianna, and Fabı́ola Puty S85 Costos de la Diabetes en Ameŕica Latina: Evidencias del Caso Mexicano Armando Arredondo and Esteban De Icaza S89 Short-Term Therapy with Enoxaparin or Unfractionated Heparin for Venous Thromboembolism in Hospitalized Patients: Utilization Study and Cost-Minimization Analysis Catia Argenta, Maria Angélica Pires Ferreira, Guilherme Becker Sander, and Leila Beltrami Moreira S93 The Burden of Moderate/Severe Premenstrual Syndrome and Premenstrual Dysphoric Disorder in a Cohort of Latin American Women Alexandre Schiola, Julia Lowin, Marion Lindemann, Renu Patel, and Jean Endicott S96 HEALTH OUTCOMES ANALYSIS Análisis de la Satisfacción con los Cuidados en Salud a Través del Cuestionario EORTC IN-PATSAT32 en Pacientes con Cáncer de Mama, Linfoma no Hodgkin y Cáncer Colo-Rectal en Diferentes Etapas Clı́nicas. Relación con las Caracterı́sticas Socio-Demográficas, Estados Co-Mórbidos y Variables del Proceso de Atención en el Instituto Mexicano del Seguro Social Luz-Ma-Adriana Balderas-Peña, Daniel Sat-Muñoz, Iris Contreras-Hernández, Pedro Solano-Murillo, Guillermo-Allan Hernández-Chávez, Ignacio Mariscal-Ramı́rez, Martha Lomelı́-Garcı́a, Margarita-Arimatea Dı́az-Cortés, Joaquı́n-Federico Mould-Quevedo, Juan-Manuel Castro-Cervantes, Oscar-Miguel Garcés-Ruiz, and Gilberto Morgan-Villela S100 Cost-Effectiveness of Supervised Exercise Therapy in Heart Failure Patients Eduardo M. Kühr, Rodrigo A. Ribeiro, Luis Eduardo P. Rohde, and Carisi A. Polanczyk S108 Estimating the SF-6D Value Set for a Population-Based Sample of Brazilians Luciane N. Cruz, Suzi A. Camey, Juliana F. Hoffmann, Donna Rowen, John E. Brazier, Marcelo P. Fleck, and Carisi A. Polanczyk S115 Factores Asociados al Incumplimiento de los Tratamientos con Antidepresivos en Santiago, Chile Marcela Jirón, Leslie Escobar, Leonardo Arriagada, Sebastián Orellana, and Ariel Castro TABLE OF CONTENTS - continued S119 Health-Related Quality of Life of Patients Recieving Hemodialysis and Peritoneal Dialysis in São Paulo, Brazil: A Longitudinal Study Mirhelen Mendes de Abreu, David R. Walker, Ricardo C. Sesso, and Marcos B. Ferraz S122 Health Status and Health Behaviors in Venezuelan Pharmacy Students Yajaira M. Bastardo S126 Influence of Organic and Functional Dyspepsia on Work Productivity: The HEROES-DIP Study Guilherme Becker Sander, Luiz Edmundo Mazzoleni, Carlos Ferrnando de Magalhães Francesconi, Giácomo Balbinotto, Felipe Mazzoleni, Andre Castagna Wortmann, Israel de Quadros Cardoso, Alexandre Luis Klamt, and Tobias Cancian Milbradt, on behalf of the Helicobacter Eradication Relief of Dyspetic Symptoms Trial Investigators S130 Evaluación de la Calidad de Vida en Pacientes con Linfoma no Hodgkin y Cáncer Colo-Rectal en Diferentes Etapas Clı́nicas Atendidos en el Instituto Mexicano del Seguro Social Luz-Ma-Adriana Balderas-Peña, Daniel Sat-Muñoz, Iris Contreras-Hernández, Pedro Solano-Murillo, GuillermoAllan Hernández-Chávez, Ignacio Mariscal-Ramı́rez, Martha Lomelı́-Garcı́a, Margarita-Arimatea Dı́az-Cortés, Joaquı́nFederico Mould-Quevedo, Ulises Palomares-Chacón, César-Adrián Balderas-Peña, Oscar-Miguel Garcés-Ruiz, and Gilberto Morgan-Villela S133 Calidad de Vida en Mujeres Mexicanas con Cáncer de Mama en Diferentes Etapas Clı́nicas y su Asociación con Caracterı́sticas Socio-Demográficas, Estados Co-Mórbidos y Caracterı́sticas del Proceso de Atención en el Instituto Mexicano del Seguro Social Daniel Sat-Muñoz, Iris Contreras-Hernández, Luz-Ma-Adriana Balderas-Peña, Guillermo-Allan Hernández-Chávez, Pedro Solano-Murillo, Ignacio Mariscal-Ramı́rez, Martha Lomelı́-Garcı́a, Margarita-Arimatea Dı́az-Cortés, Joaquı́nFederico Mould-Quevedo, Alma-Rosa López-Mariscal, Sergio-Emilio Prieto-Miranda, and Gilberto Morgan-Villela S137 The Costs of Type 2 Diabetes Mellitus Outpatient Care in the Brazilian Public Health System Luciana R. Bahia, Denizar Vianna Araujo, Beatriz D. Schaan, Sérgio A. Dib, Carlos Antônio Negrato, Marluce P.S. Leão, Alberto José S. Ramos, Adriana C. Forti, Marı́lia B. Gomes, Maria Cristina Foss, Rosane A. Monteiro, Daniela Sartorelli, and Laércio J. Franco S141 The Use of a Decision Board to Elicit Brazilian Patients’ and Physicians’ Preferences for Treatment: The Case of Lupus Nephritis Mirhelen Mendes de Abreu, Amiran Gafni, and Marcos Bosi Ferraz S147 Patrones de Tratamiento y Costo de Atención del Cáncer de Mama Avanzado Con Falla a Antraciclinas y Taxanos en 3 Hospitales Públicos de México Juan Alejandro Silva, Juan Francisco Gonzalez, Juan Enrique Bargalló, Gabriela Hernández-Rivera, Xóchitl Gómez-Roel, Sigfrido Rangel, Juan Jesús Vargas-Valencia, Jonathan Martı́nez-Fonseca, Bonnie Meyer Korenblat Donato, and Ariadna Juárez-Garcı́a S151 Reviewer Acknowledgement 1st Special Issue: Pharmacoeconomics & Outcomes Research in Latin America Special Issue Co-Editors Marcos Bosi Ferraz, MD, PhD Associate Professor, Department of Medicine, Federal University of São Paulo (UNIFESP). Director of the São Paulo Center for Health Economics (CPES), FAP UNIFESP. Director of Medical Economics of the Brazilian Medical Association. São Paulo – Brazil [email protected] Federico Augustovski, MD, MSc, PhD Director, Economic Evaluations and HTA Department, Institute for Clinical Effectiveness and Health Policy (IECS). Professor of Public Health, University of Buenos Aires. Staff Physician, Family and Community Medicine Division, Italian Hospital of Buenos Aires. Buenos Aires, Argentina [email protected] The International Society for Pharmacoeconomics and Outcomes Research recognizes the following companies for their financial support of this Value in Health Special Issue: AstraZeneca Mexico Axia.Bio Bayer de Mexico S.A. de C.V Eli Lilly and Company Merck Novartis The opinions or views expressed in this Value in Health Special Issue are those of the authors and do not necessarily reflect the opinions or recommendations of the publisher, guest editors, sponsor(s) or ISPOR. Articles may discuss pharmaceutical products and/or uses of products that have not been approved by the US Food and Drug Administration or other regulatory authorities outside the United States. VALUE IN HEALTH 14 (2011) S1–S2 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval EDITORIAL Pharmacoeconomics and Outcomes Research in Latin America: A Promising and Developing Field We are proud to present this special issue of Value in Health: Pharmacoeconomics and Outcomes Research in Latin America. We present a combination of plenary summaries presented at the International Society of Pharmacoeconomics and Outcomes Research (ISPOR) Second Latin America Conference held in Brazil during September 2009, and original research studies conducted in our region. We think this corpus gives a very rich picture of the current state of pharmacoeconomics and outcomes research in Latin America. Like the ISPOR Asia Consortium [1,2], the ISPOR Latin America Consortium promotes biennial conferences held in the region with a two-fold mission: to help develop knowledge and capacity for health economics and outcomes research in Latin America and to promote the use of health economics and outcomes research in policy-making processes, with the goal of improving resource allocation efficiency. This second ISPOR Latin American conference, with its theme “Uniting Research and Policy to Improve Health Care in Latin America,” had a major level of participation and success in the region. During the past decade or so, a large number of Latin American countries have undertaken profound reforms of their health care systems. Even though the details are country-specific, a common issue is the need to establish a mechanism that ensures efficient allocation of scarce resources, as well as guaranteeing a wide provision of health care services on the basis of local population needs and equity [3]. We are currently witnessing the adolescence of the health economics and outcomes research field in Latin America after reaching some relevant regional milestones, namely the establishing of the Latin American branch of the International Clinical Epidemiology Network during the 1980s, development of the Thematic Network on the Economic Evaluation of Health Programmes in Latin America (NEVALAT) in the early 2000s, the growth of the Cochrane Collaboration, the creation of several health technology assessment agencies in the region affiliated with the International Network of Agencies for Health Technology Assessment around 2005, as well as the creation of the ISPOR Latin America Consortium and local country chapters acting from 2006 on. In addition, regional common country alliances like MERCOSUR and the Andean Community are also advancing in harmonizing health technology assessment and economic evaluations in their member countries. It has been a great learning experience to be the editors of this special issue, the review process of which adheres to current standards for international scientific journals. Both the process and the results reflect the challenges of a region like Latin America, which is in the early phases of strengthening the health economics and outcomes research field [3]. Barriers include the scarcity of good and methodologically sound questions, difficulty in using state-of-the-art methods and tools, paucity of local primary data and therefore reliance on international data in the field of modeling studies, and a limited discussion of applicability of results to inform and influence decisions [4]. A recent study [3] concluded that to promote the use of economic evaluation in this region, two main conditions need to be fulfilled: 1) adequate resources and skills need to be available; and 2), decision-making process needs to be modified to accommodate evidence-based approaches such as economic evaluations [3]. From the 57 submissions we received, 30 articles were accepted for final publication in the current issue. We have representation from six countries, and the topics range from economic analysis (53%), to health outcomes analysis (40%) and health policy analysis (7%). Though all abstracts are in English, we decided to publish not only studies written in English (50%) but also in Spanish (40%) and in Portuguese (10%), because these are the two main languages used in Latin America. A majority of the issue reviewers were from Latin America (64%), which also shows the growth of interest in the field. Reviews were usually heterogeneous and sometimes discordant, and we made a large effort to try to reconcile them as much as possible. We are a part of the first attempt to publish a special issue of original research in the field conducted in Latin America, with plenty of examples of studies in this growing discipline. We are aware that this is an evolving field in our region, and that most of our production does not satisfy the most rigorous international standards (i.e., descriptive studies of a highly selected population, economic evaluation with scarcity of good quality local data inputs, small sample size, basic analytical approaches in observational studies, or modeling techniques in economic evaluations). Nevertheless, we think the glass is half full, with many regional efforts reflected. We hope you enjoy what to our knowledge is a milestone that reflects the current state of the art of the health economics and outcomes research field in Latin America. It is our desire that it can act as a stimulus to our community to strengthen our capacities and standards to contribute to evidence-based resource allocation decisions in this rapidly growing region. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. S2 VALUE IN HEALTH 14 (2011) S1–S2 Federico Augustovski, MD, MSc, PhD Director, Departamento de Evaluaciones Económicas y de Tecnologías Sanitarias/Economic Evaluations and Heath Technology Assessment Department, Instituto de Efectividad Clínica y Sanitaria/Institute for Clinical Effectiveness and Health Policy; and profesor de Salud Pública/Professor of Public Health, Universidad de Buenos Aires, Servicio de Medicina Familiar y Comunitaria/Family and Community Medicine Division, Hospital Italiano de Buenos Aires, Argentina. Marcos Bosi Ferraz, MD, PhD Department of Medicine, Federal University of São Paulo; Center for Health Economics, São Paulo, Brazil; and Medical Economics of the Brazilian Medical Association, São Paulo, Brazil. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.06.001 REFERENCES [1] Yang BM, Lee K. Growing application of pharmacoeconomics and outcomes research in health-care decision-making in the Asia-Pacific region. Value Health 2009;12(Suppl. 3):S1–2. [2] Liu GG, Eggleston K, Hu TW. Emerging health economics and outcomes research in the Asia-Pacific region. Value Health 2008;11(Suppl. 1):S1–2. [3] Iglesias CP, Drummond MF, Rovira J; NEVALAT Project Group.Healthcare decision-making processes in Latin America: problems and prospects for the use of economic evaluation. Int J Technol Assess Health Care 2005;21:1–14. [4] Augustovski F, Iglesias C, Manca A, et al. Barriers to generalizability of health economic evaluations in Latin America and the Caribbean region. Pharmacoeconomics 2009;27:919 –29. VALUE IN HEALTH 14 (2011) S3–S7 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval HEALTH POLICY ANALYSIS Implementing Pharmacoeconomic Guidelines in Latin America: Lessons Learned Introduction Latin America is a large and diverse region. Therefore it is no surprise that the development and implementation of pharmacoeconomic (PE) guidelines is proceeding at a different pace in different jurisdictions. The first three articles in this special issue of Value in Health focusing on Latin America outline the challenges in the development and implementation of guidelines in three country clusters. The three clusters represent countries at different stages in the development and implementation of guidelines at the time when the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) Latin American conference was held in Brazil (September 2009). However there have been rapid changes in some countries during the past year, especially in Colombia. In Argentina, Colombia, Guatemala, Uruguay, and Venezuela, PE guidelines were only under consideration at the time of the conference. In Chile and Mexico, PE guidelines were being developed, or were already developed. In Brazil, PE guidelines were already in the implementation phase and formed the basis of industry submissions to the Office of Economic Regulation at the Brazilian Health Surveillance Agency (ANVISA), the agency that decides on the regulatory approval and reimbursement of drugs and other health technologies [1,2] (see Augustovski et al. for a review and comparison of the PE guidelines currently existing in the Latin American region). Each step in the development and use of PE guidelines raises its own challenges. In addition, the characteristics of each country’s health care system have important influences on the process. Nevertheless, several common lessons were learned. Consideration of PE Guidelines (Argentina, Colombia, Guatemala, Uruguay, and Venezuela) When considering the development of PE guidelines, it is important to take account of how a new policy of requiring evidence on clinical and cost-effectiveness of drugs or other health care technologies will operate alongside the existing health care system. In particular it is important to note how health care is financed (e.g., through general taxation, social security contributions, or patient payments), how drugs and other health technologies currently receive market approval, how reimbursement is determined, how prices are set, and what existing controls are in place (e.g., hospital formularies or generic substitution). There is considerable diversity on these issues in the countries being considered in this cluster. In most countries in Latin America social security represents the largest financer and provider of health care, complemented by public and private systems. Therefore the development of PE guidelines must recognize the needs of all three sectors. Nevertheless, in most countries there is a compulsory package of benefits to be provided by social security, an approved list of drugs for public subsidy, and in some countries a special fund for high-cost technologies. Although no fourth hurdle is currently formally in place in any of the countries in this cluster, consideration of clinical and cost-effectiveness—through the application of PE guidelines—is being considered to be and could well be incorporated within the existing set of policy instruments. Development of PE Guidelines (Chile and Mexico) In Mexico a set of PE guidelines has already been developed and a law established requiring a PE dossier before inclusion of a technology in the national formulary. After inclusion of a technology in the national formulary, each health care institution can then decide whether or not to purchase the technology, based on its cost, budget availability, previous experience with the technology at the institution concerned, and priority of disease. In Chile, clinical guidelines have been established in several different pathologies based on a prioritization process. A particular concern is to obtain more efficiency without infringing on some of the basic principles of community solidarity and equal access to health care. In both Mexico and Chile the advantages of considering clinical effectiveness and cost-effectiveness, through the implementation of PE guidelines, are well recognized. These include establishing a more objective basis for purchasing decisions and enabling decision makers to have a better estimate of the budgetary effects of adopting new technologies. Several potential disadvantages, however, have also been identified. These include differences of opinion on the need for guidelines, the lack of adequately trained personnel to submit and review PE dossiers, and the fact that guidelines might be the subject of controversy between the government and health care industries. Implementation of PE Guidelines (Brazil) So far Brazil is the Latin American country with the most experience in the implementation of PE guidelines. The implementation is the responsibility of ANVISA and the Ministry of Health. PE is applied in pricing decisions for new drugs. If the new drug is no better than existing care a cost-minimization analysis is used. If the drug is found to be superior to existing care, the ceiling price is set at the lowest price among nine reference countries. To develop and implement the PE guidelines, an expert working group was established in 2006. After revision of the first draft of the guidelines, two workshops were convened, involving experts in economic evaluation and representatives from the Ministries of Health and Finance. Discussions were also held with a broader Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. S4 VALUE IN HEALTH 14 (2011) S3–S7 group of participants at the second ISPOR Brazilian Chapter Congress before publication of the guidelines in 2009. The development and implementation of guidelines has proved to be a complex task. The main lessons learned are: 1) the process must involve a broad range of stakeholders, including experts in the field and the researchers who will carry out the studies; 2) the process is as important as the final product; and 3) it is important to undertake periodic revisions of the recommendations and this will take place after 2 years, based on the feedback given by users. there is no independent regulatory agency (there is a division in the Ministry of Health known as DIGESA) nor a fourth hurdle system although a presidential decree issued after the plenary recommends the compulsory use of economic evaluations. The Ministry of Health regulates new inclusions in the PIAS, and in the therapeutic drug formulary. For high-cost technologies the FNR performs reviews of the economic evaluations and budget effects (although there is no formal guideline in place). In Venezuela all new drugs and devices are approved by the Ministry of Health. There is no fourth hurdle system in place. Country Groups Pricing policies Argentina, Colombia, Guatemala, Uruguay, and Venezuela The organizing team started working long before the plenary for this activity, undertaking projects from grouping the countries to deciding the contents. The report of the first country group, “When to consider pharmacoeconomic guidelines” [2] was so titled to reflect the position of countries that are thinking about developing guidelines or are currently developing them in earlier phases. Five countries are presented in this cluster. What is the framework in each country? Argentina’s framework is a multitier system divided into three large sectors: public, social security, and private. It has a compulsory medical package called the Plan Médico Obligatorio from the social security system, which has 24 provinces. Each province has its own health care system in place. Colombia has a social security system based on health insurance with two main schemes: the contributive scheme, with wider coverage for technologies, and the subsidized scheme, mainly for poor citizens. It has a compulsory plan of health called Plan Obligatorio de Salud (POS). Guatemala has social insurance that provides health care services for workers and pensioners. The uninsured population has access to free consultations and tests via the public network. Uruguay has recently changed its legislation into what is called the Nationally Integrated Health System, which includes a national health insurance regulated by a national health insurance body (Fondo Nacional de Salud [FONASA]) and a national board of health (Junta Nacional de Salud [JUNASA]). The federal government provides coverage to the entire population via Plan Integral de Atención a la Salud (PIAS), an integrated plan and a mandatory health benefit package, and Fondo Nacional de Recursos (FNR), an agency that specializes in high-complexity, high-cost technologies. Venezuela has two contributive government programs, the solidario health system with compulsory affiliation, and a complementary system with voluntary affiliations. The main players in the Venezuelan health care field are the Ministerio de Sanidad y Asistencia Social, the Instituto Venezolano de los Seguros Sociales, the Instituto de Previsión Social del Ministerio de Educación, and the Instituto de Previsión Social de las Fuerzas Armadas. Approval policies regarding health technologies Argentina has a typical, classic licensing agency, the Agencia Nacional de Medicamentos Alimentos y Tecnologías (ANMAT) and there is no formal fourth hurdle in place. There’s currently no need of any PE information for approval or reimbursing of new drugs or devices. The superintendence of health services is responsible for maintaining the benefit package. In Colombia all new drugs and medical devices that have to be or want to be included in the benefit package must be approved by the Comisión Reguladora de Salud, which is advised by the Committee of Assessments of Medicines and Health Technology. This group has recently been involved in developing health economic and budget impact guidelines. Guatemala has no independent licensing agency; it is within the Ministry of Health. It has no fourth hurdle in place. In Uruguay In Argentina there is no formal price regulation and insurance companies make arrangements for discounts with the pharmaceutical and device industry depending on their scale. In Colombia the Ministry of Commerce defines the top price of each medication package. Insurance companies, clinics, and hospitals make private arrangements. In Guatemala there are multilateral agreements for open contracting and a bidding process for essential drugs. In Uruguay there is no price regulation. The Direction of Commercee controls prices in pharmacies and drugstores and allows a maximum 25% discount. The FNR acts as a monopsony. Venezuela has had a mixed price system since 1994 and a list of essential drugs with prices is published by the Ministry of Commerce. There is no price regulation for other technologies. Reimbursement policies In Argentina a group of essential drugs is delivered to all primary care centers in the country. Ambulatory drugs are subsidized in a variable proportion and there is a special fund (Administración de Programas Especiales [APE]) that reimburses the social security system for most high-cost technologies. In Colombia insurance companies provide all services included in the POS. In most countries when an insurance company denies the provision of health care, a patient can go to the courts; Colombia has a special fund for reimbursing these cases (Fondo de Garantía de Pagos [FOSYGA]). In Guatemala there is no reimbursement system in place for the public sector and all reimbursement occurs in the private sector. In Uruguay drugs in the compulsory formulary are bought by each provider. Consumers have some co-pay in the private sector. Those technologies covered and funded by FNR have no co-pay. In Venezuela reimbursement only occurs in the case of high-cost drugs for chronic and end-of-life illnesses for which coverage reaches all citizens. Financial control policies In Argentina there is national reference pricing for approximately 200 essential drugs. Recently there was a strong stimulus for prescribing drugs using the generic name. The benefit package (Plan Médico Obligatorio) has currently no clear system in place for updating the package; but before 2007 there was a fourth hurdle system in place with PE requirements. In Colombia, the Comisión Reguladora de Salud defines and updates the per capita unit payment each year, as well as its benefits. Also, guidelines have been published that make economic evaluation of budget impact compulsory. In Guatemala the process has begun of elaborating national guidelines that will include health and economic information as well as efficiency principles to update or to change the national formulary. In Uruguay all pharmacies and pertinent parties have to report selected indicators to the information technology system on a monthly basis. The FNR is externally audited and results are publicly available. Venezuela has no current financial control policies in place. There’s a strong preference for generics in case there is a choice. Although these countries were grouped in the same cluster these are very heterogeneous countries in many aspects. Most have a rather fragmented health care system, and have designed a compulsory benefit package for the social security system. There VALUE IN HEALTH 14 (2011) S3–S7 is no formal fourth hurdle or PE system in place, but methodologic guidelines are being considered, developed, or advanced in some of them (e.g., Colombia). There’s also no formal drug price regulation, except for reference prices for essential drugs in some countries, and many countries have specific reimbursement policies for selected high-cost technologies (e.g., Argentina, Uruguay, and Venezuela). Also, most have a list of essential drugs that are provided or subsidized by the government or social insurance. Finally, the financial control policies in these countries range from reference pricing or providing a mandatory positive list to open bidding processes. The incorporation of formal fourth hurdle systems is also under consideration. Chile and Mexico During the past 60 years life expectancy in Latin-American countries extended from 51 to 74 years [3]; this gain in the population’s longevity represents several challenges for the regional health systems: 1) predominant causes of death are transitioning from infectious (less expensive to treat) to chronic degenerative diseases (more expensive to treat); 2) aging of the population results in a number of disabilities requiring dedicated assistance from the family or from the health system to perform tasks of day-to-day living; and 3) the increasing demand for health care is stretching the existing health system budgets to their maximums and allocation in a wise manner becomes compulsory to optimize the limited available resources in the countries. The science of economic assessment of health technology has appeared during a time when clear statements of what and how to evaluate are most important. The techniques facilitate a more objective approach to the assessment of the best options for the society to invest in, aligned with the specific necessities of jurisdictions. The existence of health technology assessment guidelines eases communication among involved parties because all clearly understand the rules in the appraisal process. Latin American countries are now at three different stages in the implementation of guidelines for health technology economic evaluation. Most advanced countries (mainly Brazil) have already developed and implemented their own guidelines for conducting economic assessments. Other countries, such as Chile and Mexico, are just embarking on guidelines implementation and intend to empower decision makers with a clearer and more transparent methodology that bring certainty to all participants in the acquisition of new technology. Mexico has recently released its PE guidelines [4], anticipating that the submission protocol of new products before the Health General Council for approval into the national formulary be clearer for all stakeholders. The exercise took about 2 years from the conception of the guidelines to the final deliverable document. The Chilean government is seeking to reduce the inequality gap in health care by guaranteeing access for all members of the population to treatment for a specific portfolio of diseases, selected by the authorities. To finance such an ambitious goal, the government is planning to increase the VAT general tax (value added tax) 1% from the current value and the surplus will be dedicated to ensuring that this health access policy is properly implemented. Finally, there are countries that have not yet incorporated a formal framework of economic assessment of health technology guidelines to help better allocate their resources. Some examples from the first cluster include Venezuela or Costa Rica. The spectrum of cases addressed here are to help those countries thinking of carrying out their own health technology economic evaluation guidelines. To ensure that economic evaluation guidelines are carried out successfully in Latin America, it is necessary to have all parties agree that the guidelines will benefit everybody involved, as well as facilitate the decision-making process for all stakeholders— S5 from regulators to the health industry— otherwise the guidelines will not be adopted. It is necessary to have in mind several questions at the conceptual phase of the project: 1. What are the benefits for society if economic evaluation guidelines are implemented in the health system? 2. Is the economic evaluation compulsory by law, is there a practical need to have it, or is it being explored based on convictions of all parties involved? 3. Who wants the guidelines to be implemented? 4. Have all potential stakeholders been identified and reached about the economic evaluation guidelines implementation? 5. Do all stakeholders agree on having the guidelines implemented? 6. Is there a sufficient number of experts on health technology economic assessment in the country to have a balanced number of them working among the stakeholders? It is also necessary to have in mind several questions during the execution phase of the project: 1. Has a committee with representation from all parties been set to write the guidelines? 2. Has every important issue been considered in the content of the guidelines? Perspectives Horizons Acceptable discount rates Acceptable willingness-to-pay threshold Equity Results presentation Support required 3. Who will review and give feedback, beyond the committee, on the guidelines? 4. Who is the audience for the guidelines? 5. How will financial support to complete the guidelines implementation be obtained? 6. Is there general consensus among the stakeholders on the final draft of the guidelines? 7. Have particular issues of the guidelines been discussed head to head and clarified with those who the guidelines will affect? 8. What is the scope of the guidelines? Pharmaceutical drugs only? Pharmaceuticals, equipment, devices, reagents, and other? 9. What methodologies of economic evaluation are best suited for the country? Cost minimization Cost-effectiveness Cost utility Cost benefit Some of these All of these 10. Will the guidelines be used as a tool to align and standardize economic evaluations submission for decision makers regarding new technology approval? Finally there are questions to be addressed after the guidelines have been set up: S6 VALUE IN HEALTH 14 (2011) S3–S7 1. Has a committee been appointed to monitor guidelines performance? 2. Do the guidelines require a short-term review to update or amend? 3. What is the life expectancy of the guidelines before a new, thorough review will be required? 4. Do the guidelines meet expectations for all stakeholders? 5. After a certain period of time, have the economic evaluation guidelines made a difference in the appraisal process? Setting up economic evaluation guidelines is equivalent to all parties acknowledging that guidelines are the most suitable road map to developing and presenting economic evaluations of new technologies to key decision makers to facilitate their resolutions. Guidelines are like any instruction manual that is carefully revised to explain the rules for decision making regarding new technology to be included in the health system. The eventual goal is to make a decision on if a new technology is worthy of being acquired and how much the health system is willing to pay for it, and if it represents additional benefits to the health status of the society. It is advisable to get counselling from groups in countries where the guidelines have already been implemented because these individuals can provide important feedback and information derived from their own experience The final benefit of economic evaluation guidelines is for the entire society because better decision making might bring both efficiency and equity in opportunities for the whole population. Brazil In this article, the Brazilian experience in developing the Methodological Guidelines for Studies in Economic Assessment of Health Technologies is presented. Although one of the authors (MD) participated as a collaborator in the development of these guidelines, credit for the elaboration and revision of the guidelines is, in large part, due to Flávia Elias and her team at the Ministry of Health; to Cid Vianna and Rosángela Caetano, who prepared the basic text that was submitted to a robust discussion process; and to Everton Nunes, who developed the final revision of the guidelines and collaborated in the presentation that gave origin to this article. It is important to highlight the Brazilian context of health care decision making as it relates to health technologies. Two types of decisions merit emphasis: 1) decisions related to pricing of new medications; and 2) decisions related to incorporation of technologies into the Unified Health System (SUS). In Brazil, drug prices have been regulated since the end of 2000. The regulation policies are defined by the Chamber of Regulation of Drugs Market (Cámara de Regulação do Mercado de Medicamentos), which is composed of five ministries and headed up by the Ministry of Health. ANVISA, by means of its Office of Economic Regulation, is responsible for policy implementation. With regard to new drugs, it is possible to say that there is an evidence-based pricing regulation policy. The model could be summarized as follows: if the new medication does not present benefits compared to the best therapeutic option available, then its maximum price will be defined based on a cost-minimization analysis. If the scientific evidence supports an additional therapeutic benefit in relation to the chosen comparator, then the peak price will be fixed based on international prices, and cannot be higher than the lowest price from a group of nine reference countries from different regions of the world.1 In addition, for a specific list of products considered highly important, which are generally high in cost, there is also a mandatory discount, reviewed annually, based on the difference in the revenue index between Brazil and the nine countries that comprise the Human Development Index. Currently this discount is 22.85%. The second type of decision refers to incorporation of new technologies. At the end of 2006, a commission was created with the purpose of making recommendations to the Ministry of Health regarding the incorporation of technologies. Based on scientific evidence, cost-effectiveness studies, and estimates of effects on budget, this commission makes recommendations regarding the incorporation of a submitted technology to the Ministry of Health, with the final decision being at Ministry’s discretion. Economic evaluation is one of the requirements established by legislation. This indicates the importance of guidelines for economic evaluation for decision making. Developing economic evaluation studies is no trivial task. One has to appropriately choose the type of evaluation, the analytical models to be used, the study perspective, the time horizon, the sensibility analysis, and the discount rate. Several countries have followed the path of developing guidelines for economic evaluation to contribute to results comparability and better practical applications for decision making. The process of developing the Brazilian guidelines started in 2006, with the Health Technologies Evaluation Workgroup (Grupo de Trabalho de Avaliação de Tecnologias em Saúde), of the Ministry of Health. This group congregates all governmental units working with Health Technology Assessment (HTA). Such integration has been very important for advancing HTA in Brazil. This group started, at that time, a discussion on the necessity of specific guidelines for economic evaluation. The first version was then produced, and two workshops were conducted for its evaluation. The workshops included representatives from several government entities and universities. A revised and still preliminary version of the guidelines was presented for the first time to the public at the ISPOR Second Brazilian Chapter Congress in 2008. After that, the guideline development process was opened to the public through a public consultation process. Including worthy contributions that resulted from the public consultation, the final revision was accomplished and, in September of 2009, the guidelines were officially launched after a 3-year process. The objective of a publication such as this is to establish methodologic standards that facilitate the application of economic evaluations in health care decision making. Considering the scarce literature available on this topic, its publication is very important in closing the gap of Portuguese written materials in this area. The objective was not to supply didactic material, but to ensure that economic evaluations become important tools for health care decision making. The public consultation process was very interesting because it allowed for open evaluation throughout the process. The public consultation participation rate was as follows: 37% from government entities, 37% from universities, 19% from industry; and 7% from other sectors. Out of the 23 topics presented in the consultation, all presented a concordance of at least 50%. Only three topics presented a concordance level lower than 60%: the study perspective, the intervention description, and the discount rate. In these topics was the approval rate was between 50% and 60%. Guideline use is already a reality. By the end of 2010 there were about 60 studies granted by the Ministry of Health that followed the guidelines. A multiplier effect is expected, which will be very important to increase awareness of the importance of the guidelines among more and more stakeholders—not only those who develop the studies (either in the university or industry) but also for decision makers. There is a clear trend toward growth in the number of economic evaluations supporting health care decision making, and, therefore, greater utilization of the guidelines. In the international arena, it is important to highlight the recent adoption of Methodological Guidelines for Economic Evaluation by the Mercado Común del Sur (Southern Common Market or MERCOSUR). Those guidelines were based on the Brazilian text. It is necessary to emphasize the importance of a participative process in developing guidelines. This process was as important as its final result. An otherwise excellent text, deprived of a participative elaboration process, would not bring the benefit of facil- VALUE IN HEALTH 14 (2011) S3–S7 itating adherence to the guidelines because there would not be any involvement from the main stakeholders from the beginning of the process. During the 3 years of discussion that went into developing the Brazilian guidelines the process benefited greatly from the debate among all stakeholders involved. This participative process was very enriching and, surely, will contribute to the future success of the guidelines. Conclusions The fragmentation in Latin American health care systems means that the development and implementation of PE guidelines is a complex task. It has to be consistent with other health care reforms in the countries concerned and involve a broad range of stakeholders, including academic experts, industry, and decision makers in the different sectors of health care (i.e., government, social security, and private as well as citizens). Finally, the implementation of PE guidelines should take account of the need for good local data on resource use and cost, and the need for adequately trained personnel to submit and evaluate PE dossiers. Nevertheless, several countries have decided to embark of this task and are moving forward at different speeds. No one is pretending that this will be easy, but at least in principle it appears that existing health care decision-making processes in Latin America could accommodate PE guidelines, should there be interest in implementing them. Federico Augustovski, MD, MSc, PhD Director, Departamento de Evaluaciones Económicas y de Tecnologías Sanitarias/Economic Evaluations and HTA Department, IECSInstituto de Efectividad Clínica y Sanitaria/Institute for Clinical Effectiveness and Health Policy; Profesor de Salud Pública/Professor of Public Health, Universidad de Buenos Aires, Servicio de Medicina Familiar y Comunitaria/Family and Community Medicine Division Hospital Italiano de Buenos Aires, Argentina S7 Guillermo Melendez, MD, MSc Mexican Health Foundation, Mexico City, Mexico Alexandre Lemgruber Office of Economic Evaluation of New Technologies, Brazilian National Health Surveillance Agency (ANVISA), Brasilia, Brazil Michael Drummond, PhD Centre for Health Economics, University of York, York, UK 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.036 REFERENCES [1] Augustovski F, Garay OU, Pichon-Riviere A, et al. Economic evaluation guidelines in Latin America: a current snapshot. Expert Rev Pharmacoecon Outcomes Res 2010;10:525–37. [2] Implementing pharmacoeconomic guidelines in evidence-based decision making in Latin America: lessons learned? Available from: http://www.ispor.org/conferences/RiodeJaneiro0909/documents/ISPOR_ 2ndLAConferenceMatrix_pg.pdf. [Accessed March 16, 2011]. [3] Gee EM. Trends in life expectancy in less developed countries. Encyclopedia of death and dying. Available from: http:// www.deathreference.com/Ke-Ma/Life-Expectancy.html. [Accessed September 27, 2010]. [4] Consejo de Salubridad General. Guía para la conducción de estudios de evaluación económica para la actualización del Cuadro Básico de Insumos del Sector Salud en México (Guideline for the conduction of economic evaluations for the updating of the Basic Technologies in the Healthcare Sector). Available from: http://www.csg.salud.gob.mx/ descargas/pdfs/cuadro_basico/GUxA_EVAL_ECON25082008_2_ech.pdf. [Accessed September 27, 2010]. VALUE IN HEALTH 14 (2011) S8 –S12 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Is Equity of Access to Health Care Achievable in Latin America? Equity is defined as the ability to impartially recognize the right of every person, a sense of justice and impartiality being its guiding principles. Access is the establishment of communication to obtain information or use of available resources. Therefore, from the social perspective, equity of access means equal rights to information and available resources. For equity of access to be attainable in Latin America or any other global region, it is necessary to take into account some fundamental principles that aim to ensure a health system is appropriate given the social, political, epidemiologic, and economic environment of its region. The establishment of a health budget should be linked to population necessities and regularly revised; decisions about health priorities should be based on necessities with a sufficient health care services structure that is effectively equipped to answer the health demand generated by the identification of these necessities, and based on coherent legislation supporting the decisions established in the health care arena. Coherent and humane treatment, as well as knowledge of a population’s specific health reality, cultural issues, and health necessities are also fundamental requisites to be taken into account to transform the concept of equity of access into something attainable. Access to health care, although it has to do with the same fundamental principle, varies depending on the perspective by which it is evaluated. From a patient’s perspective, access is to have his/her treatment delivered in the best way, with the best therapeutic options, and at the lowest possible cost— or at no cost through a public health care system. From a physician’s perspective, access to health care means service for all patients in a way that meets their treatment needs. For medical devices manufacturers, to improve access means to provide the market with innovative technologies that could benefit patients’ health results. In this article, analyses developed by specialists in equity of access to health care in Argentina, Brazil, Colombia, and Mexico are presented. Argentina The Argentine health system is tripartite: public, social insurance, and private. The first guarantees universal health care for nearly 16 million people through primary care centers and hospitals (although these services are used mainly by persons in lower income groups). This sector is financed mainly through taxes but users can be asked for a minimal fee for service. A very important free provision of drugs program operates through these providers. The social insurance sector comprises more than 300 institutions formed mainly by labor unions organized at national and subnational levels covering nearly 17 million people. Health coverage level is fixed by law in the Mandatory Medical Program (PMO). This sector is financed by compulsory contributions made by employees (3%) and employers (6%). Finally, the private sector is mainly financed through organized prepaid medical plans, and covers about 3.2 million people. These operate similarly to social insurance, using PMO as a reference standard of minimum level of coverage. National health policy regarding drugs rests on several topics. Since 2002 a national law has mandated the prescription of pharmaceuticals by their generic name; the objectives of this policy are to improve quality of prescriptions and promote price competition. Also, Argentina has several national programs providing free medicines. REMEDIAR and other free provision programs are highly centralized, achieving important purchasing power in the prices of many drugs (i.e., antiretroviral therapy). Finally, the health authority describes different levels on the financial coverage of health care in the PMO. For drugs, this means 40% coverage for 178 commonly used drugs, 70% coverage for 112 drugs prescribed for chronic conditions, and 100% coverage of drug costs in some particular cases such as insulin, cancer, and tuberculosis drugs. Two nationally representative surveys are useful when analyzing the health care access issue in terms of utilization and expenditures in Argentina. The National Expenditures Survey (NES) 2005 and the Health Care Utilization and Expenditure Survey (HCUES) 2005. Using NES to construct standardized figures for the share of health expenditure in total household expenditure and for the presence of health coverage across income quintiles, one can find a regressive pattern for the first indicator (15% for the poorest quintile, 11% for the richest quintile) and also an inverse relationship with health coverage (50% for the poorest, 89% for the richest). There are some interesting points derived from HCUES in relation to the access problem for ambulatory consultations across income quintiles. Although there were no important differences on the utilization rate (near 40%), there were differences on the proportion of people who pay for the access (17% for the poorest, 41% for the richest). This reflects a nonregressive pattern on consultations. Also, one can find a remarkable selection pattern that indicates that for any socioeconomic condition there is always a place where one can find health care services. Regarding hospitalizations, HCUES shows that poor people are more frequently hospitalized than other groups and often without paying a fee. There is not a clear trend for drug utilization rates among income groups and in most cases people pay to get medications. Comparing the share of health expenditure versus drug expenditure in total household expenditure across income groups, the analysis of NES depicts the presence of a regressive pattern for both indicators; also drugs are the main component for lower income people. So, why do poor people have a bigger Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. VALUE IN HEALTH 14 (2011) S8 –S12 proportion of drug expenditure? A possible answer could be that they are very exposed to illness and chances of prevention are minimal, so they cannot avoid the use of and expense for pharmaceuticals. This regressive pattern is found for several classes of drugs. For example, poor people spend a bigger proportion of health expenditure than rich ones for: 1) fever or pain (57% vs. 32%); 2) cardiac problems (53% vs. 32%); and 3) antibiotics (58% vs. 38%). According to NES (1997 vs. 2005), the incidence of catastrophic drug expenditure has decreased, especially on the poorest people (29% in 1997 vs. 17% in 2005). This fact reflects an improvement on equity of access to pharmaceuticals during these years. The aim, however, is to achieve equity of access to treatment. Not only medicines; treatment is the composite good that satisfies needs, and need of health care is central on every equity criteria on health. Achieving equity of access should not be thought as a trade-off problem, but instead the real challenge is to achieve equity in health care access. In Argentina and within the public sector, improving equity in access to health care means focusing on: 1) improving capacity, structure, and management, and 2) a generalization of evidence-based practice. The aim of improving equity in access should be to ensure an adequate program based on a health technologies assessment (HTA) criteria. Argentina has an extreme need for more transparent policies regarding the uses of HTA studies. Although there were several actions implemented by the national health authorities during the past decade, no one group defines a clear role for HTA. Indeed, in Argentina we must discuss: 1) the implementation of national/subnational agencies in charge of HTA and/or the regulation (and/or the coexistence) of private and independent agencies, and 2) the formulation of national HTA guidelines. The problem of access to health care for Argentineans has been always a matter of perspective. Health systems must center on equity of access, but we know that partial analysis results in partial results; only by getting the complete picture is it possible to get full, general, and useful results. Brazil Equity in health care, the aim of every health manager, should begin with a few fundamental concepts. In Latin America there is a need to establish specific parameters that will underpin the strategies employed in achieving equity of access to health care. Brazil, for example, shows very marked differences defined by demographic factors (access in large cities and in the south of the country is better than in the north and northeast or in small towns) as well as economic ones (people with private health insurance have prompt and more complete access than patients who depend on the public system). To understand this market, an analytic separation can be made between Brazilians who have private health insurance— around 40 million people—and the 150 million who depend on the public system [1]. Even in the private system, the rules of operation are regulated by the state, which is also responsible for supervising financial viability within a country. We believe all health care plans should follow the same rules. When a health technology is evaluated for adoption and pricing according to the rules laid down by the Ministry of Health, the process is practically the same independent of who will pay in the future. Although it may take longer, this process is generally complete in 24 months. The product must meet a series of scientific and economic specifications to be approved for eventual use in the country. The major differences in processes become apparent after this stage. The model of payment in the public system is basically a fixed budget based on the disease. For example, if a patient has S9 breast cancer and receives first-line treatment, the government pays a predetermined amount of money and leaves decisions regarding the appropriate medical protocol to the health care provider, who works within the value of the reimbursement. Meanwhile, the private system is maintained on the fee-forservice model, with remuneration to providers based on a published list of prices. Discounts from a supplier to a provider generate a significant component of the financial result of this operation. This model is the subject of extensive criticism because it does not stimulate the optimization of resources [2]. Private operators (health care plans), in turn, use some forms of analysis with a view to generating criticisms of new procedures when faced with requests for their adoption in health care. A little of how each of these functions in Brazil is described here. Scientific Critical Analysis The goal of financing institutions is the ideal form of scientific evidence, represented by meta-analyses and randomized clinical trials. A prescriber, on the other hand, does not always manage to provide this level of evidence and, ultimately, he or she adopts a technology in a less critical way than required by the financing institution. Several examples of this debate exist. For example, the use of abstracts, without a thorough analysis of the full text of articles, as well as interpretations of inappropriately designed studies, represent some of the causes of discussion [3]. The pertinent debate, perhaps, is over the lack of information to which the patient has access, which ultimately creates unrealistic expectations about the results of some treatments. Clinically irrelevant changes may be understood and used in different ways among the players. Legal Analysis According to the law that regulates the operators of health care plans (known as Law 9656, dating from 1998), drugs that do not have the necessary approval from the Ministry of Health for a specific indication are considered off-label and do not need to be covered by the insurance. Even if the universal application of this rule is difficult, it is widely used for drugs of high cost. The tendency is that when there is unequivocal support for the importance of a determined medicine, there is less strictness in the application of the rule. The arguments over oral drugs, especially in cancer treatment, continue to grow. It is well known that very expensive oral drugs will only become a reality if financing institutions begin to pay for them. There is very little experience with pharmaceutical benefit programs in Brazil. In this scenario there is a great deal of tension and judicial demands and very little useful experience [4]. Pharmacoeconomics Analysis With strong growth in interest during recent years, this theme has begun to be addressed by stake-holders in a more professional way. Although this is still at the stage of defining basic concepts applicable to the country, there is a strong drive to broaden the debate, although it remains difficult to translate these concepts to the bedside in Brazil, as it is throughout the world [5]. Understanding of the effects on budgets and the need to create a list of priorities in the allocation of resources is no longer a theoretical discussion. Suppliers of medicines and devices already recognize that they need to focus on this area and offer sophisticated subsidies for discussion. The government fears, however, stimulating growth in this area; financing institutions have been critical of this and have preferred providers who are aware of the problem. So, is equity of access to health care in Latin America achievable? When we invert the current reasoning, which is to distrib- S10 VALUE IN HEALTH 14 (2011) S8 –S12 ute resources that are insufficient for the actual needs of the system, in such a way that we can define what kind of medicine we want and, based on this, request a realistic, appropriate quote, there is a real possibility that we can reduce differences in our countries. Colombia Colombia is one of the few countries in Latin America that has implemented a nationwide social health insurance program targeted to the poor. This program, known as the Subsidized Regime (SR), began to be implemented in 1996 as part of one of the most ambitious and innovative health care reforms in the region, enacted by Law 100 in 1993. The reforms sought to improve health outcomes and protect families from the economic consequences of poor health through mandated social health insurance coverage for all the population, to increase quality and efficiency of health care services by introducing competition in health insurance and health care provision markets and supply-side reforms, and to augment equity through better targeting of public services. The SR allocates public subsidies to individual insurance premiums for the poor according to a proxy-means testing index. The SR is financed through a combination of resources obtained from a solidarity contribution from formal workers’ mentioned above, national tax revenues and social investment transfers to municipalities earmarked for health, and local tax revenues. Resources allocated to the SR have reached considerable amounts, almost 3% of the gross domestic product in 2009 [6]. The benefits package of the SR emphasizes coverage for preventive and basic ambulatory care services, and catastrophic care. Medications within a national listing and medical transportation expenses are also covered. There is no coverage, however, for specialist care and there are important gaps in coverage for hospital care except for some surgical procedures and orthopedics. Children aged less than 1 year and pregnant women have almost no restrictions in coverage. There are no copayments and out-of-pocket expenses are restricted to noncovered services. The premium for the SR benefits package was valued at US$142 in 2010. There is evidence that the SR has contributed to equity improvement in several aspects. One of these is health insurance coverage. The growth of enrollment in the SR has been amazing, from 0% to almost 51% of the total population in 16 years (1993– 2010). Currently, about 24 million poor people have SR insurance coverage [7]. A key feature of SR insurance has been reaching the people most vulnerable to economic shocks. In 2008, 81% of the population in the lowest quintiles had health insurance. A comparison of the proportion of insured individuals by income quintile in 1992 and 2003 shows an increase of 37 percentage points or a variation of 444% for the first quintile (9% in 1992 to 48% in 2003), whereas in the fifth quintile the increase was 21 percentage points (60% in 1992 to 81% in 2003 [8]. Rural and urban disparities in insurance coverage have been progressively bridged. In 1993 there was a rural– urban insurance coverage difference of 26% that was reduced by 13 points in 2003, and health insurance coverage in rural areas currently is mainly at the expense of the SR. The SR adds to equity by providing financial protection from catastrophic health expenditures. Miller et al. [9] found that compared to uninsured persons, average and large outliers in health care expenditures in SR enrolees decrease. The SR has also played a role in poverty alleviation. Household survey data analyses reveal that between 1997 and 2003 health subsidies reduced poverty levels and income inequality by more than two and three percentage points, respectively [10]. Quasiexperimental studies have also found that the SR has made a difference in access and use of services, particularly for the poor and population in rural areas. For example, Giedion et al. [11] found that compared to the uninsured, the insured poor have a 40% higher probability of ambulatory consultations, a 17% higher probability of taking a child with diarrhea to a health care institution, a 23% higher probability of taking a child with respiratory infection to a health care institution, and a 7% higher probability of birth attendance by a health professional or at a formal institution. Miller et al. [9] also found a higher likelihood of preventive visits, and of growth monitoring and well-child visits. Regarding improvements in equity in health outcomes a few studies point toward positive results, although on very broad health indicators. Miller et al. [9] found lower probabilities of cough, fever, or diarrhea and number of days unable to carry out daily activities due to illness in children younger than age 5 years covered by the SR. Other research finds that infant mortality decline has been larger amongst the poor, and between 1995 and 2005 overall differentials in infant mortality between wealth groups slightly decreased, from a 2.5 to a 2.2 times larger mortality rate in the lower quintiles with respect to the higher [12]. The latter results, however, do not establish a causal relationship to the SR. The Colombian health care reforms, particularly the provision of health insurance for the poor, have contributed to overall equity improvement in financial protection, access to and use of health services, and in some health outcomes. Nevertheless, there is evidence of differentials in these indicators by sociodemographic and geographic groups, which may require tailoring health insurance policies to address specific needs [10]. Further progress is to be expected given that there is a recent mandate issued by the Colombian Constitutional Court to reach universal health insurance and to upgrade the breadth and scope of the SR benefits package in the next years. Implementation of this policy raises the challenge of designing an effective package under budget constraints. Health technology assessment can be a key instrument to guide the selection of interventions to be included in the new package, but Colombia will need to strengthen both public and private sector capacity to generate and use economic evaluation for priority setting. Mexico Article four of the Constitution of Mexico [13] clearly states that all Mexicans have the right to health care, but this has not been possible due to factors ranging from the purely economic (insufficient resources) to the geographic (communities in remote areas with difficult access). Demographically [14], Mexico, as with many countries in the world, has an increasingly aging population, and with it an increase in prevalence of degenerative chronic diseases, the treatment of which is generally lifelong, thus putting pressure on the health care budget, which is 5.9% of the gross domestic product. On the other hand, given the socioeconomic characteristics of this country, contagious diseases still persist and are common among the poor, whose treatment is also costly in light of the need for new generations of antibiotics to address the fact of ever-increasing bacterial resistance, a product of the self-prescription which for many years was a part of cost containment for social security. On one hand, the low health care budget is a result of the chronic economic crisis with periodic exacerbations that affect the national economy combined with costly and complicated administrations. According to numbers from the Organization VALUE IN HEALTH 14 (2011) S8 –S12 for Economic Cooperation and Development [15], the health administration consumes 10% of the budget destined to health care; thus, a percentage of the health expenditure has fallen back directly onto families. It is calculated that 3.6% of total family income goes to health care and from that approximately 43.3% goes to the purchase of medicines. There are various health care institutions in Mexico, which together cover 56.2% of the population. Among them is the Mexican Institute of Social Security (IMSS), which covers the majority of social security recipients in Mexico (42.8%), the beneficiaries of which are, for the most part, workers in the private sector. It is an obligatory insurance plan: all employers by law have to insure their employee(s) via this institution, which is financed in three ways: 1) contributions from employers; 2) from federal funds; and 3) by employees who contribute a small part of their salary. The Social Security Institute for State Workers covers 9.6% of the population. In this case, the beneficiaries are workers in the public sector and it is financed with public resources as well as worker contributions. The remaining 3.8% is covered by a series of state institutions that function as a captive insurer, which is to say the beneficiaries of medical coverage are the workers and their families of Mexican Petroleum or government entities with special characteristics such as the Secretary of National Defense or the Secretary of the Navy. The other 43.8% of the population does not have access to social security and responsibility for this population falls directly to the Ministry of Health through hospitals, local and state agencies, and the National Institutes of Health, and is operated by the newly created Popular Insurance, which has grown rapidly and to which a good amount of resources have been devoted— mostly for the purchase of medicines and other health supplies—leaving almost nothing for the infrastructure of clinics and hospitals. This arrangement functions as an open health plan in which a beneficiary pays with an extremely low cost policy (the resources come primarily from the public sector). The objective is to provide access to services to people who have established small businesses, working in an informal economy that once was under the auspice of public hospitals run by the Secretary of Health. Access to this system is currently limited to those with the 110 most frequent illnesses and, since consolidating this system, the Secretary of Health acts only to guide health policy without providing actual services. There are other specific programs that have been implemented, such as IMSS Opportunities, directed toward unprotected populations and funded by public resources. It is interesting to note that a family can have access to two systems of health coverage, which is to say, they can be doubly insured. For example, people can be covered through the Social Security Institute for State Workers because they have a family member (e.g., a wife) who works in the public sector and a family member (e.g., a husband) has the right to IMSS because he is employed in the private sector. This creates duplicate access to health care and also doubles family contributions. The level of prescriptions dispensed also represents a problem in the dispensation of medicines in the public sector. A patient must purchase medicines in pharmacies where the price is three or four times higher than the original value. On the other hand, there is no customized system for drug supplies, which is why the prescribed amount is either insufficient to complete the treatment indicated or is overly prescribed, resulting in a waste of important resources. Medicines are dispensed in private pharmacies by individuals who often have not even received basic education. Some even make recommendations on products to “alleviate the symptoms of the patient.” It is not uncommon for pharmacy S11 workers to change medical prescriptions according to their convenience. Pharmacies and commercial centers commonly have doctors incorporated within stores to see patients; evidently, prescriptions are made out in favor of the products that the commercial chain or establishment is selling. Who pays the costs of complications resulting from these unscrupulous practices? Gabriela Tannus Branco Araújo, MSc Axia.Bio Consulting, Sao Paolo, Brazil. Joaquín E. Caporale, BEc, MSc Institute for Clinical Effectiveness and Health Policy, Buenos Aires, Argentina. Stephen Stefani, MD UNIMED, Porto Alegre, Brazil. Diana Pinto, MD, MHA, DSc Javeriana University, Bogata, Colombia. Antonio Caso, MD, MEd National University of Mexico, Mexico City, Mexico. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.037 REFERENCES [1] Agencia Nacional de Saude Suplementar [National agency of supplementary health]. Available from: www.ans.gov.br. [Accessed March 15, 2011]. [2] Stefani SD. O tumultuado cenário das relações de prestadores e fontes pagadoras [The tumultuous scene of relationships between providers and payers]. Rev Bras Oncol Clínica, 2005;2:19 –24. [3] Greenhalgh T. How to Read a Paper—The Basics of Evidence Based Medicine. (3rd ed.). London: BMJ Books, 2006. [4] Reis Neto JP, Frenkel J, Stefani SD. Five-year Brazilian Pharmaceutical Assistance for Chronic Disease Management. 12th Pharmaceutical Benefit Management Meeting, Arizona, 2006. [5] Desch C. Pharmacoeconomics: a scientific approach to resource allocation at the bedside. ASCO Ed Book 1997;33:180 –3. [6] Santa María M, Zapata JG, Arteaga C, Reyes CE. Descentralización, el financiamiento de la salud y la educación y los departamentos: ¿Cuáles son las alternativas? [Decentralization, funding health care and eduction departments: what are the alternatives?]” Bogotá: Mimeo. Fedesarrollo y Federación Nacional de Departamentos, 2009. [7] República de Colombia, Ministerio de la Protección Social, Sector de la Protección Social. 2009. Informe de Actividades 2008-2009 al Honorable Congreso de la República [Republic of Colombia, Ministry of Social Protection, Sector of Social Protection. 2009. Activities Report 2008-2009 to the Honorable Congress of the Republic]. Available from: www2.igac.gov.co. [Accessed December 1, 2010]. [8] Pinto D. Colombia: good practices in expanding health care coverage. In: Gottret P, Schieber GJ, Waters HR (eds.). Good Practices in Health Financing: Lessons from Reforms in Low and Middle-Income Countries. Washington, DC: The International Bank for Reconstruction and Development/The World Bank, 2008. S12 VALUE IN HEALTH 14 (2011) S8 –S12 [9] Miller G, Pinto D, Vera M. High-powered incentives in developing country health insurance: evidence from colombia’s régimen subsidiado. NBER Research Working Paper. Available from: http://www.nber.org. [Accessed December 1, 2010]. [10] Flórez CE, Acosta OL. Avances y desafíos de la equidad en el sistema de salud colombiano [Progress and challenges of equity in the Colombian health care system]. Working Document No. 15. Bogotá: Fundación Corona, 2007. [11] Giedion U, Díaz BY, Alfonso EA, Savedoff W. The impact of subsidized health insurance on health status and on access to and use of health services. In: Glassman et al. (eds.). From Few to Many. Ten Years of Health Insurance Expansion in Colombia. Washington, DC: Interamerican Development Bank, 2009. [12] Flórez CE, Soto V. Avances y desafíos de la equidad en el Sistema de Salud Colombiano [Progress and challenges of equity in the Colombian health care system]. Working Document. Bogotá: Fundación Corona, 2007. [13] Constitution of the United States of Mexico-Last Reform DOF 29-072010 article 2/b/111 - Chamber of Deputies of the Congress of the Union, General secretary, Secretary of parliamentary services, Documentation Center, information and analyses. DOF 29-07-2010. [14] Guilera N, et al. Desigualdad en salud en México: los factores determinants. Comercio Exterior México 2006;56:106 –13. [15] Wirtz VJ, Russo G, Kageyama-Escobar ML. Access to medicines by ambulatory health service users in Mexico: an analysis of the national health surveys 1994 to 2006. Salud pública Méx 2010;52:30 – 8. VALUE IN HEALTH 14 (2011) S13–S15 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Insurance Status and Demographic and Clinical Factors Associated with Pharmacologic Treatment of Depression: Associations in a Cohort in Buenos Aires Gerardo Machnicki, MSc, PhD1, Carol Dillon, MD, PhD1, Ricardo F. Allegri, MD, PhD1,2,* 1 Memory Research Center, Department of Neurology, Zubizarreta General Hospital, GCBA Buenos Aires, Argentina; 2Department of Cognitive Neurology, Instituto de Investigaciones Neurológicas Raúl Carrea, Buenos Aires, Argentina A B S T R A C T Objective: There is a paucity of evidence about insurance status and the likelihood of receiving medical services in Latin America. The objective of this analysis was to examine the association between insurance status and pharmacologic treatment for depression. Methods: Patients referred to a memory clinic of a public hospital in Buenos Aires, Argentina, and identified with any of four types of depression (subsyndromal, dysthymia, major, and due to dementia) were included. Age, years of education, insurance status, Beck Depression Inventory score, and number of comorbidities were considered. Associations between these factors and not receiving pharmacologic treatment for depression were examined with logistic regression. Use of prescription neuroleptics, hypnotics, and anticholinesterase inhibitors was also explored. Results: Out of 100 patients, 92 with insurance status data were used. Sixty-one patients (66%) had formal insurance and 31 patients (34%) lacked insurance. Twenty-seven (44%) insured patients and 23 (74%) uninsured pa- Introduction Depression represents a concern to public health because it is one of the most burdensome diseases [1] and it also exacts a considerable economic [2] and humanistic burden [3]. Pharmacologic and other mental health treatment rates remain low and variable in international comparisons [4]. Out-of-pocket costs [4] and lack of insurance [5] have been identified as factors associated with lower treatment rates. A Latin American city involved in an international study ranked among the lowest in terms of antidepressant therapy rates and was among the first citing financial barriers as a reason for not receiving treatment [4]. More than 40% of Argentineans rely on public infrastructure for their health care [6]. Although during the past decade federal coverage of essential medicines has been developed [7], the scope of the plan is limited. Evidence supports a link between lack of insurance (or reduced coverage) and reduced access to pharmacologic or nonpharmaco- tients did not receive antidepressants (P ⫽ 0.001). Controlling for other factors, uninsured patients had 7.12 higher odds of not receiving treatment compared to insured patients (95% confidence interval 1.88 –28.86). Older patients and those with more comorbidities had higher odds of not receiving treatment. More educated patients, those with higher Beck Depression Inventory score, and those without subsyndromal depression had lower odds of not receiving treatment. None of those associations were statistically significant. Conclusions: These results suggest a potential negative effect of the lack of formal insurance regarding pharmacologic treatment for depression. These findings should be confirmed with larger samples, and for other diseases. Keywords: antidepressants, depression, insurance, Latin America. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. logic medical treatment, as well as with impaired medication persistence. Latin American data coincide with these findings. An analysis of Mexican elderly patients [8] showed that presence of health insurance was associated with improved access to medications. Lack of insurance was identified as a risk factor for catastrophic health expenditures in Mexico [9] and Colombian data support the link between no insurance or lower quality insurance and reduced human immunodeficiency virus/autoimmune deficiency syndrome medication adherence [10]. To our knowledge few investigations addressed care for depression in low income, uninsured individuals in Latin America [11–14]. The primary objective of this research was to assess the association between antidepressant pharmacotherapy and insurance status, controlling for demographic and clinical factors. The secondary objective was to explore the associations between insurance status and other factors and treatment with prescription neuroleptics, hypnotics, and anticholinesterase inhibitors. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Ricardo F. Allegri, Department of Cognitive Neurology, Instituto de Investigaciones Neurológicas Raúl Carrea, Montañeses 2325, (C1428AQK) Buenos Aires, Argentina. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.014 S14 VALUE IN HEALTH 14 (2011) S13–S15 Methods Patient population The study population consisted in patients evaluated at the Memory Clinic at Zubizarreta Hospital in Buenos Aires, Argentina. A clinical database with the data collected as part of the initial visit of the patient was used [15–16]. This was a cross-sectional study. Patients referred for impaired memory and depressive symptoms and assessed in the period 2005 to 2009 were included. All of them were assessed using a semistructured neuropsychiatric interview (administered by specialized psychiatrists and neurologists). Depressive syndromes were categorized into different diagnoses according to the DSM IV [17] and ICD-10 criteria [18] using SCAN 2.1: Schedules [19] for clinical assessment in neuropsychiatry. Four types of depression were confirmed. These were: 1) subsyndromal depression (diagnosed when patients experience depressive symptoms but do not meet standard diagnostic criteria); 2) dysthymia (a syndrome of depression of mild or moderate severity that lasts at least 2 years); 3) major depression; and 4) depression associated with dementia (Alzheimer disease and vascular and mixed dementia). Patients were included if they had a minimum age of 55 years and were younger than 80 years old, if they presented depressive symptoms due to psychiatric causes or related with dementia (with clinical dementia rating scale [CDR] [20] score of 1) and if they had a Hamilton Depression Scale [21] score greater than 9 points. Exclusion criteria covered: 1) drug or alcohol abuse; 2) presence of neurologic diseases (except dementia); 3) dementia and CDR score of 2 or CDR score of 3; and 4) schizophrenia or schizoaffective disorder. Written informed consent was obtained from each subject after they had been given a full explanation of the study. The research was performed in accordance with the International Conference on Harmonisation Good Clinical Practice guidelines [22], the latest revision of the 1964 Helsinki Declaration [23], and the Buenos Aires Government Health Authorities. Study variables Patients’ demographic data included age, sex, education level, and income. Health insurance status was recorded as: PAMI (Medicarelike system for retired persons aged ⬎ 65 years and special groups), sickness fund (insurance linked to employment), and privately paid insurance or public system. Patient depression severity was assessed with the Beck Depression Inventory (BDI), a questionnaire with a 0 to 63 total score [24]. Higher scores indicate more depressive symptoms. The Mini-Mental Scale Examination (MMSE) for cognitive impairment was used (0 to 30 score, with lower scores indicating higher cognitive impairment) [25]. Patient comorbidities were recorded and included hypertension, diabetes, cardiovascular and cerebrovascular disease, dyslipidemia, tobacco use and alcohol abuse, hypo- and hyperthyroidism, encephalo cranial traumatism, epilepsy, Parkinson’s disease, and other concomitant diseases. Assessed medications included antidepressants (trycyclics and serotonin-reuptake inhibitors), anticholinesterase inhibitors, neuroleptics, and hypnotics. Statistical analysis All analyses were performed with R version 2.11.1 [26]. Patients were aggregated in two groups. Group 1 were those with PAMI, sickness-fund, or private insurance and group 2 were those relying on public services (labelled as uninsured due to the lack of formal insurance services). Depression diagnosis was collapsed into two groups: subsindromal depression and all other diagnoses. Bivariable analysis used Wilcoxon tests for continuous variables and Pearson’s chi-square tests for categorical variables. Primary multivariable analyses used as outcome not being prescribed an antidepressant. Independent variables included insurance status, age in years, education in years, depression diagnosis, total BDI, and count of comorbidities. Models were repeated with use of neuroleptics and hypnotics. A separate model was used for anticholinesterate inhibitors (MMSE was included as independent variable while BDI and depression diagnosis were dropped). Results One hundred patients were initially included in the study and 92 were retained due to missing insurance data. Patients not insured were younger (median 59 vs. 70 years, P ⬍ 0.001, mean 70.3 vs. 60.3), which is explained by the proportion of elderly insured patients. A higher proportion of uninsured patients had subsyndromal depression (26% vs. 21%) and major depression (39% vs. 23%) (Table 1, in Supplemental Material found at: doi:10.1016/j.jval.2011.05.014). A higher proportion of uninsured patients were not receiving antidepressants compared to insured patients (74% vs. 44%; P ⫽ 0.006). The proportion of patients not treated with anticholinesterase inhibitors was similarly high among uninsured and insured (92% vs. 94%; P ⫽ 0.76). Eighty-one percent of uninsured patients and 89% of insured patients were not receiving neuroleptics, whereas 45% of uninsured patients and 34% of insured patients were not receiving hypnotics. A logistic regression model indicated that patients without insurance had 7.12 (95% confidence interval 1.88 –26.86) higher odds of not being prescribed an antidepressant compared to patients with insurance (Table 2 in Supplemental Material found at: doi: 10.1016/j.jval.2011.05.014). No other point estimate reached statistical significance. Increasing age and more comorbidities were associated with higher odds of not receiving treatment whereas more years of schooling, diagnosis other than subsyndromal depression, and higher BDI were associated with lower odds of not receiving treatment. No health care insurance was also associated with higher odds of not being prescribed neuroleptics or hypnotics and with lower odds of being prescribed anticholinesterase inhibitors, although the results were not statistically significant. Older age was associated with higher odds of not being prescribed hypnotics and lower odds of not being prescribed anticholinesterase inhibitors. More years of education were associated with lower odds of not being prescribed neuroleptics, hypnotics or anticholinesterase inhibitors (and it was the only statistically significant finding for this endpoint). Higher BDI was associated with lower odds of not receiving neuroleptics or hipnotics. A higher MMSE score was associated with lower odds of being prescribed an anticholinesterase inhibitor. Conclusions The associations between insurance status and pharmacologic treatment for depression in a clinic in Buenos Aires were investigated. To our knowledge, this is the one of the first reports about this topic in South America. Previously, a study in Botucatu, Brazil [14], showed that patients earning less than the minimum wage had five times the odds of not receiving antidepressants compared to people earning more than four times the minimum wage, even though essential medicines were in principle accessible to all socioeconomic strata. In a multivariable model that accounted for demographic and clinical factors uninsured patients had 7.12 higher odds of not receiving antidepressant therapy, and the result was statistically significant. This association was more important than the lower odds of receiving treatment resulting from varying patient age from 59 to 74 years or from the effect of being a subsindromal patient compared to any other diagnosis (Fig. 1 in Supplemental Material found at: doi: 10.1016/j.jval.2011.05.014). The mean annual treatment rate for chil- VALUE IN HEALTH 14 (2011) S13–S15 dren and adolescent with depression in the United States was approximately half the rate of insured children [5]. The relative treatment proportions reported in this study are of similar magnitude. Although comparability is impaired by different endpoints and settings, this odds ratio of 7.12 is similar to the odds ratio for reporting cost-related non adherence for uninsured patients [27] and lower than the odds in the city of Botucatu [14]. Even if not statistically significant, other associations had the expected direction. Older patients and those with more comorbidities had higher odds of not receiving antidepressant treatment, which may be due to the complexity of dealing with several conditions. Low treatment rates for older patients have been reported in Latin America [28]. Patients with a higher BDI (more severe depression) had lower odds of not being treated and patients with subsindromal depression had higher odds of not being treated. More educated patients had lower odds of not being treated. This may be a related to higher likelihood of insurance, higher income, or more health-oriented behavior. This study had some limitations. Sample sizes were small but in contrast to a bigger sample of a single subsystem, this cohort had a variety of exposure to different insurance status. Selection bias was present. Representativeness of the city of Buenos Aires or beyond may be in doubt even for the studied age group. It was not possible to estimate the same relationships regarding nonpharmacologic treatment for depression. Confounding bias may have affected the findings. Some of the differences may be due to observed or unobserved patient imbalances; however, it is unlikely that unobserved confounders would change such a strong relationship between lack of insurance and lack of antidepressant treatment. Finally, limitations arise from the crosssectional design. Depression history was not assessed. In the extreme, if all uninsured cases were detected at the clinic and the insured depressive cases were already diagnosed, higher treatment rates may be explained by the higher depression diagnosis rate for insured patients. This may dilute part of the association but it is unlikely that all of the association observed would be due to this. This was probably not the first interaction with the health care system for the uninsured patients in this study, because these patients were referred to the clinic. There were also some imbalances in the type of depression but these seem to be in a direction that would favour more treatment in the uninsured group, because a higher proportion of patients had major depression, although this was balanced with a higher proportion of insured patients with dysthymia. Measurement bias leading to misclassification may have occurred for medication use if patients reported no use of medication but they dropped out of current treatment or were not adherent to it. It is unlikely that such bias would greatly affect the main finding, given its numerical strength. Interventions to enhance access and improve outcomes for patients with depression can succeed. A primary care program designed for low-income women in Chile showed its effectiveness and efficiency [11–12]. In the past 20 years, Brazil has made important progress toward ensuring appropriate mental health care, including enhanced access to essential psychotropic medication [29]. There is a need to monitor such policies to evaluate their effects and detect remaining access gaps. Lack of insurance with comprehensive pharmaceutical coverage was associated with much higher odds of not receiving pharmacologic treatment for depression in a cohort in Buenos Aires. Future studies should explore this relationship in larger samples with broader geographical scope, and into other areas of medical care. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi: 10.1016/j.jval.2011.05.014 or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). S15 REFERENCES [1] Murray CJ, Lopez AD. Alternative projections of mortality and disability by cause 1990 –2020: Global Burden of Disease Study. Lancet 1997;349:1498 –504. [2] Luppa M, Heinrich S, Angermeyer MC, et al. Cost-of-illness studies of depression: a systematic review. J Affect Disord 2007;98:29 – 43. [3] Scazufca M, Menezes PR, Almeida OP. Caregiver burden in an elderly population with depression in Sao Paulo, Brazil. Soc Psychiatry Psychiatr Epidemiol 2002;37:416 –22. [4] Simon GE, Fleck M, Lucas R, Bushnell DM. Prevalence and predictors of depression treatment in an international primary care study. Am J Psychiatry 2004;161:1626 –34. [5] Olfson M, Gameroff MJ, Marcus SC, Waslick BD. Outpatient treatment of child and adolescent depression in the United States. Arch Gen Psychiatry 2003;60:1236 – 42. [6] Cavagnero E. Health sector reforms in Argentina and the performance of the health financing system. Health Policy 2008;88:88 –99. [7] Homedes N, Ugalde A. Improving access to pharmaceuticals in Brazil and Argentina. Health Policy Plan 2006;21:123–31. [8] Maurer J. Assessing horizontal equity in medication treatment among elderly Mexicans: which socioeconomic determinants matter most? Health Econ 2008;17:1153– 69. [9] Sesma-Vazquez S, Perez-Rico R, Sosa-Manzano CL, Gomez-Dantes O. [Catastrophic health expenditures in Mexico: magnitude, distribution and determinants]. Salud Publica Mex 2005;47(Suppl. 1):S37– 46. [10] Arrivillaga M, Ross M, Useche B, et al. Social position, gender role, and treatment adherence among Colombian women living with HIV/AIDS: social determinants of health approach. Rev Panam Salud Publica 2009;26:502–10. [11] Araya R, Flynn T, Rojas G, Fritsch R, Simon G. Cost-effectiveness of a primary care treatment program for depression in low-income women in Santiago, Chile. Am J Psychiatry 2006;163:1379 – 87. [12] Araya R, Rojas G, Fritsch R, et al. Treating depression in primary care in low-income women in Santiago, Chile: a randomised controlled trial. Lancet 2003;361:995–1000. [13] Arredondo A, Ramos R, Zuniga A. [Economic evaluation of the demand of medical care for mental health in Mexico: schizophrenia and depression, 1996-2000]. Rev Invest Clin. 2003;55:43–50. [14] Lima MC, Menezes PR, Carandina L, et al. [Common mental disorders and the use of psychoactive drugs: the impact of socioeconomic conditions]. Rev Saude Publica 2008;42:717–23. [15] Allegri RF, Butman J, Arizaga RL, et al. Economic impact of dementia in developing countries: an evaluation of costs of Alzheimer-type dementia in Argentina. Int Psychogeriatr. 2007;19:705–18. [16] Machnicki G, Allegri RF, Dillon C, et al. Cognitive, functional and behavioral factors associated with the burden of caring for geriatric patients with cognitive impairment or depression: evidence from a South American sample. Int J Geriatr Psychiatry 2009;24:382–9. [17] Diagnostic and Statistical Manual of Mental Disorders. (4th ed.). Washington, DC: American Psychiatric Association, 2000. [18] ICD 10-Internacional Codification of Diseases. (10th ed.). Geneva: World Health Organization, 1990. [19] Wing JK, Babor T, Brugha T, et al. SCAN. Schedules for clinical assessment in neuropsychiatry. Arch Gen Psychiatry 1990;47:589 –93. [20] Hughes CP, Berg L, Danziger WL, et al. A new clinical scale for the staging of dementia. Br J Psychiatry 1982;140:566 –72. [21] Hamilton M. Rating depressive patients. J Clin Psychiatry 1980;41:21– 4. [22] ICH Guidelines. 2005. Available from: http://www.ich.org/cache/ compo/276-254-1.html. [Accessed August 14, 2010]. [23] WMA Declaration of Helsinki. 2008. Available from: http://www .wma.net/en/30publications/10policies/b3/index.html. [Accessed August 14, 2010]. [24] Steer RA, Ball R, Ranieri WF, Beck AT. Dimensions of the Beck Depression Inventory-II in clinically depressed outpatients. J Clin Psychol 1999;55:117–28. [25] Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189 –98. [26] R: A language and environment for statistical computing. Available from: http://www.R-project.org. [Accessed August 14, 2010]. [27] Kennedy J, Morgan S. Cost-related prescription nonadherence in the United States and Canada: a system-level comparison using the 2007 International Health Policy Survey in Seven Countries. Clin Ther 2009; 31:213–9. [28] Guerra M, Ferri CP, Sosa AL, et al. Late-life depression in Peru, Mexico and Venezuela: the 10/66 population-based study. Br J Psychiatry 2009; 195:510 –5. [29] Jacob KS, Sharan P, Mirza I, et al. Mental health systems in countries: where are we now? Lancet 2007;370:1061–77. VALUE IN HEALTH 14 (2011) S16 –S19 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Precios de Referencia Internacional y Análisis de Costo Minimización para la Regulación de Precios de Medicamentos en Colombia Caludia Vacca, Msc*, Angela Acosta, Msc(c), Ivan Rodriguez, Msc(c) Grupo RAM - Departamento de Farmacia, Universidad Nacional de Colombia, Bogotá, Colombia A B S T R A C T Objetives: To suggest a scheme of decision making on pricing for medicines that are part of Free Regulated Regime, a regulation way of the pharmaceutical pricing policy in Colombia. It includes two regulation tools: international reference prices and a cost minimization analysis methodology. Methods: Following the current pricing policy, international reference prices were built with data from five countries for selected medicines, which are under Free Regulated Regime. The cost minimization analysis methodology includes selection of those medicines under Free Regulated Regime with possible comparable medicines, selection of comparable medicines, and treatment costs evaluation. Results: As a result of the estimate of International Reference Prices, four medicines showed in the domestic pharmaceutical market a bigger price than the Reference Price. A scheme of decision-making was design containing two possible regulation tools for medicines that are part of Free Regulated Regime: estimate of international reference prices and cost minimization analysis methodology. This diagram would be useful to assist the pricing regulation of Free Regulated Regime in Colombia. Conclusions: As present Introducción En Colombia, desde 2006 [1] el modelo de regulación de precios de medicamentos establece tres regímenes bajo los cuales se pueden clasificar los medicamentos, en función del comportamiento de sus precios: Régimen de Libertad Vigilada, en el que se acepta el precio que fija el fabricante o el distribuidor con el compromiso de que estos reporten a la Comisión Nacional de Precios de Medicamentos (CNPM) las variaciones y la determinación de sus precios; Régimen de Libertad Regulada (RLR) en el que ingresan los medicamentos que: 1) hagan parte de una medida para proteger la salud pública así como las circunstancias de extrema urgencia o emergencia nacional; 2) Que tengan alta concentración de mercado, en ventas o unidades, respecto a su Clasificación Terapéutica Relevante, CTR, sobre estas se determinará que hay alta concentración cuando el índice Herfindhal-Hirshman (HH), tanto por valor en ventas como en unidades vendidas, sea superior a 0,45; y 3) Que carezca de sustitutos al momento de entrar al mercado. A los medicamentos que ingresen a este régimen se les calcula un results shows, international reference prices make clear when domestic prices are higher than those of reference countries. In the current regulation of pharmaceutical prices in Colombia, the international reference price has been applied for four medicines. Would be suitable to extend this methodology to other medicines of high impact on the pharmaceutical expenditure, in particular those covered by public funding. The availability of primary sources about treatment costs in Colombia needs to be improved as a requirement to develop pharmaco-economic evidence. SISMED is an official database that represents an important primary source of medicines prices in Colombia. Nevertheless, having into account that SISMED represents an important advantage of transparency in medicines prices, it needs to be improved in quality and data availability. Palabras Claves: pharmaceutical pricing policies, international reference pricing, cost minimization analysis. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. Precio Internacional de Referencia (PIR), el cual se establecerá como techo o precio máximo. Finalmente, si el precio de un medicamento en RLR supera el precio techo, ingresará entonces en el Régimen de Control Directo (RCD), en el cual la CNPM le fija directamente un Precio Máximo de Venta al Público, con base en la estimación del PIR. La CNPM es la entidad que determina el régimen al que ingresa un medicamento y establece la metodología para calcular el PIR. Para este fin, exige el reporte de los precios de venta por parte de diversos agentes en el Sistema de Información de Medicamentos (SISMED) del Ministerio de la Protección Social (MPS), plataforma de ingreso y consulta de información de precios de medicamentos en Colombia. Desde 2002 se ha observado un incremento en el uso de mecanismos de excepción – e. g. acciones judiciales- para la cobertura de medicamentos no incluidos en los planes de beneficios, ocasionando mayores reembolsos del Estado a los aseguradores en salud; Figura 1 a Materiales Complementarios en: doi:10.1016/ j.jval.2011.05.034. Los precios de medicamentos que ocasionaron estos reembolsos son comparativamente altos respecto a países Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Título corto: PRI y ACM en Regulación Farmacéutica de Precios en Colombia. * Correspondencia del autor: Claudia Vacca, Docente Departamento de Farmacia, Facultad de Ciencias, Universidad Nacional de Colombia, Edif. 450, Oficina 214, Bogotá D.C., Colombia; Tel: (57-1)- 3165000 ext. 14623; Fax: 3165000 ext. 14623. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.034 VALUE IN HEALTH 14 (2011) S16 –S19 de la Región, tal como se muestra en la Tabla 1 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.034, y más adelante con la estimación del PIR; esto exigió un papel activo por parte de los reguladores, i.e. CNPM y MPS. Como resultado de esto se emitió la Circular 02 de 2008 y las Circulares 01, 02 y 04 de 2009. La Circular 02 de 2008 [1] incorpora dentro del RLR 15 marcas comerciales cuyo índice de concentración supera el 0,45; la Circular 01 de 2009 [1] modifica el mínimo de países de referencia a incluir en la construcción del PIR, el cual se reduce a cuatro. Por su parte, la Circular 02 de 2009 [1] establece el PIR para el medicamento Lopinavir 200 mg Ritonavir 50 mg. La Circular 04 de 2009 [1] incorpora 52 marcas comerciales al RLR, dentro de las que se incluyen medicamentos de alto impacto en el gasto por reembolsos, como Trastuzumab, Dasatinib y Rituximab; entre otros., Esta medida podría resultar en la contención de los precios de estos medicamentos. En la regulación mas reciente se lee una intención de controlar el gasto causado por los reembolsos en Colombia. Por un lado, la Circular 02 de 2010 [1] da a la CNPM la facultad para fijar el precio máximo de recobro (PMR) –i. e., precio máximo para el reembolso de medicamentos a las aseguradoras en salud-; y la Circular 04 de 2010 establece los precios maximos de recobro para dieciséis (16) principios activos. Luego, el Decreto 4474 de 2010 otorga al MPS la misma facultad que previamente se otorgó a la CNPM y deroga las disposiciones que le sean contrarias. Finalmente, la Resolución 1020 de 2011 [2] fija los precios máximos de reembolso para cuarenta y siete (47) principios activos, de los cuales ocho (8) se habían incluido también en la Circular 04 de 2010. Tanto en la Circular 04 de 2010, como en la Resolución 1020 de 2011, las metodologías son discrecional es. Teniendo en cuenta esta dinámica reciente en la regulación y las responsabilidades técnicas de la CNPM, éste estudio pretende facilitar la toma de decisiones para la regulación de los medicamentos que se encuentren bajo el RLR. Para ello se sugiere la estimación del PIR y el Análisis de Costo Minimización (ACM), los cuales pueden ser mecanismos simultáneos o complementarios, que permiten construir una referencia para la fijación de precios de regulación. Metodología Estimación del PIR De acuerdo con la Circular 04 de 2006 [1], “el PIR será el promedio de los tres (3) precios unitarios más bajos de los medicamentos iguales producidos [. . .] por la misma casa matriz en los países de referencia. En la fijación del PIR, la Comisión comparará los precios de los medicamentos en el mismo nivel de la cadena de distribución en los países de referencia.” De acuerdo a la Circular 02 de 2010, los países que se toman como referencia son: Argentina, Brasil, Chile, Colombia, Ecuador, México, Panamá, Perú y Uruguay.” [1] Consecuentemente con la normatividad expuesta, se seleccionó una muestra de medicamentos de interés que a febrero de 2010 se encontraban en el RLR. Dadas las restricciones de fuentes de información de precios de medicamentos para los países de referencia, se relajó el criterio de contar con al menos información de cuatro de los países. Se estimó el PIR con información de los precios de adquisición por parte de entidades de salud – e. g. ministerios de salud, hospitales o aseguradores- de al menos tres de los siguientes países: Argentina, Brasil, Colombia y México. Para el caso de Brasil, los precios consultados corresponden al dato de la última compra de una institución del Estado; para Colombia corresponden al promedio de compra institucional para un período determinado, de acuerdo al SISMED. Para México, los precios corresponden, tanto a compras institucionales del Estado, como a precios de venta de laboratorios. Por último, para el caso de Argen- S17 tina, los datos corresponden a precios de venta de laboratorios y proveedores. Propuesta de análisis de costo minimización La aplicación de ACM para medicamentos bajo RLR se procuró desarrollar bajo las siguientes fases: Selección de medicamentos con posibles comparadores, selección de medicamentos comparadores y evaluación de costos de tratamiento. La comparación en cuanto a seguridad y eficacia para identificar medicamentos comparadores se formuló comprendiendo las siguientes etapas: 1. definición de indicaciones aprobadas en el país para cada medicamento, 2. construcción de la pregunta de investigación, según la metodología de elección de la mejor evidencia PICO (pregunta, intervención, comparador, resultados) [3]. La Tabla 2 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.034 expone la pregunta desarrollada para el quinapril, medicamento bajo RLR, 3. búsqueda de literatura científica basada en la(s) pregunta(s) clínica(s) planteada(s), con estrategias de búsqueda de estudios secundarios, en particular de revisiones sistemáticas y de guías de práctica clínica. En segunda instancia, se procede a buscar estudios primarios como ensayos clínicos aleatorizados. 4. evaluación de la calidad de la evidencia científica disponible mediante diversas escalas y listas de chequeo que han sido propuestas para ello [4-9]. No se pretende realizar una revisión exhaustiva de la calidad de cada estudio, se busca eliminar estudios con alto riesgo de sesgo que pudieran comprometer los resultados. 5. resumen de los hallazgos para establecer alternativas terapéuticas farmacológicas existentes, de acuerdo a la revisión de la literatura realizada, se integran los resultados y se establecen las alternativas farmacológicas existentes. Los hallazgos podrán incluir las siguientes variables: intervención, comparadores, condición o patología, eficacia y seguridad. 6. elección del medicamento comparador, se define según los resultados sobre seguridad y eficacia de las alternativas farmacológicas identificadas para el medicamento de interés, teniendo en cuenta los siguientes criterios: Mejor o igual desempeño en cuanto a seguridad y a eficacia del medicamento de interés, indicación evaluada igual a la indicación del medicamento de interés, y que el medicamento comparador tenga relevancia terapéutica en el país. Si bien, para el caso del quinapril las alternativas terapéuticas se encuentran en revisiones sistemáticas y metanálisis, en los que se recoge evidencia terapéutica de priles en general, los desenlaces en cuanto a seguridad y eficacia terapéutica no revelan diferencias entre unos y otros, por lo cual, se consideran entre sí con diferencias no relevantes y, aunque existen comparaciones con otros principios activos, se seleccionó el enalapril como medicamento comparador dada su relevancia terapéutica local. La evaluación de costos de tratamiento comprende tres opciones: 1).si el medicamento es componente fundamental del costo, se comparan precios nacionales e internacionales con el precio del comparador; 2) evaluación de costos integrales de tratamiento; 3) de no existir datos de costos, pedir al laboratorio evidencia sobre ahorros, en este caso se pueden asignar transitorios y esperar evidencia. Siguiendo la opción uno para el caso del quinapril, se construyeron los costos directos de tratamiento farmacológico con los precios locales e internacionales de los medicamentos de interés. El objeto final es calcular el precio día/ataque evitado, se asume también que el costo de otra medicación asociada al tratamiento S18 VALUE IN HEALTH 14 (2011) S16 –S19 antihipertensivo es la misma para los dos medicamentos en evaluación. Esquema de toma de decisión para medicamentos bajo RLR Tanto la estimación del PIR, como el ACM, se integran en una propuesta de esquema que facilita la toma de decisión por parte de la CNPM para la regulación de precios de medicamentos en Colombia, de igual manera, en los resultados se enumeran los posibles desenlaces en la aplicación del esquema. Fuentes de información y calidad de datos Para la estimación del PIR, se contó con la información de cinco países consultados a través de las siguientes vías: contactos directos institucionales para Argentina, México y Colombia; y consultas en portales de precios de medicamentos para Brasil [10] y Colombia [11]. No se incluyeron otros países de referencia, en portales de información como el Observatorio de Precios de medicamentos de Perú y CENABAST de Chile [12] no se encontraron precios publicados para los medicamentos de interés. Por su parte, Ecuador maneja compras en instancias como el Ministerio de Salud y la Sociedad de Lucha contra el Cáncer, cuyos precios de compra no son públicos. En cuanto a la calidad de la información recogida por el SISMED, la principal falla es la ausencia de estandarización. Algunos reportes muestran el precio por presentación, mientras otros el precio por tableta o un precio que no corresponde a descripción alguna del producto. Resultados PIR Rituximab, Bevacizumab, Infliximab y Adalimumab, cuatro de los medicamentos incluidos en la Circular 04 de 2010 [1] cuentan con un PMR asignado por la CNPM muy cercano al PIR establecido en este estudio, Figura 2 a Materiales Complementarios en: doi: 10.1016/j.jval.2011.05.034. La Tabla 3 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.034 relaciona los precios de países de referencia utilizados para la estimación de este estudio. ACM El precio promedio del enalapril, incluyendo precios locales e internacionales es de 4.86 centavos de dólar por unidad, para el quinapril es de US$ 1.02, Tabla 4 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.034. El precio día/Ataque HTA evitado para el quinapril es de U 2.17, y enalapril es de U 9.72 centavos. De acuerdo a las posibles decisiones contemplados en la rama de ACM, el medicamento comparador determinaría el precio a asignar para el quinapril, 4.86 centavos de dólar unidad de tableta. Se diseña un algoritmo de decisión coherente con la aplicación del ACM para los medicamentos bajo RLR, Figura 3 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.034, con sus respectivos desenlaces de decisión, Tabla 5 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.034. Análisis El esquema de regulación por precios de referencia a nivel exfábrica es aplicado en la mayoría de países de la Unión Europea. En Holanda, los precios máximos permitidos se determinan calculando un precio medio agregado con base en los precios de Bélgica, Alemania, Reino Unido y Francia. Dinamarca utiliza esta metodología definiendo como referencia todos los países europeos, excepto Grecia, Portugal, España y Luxemburgo. Portugal e Italia utilizan precios de referencia a nivel ex-fábrica, aunque mientras el primero toma el precio mínimo en sus países de referencia, el segundo toma el promedio ponderado [13]. Los retos metodológicos a tener en cuenta en un sistema de precios de referencia como el que busca aplicar Colombia se podrían restringir a: 1) Identificación de medicamentos idénticos en términos de las formas farmacéuticas, concentraciones y nombres de marca; 2) Identificar punto de la cadena de valor del medicamento, en que se aplican los precios de referencia; 3) Se requiere de información adecuada y actualizada de los precios de los medicamentos en otros países y de sistemas de información (tecnologías de información y comunicación) robustos, para el procesamiento oportuno de los datos [14]. Resolviendo estos desafíos metodológicos, varios países han identificado el comparador adecuado y han logrado un control de precios efectivo. Fuentes de información El relevamiento de información de precios de compra de medicamentos en diferentes países presentó obstáculos como deficiencias en las fuentes de consulta pública, falta de estandarización en la descripción de los productos farmacéuticos, imprecisión de los períodos de compra, entre otros. Este estudio permite identificar aspectos útiles a tener en cuenta en plataformas de intercambio de información de precios de medicamentos para países de América Latina: estandarización para la descripción de producto farmacéutico, fuentes de información oficiales, períodos y tasas a las que corresponden los datos de compra registrados. Con el avance logrado bajo esta experiencia, se podría construir una plantilla para comparación de precios de medicamentos útil para la toma de decisiones de la CNPM, del MPS y del Ministerio de Comercio Indsutria y Turismo en Colombia. La misma podría actualizarse o modificarse desde usuarios internos y externos autorizados. No fue posible evaluar la calidad de información de costos de tratamiento para la aplicación de ACM. PIR Según se observa en la Tabla 3 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.034, para tres de los cuatro medicamentos para los cuales se sugiere PIR, Colombia presenta precios que son entre un 19 y un 37% más altos que el precio de referencia más bajo, todos correspondientes a Brasil. El levantamiento de información de precios de medicamentos requiere de esfuerzos de cooperación entre los países de América Latina, al respecto es necesario superar retos metodológicos como la estandarización en la descripción de medicamentos y facilitar herramientas tecnológicas que faciliten la transparencia en el manejo e intercambio de información de precios de medicamentos en la Región. Herramienta para facilitar la toma de decisión de los medicamentos clasificados en el RLR ACM Se diseña un diagrama, Figura 3, como herramienta útil en la regulación de medicamentos bajo el RLR, el cual incorpora dos de los esquemas de regulación encontrados en la revisión de literatura: PIR y ACM. En Colombia y en países de la región, son pocas las fuentes y estudios asociados al costo de tratamientos médicos. En la revisión realizada, no se encontraron publicaciones, ni portales públicos de consulta de tarifas de servicios razón por la cual solo se tuvieron VALUE IN HEALTH 14 (2011) S16 –S19 en cuenta los costos directos de tratamiento relacionados con los medicamentos de interés, aunque en rigor estos deberían estar incluidos para el ACM. S19 www.valueinhealthjournal.com/issues (seleccione el volumen, número y artículo). REFERENCIAS Conclusiones Para los medicamentos que se encuentran en RLR en Colombia pueden ser aplicados dos esquemas técnicos de regulación: PIR y ACM. Los PIR podrían ser útiles en la regulación de precios de todos los medicamentos bajo RLR, actualmente solo se ha utilizado para la fijación de precios de 6 medicamentos de recobro y solo uno bajo RLR, i. e. Lopinavir-Ritonavir. El esquema de ACM requiere optimización en el cálculo de los costos de tratamiento, se requiere mayor disponibilidad de fuentes primarias de información para el desarrollo de evidencia farmaco-económica. El SISMED es un valioso insumo para la regulación de precios de medicamentos en Colombia, algunas de las inconsistencias y hallazgos de la revisión de datos podrían subsanarse con la realización de auditorias periódicas que garanticen la validez y el cumplimiento de los requisitos de reportes. El diagrama para la toma de decisión es una herramienta técnica que puede facilitar y mejorar la regulación de precios de medicamentos bajo RLR en Colombia. Agradecimientos A Ludovic Reveiz (Promoción y Desarrollo de la Investigación de la Organización Panamericana de Salud) por su apoyo y aporte en el desarrollo de esta propuesta. Fuentes de financiamiento: Universidad Nacional de Colombia, Ministerio de Comercio, Industria y Turismo de Colombia. Materiales Complementarios Material complementario que acompaña este artículo se puede encontrar en la versión en línea como un hipervínculo en doi: 10.1016/j.jval.2011.05.034 o si es un artículo impreso, estará en [1] MCIT. Ministerio de Comercio Industria y Turismo, Portal Regulación de Precios de Medicamentos en Colombia. Disponible en: http://www .mincomercio.gov.co/eContent/newsdetail.asp?id⫽2674&idcompany⫽ 23. [Accessed 22 de Febrero 2011]. [2] MPS. Resolución 005 de 2011. Disponible en: http://consultorsalud .com/index.php?option⫽com_content&view⫽article&id⫽411% 3Aresolucion-005-de-2011-mps&catid⫽37%3Aflash-noticias& Itemid⫽10. [Accessed Febrero 22 de 2011]. [3] Mamédio da Costa Santos C, Roberto Cuce Nobre M. Estrategia pico para la construcción de la pregunta de investigación y la búsqueda de evidencias. Rev Latino-am Enfermagem 2007;15:1038 – 42. [4] Moher DCD, Eastwood S, Olkin I, et al. Improving the quality of reports of meta-analyses of randomised controlled trials: the QUORUM statement. Lancet 1999;354;1896 –900. [5] Katrak PBA, Massy-Westropp N, Kumar S, Grimmer KA. A systematic review of the content of critical appraisal tools. BMC Medical Research Methodology 2004;4:22. [6] Olivo SA, Gadotti IC, Fuentes J, et al. Scales to assess the quality of randomized controlled trials: a systematic review. Phys Ther 2008;88: 156 –75. [7] TheCochrane C. Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.1 [updated September 2008]. Higgins JPT, Green S (eds.). [8] Tooth L, McCluskey A, Hoffmann T, et al. Appraising the quality of randomized controlled trials: inter-rater reliability for the OTseeker evidence database. J Eval Clin Pract 2005;11:547–55. [9] Vlayen J, Hannes K, Sermeus W, Ramaekers D. A systematic review of appraisal tools for clinical practice guidelines: multiple similarities and one common deficit. XIV Cochrane Colloquium, 2006. October 23–26; Dublin, Ireland. [abstract]. [10] Brasil, B.d.P.e.S.d. Disponible en: http://portal2.saude.gov.br/BPS/ visao/consultapublica/index.cfm 2009 [cited 2009 diciembre]. [11] Sistema Integral de Información de la Protección Social, SISPROSistema de Información de Precios de Medicamentos, SISMED. Disponible en: http://www2.sispro.gov.co/Paginas/Publicaciones.aspx [cited 2011 julio]. [12] Cenabast, C. http://www.cenabast.cl/ListadodePrecios/ 2009 [cited 2009 diciembre]. [13] OECD, Pharmaceutical pricing policies in a Global Market. 2008: OECD Health Policy Studies [cited 2009 diciembre]. [14] Mossialos M, Regulating pharmaceutical prices in the European Union ed. M.M.T.W. 2004 [cited 2009 diciembre]. VALUE IN HEALTH 14 (2011) S20 –S23 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval ECONOMIC ANALYSIS Hospitalization Costs for Heart Failure in People with Type 2 Diabetes: Cost-Effectiveness of its Prevention Measured by a Simulated Preventive Treatment Joaquín E. Caporale, MPF1, Jorge Elgart, MPF1, Guillermina Pfirter, MD1, Pablo Martínez, MD2, Gloria Viñes, MD2, Jorge T. Insúa, MD3, Juan J. Gagliardino, MD, PhD1,* 1 Centro de Endocrinología Experimental y Aplicada, PAHO/WHO Collaborating Centre for Diabetes, La Plata, Argentina; 2Hospital Privado de Comunidad, Mar del Plata, Argentina; 3Departamento de Medicina, Universidad Austral, Buenos Aires, Argentina A B S T R A C T Objectives: To estimate the cost-consequence of interventions to prevent hospitalizations for heart failure (HF) in people with type 2 diabetes. Methods: In HF events (63) from type 2 diabetes-related hospitalizations (N ⫽ 462) recorded in an Argentine hospital (March 2004 –April 2005), we verified 1) the presence of one metabolic HF predictor (glycosylated hemoglobin [HbA1c] value) before hospitalization; and 2) in a simulation model, the resources needed for its prevention controlling such predictor during 6 months before and after the event. Sensitivity analysis of HF risk reduction, hospitalization cost, and cost of different treatments to achieve HbA1c 7% or less was performed with a Monte Carlo simulation (10,000 iterations). Results: HF represented 14% of hospitalizations, with a Introduction Heart failure (HF) represents a major public health concern because of its continuous incidence rise, hospitalization rate, and care costs. The United States has approximately 670,000 new HF cases per year in persons older than age 45 years [1,2] and its hospitalization rate has tripled between 1979 and 2004, partly due to the aging population and the efficiency of cardiovascular therapy [3]; the estimated HF cost burden in the United States in 2009 was $37.2 billion [2]. The Framingham study established that a clinical history of diabetes was independently associated with risk of developing HF [4]. More recent studies [5] have reported higher annual incidences of HF in the diabetic population. Further, the Heart and Estrogen/Progestin Replacement Study demonstrated that diabetes was the strongest independent risk factor for HF development (adjusted hazard ratio 3.1) [6]. In people with diabetes, glycosylated hemoglobin (HbA1c) value is associated with HF 44% rehospitalization rate for the same cause. Due to the total estimated cost for an HF hospitalization event was $437.31, the prevention attained using our simulated treatment was $2326.51. The number needed to treat to prevent an HF event under any of the proposed alternatives to reduce HbA1c would be 3.57 (95% confidence interval 2.00 –16.67). The additional cost of the simulated treatment versus the real one oscillates between $6423.91 and $8455.68. Conclusions: HbA1c control to reduce the number of HF events would be economically beneficial for health care payers. Keywords: cost analysis, diabetes, heart failure, prevention and control. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. risk [7–9]. In a cohort study of 48,858 adult patients with predominantly type 2 diabetes, Iribarren et al. [7] showed that each 1% increase in HbA1c was associated with an 8% increased risk of HF hospitalization or death, even after adjusting for demographics, medical history, medications, and other risk factors. In the Atherosclerosis Risk in Communities study [9], the risk of HF also increased proportionally with HbA1c among people with diabetes and no evidence of previous HF [9]. Despite this strong evidence on the relationship between HF and HbA1c levels, the latter are above target values in most patients worldwide, including Argentina, [10,11]. Therefore, many HF events could be prevented in people with diabetes by improving their metabolic control, with the consequent beneficial effect for patients and the health care system. To test this hypothesis, we carried out a cost-consequence study comparing the cost of HF events in people with type 2 diabetes with that of a simulated intensive preventive treatment of hyperglycemia. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Juan J. Gagliardino, CENEXA (UNLP-CONICET), Facultad de Ciencias Médicas, UNLP - Calles 60 y 120, 1900 La Plata, Argentina. E-mail: [email protected], [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.018 VALUE IN HEALTH 14 (2011) S20 –S23 Methods To study the simulated cost of early intensive metabolic control to prevent hospitalization events caused by HF episodes in people with type 2 diabetes we adopted the following data collection scheme: we recorded all the hospitalization events of people with type 2 diabetes at the Hospital Privado de Comunidad (HPC), Mar del Plata, Argentina, from March 2004 to April 2005; thereafter, we identified those with HF events (International Classification of Diseases, Ninth Revision code 428), and evaluated their clinical and metabolic state. HbA1c levels (HF event predictor) and resource utilization rates during hospitalization were also recorded 6 months before and after the event. Diabetes and associated obesity, hypertension, and dyslipidemia were identified using American Diabetes Association criteria [12]. According to Stratton et al. [8], we defined an HF event as preventable when the patient had no HF antecedents and a HbA1c value greater than 7%; the benefits of our intervention were also measured according to these authors: a 14% HF risk reduction for each 1% decrease in HbA1c independent of the treatment used to attain such value. The study was implemented according to the Good Clinical Practice Recommendations (International Harmonization Conference), the 5330/97 regulation of the Administración Nacional de Medicamentos, Alimentos y Tecnología Médica, National Law 25.326 of Personal Data Protection, and the ethical Helsinki Declaration guidelines. The study protocol was approved by the HPC Ethics Committee. Simulated pharmacologic interventions to control HbA1c Before the HF event Alternative I [12]: 1) Metformin (2,500 mg/day) plus glibenclamide (15 mg/day) in patients treated with both drugs; 2) metformin (850 mg/day) plus glibenclamide (10 mg/day) in patients without previous drug treatment; and 3) insulin (40 IU/day) in patients previously treated with insulin. Alternative II [12]: 1) Metformin (2,500 mg/day) plus glibenclamide (15 mg/day) in patients treated or not with both drugs, and 2) insulin (40 IU/day) in patients previously treated with insulin. Alternative III (as in the United Kingdom Prospective Diabetes Study [13]): Metformin (2,500 mg/day) in 32% of cases and metformin (1,700 mg/day) plus glibenclamide (10 mg/ day) in the remaining 68% of patients. Self-monitoring blood glucose Because there is no general agreement on the appropriate number of strips, we established the following arbitrary number for strip use: 1) 40 to 72 strips/patient/month for insulin-treated patients; and 2) 24 strips/patient/month for those treated with oral agents and with no previous strip consumption. The glucometer cost was not included in the estimation because in general is provided free of charge. After the HF event Insulin administration (daily 40 IU/day) in all cases and 72 strips/ patient/month. Costs We considered direct medical costs from the health payers’ perspective. Because we do not have the real cost of an acute event, we adopted the values of the largest social security health care payer (Instituto de Obra Médico Asistencial) in hospitals with similar characteristics to the HPC. Ambulatory care costs (doctor visits, laboratory tests, and other medical practices) were estimated using the National Care Nomenclature Values. S21 Pharmacotherapy cost was based on a microsting approach using a mean unit retail price per milligram of each drug or per insulin units included in the study (recorded or proposed), and the corresponding daily dose (recorded or proposed). With these data, we estimated the mean daily cost for each drug for a 6-month period before and after the HF event. We compared thereafter the cost of the proposed treatment minus the real treatment versus the cost of the hospitalization events minus that of the prevented events (including ambulatory care costs before and after such event). Costs were calculated on Argentinean pesos and converted to US dollars at the average official exchange rate for the period March 2004 to April 2005 ($1 ⫽ 2.94 Argentinean pesos). Sensitivity analysis The probabilistic sensitivity analysis included: 1) the total cost of the HF event; 2) the unitary cost of drugs and strips; and 3) the relative risk reduction (percentage) to develop a HF event. A Monte Carlo simulation was carried out (10,000 iterations), assuming 1) a uniform distribution (minimum ⫽ $246.26; maximum ⫽ $724.49) based on pre-established values for HF events from many possible scenarios defined by Instituto de Obra Médico Asistencial; 2) selfgenerated probability distributions using monthly observations of mean price per milligram (for each drug either used or proposed) and mean price per unit (for each strip) in the Argentine pharmaceutical market at six months before and after the event; and 3) a normal distribution with a mean of 14% and a standard deviation of 2% for the HF relative risk reduction from hyperglycemia treatment that allowed us to achieve the 95% confidence interval reported by Stratton et al. [8]. We used Monte Carlo iterations to calculate Pearson’s coefficient to assess the level of association between these assumptions and the result (additional total cost for each of the alternatives considered). Also, we assumed that 1) the antihyperglycemic therapy implemented could reduce the relative risk for non fatal HF with a comparable effectiveness to that recorded in the United Kingdom Prospective Diabetes Study [8,14], despite our population hospitalized for HF was older than that of the United Kingdom Prospective Diabetes Study; and 2) the decreased relative risk for HF would be linear; that is, a 14% risk decrease by each 1% HbA1c decrease. All calculations were performed in MS-Excel 11.0 (Microsoft Corp., Redmond, WA) with add-on Crystal Ball Trial Version (Decisioneering (R), Inc., Denver, CO). Results Out of a total of 462 hospitalized patients with type 2 diabetes, 38% of admissions were related to cardiovascular disease, HF being the most frequent cause (14%); 44% of the HF events were rehospitalized for the same cause. Forty-nine percent were women, with a mean age of 77.1 ⫾ 8.4 years; 80% were obese (body mass index ⬎ 30); 77% had hypertension; and 71% had hypercholesterolemia. HbA1c levels were between 7.6% and 8.6%. Thirty percent of the population had microangiopathic complications (e.g., neuropathy, retinopathy, or nephropathy) and 29% had macroangiopathic signs/events (e.g., acute myocardial infarction, stroke, or lower-limb claudication) (Table 1 in Supplemental Materials found at: doi:10.1016/j.jval. 2011.05.018). Sixteen out of the total 63 HF events were preventable by tight control of HbA1c (criteria mentioned above). Thirty-one percent of the patients hospitalized for HF events and with HbA1c of 7% or greater (n ⫽ 16) received no antidiabetic drug treatment before the event; 55% of those treated received oral monotherapy (metformin or glibenclamide), 18% received combined therapy, and the remaining 27% was treated with insulin. After the event, 50% of patients received no antidiabetic treatment and among those treated, 37% received monotherapy (19% some S22 VALUE IN HEALTH 14 (2011) S20 –S23 oral agent and 18% insulin), whereas 12% received combined oral therapy (Table 2 in Supplemental Materials found at: doi:10.1016/ j.jval.2011.05.018). As mentioned, 28% of the HF events would be preventable (4.48 over 16 HF events) with the antihyperglycemic pharmacologic interventions proposed; a proportional number of rehospitalizations for HF events would be also avoided (0.84 cases). Because the total estimated cost for a HF hospitalization event was $437.31, the prevention using our simulated treatment would be $2,326.51. The number needed to treat to prevent an HF event with any of the pharmacologic options proposed to reduce HbA1c would be 3.57 (95% confidence interval 2.00 –16.67). The total cost of the simulated treatments was (in US dollars): Alternative I ⫽ $13,615.09, Alternative II ⫽ $14,079.34, and Alternative III ⫽ $12,047.58; the real treatment was $3297.16. Consequently, the additional costs were $7991.42, $8455.68 and $6423.91 for Alternatives I, II, and III, respectively (see Table 3 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.018). According to the pre-event treatment costs and with an arbitrary decision threshold of $510.20 for the net per capita additional cost of the simulated treatment, the probability to surpass would be 36%, 52%, and 2.3% for Alternatives I, II, and III, respectively. Such cost variation could be ascribed to the total minimal/maximal cost of the event (Pearson’s correlation in each of the alternatives considered ranged from ⫺0.65 to ⫺0.71), strips (Pearson’s correlation 0.49 to 0.52), insulin (Pearson’s correlation 0.32 to 0.40) and the percent reduction of HF risk (Pearson’s correlation ⫺0.31 to ⫺0.33). In all cases, the correlation coefficients had the expected sign and significance for the confidence level used (95%). Discussion As already reported, we found that cardiovascular disease was the main cause of hospitalization, with a particularly high frequency of HF [15]. Based on the reported relationship between HbA1c and HF [5,7–9], we tested the simulated cost-consequence of improving HbA1c levels to prevent HF hospitalization events in people with type 2 diabetes. Our data showed that the probability to surpass an arbitrary decision threshold of $510.20 for the net per capita additional cost of the simulated treatment was 36%, 52%, and 2.3% for the medium, highest, and lowest alternative treatment costs, respectively. Such variation would depend on the cost of the event, the strips, and the insulin treatment, as well as on the percent reduction of HF risk. These results confirm our working hypothesis that prevention of HF events in people with type 2 diabetes has a reasonable and affordable cost for payers. It should be noted that the cost of the intensive hyperglycemia treatment was high because we applied the traditional insulin treatment after the event, regardless of reported evidence showing that metformin could also be used in these patients [16]. The use of metformin rather than insulin would decrease significantly the preventive treatment cost. In addition, the low number needed to treat value plays in favor of its applicability in settings similar to the one currently described. Beyond this economic achievement, prevention of HF hospitalization events could also decrease their high recorded mortality rate (23%). In our sample, 33% of the patients hospitalized for HF events and with HbA1c 7% or greater did not receive antidiabetic drug treatment before the event and more than half of them received a single drug. Comparable undertreatment behavior was observed after the event. Our results promote a more proactive treatment attitude. Our conclusions are in line with the proposal of Karter et al. [17] about the convenience for health financing entities to provide coverage for preventive strategies now instead of complete coverage for recovery/rehabilitation strategies in the future. In Argentina, the Health Ministry provision of economic incentives to entities of the Social Security subsector that include preventive strategies in their care programs for chronic diseases, play also in favor of this concept. This policy would be particularly important in developing countries, where the expected rise in the prevalence of diseases such as type 2 diabetes will imply an increased demand of care both in the short and long term [11,18]. As with most simulation studies, our own has some limitations, namely 1) we had no direct information on glycemic selfmonitoring performance; and 2) we assumed a linear efficacy relationship between risk factor reduction and HF prevention, despite many authors have shown the appropriateness of using Weibull distributions and accelerated failure time equations to treat these relationships [7,19,20]. Nonetheless, using the Economic Assessment of Glycemic Control and Long-Term Effects of Diabetes model hazard ratios [14] for nonfatal HF, we found high and similar goodness of fit between a logarithmic (R2 ⫽ 0.998 for hyperglycaemia relative risk reduction) and a linear tendency (R2 ⫽ 0.927) to adjust hazard ratio reductions from different HbA1c values (7%–11%) (data not shown; it is available from the authors on request). Thus, although not precisely estimated, our results would still be valid, conservative, and suitable for evidence-based decision making. Conclusions Considering that no similar data have been previously reported, our results show for the first time that intensive hyperglycemia treatment to decrease the number of hospitalizations for HF events in people with type 2 diabetes would have a favorable costconsequence ratio. Thus, we believe it is important to identify inadequate HbA1c values in people with type 2 diabetes and treat them to reach values within target, as recommended by international guidelines. This preventive policy would simultaneously decrease cardiovascular complications requiring high-cost hospitalization and rehospitalization, with the consequent optimization on the use of economic resources. Acknowledgments The authors thank the authorities at Hospital Privado de Comunidad, Mar del Plata, Argentina, and Eng. Luis Buffoni for providing electronic data, Eleonora Aiello and Robert A. Gerber (Pfizer) for authorizing the EAGLE Model use, and Adriana Di Maggio for careful manuscript editing. Source of financial support: This study was partially supported with an unrestricted grant provided by Merck Sharp & Dohme of Argentina and funds provided by CONICET. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi:10.1016/j.jval.2011.05.018, or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). REFERENCES [1] Horwich TB, Fonarow GC. Glucose, obesity, metabolic syndrome, and diabetes relevance to incidence of heart failure. J Am Coll Cardiol 2010; 55:283–93. [2] Lloyd-Jones D, Adams R, Carnethon M, et al. Heart disease and stroke statistics—2009 update. A report from the American Heart Association VALUE IN HEALTH 14 (2011) S20 –S23 [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] Statistics Committee and Stroke Statistics Subcommittee. Circulation 2009;119:e21–181. Fang J, Mensah GA, Croft JB, Keenan NL. Heart failure-related hospitalization in the U.S., 1979 to 2004. J Am Coll Cardiol 2008;52:428–34. Kannel WB, Hjortland M, Castelli WP. Role of diabetes in congestive heart failure: the Framingham study. Am J Cardiol 1974;34:29 –34. Nichols GA, Gullion CM, Koro CE, et al. The incidence of congestive heart failure in type 2 diabetes: an update. Diabetes Care 2004;27:1879 – 84. Bibbins-Domingo K, Lin F, Vittinghoff E, et al. Predictors of heart failure among women with coronary disease. Circulation 2004;110:1424 –30. Iribarren C, Karter AJ, Go AS, et al. Glycemic control and heart failure among adult patients with diabetes. Circulation 2001;103:2668 –73. Stratton IM, Adler AI, Neil HAW, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ 2000;321:405–12. Pazin-Filho A, Kottgen A, Bertoni AG, et al. HbA1c as a risk factor for heart failure in persons with diabetes: the Atherosclerosis Risk in Communities (ARIC) study. Diabetologia 2008;51:2197–204. Chan JC, Gagliardino JJ, Baik SH, et al. The IDMPS Investigators. Multifaceted determinants for achieving glycemic control: the International Diabetes Management Practice Study (IDMPS). Diabetes Care 2009;32:227–33. Gagliardino JJ, de la Hera M, Siri F; Grupo de Investigación de la Red QUALIDIAB. Evaluación de la calidad de la asistencia al paciente diabético en América Latina. Rev Panam Salud Pública 2001;10:309 –17. American Diabetes Association: clinical practice recommendations. Diabetes Care 2002;25(Suppl.):S1–147. S23 [13] Turner RC, Frighi CC, Holman RR. Glycemic control with diet, sulfonylurea, metformin or insulin in patients with type 2 diabetes mellitus: progressive requirement for multiple therapies (UKPDS 49). JAMA 1999;281:2005–12. [14] Mueller E, Maxion-Bergemann S, Gultyaev D, et al. Development and validation of the Economic Assessment of Glycemic Control and LongTerm Effects of Diabetes (EAGLE) model. Diabetes Technol Ther 2006;8: 219 –36. [15] Carr AA, Kowey PR, Devereux RB, et al. Hospitalizations for new heart failure among subjects with diabetes mellitus in the RENAAL and LIFE studies. Am J Cardiol 2005;96:1530 – 6. [16] Tahrani AA, Varughese GI, Scarpello JH, Hanna FWF. Metformin, heart failure, and lactic acidosis: is metformin absolutely contraindicated? BMJ 2007;335:508 –12. [17] Karter A, Stevens M, Herman WH, et al. Out-of-pocket costs and diabetes prevention services: the Translating Research Into Action for Diabetes (TRIAD) study. Diabetes Care 2003;26:2294 –9. [18] Primera Encuesta Nacional de Factores de Riesgo. Ministerio de Salud de la Nación. Available from: http://www.msal.gov.ar/htm/Site/enfr/ index.asp. [Accessed June 23, 2006]. [19] Haider AW, Larson MG, Franklin SS, Levy D. Systolic blood pressure, diastolic blood pressure, and pulse pressure as predictors of risk for congestive heart failure in the Framingham Heart Study. Ann Intern Med 2003;138:10 – 6. [20] Clarke PM, Gray AM, Briggs A, et al. A model to estimate the lifetime health outcomes of patients with type 2 diabetes: the United Kingdom Prospective Diabetes Study (UKPDS) outcomes model (UKPDS 68). Diabetologia 2004;47:1747–59. VALUE IN HEALTH 14 (2011) S24 –S28 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Chronic Hepatitis B Treatment: The Cost-Effectiveness of Interferon Compared to Lamivudin e Alessandra Maciel Almeida, ScD1,*, Anderson Lourenço da Silva, MSc2, Mariângela Leal Cherchiglia, MD, PhD1, Eli Iola Gurgel Andrade, PhD1, Gustavo Laine Araújo de Oliveira, postgraduate student1, Francisco de Assis Acurcio, MD, ScD2 1 Department of Preventive and Social Medicine, Universidade Federal de Minas Gerais, Brazil; 2Department of Social Pharmacy, Universidade Federal de Minas Gerais, Brazil A B S T R A C T Objective: To perform a cost-effectiveness evaluation from the perspective of the Brazilian National Health System of alternatives strategies (i.e., conventional interferon, pegylated interferon, and lamivudine) for the treatment of patients with chronic hepatitis B who present elevated aminotransferase levels and no evidence of cirrhosis at the beginning of treatment. Methods: A Markov model was developed for chronic hepatitis B (hepatitis B antigen e [HBeAg] positive and negative) with 40 years’ time horizon. Costs and benefits were discounted at 5%. Annual rates of disease progression, costs due to complications, and the efficacy of medicines were obtained from the literature. One-way and probabilistic sensitivity analysis evaluated uncertainties. Results: For HBeAg positive patients, peginterferon (48 weeks) resulted in an increase of 0.21 discounted life-years gained compared to interferon (24 weeks). The incremental cost-effectiveness ratio (ICER) converted to US dollars using the 2009 purchasing power parity conversion factor was US$100,752.24 per life-year gained. For HBeAg negative patients, it was observed that interferon (48 weeks) compared with longterm lamivudine presented an increase of 0.45 discounted life-years gained and ICER of US$15,766.90 per life-year gained. In the sensitivity Introduction Hepatitis B is one of the most common infectious diseases worldwide. An estimated 350 million people worldwide are chronically infected with hepatitis B virus (HBV) [1]. In Brazil, at least 15% of the population has been in contact with HBV and 1% present with chronic disease [2]. Persistently high HBV DNA levels are associated with an increased risk of cirrhosis and hepatocellular carcinoma (HCC) [3,4], which contributes to the increase of treatment costs due to morbidity [5]. Until recently and according to the Clinical Protocols and Therapeutic Guidelines for High Cost Medications of the Brazilian Ministry of Health [6], pharmacological options for the treatment of chronic hepatitis B were restricted to interferon and lamivudine. Currently, three antiviral medications (tenofovir, entecavir, and adefovir) have extended the treatment alternatives for the control of HBV action [7]. analysis, the ICER was more sensitive to variation in the probability of transition from chronic hepatitis B to compensated cirrhosis, discount rate, and medicine prices. Cost-effectiveness acceptability curve for HBeAg positive (pegylated interferon vs. conventional interferon) and negative (conventional interferon vs. lamivudine) showed that conventional interferon was cost-effective until three times the gross domestic product per capita. Conclusions: For patients with chronic hepatitis B with elevated aminotransferase levels in the pretreatment and no cirrhosis who were HBeAg positive, pegylated interferon (48 weeks) provided more life-years gained when compared to conventional interferon (24 weeks), and the ICER surpasses the country’s buying power, which makes conventional interferon the chosen alternative. For HBeAg negative patients, conventional interferon (48 weeks) compared to lamivudine provided more life-years gained at a favorable ICER. Keywords: chronic hepatitis B, cost-effectiveness, interferon, lamivudine, peginterferon. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. In a systematic review [8], it was observed that interferon (IFN) presented the advantages of long-term response in hepatitis B antigen e (HBeAg) positive patients, a short treatment duration and absence of resistance. The main advantages of pegylated interferon (PEG-IFN) were its extended biological effect and the lower number of treatments it required. Both treatment options showed the disadvantages of limited use in patients with a lower alanine aminotransferase (ALT) level at pretreatment or with a decompensated liver, their association with several adverse events and the inconvenience of subcutaneous injection. The first nucleoside analogue to be approved and used for HBV was lamivudine (LAM) [9], which is associated with minimal adverse events, low maintenance response rates, and a need for long-term therapy [10]. Its greatest limitation is the selection of resistant mutants, with patients becoming resistant after a year of treatment [11]. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Alessandra Maciel Almeida, Prof. Antônio Aleixo Street, 760/602 – Bairro de Lourdes, Belo Horizonte, MG, Brazil. CEP 30180-150. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.011 VALUE IN HEALTH 14 (2011) S24 –S28 The objective of this study was to perform a cost-effectiveness evaluation from the perspective of the Brazilian Public Health System (SUS) of alternative strategies (IFN, PEG-IFN, and LAM) for the treatment of patients HBeAg positive and negative (i.e., antibodies to HBeAg), who present high ALT levels and no evidence of cirrhosis at the beginning of treatment. This group of patients was chosen because it is considered to represent the most prevalent and clinically relevant chronic HBV infection. Methods Decision analysis software (DATA, version 1.3.1, Tree Age software, Inc., Williamstown, MA) was used for the cost-effectiveness analysis aimed at evaluating a hypothetical cohort of patients with chronic HBV infection with a histological diagnosis of the disease, positive for serum hepatitis B surface antigen for more than 6 months, with detectable HBV DNA levels and high ALT levels (more than twofold the upper normal limit [UNL]) and no clinical or histological evidence of cirrhosis. Clinical research shows differences in the age profile among HBeAg positive and negative patients [12,13]. As a result, two models were built, considering the average age at the onset of treatment as 32 years for HBeAg positive patients [12,14] and 40 years for HBeAg negative patients [14]. A Markov model was used with 1-year cycles and both models evaluate short-course and longer duration treatments with time-horizon of 40 years, given that most of the patients in the cohort would be dead after this period. Ideally, economic analysis should be established prospectively, and together with the results of clinical research. The disease progression involves decades and it is difficult to realize prospective studies. The model parameters, including efficacy/effectiveness measures were obtained from a specific systematic review [8] and from review of selected studies. The efficacy measures used were: 1) HBeAg positive patients: HBeAg seroconversion; and 2) HBeAg negative patients: response to treatment [low levels of HBV DNA (⬍ 300 – 400 copies/mL) and normalisation of ALT levels] [13]. The long-term results were modelled using the stages in the Markov model considering the annual failure in the durability of HBeAg seroconversion and treatment response. In this study, like previous economic analyses of antiviral treatment for chronic hepatitis B [15], HBeAg seroconversion was used as a treatment-stopping criterion. However a patient could experience a relapse and return to the chronic hepatitis B stage [16]. For HBeAg negative patients, sustained remission was also not considered common [17]. The model did not assume any explicit consideration in regard to LAM resistance. However, some effects of resistance to the medication were captured with the reduction in long-term seroconversion rates [15]. HBeAg positive The model consisted of six disease stages (Fig. 1A in Supplementary Materials found at: doi:10.1016/j.jval.2011.05.011). The model structure was adapted from the model of Crowley et al. [18]. All patients in the model started at the chronic hepatitis B disease stage with no cirrhosis and received treatment alternatives. The model evaluated short-term treatment for HBeAg positive patients: IFN dosed at 9 to 10 MU three times a week (24 weeks), PEG-INF alfa 2a (180 g) once a week (48 weeks) or long-term LAM (100 mg) daily (LAM use after 4 years of treatment in patients who did not undergo HBeAg seroconversion did not bring any added benefit; however, in the Markov model, these patients continued to receive LAM and the cost was calculated in subsequent years). The model doesn’t assume rescue therapy in case of treatment failure associated with emergence of drug-resistant virus. S25 Efficacy measures were obtained for 1 year of treatment considering results from clinical research in patients with ALT levels greater than or equal to twice the UNL, HBeAg seroconversion rates for IFN (24%) [18], LAM (19%) [12], and PEG-IFN (32%) [12]. The seroconversion estimates sustained for LAM for the second, third, and fourth year of treatment were 10%, 6%, and 5%, respectively, based on observational studies [12,19,20]. The rate of seroconversion observed in the fourth year (5%) [20] was used from the fifth year onward (Table 1 in Supplementary Materials found at: doi: 10.1016/j.jval.2011.05.011). In all therapeutic alternatives, after treatment cessation, all patients could experience a relapse. Van Nunen et al. [21] and Wang et al. [22] demonstrated a 35% relapse for LAM in patients with ALT levels greater than or equal to two to five times the UNL 6 months after the treatment; that rate was considered until the fourth year of treatment. For the fifth year, a relapse of 25% was estimated considering the potential long-term impact on the durability of seroconversion [21,22]. Spontaneous seroconversion rates of 9% were considered for patients in the beginning of the treatment with PEG-IFN and IFN [18] (Table 1 in Supplementary Materials found at: doi:10.1016/j.jval.2011.05.011). There was limited published data on the annual loss of response following treatment with PEG-INF, so cconservative relapse rates of 8% were used in the analysis for IFN and PEG-IFN [21], despite PEG-IFN has showed fewer relapses than conventional IFN (Table 1 in Supplementary Materials found at: doi: 10.1016/j.jval.2011.05.011). All the efficacy assumptions for LAM and PEG-INF are similar to what was used by Veenstra et al. [14]; thus, relapse rates for INF and PEG-INF were obtained in the same study. The annual rates of disease progression or effectiveness measures were described in Table 2 in Supplementary Materials found at: doi:10.1016/j.jval.2011.05.011. HBeAg negative All HBeAg negative patients in the model started at chronic hepatitis B disease stage with no cirrhosis. The model consisted of six disease stages (Fig. 1B in Supplementary Materials found at: doi: 10.1016/j.jval.2011.05.011). The treatment alternatives considered were IFN at a dosage of 9 to 10 UM three times a week (48 weeks), PEG-IFN (48 weeks) once a week, or long-term LAM daily (until the patient responds to treatment). To obtain efficacy measures, the estimates for sustained combined response (suppression of HBV DNA and ALT normalization) 6 months after the treatment was stopped were derived from the randomized controlled trial by Marcellin et al. [13] that compared PEG-IFN with LAM (48 weeks). The response rates for PEG-IFN were 36% at the end of treatment and at follow-up and 69% for LAM at the end of treatment. There is some data in the literature on relapse rates after 6 months of treatment with LAM, but two studies pointed out that the combined response rates were around 11% to 20% 1 to 2 years after treatment [23,24]. These authors reported relapse rates of 83% after 6 months and after a year of follow-up [23,24]. Based on these data, a conservative annual rate of relapse of 80% for LAM was considered. For IFN, a complete response rate to the treatment of considered 60% [25]. There are no long-term data on combined response for PEG-IFN and there are few for IFN. One study suggested that 50% of patients treated with IFN experienced relapse between 6 months and 32 months post treatment [26]. Considering that there were no relapses 6 months after treatment with PEG-IFN [13], a relapse of 25% was assumed after 6 months of treatment [27]. Spontaneous relapses of 6% were considered [28,29]. These estimates were obtained from Veenstra et al. [27]. The annual rates of disease progression or effectiveness measures are described in Table 2 in Supplementary Materials found at: doi:10.1016/j.jval.2011.05.011. S26 VALUE IN HEALTH 14 (2011) S24 –S28 The costs Only direct costs were taken into account in this study. All costs were originally calculated in the national currency (Brazilian real [BRL]). These values were converted to US dollars using the 2009 purchasing power parity (PPP) conversion factor according to the International Monetary Fund. It was assumed 2009 PPP conversion factor (US$1 ⫽ 1.56 BRL). Prices of the medication therapies Prices were based on the list of medication prices from the Medication Market Regulating Chamber (Câmara de Regulação do Mercado de Medicamentos) on November 13, 2009. The average factory price was used with a state tax on goods and services of 0% and the price adjustment coefficient of 24.92%, which is a mandatory minimum discount that affects the factory price of some medications purchased by public entities. The average prices were of US$51.42 (US$87.87–US$110.85) for IFN, US$578.69 (US$338.69 – US$730.02) for PEG-IFN and US$1.39 (US$0.43–US$2.16) for LAM. The calculation of the minimum and maximum prices used the lowest and the highest price found in the Medication Market Regulating Chamber table and then applied the price adjustment coefficient. Because there was no price variation for the PEG-IFN, the same variation found for the IFN was used. Annual costs due to chronic hepatitis B complications Annual costs per patient with compensated cirrhosis (CC), decompensated cirrhosis, and HCC were obtained from the study by Castelo et al. [30] that evaluated the chronic hepatitis B costs in 2005 in Brazil with a Delphi panel of specialists. The direct costs included those generated by medical fees, lab exams, diagnostic and therapeutic procedures, hospitalizations, and medications. Data on costs were predominantly obtained from SUS billing tables and medication prices. The annual costs of the evolving chronic hepatitis B stages were updated to 2009 and estimated as follows: chronic hepatitis B (US$870), CC (US$1243), decompensated cirrhosis (US$7763), and HCC (US$1679). Cost-effectiveness analysis The Markov model was used to estimate the clinical benefits in life-years gained (LYG) and the costs of the medication alternatives in the time horizon period. The comparison among the treatment alternatives was measured by the incremental cost-effectiveness ratio (ICER). The cost-effectiveness threshold developed by the World Health Organization [31] is one to three gross domestic product (GDP) per capita for an additional disability adjusted life year prevented. A discount rate of 5% per year was adopted for the costs and results. A unidirectional sensitivity analysis was conducted to determine the impact in the ICER estimate. This analysis was carried out by changing individual inputs: therapy response at the first year of treatment (interval confidence), prices of the medication (lowest, highest), discount rates (0%, 5%, and 10%) or magnitude of treatment effectiveness in a Tornado analysis. A probabilistic sensitivity analysis was conducted and generated a cost-effectiveness acceptability curve using Monte Carlo simulation methods. Triangular distributions were assigned to probability based on the parameter ranges (minimum, maximum) listed in Table 2 in Supplementary Materials found at: doi:10.1016/j.jval.2011.05.011. Results The clinical results and economic estimates for each medication alternative (IFN, PEG-IFN, and LAM) for chronic hepatitis B treatment are presented in Table 3 in Supplementary Materials found at: doi:10.1016/j.jval.2011.05.011. In HBeAg positive patients, it was observed that PEG-IFN (48 weeks) resulted in more LYG compared to IFN (24 weeks), with a difference of 0.21. The ICER was US$100,752.24 per LYG. The LAM strategy was dominated. In the case of HBeAg negative patients, it was observed that IFN (24 weeks) presented an ICER of US$15,766.90 per LYG compared to LAM (lifetime), with a difference of 0.45 LYG. The LAM strategy and PEG-INF were dominated. For HBeAg positive patients, the treatment with LAM resulted in an average of 13.58 LYG compared to 14.25 LYG (IFN) and 14.46 LYG (PEG-IFN). The accumulated incidence of CC across 10 years was 18%, 15%, and 14%, respectively. For HBeAg negative patients, the treatment with IFN results in an average of 13.12 LYG compared to 12.93 LYG (PEG-IFN) and 12.67 LYG (LAM). The incidence accumulated of CC in 10 years was 22.7%, 23.3%, and 26%, respectively. Sensitivity analysis For both groups, when the discounts (0%, 5%, and 10%) were applied to the costs, the ICER estimates decreased. For HBeAg positive patients, when the worst and the best scenarios (minimum or maximum values of the seroconversion rates) were modelled for the first year of treatment, there was no affect on the ICER comparing PEG-INF versus INF. For HBeAg negative patients comparing INF versus LAM, when the best scenario (maximum value of response) was used for the first year of treatment, we observed a reduction in the ICER US$4894.87. Concerning the worst scenario, the ICER was not altered. For HBeAg positive patients comparing PEG-INF versus INF, when medicines minimum prices were applied to the medications, there was a decrease in the ICER US$57,341.40 and the ICER increased (US$128,248.29) when maximum prices were applied. The same phenomenon was observed for the HBeAg negative patients comparing INF versus LAM, with ICERs of US$4895.94 and US$24,507.22, respectively. The Tornado analysis demonstrated that the ICER estimates were more sensitive to the variation in the probability of chronic hepatitis B evolving to CC, seroconversion to CC and response to CC (Figs. 2 and 3 in Supplementary Materials found at: doi:10.1016/ j.jval.2011.05.011). For HBeAg positive patients, the cost-effectiveness acceptability curve generated from the PSA for the discounted incremental cost-effectiveness ratio indicated that INF was cost-effective compared with PEG-INF until three GDP per capita. To willingness to pay up than US$104,487.20, using the 2009 PPP conversion factor, PEG-INF has up than 50% to be cost-effective compared with INF. For HBeAg negative patients, INF was cost-effective compared with LAM at the Brazilian threshold. Discussion In this study, for HBeAg positive patients, PEG-IFN when compared to the IFN showed a little better result, but presented an ICER above the current Brazil cost-effectiveness threshold. The ICER was more sensitive to variation in the progression probability of the chronic hepatitis B to CC and seroconversion for CC. Therapy response did not impact the sensitivity analysis and can be related to the small seroconversion interval observed in the literature. Greater variation in the ICER was noticed when medication price and discount rate was varied. In probabilistic sensitivity analysis, INF was cost-effective compared with PEG-INF until three GDPs per capita. Using a Markov model system and data from clinical research, Sullivan et al. [32] evaluated the ICER of treatment with PEG-IFN compared to that of LAM in HBeAg positive patients from the perspective of Taiwan. Treatment with PEG-IFN was considered more cost-effective considering factors such as disease progression, VALUE IN HEALTH 14 (2011) S24 –S28 LYG, therapy cost, and effectiveness. The ICER was sensitive to the same variations observed in this study and was considered favorable because it was within the buying power parameters of that country. Veenstra et al. [14] observed that even though PEG-IFN was a more expensive alternative than LAM, it provided better results in terms of health and cost-effectiveness within the buying power constraints of the United Kingdom. In our study, for HBeAg negative patients, INF compared to LAM demonstrated more LYG and ICER within the buying power of the country. The PEG-IFN alternative was dominated. In the sensitivity analysis, we observed that varying the interval of the transition probabilities caused less variation in the ICER. Greater variation in the ICER was observed when medications price and discount rate varied. In PSA, INF was cost-effective even for values up to three GDPs per capita. The results of our study show that the progression to CC in HBeAg negative patients was higher than in HBeAg positive patients and that LAM resulted in greater progression in both groups. Lacey et al. [16] observed similar results and concluded that shortterm treatment with IFN, PEG-IFN, or LAM presented limited influence on the disease progression. Comparing PEG-IFN with LAM in HBeAg negative patients, Veenstra et al. [27] demonstrated that PEG-IFN had incremental benefits on life expectancy and quality of life with an acceptable ICER in Taiwan. Kanwal et al. [28] verified that IFN was more costeffective compared to LAM in HBeAg negative patients. The authors emphasized that IFN could reduce costs because it eliminated the need for longer therapy and that it could be effective when it did not present viral resistance. Some limitations to our study can be identified and they pertain to treatment compliance, patient profiles, natural history of the disease, annual costs, time-horizon, no natural mortality rates, and the estimates obtained in the literature. All long-term modeling studies are inherent uncertainties in projecting longterm results. The modelling did not consider the occurrence of problems in compliance with LAM long-term antiviral treatment, which can generate worse results than those observed in the clinical trials. The same can be considered in the treatment with IFN due to adverse events. Low adherence rates to the treatment can reduce the therapy response, thereby risking the treatment’s effectiveness on the disease progression. The profile of the patients included in our study does not allow for the extrapolation of the results to patients with a different profile. However, that choice was considered the most prevalent and clinically relevant form of chronic HBV infection. High levels of ALT in the pretreatment is a predictable factor for the response at the end of the treatment and that IFN may be adverse to patients in advanced stages of the disease [18].. We assumed that patients in different stages of chronic hepatitis B present the same clinical course and progression rates as nontreated patients. This condition is consistent with several published analyses of cost-effectiveness, where the patients without response and nontreated patients progress in a similar manner [28,32,33]. In addition, the estimates in the literature are limited and international studies that are mainly focused on Asian populations are the main sources of findings on the medications used in the treatment of chronic hepatitis B. Conclusions This analysis suggests that for HBeAg positive patients with high levels of ALT in the pretreatment and without cirrhosis or HCC at the beginning of the treatment, PEG-IFN (48 weeks) provided more LYG when compared to IFN (24 weeks), but the ICER surpasses the country’s buying power, which makes IFN the chosen alternative S27 for those patients. For HBeAg negative patients, INF (48 weeks) compared to LAM (long-term) provided more LYG at an acceptable ICER to the country. The sensitivity analyses show that the ICER was more sensitive to variation in the probability of transition from chronic hepatitis B to CC, discount rate, and medicine prices. Our findings suggest that interferon could be considered the chosen alternative in health care systems with limited resources. Acknowledgements The authors thank the members of the Pharmacoepidemiology Research Group, the Health Economy Research Group of the UFMG, and infectious disease specialist Dr. Ricardo Andrade Carmo for valuable contributions. Sources of financial support: Conselho Nacional de Desenvolvimento Científico e Tecnológico – CNPq (process No. 551412/ 2007-0) Edital MCT/CNPq/MS-SCTIE-DECIT/CT-Saúde No. 033/ 2007, and Fundação de Amparo à Pesquisa do Estado de Minas Gerais – FAPEMIG (Process No. 4611-5.01/07). Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi:10.1016/j.jval.2011.05.011 or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). REFERENCES [1] Ayoub WS, Keeffe EB. Review article: current antiviral therapy of chronic hepatitis B. Aliment Pharmacol Ther 2008;28:167–77. [2] Costa AM, L ’italien G, Nita ME, Araujo ES. Cost-effectiveness of entecavir versus lamivudine for the suppression of viral replication in chronic hepatitis B in Brazil. Braz J Infect Dis 2008;12:368 –73. [3] Chen CJ, Yang HI, Su J, et al. Risk of hepatocellular carcinoma across a biological gradient of serum HBV DNA level. JAMA 2006;295:65–73. [4] Iloeje UH, Yang H, Su J, et al. Predicting cirrhosis risk based on the level of circulating hepatitis B viral load. Gastroenterology 2006;130: 678 – 86. [5] Lok AS, McMahon BJ. Practice Guidelines Committee, American Association for the Study of Liver Diseases. Chronic hepatitis B. Hepatology 2001;34:1225– 41. [6] Resolution nº 860, Health Care Assistance Department, November 12, 2002. The Department of Health Care Assistance of Ministry of Health Approves Clinical Protocol and Therapeutic Guidelines of chronic viral hepatitis B. Official Journal of Federative Republic of Brazil, Brasília, Federal District, Union Official Journal, 2002;1:84 – 6. [7] Ministry of Health Resolution nº. 2561/GM, October 28, 2009. Approves the Clinical Protocol and Therapeutic Guidelines - Viral Hepatitis B and chronic co-infections. Union Official Journal. 2009;1:59. [8] Almeida AM, Silva DI, Guerra AA Jr, et al. Efficacy of interferon (conventional, pegylated) and lamivudine for treatment of chronic hepatitis B: A systematic review. Cad Saude Publica 2009;25:1667– 77. [9] Lok AS. Lamivudine therapy for chronic hepatitis B: is longer duration of treatment better? Gastroenterology 2000l;119:263– 6. [10] Liaw YF, Leung NW, Chang TT, et al. Effects of extended lamivudine therapy in Asian patients with chronic hepatitis B. Asia Hepatitis Lamivudine Study Group. Gastroenterology 2000;119:172– 80. [11] Lai CL, Ratziu V, Yuen MF, Poynard T. Viral hepatitis B. Lancet 2003;20; 362:2089 –94. [12] Lau GK, Piratvisuth T, Luo KX, et al. Peginterferon Alfa-2a, lamivudine, and the combination for HBeAg-positive chronic hepatitis B. N Engl J Med 2005;352:2682–95. [13] Marcellin P, Lau GK, Bonino F, et al. Peginterferon alfa-2a alone, lamivudine alone, and the two in combination in patients with HBeAg-negative chronic hepatitis B. N Engl J Med 2004;351:1206 –17. [14] Veenstra DL, Sullivan SD, Dusheiko GM, et al. Cost-effectiveness of peginterferon alpha-2a compared with lamivudine treatment in patients with HBe-antigen-positive chronic hepatitis B in the United Kingdom. Eur J Gastroenterol Hepatol 2007;19:631– 8. S28 VALUE IN HEALTH 14 (2011) S24 –S28 [15] Shepherd J, Jones J, Takeda A, et al. Adefovir dipivoxil and pegylated interferon alfa-2a for the treatment of chronic hepatitis B: a systematic review and economic evaluation. Health Technol Assess 2006;10:iii–iv, xi–xiv, 1–183. [16] Lacey LF, Gane E. The cost-effectiveness of long-term antiviral therapy in the management of HBeAg-positive and HbeAg-negative chronic hepatitis B in Singapore. J Viral Hepat 2007;14:751– 66. [17] Fattovich G. Natural history of hepatitis B. J Hepatol 2003;39(Suppl. 1): S50 – 8. [18] Crowley SJ, Tognarini D, Desmond PV, Lees M. Cost-effectiveness analysis of lamivudine for the treatment of chronic hepatitis B. Pharmacoeconomics 2000;17:409 –27. [19] Leung NW, Lai CL, Chang TT, et al. Extended lamivudine treatment in patients with chronic hepatitis B enhances hepatitis B e antigen seroconversion rates: results after 3 years of therapy. Hepatology 2001; 33:1527–32. [20] Lok AS, McMahon BJ, Practice Guidelines Committee, American Association for the Study of Liver Diseases (AASLD). Chronic hepatitis B: update of recommendations. Hepatology 2004;39:857– 61. [21] van Nunen AB, Hansen BE, Suh DJ, et al. Durability of HBeAg seroconversion following antiviral therapy for chronic hepatitis B: relation to type of therapy and pretreatment serum hepatitis B virus DNA and alanine aminotransferase. Gut 2003;52:420 – 4. [22] Wang CC, Kowdley KV. Built to last? Durability of lamivudine-induced seroconversion in patients with chronic hepatitis B. Gastroenterology 2004;126:1917–9. [23] Santantonio T, Mazzola M, Iacovazzi T, et al. Long-term follow-up of patients with anti-HBe/HBV DNA-positive chronic hepatitis B treated for 12 months with lamivudine. J Hepatol 2000;32:300 – 6. [24] Tassopoulos NC, Volpes R, Pastore G, et al. Post lamivudine treatment follow-up of patients with HBeAg negative chronic hepatitis B. J Hepatol 1999;30(Suppl 1):117. [25] Fattovich G, Farci P, Rugge M, et al. A randomized controlled trial of lymphoblastoid interferon-alfa in patients with chronic hepatitis B lacking HBeAg. Hepatology 1992;15:584 –9. [26] Lin CC, Wu JC, Chang TT, et al. Long-term evaluation of recombinant interferon alpha2b in the treatment of patients with hepatitis B e antigen-negative chronic hepatitis B in Taiwan. J Viral Hepat 2001;8: 438 – 46. [27] Veenstra DL, Sullivan SD, Lai MY, et al. HBeAg-negative chronic hepatitis B: cost-effectiveness of peginterferon alfa-2a compared to lamivudine in Taiwan. Value Health 2008;11:131– 8. [28] Kanwal F, Gralnek IM, Martin P, et al. Treatment alternative for chronic hepatitis B virus infection: a cost-effectiveness analysis. Ann Intern Med 2005;142:821–31. [29] Wong JB, Koff RS, Tinè F, Pauker SG. Cost-effectiveness of interferonalpha 2b treatment for hepatitis B e antigen-positive chronic hepatitis B. Ann Intern Med 1995;122:664 –75. [30] Castelo A, Pessoa MG, Barreto TCBB, et al. Estimativas de custo da hepatite B crônica no sistema único de saúde brasileiro em 2005. Rev Assoc Med Bras 2007;53:486 –91. [31] Cost-effectiveness thresholds. Available from: http://www.who.int/ choice/costs/CER_thresholds/en/index.html. [Acessed June 8, 2010]. [32] Sullivan SD, Veenstra DL, Chen PJ, et al. Cost-effectiveness of peginterferon alpha-2a compared to lamivudine treatment in patients with hepatitis B e-antigen positive chronic hepatitis B in Taiwan. J Gastroenterol Hepatol 2007;22:1494 –9. [33] Fattovich G, Brollo L, Giustina G, et al. Natural history and prognostic factors for chronic hepatitis type B. Gut 1991;32:294 – 8. VALUE IN HEALTH 14 (2011) S29 –S32 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Análise de Custo-Efetividade da Sinvastatina versus Atorvastatina na Prevenção Secundária de Eventos Cardiovasculares no Sistema Único de Saúde Brasileiro Denizar Vianna Araujo, MD, PhD1,*, Camila Pepe Ribeiro de Souza, MSc1, Luciana Ribeiro Bahia, MD, PhD1, Helena Cramer Veiga Rey, MD, MSc2, Braulio dos Santos Junior, MD2, Bernardo Rangel Tura, MD, PhD2, Otavio Berwanger, MD, PhD3, Anna Maria Buehler, PhD3, Marcus Tolentino Silva, MSc4 1 Universidade do Estado do Rio de Janeiro (UERJ), Rio de Janeiro, Brazil; 2Instituto Nacional de Cardiologia do Ministério da Saúde Rio de Janeiro, Brazil; Instituto de Ensino e Pesquisa do Hospital do Coração - IEP-Hcor, São Paulo, Brazil; 4Departamento de Ciência e Tecnologia (DECIT) do Ministério da Saúde Brasília, Brazil 3 A B S T R A C T Objective: The objective of this study is to perform an economic evaluation analyzing the treatment with atorvastatin and simvastatin in comparison to placebo treatment, within the Brazilian Public Healthcare System (SUS) scenario, for patients with high risk of cardiovascular disease; analyzing if the additional cost related to statin treatment is justified by the clinical benefits expected, in terms of cardiovascular event and mortality reduction. Methods: Cardiovascular event risk and mortality risk were used as outcomes. Statin efficacy at LDL-c and cardiovascular events levels lowering data was obtained from a systematic review of literature. A decision analytic model was developed to perform a cost-effectiveness analysis comparing atorvastatin 10mg/day and simvastatin 40mg/day to placebo treatment in patients with dyslipidemia in Brazil. The target population of this study was a hypothetic cohort of men and women with a mean age of 50 years old and high risk of cardiovascular disease. The model includes only direct costs obtained from Ambulatory and Hospital Information System and Price Database of Brazilian Ministry of Health. The comparative cost-effectiveness analysis itself was done through Excel spreadsheets covering a 5 -years time horizon. Results: The result shows that atorvastatin 10mg/day in comparison to placebo has higher cost with higher effectiveness in the time horizon of 5 years (Incremental Cost Effectiveness Ratio of R$ 433.065,05 per life year gained). In this scenario atorvastatin is not cost effective in comparison to placebo. The simvastatin 40mg/day appears to be a strategy with lower cost and higher effectiveness in comparison to placebo, in the time horizon analyzed (5 years). In the multivariate probabilistic sensitivity analysis, simvastatin showed 53% of the results in the quadrant with greater effectiveness and lower cost. Conclusions: This study is an important tool for public decision makers. The study can be used in the decision process of increasing cardiovascular disease treatment access with budgetary sustainability for Ministry of Health. In comparison to placebo, the results show that sinvastatin is a cost saving strategy while atorvastatin is not cost effective. Palabras Claves: cardiovascular disease, cholesterol lowering, cost-effectiveness, secondary prevention, statin therapy. Introdução Métodos A doença cardiovascular de etiologia isquêmica (DCV) é responsável por expressiva carga de morbi-mortalidade na população brasileira [1]. Em estudo multicêntrico brasileiro com 81.262 indivíduos, 40% da amostra avaliada apresentava níveis de colesterol superior a 200 mg/dl e 13 % níveis superiores a 240mg/dl [2]. No Brasil, o gasto do setor público com as estatinas, no ano 2009, foi de cerca de R$ 92 milhões, sendo que, deste valor, 96% representou o gasto somente com a atorvastatina [3]. O objetivo deste estudo foi à elaboração de análise de custo-efetividade da sinvastatina e atorvastatina versus placebo para o cenário do Sistema Único de Saúde (SUS) em pacientes com alto risco de DCV. Revisão sistemática de ensaios clínicos controlados randomizados. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. Critérios de elegibilidade Estudos Ensaios clínicos randomizados controlados de estatinas em pacientes com alto risco de eventos cardiovasculares, sem limites de data de publicação ou idioma. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Título resumido: Cost-effectiveness of statins in Brazil. * Autor de Correspondência: Denizar Vianna Araujo - Avenida Visconde de Albuquerque, n° 1400/501 - Leblon, Rio de Janeiro, Brazil 22450-000; Tel: ⫹55 21 8871-6249; Fax: ⫹55 21 2274-0856. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.024 S30 VALUE IN HEALTH 14 (2011) S29 –S32 Participantes Dados probabilísticos Adultos com doença arterial coronariana, cerebral ou periférica, ou Diabetes Mellitus. O risco de ocorrência do IAM fatal e não fatal, AVC fatal e não fatal, revascularização e mortalidade por qualquer causa nos pacientes sem tratamento com estatina foi obtido do ensaio clínico Heart Protection Study (HPS) [12]. Os riscos relativos e os riscos esperados no grupo em tratamento com sinvastatina e atorvastatina estão apresentados na Tabela 1 o Material Suplementar: doi:10.1016/j.jval.2011.05.024. Como este modelo contemplou horizonte de tempo de cinco anos com ciclos de duração anual, foi necessário transformar estas taxas de cinco anos em taxas anuais. Para obter a taxa média anual de evento utilizou-se a fórmula abaixo [13]: Intervenções Atorvastatina e sinvastatina, comparadas com placebo. Desfechos 1) Eficácia: evento cardiovascular [morte, Infarto Agudo do Miocárdio (IAM), Acidente Vascular Cerebral (AVC) ou revascularização miocárdica cirúrgica ou percutânea], e 2) Segurança: efeitos adversos (mialgia, miopatia, rabdomiólise, hepatopatia ou insuficiência hepática) ou interrupção do tratamento. Exclusão Artigos com violação do sigilo de alocação, tempo de seguimento inferior a 24 meses ou perda superior a 20%. tp1 ⫽ 1 ⫺ 共1 ⫺ tpt兲 1⁄t onde: tp1 ⫽ taxa de evento a ser obtida no ciclo de 1 ano tpt ⫽ taxa total durante o período t t ⫽ período em que está apresentada a taxa total Tipos e unidades de custos Fontes de informação MEDLINE, Embase, LILACS, Cochrane Database of Systematic Reviews, Cochrane Clinical Trials, Database of Abstracts of Reviews of Effectiveness. As buscas em bases de dados foram atualizadas até dezembro de 2008. Estratégia de busca da literatura (Vocabulário controlado OR Termo livre de cada medicamento) AND Filtro para ensaios clínicos Os filtros para ensaios clínicos utilizados foram: 1) MEDLINE: filtro da Cochrane Collaboration [4]; 2) LILACS: filtro da Cochrane Collaboration adaptado por Castro et al [5,6]; and 3) Embase: filtro sensível do Hedges Team [7]. Seleção dos estudos Dois investigadores independentes selecionaram os artigos e as discordâncias foram resolvidas por um terceiro investigador. Qualidade dos estudos Os aspectos metodológicos analisados foram: sigilo da lista de alocação, procedimento de geração da randomização, mascaramento da alocação, perda de seguimento e análise por intenção de tratamento. O escore de Jadad [8] foi calculado para cada ensaio. Análise dos dados Foi utilizado risco relativo como medida de efeito. Foi realizada metanálise de efeitos aleatórios pelo método de DerSimonian e Laird [9], com ponderação pelo inverso da variância. A heterogeneidade entre os estudos foi analisada pelos testes Q de Cochran e I [10]. Na presença de heterogeneidade significativa foi realizada meta-regressão; quando a origem da heterogeneidade foi identificada, realizou-se estratificação da metanálise. O risco de viés foi avaliado informalmente pela inspeção do gráfico de funil e formalmente pelo teste de Egger [11]. Os programas utilizados para a análise foram o Stata versão 11 (StataCorp, 2009) e o R versão 2.10.1 (The R Foundation for Statistical Computing, 2009). Delineamento do modelo de custo-efetividade Perspectiva e horizonte de tempo do estudo A perspectiva adotada foi a do SUS. O horizonte de tempo analisado no caso base foi de cinco anos. Este foi o maior intervalo temporal encontrado em ensaios clínicos controlados randomizados com estatinas. O custo de aquisição das estatinas foi disponibilizado pelo Departamento de Assistência Farmacêutica (DAF) do Ministério da Saúde, com base no preço médio de cada um dos medicamentos nas últimas compras praticadas pelos hospitais da rede pública, no ano 2009, ponderado pela quantidade de comprimidos adquiridos. Sinvastatina, R$ 0,10 por comprimido de 40 mg/dia, custo anual de R$ 36,50. Atorvastatina, R$ 2,32 por comprimido de 10 mg/dia, custo anual de R$ 846,80. Os custos unitários dos exames laboratoriais e de imagem, consultas médicas e fisioterapia foram obtidos do software Sigtap Desktop (Sistema de Gerenciamento da Tabela de Procedimentos, Medicamentos e OPM do SUS), com valores baseados na competência de junho de 2009. O custo do episódio de IAM e o custo anual de acompanhamento ambulatorial foram obtidos do estudo de Ribeiro et al. [14]para o ano de 2002.e ajustados pela inflação de saúde [15], sendo R$ 2.111,71 para o tratamento hospitalar e R$ 2.523,88 para o acompanhamento ambulatorial anual. O custo do episódio de AVC, no cenário brasileiro, foi obtido do estudo de Cristensen et al. [16] para o ano de 2007. O estudo publicou o custo em dólar, convertendo os valores em Reais para esta moeda através do Purchasing Power Parity (PPP) do ano de 2005 [17]. Assim, convertemos o valor publicado em Dólar para Real usando esta mesma taxa de conversão e ajustamos o valor pela taxa de inflação de saúde [15]. O custo do episódio agudo considerado foi de R$ 2.994,94. O custo da cirurgia de revascularização cirúrgica do miocárdio considerado foi obtido do estudo de Girardi et al. [18] para o ano de 2006 e ajustados pela inflação de saúde [15]. O custo de cirurgia de revascularização do miocárdio considerado no modelo foi de R$ 7.201,59. O reembolso do procedimento de angioplastia coronariana com implante de stent no SUS foi de R$ 4.297,11. Foram realizados 62.488 procedimentos de revascularização do miocárdio (RVC) em 2007 no SUS (último ano com dados compilado disponíveis no DATASUS). Do total, angioplastia coronariana representou 66% dos procedimentos (41.144) e cirurgia de revascularização do miocárdio 34% (21.344). O valor final ponderado do desfecho RVC foi de R$ 5.284,63. A conduta de tratamento para o acompanhamento dos pacientes após evento foi obtida do primeiro consenso brasileiro do tratamento da fase aguda do AVC [19]. Os itens de custo foram valorados pelo SIGTAP conforme descrito no item anterior. O custo do 1° ano foi de R$ 1.901,84. O custo de acompanhamento nos anos seguintes, correspondeu a uma consulta médica a cada 2 meses, totalizando R$ 60,00 ao ano. VALUE IN HEALTH 14 (2011) S29 –S32 S31 Desenho do modelo Sinvastatina versus placebo Para o desenvolvimento do modelo foram elaboradas duas comparações: Os resultados estão apresentados na Figura 2 o Material Suplementar: doi:10.1016/j.jval.2011.05.024. Cem por cento dos resultados ficaram posicionados no Quadrante 3. Conclui-se pela análise do gráfico acima, que assumindo um Willingness to Pay (WTP) de sessenta mil reais, temos os resultados de todas as iterações simuladas dentro do limite estabelecido (RCEI ⬍ R$60.000). Lembramos que nesta análise uma efetividade incremental ⬍ 0 representa uma redução no número de eventos. Assim, temos no Quadrante 3 um cenário onde o tratamento com sinvastatina apresenta um perfil melhor que o grupo em tratamento com placebo (redução de eventos) e com relação aos custos, o tratamento com sinvastatina tem um custo inferior ao tratamento com placebo. Œ Œ Tratar com sinvastatina (40mg/dia) versus placebo; Tratar com atorvastatina (10mg/dia) versus placebo. Os ciclos do modelo de Markov ocorreram em intervalos de 1 ano. Os pacientes entraram no modelo em um único estado: como paciente de alto risco cardiovascular. Deste estado, com o passar de um ciclo, os pacientes poderiam permanecer, morrer por outras causas que não um evento cardiovascular (estado ‘Morte por outras causas’) ou transitar para quatro outros estados de acordo com as probabilidades de transição definidas. Estes estados foram: Œ Œ Œ Œ Œ Pós IAM não fatal Pós AVC não fatal Pós Revascularização do miocárdio IAM fatal AVC fatal A Figura 1 o Material Suplementar: doi:10.1016/j.jval.2011.05.024 sumariza a representação esquemática dos estados do modelo de Markov. Foi utilizada taxa de desconto de 5% ao ano. Resultados Sinvastatina versus placebo Os resultados de efetividade e custo da sinvastatina versus placebo estão apresentados na Tabela 2 o Material Suplementar: doi: 10.1016/j.jval.2011.05.024. Observa-se que o custo total de tratamento com sinvastatina é inferior ao custo do tratamento com placebo no horizonte de tempo analisado. Em relação à eficácia, a sinvastatina apresentou um perfil melhor que o grupo em tratamento com placebo (cenário dominante). Atorvastatina versus placebo Os resultados de efetividade e custo da atorvastatina versus placebo estão apresentados na Tabela 3 o Material Suplementar: doi: 10.1016/j.jval.2011.05.024. Nota-se que o custo total de tratamento com atorvastatina é superior ao custo do tratamento com placebo no horizonte de tempo analisado. Em relação à eficácia, a atorvastatina apresentou um perfil melhor que o grupo em tratamento com placebo, exceto em relação ao IAM fatal, pois em nenhum estudo clínico que comparou atorvastatina com placebo foi apresentado este desfecho isolado, apenas combinado ao número de eventos não fatais. Os resultados da razão de custo-efetividade estão apresentados na Tabela 4 o Material Suplementar: doi:10.1016/j. jval.2011.05.024. Análise de sensibilidade probabilística multivariada A análise de sensibilidade probabilística considerou variações de diversos parâmetros por vez, e foi realizada através da atribuição de uma distribuição apropriada para cada um dos parâmetros analisados, apresentados na Tabela 5 o Material Suplementar: doi: 10.1016/j.jval.2011.05.024. A análise de sensibilidade probabilística foi calculada com 1.000 simulações. Os resultados foram avaliados e classificados em: Quadrante 1 (efetividade incremental ⬎ 0 e custo incremental ⬎ 0); Quadrante 2 (efetividade incremental ⬍ 0 e custo incremental ⬎0); Quadrante 3 (efetividade incremental ⬍0 e custo incremental ⬍ 0) e Quadrante 4 (efetividade incremental ⬎ 0 e custo incremental ⬍ 0). Atorvastatina versus placebo Os resultados estão apresentados na Figura 3 o Material Suplementar: doi:10.1016/j.jval.2011.05.024, com 100% dos resultados situados no Quadrante 2. Conclui-se pela análise do gráfico acima, que assumindo um Willingness to Pay (WTP) de sessenta mil reais, temos os resultados de todas as iterações simuladas dentro do limite estabelecido (RCEI ⬍ R$60.000). Lembramos que nesta análise uma efetividade incremental ⬍ 0 representa uma redução no número de eventos. Assim, temos no Quadrante 2 um cenário onde o tratamento com atorvastatina apresenta um perfil melhor que o grupo em tratamento com placebo (redução de eventos) e com relação aos custos, o tratamento com atorvastatina tem um custo superior ao tratamento com placebo. A Figura 4 o Material Suplementar: doi:10.1016/j.jval.2011.05.024 apresenta o resultado da análise de sensibilidade univariada do custo unitário do comprimido de 10mg de atorvastatina. Observa-se que para que o custo total do tratamento com atorvastatina seja equivalente ao custo total do tratamento com placebo, o custo unitário de um comprimido de atorvastatina é de R$0,13; inferior ao valor real deste medicamento. Conclui-se que se o comprimido de 10mg de atorvastatina for adquirido por um custo inferior a R$0,13, temos um resultado onde o tratamento com a atorvastatina é dominante em comparação ao tratamento com placebo. Como pode ser observado na Figura 5 o Material Suplementar: doi:10.1016/j.jval.2011.05.024, o parâmetro que mais influenciou o resultado de custo-efetividade foi o custo de aquisição da atorvastatina. Conclusão O tratamento da dislipidemia com atorvastatina 10mg/dia é mais efetivo, mas requer custo incremental em comparação ao placebo. A sinvastatina 40mg/dia mostrou ser uma estratégia de menor custo e maior efetividade, em comparação ao placebo, em todos os desfechos analisados. A análise de sensibilidade probabilística mostrou que, assumindo um Willingness to Pay (WTP) de sessenta mil reais por desfecho evitado, obteve-se todos os resultados das iterações simuladas superiores ao limite estabelecido (RCEI ⬍ R$60.000 por desfecho evitado) para a análise que compara atorvastatina ao placebo e todos os 1.000 resultados abaixo deste limite para a análise que compara sinvastatina ao placebo. Fontes de financiamento: CNPq - Conselho Nacional de Desenvolvimento Científico e Tecnológico (The National Council for Scientific and Technological Development). Material Suplementar O material suplementar que acompanha este artigo pode ser encontrado na versão na linha como um hiperlink no doi: 10.1016/j.jval.2011.05.024, ou se cópia impressa do artigo, em S32 www.valueinhealthjournal.com/issues número e artigo). VALUE IN HEALTH 14 (2011) S29 –S32 (Selecione: volume, REFERÊNCIAS [1] Araújo DV, Ferraz MB. Impacto Econômico do Tratamento da Cardiopatia Isquêmica Crônica no Brasil. O Desafio da Incorporação de Novas Tecnologias Cardiovasculares. Arq Bras Cardiol 2005;85:1–2. [2] Martinez TL, Santos RD, Armaganijan D et al. National alert campaign about increased cholesterol. determination of cholesterol levels in 81,262 Brazilians. Arq Bras Cardiol 2003;80:635–38. [3] Boletim Brasileiro de Avaliação de Tecnologias em Saúde, Ano IV, n° 9, setembro de 2009. Disponível em: http://200.214.130.94/rebrats/ Publicacoes.html. [Acesso em 17 de Janeiro de 2010]. [4] The Cochrane Collaboration. Cochrane Handbook for Systematic Reviews of Interventions v. 4.2.6. Disponível em: http://www.cochrane .org/sites/default/files/uploads/Handbook4.2.6Sep2006.pdf. [Acesso em 12 de Agosto de 2009]. [5] Castro AA, Clark OA, Atallah AN. Optimal Search Strategy for Clinical Trials in the Latin American and Caribbean Health Science Literature Database (LILACS). Sao Paulo Med J/ Rev Paul Med 1997;115:1423– 6. [6] Castro AA, Clark OA, Atallah AN. Optimal search strategy for clinical trials in the Latin American and Caribbean Health Science Literature Database (LILACS database): Update [Letter]. Sao Paulo Med J/Rev Paul Med 1999;117:138 –9. [7] Wong SS-L, Wilczynski NL, Haynes RB. Developing optimal search strategies for detecting clinically sound treatment studies in EMBASE. J Med Libr Assoc 2006;94:41–7. [8] Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of randomized clinical trials: Is blinding necessary? Control Clin Trials 1996;17:1–12. [9] DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177– 88. [10] National Cholesterol Education Program. Final report of the expert panel Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). NIH Pub. No. 02-5215. Bethesda, MD: National Heart, Lung, and Blood Institute 2002;284. [11] Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. Br Med J 1997;315:629 –34. [12] Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20 536 high-risk individuals: a randomized placebo-controlled trial. Lancet 2002;360:7–22. [13] Briggs A, Sculpher M. An introduction to markov modelling for economic evaluation. Pharmacoeconomics 1998;13:397– 409. [14] Ribeiro RA, Mello RG, Melchior R, et al. Custo Anual do Manejo da Cardiopatia Isquêmica Crônica no Brasil. Perspectiva Pública e Privada. Arq Bras Cardiol 2005;85:3– 8. [15] Fundação Instituto de Pesquisas Econômicas: FIPE. Disponível em: www.fipe.org.br. [Acesso em 2009 January 23]. [16] Christensen MC, Valiente R, Silva GS, et al. Acute Treatment Costs of Stroke in Brazil. Neuroepidemiology 2009;32:142–9. [17] World Bank. Purchasing Power Parity: PPP. Disponível em: http:// siteresources.worldbank.org/ICPINT/Resources/summary-tables.pdf. [Acesso em 06 de Abril de 2009]. [18] Girardi PBMA, Hueb W, Nogueira CRSR, et al. Custos comparativos entre a revascularização Miocárdica com e sem Circulação Extracorpórea. Arq Bras Cardiol 2008;91:369 –76. [19] Sociedade Brasileira de Doenças Cerebrovasculares. Primeiro Consenso Brasileiro do tratamento da fase aguda do Acidente Vascular Cerebral. Arq Neuropsiquiatr 2001;59:972– 80. VALUE IN HEALTH 14 (2011) S33–S38 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Costo-Efectividad del Cardiodesfibrilador Implantable en Pacientes con Factores de Riesgo de Muerte Súbita en Argentina Andrea Alcaraz, MD, Msc1,*, Jorge González-Zuelgaray, MD, PhD2, Federico Augustovski, MD, PhD1,3,4 1 Instituto de Efectividad Clínica y Sanitaria (IECS), Buenos Aires, Argentina; 2Sanatorio de la Trinidad, Buenos Aires, Argentina; 3Escuela de Salud Pública, Universidad de Buenos Aires, Buenos Aires, Argentina; 4Servicio de Medicina Familiar del Hospital Italiano de Buenos Aires, Buenos Aires, Argentina A B S T R A C T Objective: To evaluate the cost-effectiveness and cost-utility of the cardioverter– defibrillator (ICD) among patients who are at risk for sudden death in Argentina, from three scenarios: public health, social security and private sector. Methods: We developed a Markov model to evaluate the survival, quality of life and cost of the prophylactic implantation of an ICD, as compared with pharmacological therapy, among three different target populations according to clinical trials selected using a systematic review, and choosing epidemiological, effectiveness, resource use and cost parameters. A healthcare system perspective was adopted. A 3% discount rate was used. Results: The use of the ICD was more costly and more effective than control therapy. The cohort with greater benefits was represented by MADIT I study, showing an incremental cost effectiveness rate (ICER) of $8,539 (dollar Introducción La Muerte Súbita (MS) representa aproximadamente un 20% del total de las muertes por causas naturales y un 50% de las muertes por causas cardiovasculares. En Argentina, cada año existen 60.000 episodios en pacientes previamente asintomáticos [1,2]. El 80% de las MS se deben a arritmias ventriculares rápidas: la taquicardia ventricular (TV) y la fibrilación ventricular (FV); siendo el tratamiento habitual la supresión de las mismas mediante la administración de drogas antiarrítmicas (DA), o su finalización una vez aparecidas mediante la colocación de un cardiodesfibrilador implantable (CDI) [2]. Existen numerosos estudios realizados en grupos de pacientes con diferente riesgo de MS en los cuales se ha demostrado la efectividad de la colocación de un CDI [3]. Debido a que los costos de implementar esta tecnología son elevados, es importante identificar si los CDIs son costo-efectivos, o si lo son en alguno de los subgrupos en los que esta tecnología se utiliza. Evaluaciones económicas realizadas en países desarrollados arrojan resultados muy variables según el subgrupo de pacientes analizados, con rangos de costo-efectividad incremental que varían entre U$S 8.000 y 145.000 por año de vida salvado y 2009) for public, $9,371 for social security and $10,083 for private sector. ICERs for secondary prevention population were $21,016, $22,520 and $24,012, and for MADIT II population were $17,379, $18,574 and $19,799. The analysis was robust to different deterministic and probabilistic sensitivity analyses, except for the cost of ICD and for battery life. Conclusions: The results varied considerably depending on the cohort and discretely according to the health system. ICD could be cost-effective in Argentina, mainly in the MADIT I patients. Palabras Claves: cardioverter-defibrillator, cost-effectiveness, cost-utility, health care utilization, sudden death. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. entre U$S 32.000 y 163.000 por años ajustados a calidad de vida (de la sigla en inglés QALY, Quality Adjusted Life Years) [4-8]. Dichas evaluaciones económicas son difícilmente extrapolables al contexto argentino debido a que los efectos, los patrones de práctica y los costos son diferentes [9]. El objetivo de este estudio es estimar los costos y beneficios de la utilización del CDI en pacientes con factores de riesgo para muerte súbita, comparado con el cuidado estándar, desde la perspectiva de los principales financiadores de la salud en Argentina: sistema público de salud, sistema de seguridad social y sistema privado; determinando el costo por año de vida ganado, el costo por QALY ganado, e identificando las poblaciones en las que el CDI resulta más costo-efectivo. Métodos Diseño del estudio Se trata de una evaluación económica de costo-efectividad y costoutilidad, que utiliza un modelo de decisión analítico para estimar la sobrevida, la calidad de vida y los costos de dos estrategias (CDI vs. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Autor de correspondencia: Andrea Alcaraz, Instituto de Efectividad Clinica y Sanitaria – IECS. Dr. Emilio Ravignani 2024. Ciudad de Buenos Aires (C1414CPV). Argentina Tel/Fax: (⫹54) 11 4777 8767 E-mail: [email protected] 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.030 S34 VALUE IN HEALTH 14 (2011) S33–S38 Se utilizó una tasa de descuento del 3% anual tanto para los beneficios como para los costos futuros, dado que es una tasa promedio generalmente aceptada para este tipo de modelo [11]. También se reporta un escenario no descontado. Efectividad y utilidad del CDI Se realizó una búsqueda sistemática en las bases de datos PubMED, EMBASE, LILACS y colaboración Cochrane, y actas de conferencias nacionales e internacionales, hasta febrero de 2006. Se seleccionaron revisiones sistemáticas y estudios clínicos controlados aleatorizados contra tratamiento habitual. No fueron estudios ciegos debido a las características del dispositivo. El comparador fue la amiodarona ya que de los tratamientos utilizados hasta el presente es la que demostró mayor eficacia [10]. De los 16 estudios que arrojó la estrategia de búsqueda, se seleccionaron tres como fuentes de los parámetros para el presente modelo, representados por las siguientes cohortes hipotéticas de pacientes con alto riesgo de muerte súbita: Fig. 1 – Estados posibles de transición en el modelo Markov. Las flechas indican las transiciones permitidas. cuidado usual) en tres poblaciones de pacientes con diferente riesgo, similares a las de diferentes ensayos clínicos aleatorizados seleccionados luego de una revisión sistemática [3,12,13]. El cuidado usual fue considerado como aquel que incluye la administración de betabloqueantes y amiodarona por vía oral [10]. El horizonte temporal fue la expectativa de vida de los pacientes. Se realizó un modelo de Markov por considerarse adecuado para enfermedades crónicas. Se utilizó el paquete Tree Age Pro 2009® y la planilla de cálculo Microsoft Office Excel 2007®. Durante cada ciclo anual los pacientes se encontraban en riesgo de muerte por arritmia, por causas cardíacas y por causas no cardíacas. Además en la cohorte con CDI se contempló la mortalidad relacionada con el dispositivo y las complicaciones del mismo. Los individuos entran al modelo a la edad promedio del estudio correspondiente. Son seguidos mediante ciclos anuales que siguen a las cohortes con y sin CDI a lo largo de sus vidas. Los estados considerados y las transiciones entre los mismos pueden observarse en la Figura 1. El caso base fue considerado desde la perspectiva del financiador de salud para el sistema público, considerando los gastos médicos directos relacionados con la atención de los pacientes con CDI o con tratamiento antiarrítmico. Se incluyeron los costos relacionados con la colocación del CDI y recambio del mismo, los de las eventuales complicaciones de la colocación del CDI, el costo del estudio electrofisiológico en el caso de realizarse y el costo de la terapia médica y el seguimiento ambulatorio para ambos grupos de pacientes. Debido a la heterogeneidad de costos en el sistema de salud argentino, también se evaluaron escenarios contemplando la perspectiva del sistema de seguridad social y los seguros privados de salud. 1. Pacientes adultos resucitados por taquicardia ventricular (TV) o fibrilación ventricular (FV), o pacientes con antecedentes de TV sincopal, o TV con fracción de eyección del ventrículo izquierdo (Fey) menor a 35% con compromiso hemodinámico, correspondiente a los pacientes en prevención secundaria de muerte súbita [3]. 2. Pacientes adultos con enfermedad coronaria previa no revascularizable y con Fey menor a 35%, que hubieran presentado episodios espontáneos de TV no sostenida y TV sostenida inducible en el estudio electrofisiológico, correspondiente a pacientes en prevención primaria de muerte súbita, representados en el estudio MADIT I [12]. 3. Pacientes adultos con antecedentes de infarto agudo de miocardio (IAM) y Fey menor a 30%, correspondientes a pacientes en prevención primaria representados en el estudio MADIT II [13]. Para calcular las probabilidades de transición anuales de muerte arrítmica y cardíaca utilizadas en el modelo en base a los estudios relevados, se siguió el método DEALE (Declining Exponential Approximation of Life Expectancy) que asume un declinamiento exponencial de la expectativa de vida mediante la fórmula probabilidad⫽1– e–tasa * tiempo. El beneficio asociado al uso del CDI se modeló a partir de los riesgos relativos evidenciados en los estudios de prevención secundaria, MADIT I y MADIT II. Se asumió que el beneficio del CDI continúa durante todo el tiempo en que se encuentre colocado. La tasa de mortalidad no cardíaca anual fue obtenida de las tablas de mortalidad del Ministerio de Salud de la Nación Argentina (Tabla 1) [14]. En cuanto a la mortalidad relacionada con el implante del CDI no se hallaron datos locales. Luego de revisar la literatura internacional en busca de datos observacionales y en consenso con expertos, se recurrió finalmente a datos de un registro de la población cubierta por Medicare de Estados Unidos, que incluyó 30.984 pacientes con colocación de CDI, reportando una mortalidad del 0,9% anual [15]. No se encontraron datos de utilidades provenientes de Argentina ni de Latinoamérica. Se decidió incorporar al modelo los datos usu- Tabla 1 – Probabilidades anuales de transición por mortalidad incluidas en el modelo para las distintas poblaciones. Probabilidad anual Mortalidad arrítmica Mortalidad cardíaca Mortalidad total* PREVENCION SECUNDARIA [3] MADIT I [12] MADIT II [13] Sin CDI Con CDI Sin CDI Con CDI Sin CDI Con CDI 5,87% 5,21% 12,30% 2,51% 5,07% 8,80% 5,72% 10,28% 17,16% 1,40% 4,45% 7,01% 6,29% 4,76% 11,88% 2,12% 4,86% 8,52% *Los datos de mortalidad no cardíaca varían en función del ciclo y fueron adaptados de las tablas de mortalidad por causa y edad del Ministerio de Salud de la Nación Argentina, correspondientes al año 2008 [14]. S35 VALUE IN HEALTH 14 (2011) S33–S38 Tabla 2 – Costos utilizados en el modelo. Público (AR$) Costo de la colocación del CDI Costo del recambio de generador Costo anual del seguimiento sin CDI Costo anual del seguimiento con CDI prevención secundaria Costo anual del seguimiento con CDI prevención primaria Costo promedio de las complicaciones S. Social (AR$) Privado (AR$) 62.466 65.838 64.791 68.350 69.287 70.476 817 2.194 2.385 787 2.114 2.369 767 2.046 2.301 2.385 5.112 7.062 AR$: pesos argentinos 2009 (tasa de cambio: AR$3,725 ⫽$1). S.: Seguridad. ales de calidad de vida en pacientes con enfermedades cardíacas relevados en el registro CEA, dado que sus datos son ampliamente utilizados en estudios de costo-utilidad en el mundo [6-8,16]. Se consideró una utilidad de 0,88 para los estados “sin CDI” y “CDI sin complicaciones” y de 0,85 para el estado “CDI con complicaciones”, dato que surge de ponderar una utilidad de 0,50 para el mes en que se produjo la complicación y 0,88 para el resto del ciclo. El modelo no permite pasar más de un ciclo en el estado “CDI con complicaciones”, luego se pasa a los estados “mortalidad por CDI” o “CDI sin complicaciones”. realizó una búsqueda sistemática de fuentes publicadas e inéditas de Argentina, asesorada por un economista de la salud. Dicha información se complementó con datos provenientes del Nomenclador Nacional. Para el resto de los costos fue necesario realizar consultas a los representantes de los sistemas de seguridad social y privados, así como a los proveedores de los dispositivos. Para los datos de utilización de recursos se realizó una encuesta a 12 médicos expertos en electrofisiología de todo el país. Los valores incluidos en el modelo se observan en la Tabla 2 [17-19]. Los resultados se reportan en pesos argentinos 2009 (AR$). La tasa de cambio promedio de 2009 fue de 3,725 pesos argentinos por dólar ($). Análisis de sensibilidad Se realizaron diversos análisis de sensibilidad para valorar las asunciones del modelo y la incertidumbre en los resultados. Para las variables de mortalidad se realizó un análisis univariado utilizando un rango de incertidumbre que representa los intervalos de confianza al 95% observados en los estudios que sirvieron como fuente de datos para el modelo. Para la utilidad, se realizó un análisis de sensibilidad probabilístico de tipo Monte Carlo de segundo orden utilizando como media los valores del modelo (0,88 y 0,85) y como desvío estándar 0,05. Dado que los costos del CDI pueden variar ampliamente en función de variables de mercado y del mejoramiento de la tecnología, se variaron los mismos en un amplio rango entre $20.000 y $100.000. Con la idea de evaluar el impacto global del resto de los costos se realizó un análisis de sensibilidad probabilístico, utilizando una distribución de tipo gamma por considerarse apropiada para costos [20]. Costos y utilización de recursos Resultados Se estimaron los costos médicos directos de cada estrategia desde la perspectiva del financiador de salud utilizando datos locales. Se Este modelo resultó tener una adecuada calibración, ya que predijo las tasas de mortalidad asociadas con los grupos “CDI” y Tabla 3 – Resultados en Efectividad, Utilidad y Costos del implante profiláctico del CDI comparado con tratamiento antiarrítmico. Tasa de descuento del 3% anual. Cohorte/Sistema de Salud PREVENCION SECUNDARIA Público S. Social Privado MADIT I Público S. Social Privado MADIT II Público S. Social Privado Estrategia Costo AR$ Sin CDI Con CDI Sin CDI Con CDI Sin CDI Con CDI 5.038 122.726 13.532 139.362 14.709 148.875 Sin CDI Con CDI Sin CDI Con CDI Sin CDI Con CDI 3.697 138.498 9.931 157.939 10.795 170.046 Sin CDI Con CDI Sin CDI Con CDI Sin CDI Con CDI 5.123 127.957 13.762 145.223 14.959 155.089 Costo Expectativa de EV increm. ICER QALY QALY ICUR increm. vida años años AR$/año increm. AR$/Qaly AR$ años 117.688 125.830 134.166 134.800 148.007 159.251 122.843 131.461 140.130 6,17 7,67 6,17 7,67 6,17 7,67 4,53 8,76 4,53 8,76 4,53 8,76 6,27 8,17 6,27 8,17 6,27 8,17 1,5 78.284 1,5 83.887 1,5 89.444 4,24 31.806 4,24 34.907 4,24 37.559 1,9 64.735 1,9 69.190 1,9 73.753 5,43 6,73 5,43 6,73 5,43 6,73 3,98 7,69 3,98 7,69 3,98 7,69 5,52 7,17 5,52 7,17 5,52 7,17 1,3 90.375 1,3 96.627 1,3 103.028 3,71 36.377 3,71 39.941 3,71 42.975 1,65 74.537 1,65 79.772 1,65 85.032 CDI: cardiodesfibrilador implantable. S.:Seguridad. ICER: Tasa de costo-efectividad incremental. ICUR: Tasa de costo-utilidad incremental. QALY: años ajustados por calidad de vida. EV: expectativa de vida. Increm: incremental. AR$: pesos argentinos 2009 (Tasa de cambio: AR$3,725 ⫽$1). S36 VALUE IN HEALTH 14 (2011) S33–S38 Tabla 4 – Análisis de sensibilidad. Variación de la tasa de costo-efectividad incremental en función de la mortalidad y los costos. PREVENCIÓN SECUNDARIA Mortalidad arrítmica (2,28-3,12%) Mortalidad cardíaca (4,61-6,31%) Mortalidad por CDI (0-3%) Costo de colocación y recambio de CDI (AR$20.000-100.000) Costo de seguimiento y complicaciones (50-200%) MADIT I Mortalidad arrítmica (0,89-2,81%) Mortalidad cardíaca (2,83-8,93%) Mortalidad por CDI (0-3%) Costo de colocación y recambio de CDI (AR$20.000-100.000) Costo de seguimiento y complicaciones (50-200%) MADIT II Mortalidad arrítmica (1,51-2,75%) Mortalidad cardíaca (3,46-6,30) Mortalidad por CDI (0-3%) Costo de colocación y recambio de CDI (AR$20.000-100.000) Costo de seguimiento y complicaciones (50-200%) SISTEMA PÚBLICO SEGURIDAD SOCIAL SISTEMA PRIVADO ICER (AR$/año) ICER (AR$/año) ICER (AR$/año) Caso base (78.284) Caso base (83.887) Caso base (78.284) Inferior Superior Inferior Superior Inferior Superior 73.093 68.550 73.088 25.739 96.456 126.835 94.092 121.360 78.240 73.462 78.126 28.276 102.809 134.755 100.644 123.896 83.413 78.310 83.246 29.703 109.656 143.776 107.482 125.324 78.167 78.633 83.369 83.624 84.207 89.857 Caso base (31.806) Caso base (34.907) Caso base (37.559) Inferior Superior Inferior Superior Inferior Superior 30.217 27.284 31.075 10.839 37.267 59.570 33.683 48.821 33.240 30.134 34.106 13.891 40.704 64.311 37.014 51.873 35.743 32.360 36.672 13.891 43.870 69.555 39.892 51.873 31.355 32.352 33.252 36.179 36.191 39.284 Caso base (64.735) Caso base (69.190) Caso base (73.753) Inferior Superior Inferior Superior Inferior Superior 60.316 50.641 61.130 23.314 86.366 120.367 75.224 109.604 59.934 51.111 65.411 23.482 82.725 110.519 80.541 102.366 63.865 54.439 69.678 24.705 88.210 117.898 85.986 103.589 64.336 64.815 68.524 69.510 72.770 74.335 Para cada variable analizada se indica entre paréntesis el rango de variación utilizado para el análisis de sensibilidad. ICER: tasa de costo efectividad incremental. AR$: pesos argentinos 2009. $: dólares (tasa de cambio: AR$3,725 ⫽$1). “sin CDI” con mínimas diferencias con respecto a los valores de los estudios individuales (menores de 0,3%). Análisis del caso base La efectividad, la utilidad y los costos variaron ampliamente según la cohorte considerada, y medianamente según el subsistema de salud, tal como puede observarse en la Tabla 3. En las tres poblaciones y para todos los subsistemas de salud los costos de la estrategia CDI resultaron marcadamente más altos que los de la estrategia de la terapia antiarrítmica. La estrategia CDI se asoció también a mayores beneficios en salud. Para todas las cohortes los costos más bajos se observaron en el sistema público y los más elevados en el sistema privado (Tabla 3). La población más beneficiada por el CDI fue la correspondiente a la cohorte representada por el estudio MADIT I, con un aumento de la expectativa de vida (EV) de 4,24 años y un aumento de los QALYs de 3,71, correspondiéndose con una tasa de costo-efectividad incremental (ICER) de AR$31.806 para el sistema Público, AR$34.907 para el sistema de seguridad social y AR$37.559 para el sistema privado y con una tasa de costo-utilidad incremental (ICUR) de AR$36.377, AR$39.941 y AR$42.975 respectivamente (Tabla 3). La población correspondiente a prevención secundaria mostró los beneficios menores en la EV, con un aumento de la misma en 1,5 años y 1,3 QALYs, observándose valores más elevados de ICER (AR$78.284, AR$83.887 y AR$89.444) y de ICUR (AR$90.375, AR$96.627 y AR$103.028) para los sistemas público, seguridad social y privado respectivamente. La población representada por el estudio MADIT II mostró beneficios en salud y costos intermedios con respecto a las otras dos cohortes, con un aumento de 1,9 años en la EV y de 1,65 QALYs. Los valores de ICER fueron de AR$64.735, AR$69.190 y AR$73.753 y las ICUR de AR$74.537, AR$79.772, AR$85.032 para los tres sistemas de salud. En el escenario sin descuento los resultados fueron similares, con EV y QALYs levemente mayores y costos levemente menores para todas las cohortes y sistemas de salud, observándose ICER e ICUR discretamente inferiores al escenario descontado. Análisis de sensibilidad Se evaluó la variación de la ICER en función de los intervalos de confianza de mortalidad de los estudios seleccionados. Se observó un alto impacto, pudiendo incluso duplicar los valores del caso base, principalmente en el caso de mortalidad cardíaca. La mortalidad asociada a CDI se varió entre 0% y 3%, observándose una variación discreta de la ICER (Tabla 4). En el análisis de sensibilidad de los costos de la colocación y el recambio del generador de CDI, se variaron los mismos hasta un VALUE IN HEALTH 14 (2011) S33–S38 mínimo de AR$20.000 y un máximo de AR$100.000 cada uno. Se evidenció así una variabilidad de la ICER entre AR$25.739 y AR$121.360 para la cohorte de prevención secundaria, entre AR$10.839 y AR$48.821 para la cohorte MADIT I y entre AR$23.314 y AR$109.604 para MADIT II. Los costos de seguimiento de los distintos grupos de pacientes y el costo de las complicaciones no tuvieron un gran impacto sobre la ICER, a pesar de variarse los mismos entre un 50% y un 200% del valor tomado para el caso base (Tabla 4). En el análisis de sensibilidad probabilístico de las utilidades se evidenció que al variar las mismas con un desvío estándar de 0,05 puntos, los QALYS del grupo CDI variaron entre 6,16 y 7,30 para la cohorte de prevención secundaria, entre 7,01 y 8,34 para la cohorte MADIT I y entre 6,58 y 7,76 para la cohorte MADIT II. Esta variabilidad provocó una dispersión importante en las ICUR observándose valores entre AR$62.935 y AR$178.888 para la cohorte de prevención secundaria, entre AR$30.918 y AR$52.558 para MADIT I y entre AR$54.841 y AR$132.198 para MADIT II. Dado que surgen recientemente en el mercado CDI con una duración nominal de la batería de 7 años, se realizó un escenario incorporando este parámetro en lugar de los 5 años seleccionados para el caso base [21]. Se observó que los ICER disminuyeron entre 17% y 19%. Discusión Este estudio demuestra que el implante preventivo del CDI en pacientes con alto riesgo de muerte súbita en Argentina, tiene una ICER por debajo de AR$89.444 ($24.012) y una ICUR por debajo de AR$103.028 ($27.659) para todos los escenarios considerados. Puede observarse que el CDI incrementa la expectativa de vida entre 1,5 y 4,24 años y los QALY entre 1,3 y 3,71. Este incremento es sustancial en comparación con el proporcionado por otras intervenciones médicas [16]. Para cada una de las cohortes las ICER son más bajas para el sistema público y más elevadas para el privado. Sin embargo, independientemente del sistema de salud considerado, las ICER más bajas corresponden a la cohorte MADIT I y las más altas a la cohorte de prevención secundaria. En Argentina, no existe un umbral de voluntad de pago por las intervenciones médicas que sea ampliamente aceptado por los proveedores de servicios de salud, la decisión se encuentra influenciada por la disponibilidad de recursos existentes. Debiera priorizarse la inclusión de aquellos tratamientos que han demostrado ser muy efectivos a un bajo costo. La Comisión de Macro-Economía de la Organización Mundial de la Salud (OMS) sugiere que una ICER menor a 1 PIB (Producto Interno Bruto) per cápita es muy costo-efectivo y que entre 1 y 3 PIB per cápita probablemente sea costo-efectivo. Según el Fondo Monetario Internacional, el PIB per cápita de Argentina para el año 2009 fue de AR$35.890 ($9.635). Si consideramos este dato, el CDI sería costo-efectivo para este país en la cohorte MADIT I y podría serlo en las cohortes MADIT II y prevención secundaria. Si bien se reportan los costos y beneficios discriminados para el sistema de salud público, de seguridad social y privado, en todos los casos la cohorte MADIT I fue la más beneficiada, principalmente debido a que estos pacientes tienen un riesgo basal de mortalidad más elevado y un beneficio en la reducción de la mortalidad atribuible al CDI más pronunciado (RR 0,46). Según la información recabada mediante encuestas a expertos en electrofisiología, en la práctica clínica de Argentina se prioriza la colocación del CDI en el grupo de pacientes en prevención secundaria de muerte súbita. Sin embargo, esta es la cohorte en la que la ICER fue más elevada en este estudio, correlacionándose con los datos de estudios de costo-efectividad más recientes alrededor del mundo [4-8]. Los pacientes con características similares al estudio MADIT II presentan un desafío para los sistemas de salud debido a que son una población numerosa y el impacto presupuestario sería muy significativo. En general en Argentina no se cubre esta indicación de colocación de CDI en los diferentes siste- S37 mas de salud, sin embargo las ICER para este grupo son significativamente inferiores a las del grupo de prevención secundaria. Al realizar los análisis de sensibilidad para evaluar la incertidumbre, se evidenció que la mortalidad por causa arrítmica y cardíaca, las utilidades, y el costo de la colocación y recambio del CDI, son variables que influencian claramente los resultados. El impacto de la mortalidad por CDI y de la duración de la batería fue moderado, y los costos del seguimiento y de las complicaciones tuvieron poco impacto en la variabilidad. Es importante destacar que los resultados son muy dependientes de la eficacia del CDI para disminuir el riesgo relativo de muerte en cada estudio, así como de los valores de utilidad considerados. Factores locales que pudieran modificar estos valores podrían provocar variaciones importantes en las ICER e ICUR, por lo que se destaca la importancia de contar con datos locales. En cuanto a la influencia de la mortalidad por CDI se observó una variación discreta, siendo la misma más acentuada en el grupo de prevención secundaria; las ICER aumentan marcadamente solo al acercarse al punto extremo de 3% de mortalidad por CDI lo que tornaría al procedimiento dudosamente aceptable en la práctica clínica. Se observa que la influencia de los costos de la colocación y recambio del generador en las ICER fue marcada en todas las cohortes; la disminución en el costo del CDI en el mercado podría provocar que las cohortes menos beneficiadas entraran dentro de rangos atractivos de costo-efectividad. Asimismo, las mejoras tecnológicas del dispositivo como ser el aumento de la duración nominal de la batería a 7 años, disminuyen las ICER en casi un 20%. Por otra parte, un parámetro que podría preocupar a los decisores en salud sería la variabilidad de los costos relacionados con el seguimiento de los pacientes y con las complicaciones ya que podrían variar mucho entre provincias y entre diferentes centros; a pesar de llevar estos valores a la mitad y al doble de los valores basales, no resultaron influyentes en las ICER al realizar el análisis de sensibilidad. Este estudio encontró valores inferiores de ICER e ICUR con respecto a la mayoría de los estudios realizados alrededor del mundo, si bien existe una enorme variabilidad. Por ejemplo, para la población de prevención secundaria pueden encontrarse valores tan disimiles como $17.000 y $114.000 [4-8]. Probablemente estos valores inferiores se deban, fundamentalmente, al costo de la colocación y recambio del cardiodesfibrilador implantable que en la actualidad para nuestro país es de aproximadamente la mitad del costo considerado en la mayoría de los estudios que son más antiguos, dado que los datos fueron muy sensibles al costo del dispositivo. Los CDI han causado un impacto en todo el mundo con su aparición en el mercado por tratarse de una tecnología relativamente sencilla de implementar y con beneficios marcados. Si bien el costo del dispositivo ha ido disminuyendo con el transcurso del tiempo, continua siendo elevado y la población pasible de recibirlo muy numerosa y creciente. Dependiendo de las posibilidades económicas de cada sistema de salud, el CDI se ha ido incorporando paulatinamente a la práctica diaria; sin embargo aún genera controversias entre los decisores de salud cuales deberían ser las poblaciones destinatarias. Este estudio pretende brindar una herramienta que pueda colaborar a la toma de esta decisión, considerando la perspectiva de los financiadores de salud en Argentina. Conclusiones Este estudio evidencia que la colocación de un CDI profiláctico en estas tres cohortes de pacientes con alto riesgo de muerte súbita es más costosa y más efectiva que el tratamiento antiarrítmico. Las tasas de costo-efectividad y costo-utilidad variaron considerablemente en función del grupo de pacientes seleccionados y discretamente en función del sistema de salud, con un rango de ICER entre AR$31.806 y AR$89.444 y de lCUR entre AR$36.377 y S38 VALUE IN HEALTH 14 (2011) S33–S38 AR$103.028. La cohorte más beneficiada fue la correspondiente a los pacientes del estudio MADIT I para los tres sistemas de salud evaluados. Los resultados fueron muy sensibles a la mortalidad de los estudios clínicos, a la utilidad y al costo del dispositivo; moderadamente sensibles a la mortalidad por CDI y a la duración de la batería; y poco sensibles a los costos del seguimiento y de las complicaciones. El CDI podría ser costo-efectivo en Argentina, principalmente en los pacientes representados por el estudio MADIT I. Agradecimientos Al Dr. Joaquín Caporale, al Dr. Andrés Pichón-Riviere y al Lic. Sebastián Vicente por su asistencia. Fuentes de financiamiento: None. REFERENCIAS [1] Gonzalez-Zuelgaray J. Arritmias Cardíacas (2da Ed). Buenos Aires: Editorial Inter-Médica, 2006. [2] Doval H, Tajer C. Evidencias en Cardiología III. Buenos Aires: Editorial GEDIC, 2003. [3] Lee D, Green L, Liu P, et al. Effectiveness of implantable defibrillators for preventing arrhythmic events and death: a meta-analysis. J Am Coll Cardiol 2003;41:1573– 82. [4] Sanders G, Hlatky M, Owens D. Cost-effectiveness of implantable cardioverter-defibrillators. N Engl J Med 2005;353:1471– 80. [5] Stanton M, Bell G. Economic outcomes of implantable cardioverterdefibrillators. Circulation. 2000;101:1067–74. [6] Buxton M, Caine N, Chase D, et al. A review of the evidence on the effects and costs of implantable cardioverter defibrillator therapy in different patient groups, and modelling of cost-effectiveness and costutility for these groups in a UK context. Health Technol Assess 2006; 10:1–164. [7] Bryant J, Brodin H, Loveman E, et al. The clinical and cost-effectiveness of implantable cardioverter defibrillators: a systematic review. Health Technol Assess 2005;9:1–150. [8] Sanders G, Hlatky M, Owens D. Cost-effectiveness of implantable cardioverter-defibrillators. N Engl J Med 2005;353:1471– 80. [9] Augustovski F, Iglesias C, Manca A, et al. Barriers to Generalizzability of Health Economic Evaluations in Latin America and the Caribbean Region. Pharmacoeconomics 2009:11:919 –29. [10] Heidenreich P, Keeffe B, McDonald K, Hlatky M. Overview of randomized trials of antiarrhythmic drugs and devices for the prevention of sudden cardiac death. Am Heart J 2002;144:422–30. [11] Shiroiwa T, Sung Y, Fukuda T, et al. International survey on willingness-to-pay (WTP) for one additional QALY gained: what is the threshold of cost effectiveness? Health Econ 2010;19:422–37. [12] Moss AJ, Hall WJ, Cannom DS, et al. Improved survival with an implanted defibrillator in patients with coronary disease at high risk for ventricular arrhythmia. Multicenter Automatic Defibrillator Implantation Trial (MADIT) Investigators. N Engl J Med 1996;335: 1933– 40. [13] Moss AJ, Zareba W, Hall WJ, et al. Prophylactic implantation of a defibrillator in patients with myocardial infarction and reduced ejection fraction. N Engl J Med 2002;346:877– 83. [14] Indicadores de natalidad y mortalidad. Dirección de Estadísticas e Información en Salud. Ministerio de Salud. Presidencia de Nación. Disponible en: www.deis.gov.ar. [Acceso el 26 de febrero de 2010]. [15] Reynolds MR, Cohen DJ, Kugelmass AD, et al. The frequency and incremental cost of major complications among medicare beneficiaries receiving implantable cardioverter-defibrillators. J Am Coll Cardiol 2006;47:2493–7. [16] CEA Registry. Disponible en: https://research.tufts-nemc. org/cear/overview/whatiscear.aspx. [Acceso el 4 de Abril 2010]. [17] Base de Datos de Costos Unitarios del Instituto de Efectividad Clínica y Sanitaria. Disponible vía www.iecs.org.ar. [Acceso el 4 de Abril de 2010]. [18] Manual Farmaceútico Alfabeta. Disponible en: www.alfabeta.net. [Acceso el 4 de Abril de 2010]. [19] Federico Tobar. Nota técnica Salud 002/2004. BID, Departamento de Desarrollo Sustentable, Washington, DC, enero 2004 y Plan Remediar disponible en: www.remediar.gov.ar. [Acceso el 4 de Abril de 2010]. [20] Briggs A. Probabilistic analysis of cost-effectiveness models: statistical representation of parameter uncertainty. Value Health 2005;8:1–2. [21] Teligen®. Implantable Cardioverter Defibrillator. Specifications. Boston Scientific. 2008. Disponible en: www.medtronic.com. [Acceso el 22 de Abril de 2010]. VALUE IN HEALTH 14 (2011) S39 –S42 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Costo Efectividad de Posaconazol versus Fluconazol/Itraconazol en el Tratamiento Profiláctico de las Infecciones Fúngicas Invasivas en México Kely Rely, MD, MPH, MS1, Pierre K. Alexandre, MPH, MS, PhD2, Guillermo Salinas Escudero, MS3,* 1 Economista de la salud, Investigador principal, Consultor en Farmacoeconomía, Director General de CEAHealthTech, México D.F.; 2Economista de la salud. Department of Mental Health – Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD. USA; 3Economista de la salud, Hospital Infantil de México Federico Gómez, México D.F. A B S T R A C T Cost effectiveness of posaconazole versus fluconazole/itraconazole therapy in the prophylaxis against invasive fungal Infections among high-risk neutropenic patients in Mexico. Objective: To estimate the cost effectiveness and long-term combined effects of Posaconazole versus fluconazole/itraconazole (standard azole) therapy in the prophylaxis against invasive fungal Infections among high-risk neutropenic patients in Mexico. Methods: A previously validated Markov model was used to compare the projected lifetime costs and effects of two theoretical groups of patients, one receiving Posaconazole and the other receiving standard azole. The model estimates total costs, numbers of IFIs, and QALY per patient in each prophylaxis group. To extrapolate trial results to a lifetime horizon, the model was extended with one-month Markov cycles in which mortality risk is specific to the underlying disease. Data on the probabilities of IFI were obtained from Study Protocol PO1899. Drug costs were taken from average wholesale drug reports for 2009. Cost and health effects were discounted at 5% according to the Mexican guideline. The analysis was conducted from the Mexican healthcare perspective Introducción Las infecciones fúngicas invasivas (IFIs) generan problemas de salud en los pacientes críticos y constituyen un aspecto clínico de creciente importancia [1,2]. El número de pacientes expuestos al riesgo de las IFIs ha aumentado en las últimas décadas por la aparición de la epidemia del SIDA, el uso de quimioterapia intensiva en pacientes onco-hematológicos, el empleo de fármacos que son rechazados en pacientes receptores de trasplante y mayor utilización de dispositivos intravasculares [3,4]. Se han producido avances en el diagnóstico de estas infecciones [5,6], a pesar de su baja prevalencia en la práctica hospitalaria, tienen un incremento en la tasa de morbi-mortalidad de pacientes que su sistema inmune no responde adecuadamente [7]. La infección por Candida spp. es la más frecuente (70-87%), using 2008 unit cost prices. Results: Our model projects an accumulated cost to the Mexican healthcare system per patient receiving the Posaconazol regimen of $US 5,634 compared to $US 7,463 for the standard azole regimen. The accumulated discounted effect is 3.13 LY or 2.25 QALYs per patient receiving Posaconazol, compared to 2.96 LY or 2.13 QALYs per patient receiving standard azole. Posaconazol remained the dominant strategy across each scenario. Probabilistic sensitivity analysis tested numerous assumptions about the model cost and efficacy parameters and found that the results were robust to most changes. Conclusion: Posaconazole provides modest incremental benefits compared with standard azole therapy in the prophylaxis against IFIs among high-risk neutropenic patients. Routine Posaconazole use appears a cost saving when the likelihood of IFIs or the cost of treatment medications is high. Palabras Claves. costo/efectividad, fluconazol, itraconazol, infecciones fúngicas, posaconazol. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. pero otras infecciones por Aspergillus spp. aparecen como nuevos hongos emergentes [8,9]. Las opciones del tratamiento han sido limitadas, hasta las nuevas formulaciones de anfotericina B y de itraconazol, que sumados a los nuevos triazoles de amplio espectro, como el voriconazol [10-12] y las equinocandinas, donde se encuentra el primer agente aprobado por la Food and Drug Administration (FDA) la caspofungina [13,14] y más recientemente el posaconazol [15-17]. El posaconazol es la terapia preventiva más efectiva, según un estudio publicado por Cornely et al. 2007 [15], este fármaco puede ayudar a prevenir las infecciones causadas por los patógenos más habituales. El objetivo del presente estudio es efectuar un análisis costoefectividad del posaconazol en comparación con itraconazol/ fluconazol, para la prevención de las IFIs en pacientes con neutropenia secundaria a quimioterapia por leucemia aguda o síndromes mielodisplásico. Conflicts of interest: Funding for this analysis was provided by Schering Plough, Mexico City. Título corto: Costo efectividad de posaconazol versus fluconazol /itraconazol en el tratamiento profiláctico de las infecciones fúngicas invasivas en México. * Autor de correspondencia: Kely Rely, CEAHealthTech, Republicas 1, Col. Banjidal, CP. 094500 México D.F. Tel: ⫹ (52)1 55-54-35-17-87; Fax: (52) 55-50-19-55-98. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.032 S40 VALUE IN HEALTH 14 (2011) S39 –S42 Métodos Se construyó un modelo farmacoeconomico, con el fin de integrar la información sobre las consecuencias clínicas y económicas en la profilaxis de las IFIs en los pacientes con neutropenia secundaria a quimioterapia por leucemia aguda o síndromes mielodisplásico, estimando la razón de costo-efectividad de posaconazol y itraconazol/fluconazol basándose en el ensayo clínico controlado PO1899 [15]. El análisis se realizó desde la perspectiva del Sistema Nacional de Salud (SNS), los costos se expresaron en dólares estadounidenses del 2009. Se modeló la evolución clínica de las IFIs, mediante una cohorte hipotética de pacientes con características iguales o similares del estudio PO1899. La estructura del modelo fue muy similar a la de otro publicado por Stam et al. 2008 [18]. En el análisis de decisión a corto plazo, se utilizó un árbol de decisión y se evaluaron los resultados tras finalizar los 100 días con cada tratamiento. En la figura 1 a Materiales Complementarios en:doi:10.1016/j. jval.2011.05.032, se muestra el árbol de decisiones utilizado en modelo. En la fase inicial, todos los pacientes recibieron profilaxis con cada ciclo de quimioterapia hasta la recuperación de la neutropenia y la remisión completa, hasta la aparición de una IFIs o durante un máximo de 12 semanas, según cualquiera de estas circunstancias que ocurriera primero. Se comparó la incidencia de las IFIs comprobadas o probables a ocurrir durante la terapia entre los grupos con posaconazol y con fluconazol o itraconazol, y los criterios de las valoraciones secundarias fueron: la muerte por IFIs o por cualquier otra causa y el tiempo transcurrido hasta el fallecimiento del paciente. En el modelo a largo plazo, después de los 100 días de iniciarse el tratamiento profiláctico, todos los pacientes que sobrevivieron inicialmente en los estados de “IFIs” o “no IFIs”, según la respuesta obtenida transcurrido el tiempo, entrarían en un modelo de Markov, con ciclos de 1 mes de duración. El paciente que está en “IFIs” sobrevive o transita a “muerte” que es una fase absorbente, por causas no relacionadas con las IFIs; el paciente que está en “No IFIs” sobrevive o transita a muerte. Las probabilidades de muerte por causas relacionadas a las IFIs fueron tomadas de PO 1899. La probabilidad de muerte del paciente por otras causas después de los 100 días y la utilidad se tomaron de la literatura [19]. Los parámetros de efectividad utilizados en el modelo que fueron: incidencia de las IFIs, los años de vida ganados AVG y años de vida ajustados por calidad (AVAC). El costo unitario de adquisición del posaconazol fue proporcionado por los Laboratorios Schering Plough y los costos unitarios de adquisición del itraconazol y fluconazol fueron tomados de los precios vigentes del Instituto Mexicano del Seguro Social (IMSS) del 2009, [20] reflejado en la tabla 1 a Materiales Complementarios en: doi: 10.1016/j.jval.2011.05.032. Para el costo de la terapia de las IFIs se tuvo en cuenta los resultados del estudio realizado en el IMSS por Mould et al. 2009 [21] donde se estimaron los costos del tratamiento de los esquemas disponibles (voriconazol, caspofungina y anfotericina B). De acuerdo a este estudio, el costo del tratamiento de un paciente con caspofungina fue de US$49,962, voriconazol de US$59,378 y anfotericina B de US$60,058 a precios vigentes del 2009. Se calcularon los índices de costo efectividad mediante los cocientes del costo efectividad medio de cada alternativa y el cociente costo efectividad incremental, resultante de emplear la opción con mayor efectividad y costo frente a la alternativa de menor costo y menos efectiva. Así, se pudo conocer el costo adicional por unidad extra de efectividad (infecciones evitadas y años de vida ajustados por calidad), al emplear la intervención que resultó la más efectiva. Para estimar el costo efectividad incremental por la utilización del posaconazol, se utilizó la siguiente fórmula [22]: ⌬C ⁄ ⌬E ⫽ n n i⫺1 i⫺1 兺 ⌬Ci ⁄ 兺 ⌬Ei (1) Dónde: ⌬C: es la diferencia entre el costo de tratamiento con posaconazol y la terapia con azoles ⌬E : es la diferencia entre la eficacia de tratamiento con posaconazol y la terapia con azoles Se efectuó un análisis de sensibilidad de las variables que están sometidas a una mayor incertidumbre y/o que más pueden influir en el resultado final del análisis, así mismo se realizó un segundo análisis de sensibilidad de tipo probabilístico. En la tabla 2 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.032, se muestran los valores basales utilizados para el análisis, el valor de la distribución de cada variable del análisis probabilístico y la información de las fuentes. Para el modelo probabilístico se utilizó la simulación de Monte Carlo, que permitió la modelación individual de las evoluciones clínicas de 1000 de pacientes hipotéticos. En cada simulación, se asignó una distribución probabilística a cada uno de los parámetros del modelo. El resultado de cada simulación de este modelo se resume en una razón costo-efectividad incremental (RCEI), formado a partir del costo y la efectividad incremental. La distribución de los parámetros se basó en los métodos descritos en la bibliografía [23]. En el modelo de Markov, el horizonte temporal se terminó con la muerte de la cohorte de pacientes y la duración máxima fue de 3.6 años. Como el periodo del análisis fue mayor de un año, se siguieron las recomendaciones de las guías internacionales de evaluación económica [24,25] y la mexicana de farmacoeconomía [26], y se aplicó una tasa descuento del 5%. Resultados Según el criterio de incidencia de las IFIs a los 30 días, el NNT asociado al posaconazol sería 17 pacientes y a los 100 días sería 16 casos, y la determinación del número de pacientes con neutropenia secundaria a quimioterapia necesarios a tratar con posaconazol frente a itraconazol/fluconazol para prevenir una IFI. Se seleccionaron 304 pacientes que fueron asignados aleatoriamente para recibir el posaconazol y 298 pacientes para recibir fluconazol/itraconazol (240 fluconazol y 58 itraconazol). Se evidenció que los pacientes con las IFIs comprobadas o probables de ocurrir en el grupo del posaconazol estuvieron en el 4.6 % de los casos y del grupo con fluconazol o itraconazol fueron del 11.0 % de los casos, expresados en la tabla 3 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.032. De acuerdo a los resultados de modelo el número de AVG para los pacientes que recibieron posaconazol y que sobrevivieron después de los 100 días de la profilaxis fue de 3.13 frente a 2.96 en itraconazol/fluconazol, los AVAC de 2.25 con posaconazol y 2.13 con itraconazol/fluconazol, con una diferencia de 0.121 AVG. Se estimó que el costo medio por paciente con posaconazol fue de US$5634 y US$7,463 con itraconazol/fluconazol, por lo que existe un beneficio económico de USD1829 por caso tratado. En este caso, el costo básico del tratamiento con posaconazol dominó la profilaxis con itraconazol/fluconazol, ya que resultó una terapia más eficaz con unos costos inferiores. Este plano de costo efectividad mostró que la mayor parte de los puntos representados se encuentran en el cuadrante de menor costo y mayor efectividad para el 88 % simulaciones realizadas, como muestra la figura 2 a Materiales Complementarios en: doi: 10.1016/j.jval.2011.05.032. Es de señalar, que el 100 % de los puntos se encuentran por debajo de la frontera del valor de 1 PIB/AVAC incremental. Se pudo VALUE IN HEALTH 14 (2011) S39 –S42 comprobar que la totalidad de las simulaciones son costo efectivas, y se aprecia que más del 80% de las mismas, indican que la profilaxis con posaconazol presentó un ahorro para el SNS mexicano. La curva de aceptabilidad muestra si la disposición a pagar desde el punto de vista social o del financiador público fuera de US$6000 por AVAC incremental, ya que la probabilidad de que el posaconazol frente al itraconazol/fluconazol presentara una óptima situación estaría en el 95% de los casos. Para una disposición a pagar mayor o igual a US$15,000/AVAC incremental, la probabilidad ascendería aproximadamente al 100% (figura 3 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.032) Discusión Este estudio de tipo costo efectividad es el primero que se realiza en el país, para evaluar el tratamiento con posaconazol frente al itraconazol/fluconazol. Los resultados del estudio demuestran que en el contexto del SNS mexicano, el posaconazol es un tratamiento más económico que el itraconazol/fluconazol, por que han sido comparados directamente en un ensayo clínico controlado en la prevención de IFIs asociada a neutropenia secundaria a quimioterapia, con un significativo ahorro de US$1829 por paciente tratado, considerando solamente el costo total de la profilaxis y el manejo de las IFIs. Con el posaconazol se podría obtener una efectividad mayor que la obtenida por el itraconazol/fluconazol, con una menor incidencia de las IFIs y una mayor supervivencia media, y un incremento en la prevención de estas enfermedades en los pacientes. El análisis de sensibilidad confirmó la robustez existente de las tendencias en todos los casos analizados, manteniéndose los resultados del estudio muy estables a cambios en las variables de eficacia, seguridad y costos. Los resultados de este estudio adaptado al entorno mexicano son consistentes con las evaluaciones económicas que se han realizado para el posaconazol en el contexto internacional. Se concluye que el posaconazol administrado para el tratamiento profiláctico de las IFIs, resulta una intervención costo-efectiva para el SNS [27-32]. Como fortalezas del modelo empleado, además de las que presenta el modelo de Markov para simular la historia natural de la enfermedad, se puede argumentar que él mismo fue validado por expertos que participaron en el estudio PO1899, y que la estimación de la utilización de recursos económicos, se hizo a partir de guías de práctica clínica y del propio estudio, y los costos unitarios de los recursos se obtuvieron de fuentes mexicanas, así como las principales asunciones del modelo fueron sometidas a análisis de sensibilidad, confirmando la estabilidad de los resultados obtenidos. En cuanto a las limitaciones, tenemos que una limitante fue la inclusión de algunos supuestos de la literatura internacional dentro del modelaje, aun cuando estos fueron posteriormente validados por los expertos, además de la problemática para la extrapolación de algunos supuestos del modelo a otros países, debido a las diferencias que se pueden presentar en términos de la estructura de los servicios de salud y sistemas de precios locales. Otra limitante, fue que aunque el estudio de Cornely et al. [15] cumplió con las características del modelaje (población, intervención, etc.), y del cual se retoman las efectividades del modelo y algunos supuestos, no se encontraron otros estudios adicionales, sin embargo, no es una debilidad metodológica sino más bien una limitante por carencia de otros estudios publicados. S41 Conclusiones La utilización del posaconazol como tratamiento profiláctico en pacientes con alto riesgo de sufrir una IFIs asociada con neutropenia secundaria a quimioterapia, es una intervención costo-efectiva con respecto a su uso alternativo del itraconazol/fluconazol, ya que presenta una mayor efectividad con menor incidencia y mayor supervivencia media en la prevención de estas enfermedades para los pacientes, así como disminuye el costo medio por caso intervenido con un ahorro económico importante, lo cual constituye un efecto beneficioso asociado al tratamiento para este problema de salud en México. Fuentes de financiamiento: Este estudio fue patrocinado para su realización por los Laboratorios Schering Plough de México. Declaración de conflicto de intereses: Los autores del presente trabajo, declaramos que los Laboratorio Schering Plough participó en el patrocinio del panel de expertos, sin embargo, la selección de los médicos y preguntas incluidas en los cuestionarios para identificar los patrones de utilización de recursos quedo en completamente a criterio de los autores, por lo cual no existe conflicto de intereses, con la publicación del estudio y de los resultados. Materiales Complementarios Material complementario que acompaña este artículo se puede encontrar en la versión en línea como un hipervínculo en doi: 10.1016/j.jval.2011.05.032 o si es un artículo impreso, estará en www.valueinhealthjournal.com/issues (seleccione el volumen, número y artículo). REFERENCIAS [1] Pfaller MA, Diekema DJ. Epidemiology of invasive candidiasis: a persistent public health problem. Clin Microbiol Rev 2007;20:133– 63. [2] Nucci M, Queiroz-Telles F, Tobón AM, et al.Epidemiology of opportunistic fungal infections in Latin America. Clin Infect Dis 2010; 51:561–70. [3] Antinori S, Nebuloni M, Magni C, et al.Trends in the postmortem diagnosis of opportunistic invasive fungal infections in patients with AIDS: a retrospective study of 1,630 autopsies performed between 1984 and 2002. Am J Clin Pathol 2009;132:221–7. [4] Naba MR, Kanafani ZA, Awar GN, et al. Profile of opportunistic infections in HIV-infected patients at a tertiary care center in Lebanon. J Infect Public Health 2010;3:130 –3. [5] Person AK. Kontoyiannis DP. Alexander BD. Fungal infections in transplant and oncology patients.Infect Dis Clin North Am 2010;24:439 –59. [6] Smith JA. Kauffman CA. Recognition and prevention of nosocomial invasive fungal infections in the intensive care unit. Crit Care Med 2010;38(8 Suppl.):S380 –7. [7] Menichetti F. Infectious complications in neutropenic cancer patients CIntern Emerg Med 2010;5(Suppl. 1):S21–5. [8] Ostrosky-Zeichner L. Pappas G. Invasive candidiasis in the intensive care unit. Crit Care Med 2006;34:857– 63. [9] Pappas PG, Rex JH, Sobel JD, et al. Guidelines for treatment of candidiasis. Clinl Infect Dis 2004;38:161– 89. [10] Ally R. Schurmann D, Kreisel W, Carosi G, et al. A randomized, doubleblind, double dummy, multicenter trial of voriconaloze andl fluconazole in the treatment of esophageal candidiasis in imnmunocompromised patients. Clin Infect Dis 2001;33:1447–54. [11] Johnson LB, Kauffman CA. Voriconazole: a new triazole antifungal agent. Clin Infect Dis 2003;36:630 –7. [12] Perfect JR, Marr KA, Walsh TJ, et al. Voriconazole treatment for lesscommon, emerging, or refractory fungal infections. Clin Infect Dis 2003;36:1122–31. [13] Kontoyiannis DP, Hachem R, Lewis RE, et al. Efficacy and toxicity of caspofungin in combination with liposomal amphotericin B. As primary or salvage treatment of invasive aspergillosis in patients with hematologic malignances. Cancer 2003;98:292–9. [14] Keating G, Figgitt D. Caspofungin: a review of its use in oesophageal candidiasis, invasive candidiasis and invasive aspergillosis. Drugs 2003;63:2235– 63. S42 VALUE IN HEALTH 14 (2011) S39 –S42 [15] Cornely OA, Maertens J, Winston DJ, et al. Posaconazole vs. Fluconazole or Itraconazole prophylaxis in patients with neutropenia. N Engl J Med 2007;356:348 –59. [16] Schiller DS, Fung HB. Posaconazole: an extended-spectrum triazole antifungal agent. Clin Ther 2007;29:1862– 86. [17] Ullmann AJ, Lipton JH, Vesole DH, et al. Posaconazole or fluconazole for prophylaxis in severe graft-versus-host disease. N Engl J Med 2007; 356:335– 47. [18] Stam WB, O’Sullivan AK, Rijnders B, et al. Economic evaluation of posaconazole vs. standard azole prophylaxis in high risk neutropenic patients in the Netherlands. Eur J Haematol 2008;81:467–74. [19] Dalziel K. Round A. Garside R. Stein K. Cost effectiveness of imatinib compared with interferon-alpha or hydroxycarbamide for first-line treatment of chronic myeloid leukaemia. Pharmacoeconomics 2005; 23:515–26. [20] Instituto Mexicano del Seguro Social. Portal de Transparencia. IMSS compro. Precios promedio 2009. Disponible en: http://transparencia.imss.gob.mx/trnsp/ncompro.aspx?c⫽1. [Accesso Abril 30, 2010]. [21] Mould QJF, Contreras HI, Gómez ME, et al. Tratamiento antimicótico empírico de pacientes inmunocomprometidos con neutropenia y fiebrepersistente con sospecha de aspergilosis sistémica: análisis de costo-efectividad en México. Bol Med Hosp Infant Mex 2009;66:241–52. [22] Antoñanzas F. Evaluación económica aplicada a los medicamentos: características y metodología. In: Sacristán J. Badía X. Rovira J. (eds.). Farmacoeconomía: evaluación económica de los medicamentos. Madrid: Editores Médicos, 1995. [23] Briggs AH. Goeree]?R. Blackhouse G/ce:given-nameO’Brien B. Probabilistic analysis of cost-effectiveness models: choosing between treatment strategies for gastroesophageal reflux disease. Med Decis Making 2002;22:290 –308. [24] Canadian Agency for Drugs and Technologies in Health (CADTH).Guidelines for the Economic Evaluation of Health Technologies: Canada. (3rd ed.). Ottawa: CADTH. 2006. [25] Pharmaceutical Benefits Advisory Committee (PBAC). Guidelines for preparing submissions to the Pharmaceutical Benefits Advisory Committee. Australia. Canberra: PBAC, 2007. [26] Instituto Nacional de Salud Pública de México. Propuesta de guías para la conducción de estudios de evaluación económica como parte de la actualización de los cuadros básico de insumos para la salud. Consejo de Salubridad General. Cuidad de México: Instituto Nacional de Salud Pública, 2007. [27] Tahami AA. O’Sullivan AK. Papadopoulos G. Economic evaluation of posaconazol vs standard azole therapy in the prophylaxis against invasive fungal infections in patients with prolonged neutropenia in Canada. Value Health 2008;11:A.PIN13. [28] Grau S, de la Cámara R, Sanz M, et al. Cost-effectiveness of posaconazole versus standard azole treatment (fluconazole or itraconazole) in the prevention of invasive fungal infections among high-risk neutropenic patients in Spain. 18th European Congress of Clinical Microbiology and Infectious Diseases,19 –22 April, Barcelona, Spain; 2008. [29] Greiner RA. Meier Y. O’Sullivan A. Imhof A. Cost-effectiveness of posaconazole versus standard azole therapy for the prevention of invasive fungal infections in high-risk patients in Switzerland. 34th Annual Meeting of the European Group for Blood and Marrow Transplantation (EBMT), March 30-April 2, 2008, Florence, Italy. [30] Thalheimer M, Cornely OA, Hoppe-Tichy T, et al. Pharmaco-economic analyis of posaconazole versus standard azole prophylaxis in highrisk neutropenic AML/MDS patients in Germany. 18th European Congress of Clinical Microbiology and Infectious Diseases, 19 –22 April Barcelona, Spain; 2008. [31] O’Sullivan AK, Pandya A, Papadopoulos G, et al. Cost-effectiveness of posaconazole versus fluconazole or itraconazole in the prevention of invasive fungal infections among neutropenic patients in the U.S. Value Health 2009;12:666 –72. [32] Collins CD. Ellis JJ. Kaul DR. Comparative cost-effectiveness of posaconazole versus fluconazole or itraconazole prophylaxis in patients with prolonged neutropenia. Am J Health Syst Pharm 2008;65: 2237– 43. VALUE IN HEALTH 14 (2011) S43–S47 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Costo-Efectividad del Tratamiento de Salmeterol/Fluticasona en Comparación con Leucotrieno Montelukast para el Control del Asma Infantil Kely Rely, MD, MPH, MS1,*, Sebastián Emanuel González McQuire, MS2, Pierre K. Alexandre, MPH, MS, PhD3, Guillermo Salinas Escudero, MS4 1 Economista de la salud, Investigador principal, Consultor en Farmacoeconomía, Director CEAHealthTech, México D.F., México; 2Glaxo SmithKline México D.F., México; 3Economista de la salud, Department of Mental Health – Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD, USA; 4Economista de la salud, Hospital Infantil de México Federico Gómez, México D.F., México A B S T R A C T Objective: To assess the incremental cost-effectiveness of SFC compared with MON for the control of persistent asthma in children. Methods: We conducted an economic evaluation on a 12-week prospective randomized open-label parallel-group comparison of SFC versus MON in children with symptomatic asthma receiving inhaled corticosteroids and short-acting 2-agonists. Asthma-related medication, unscheduled physician contacts and hospitalizations were collected prospectively. The main effectiveness measure was percentage of asthma-controlled week with no short-acting 2-agonist use during the study period. The analysis was conducted from the Mexican healthcare perspective using 2010 unit cost prices, and only direct costs were considered, all costs are reported in US dollar. . The model was made fully probabilistic to reflect the joint uncertainty in the model parameters. Results: Over the whole treatment period, the median percentages of asthma-controlled weeks were 83.3% in the SFC group and 66.7% in the MON group (SFC-MON difference, 16.7%; 95% CI, 8.3– 16.7; P ⬍ 0.001 in favor of SFC). The mean total cost of the SFC regimen Antecedentes El asma se considera un problema de salud, ya que afecta a 300 millones de personas en el mundo [1]. Actualmente, existe un aumento de su prevalencia y gravedad [2,3]. El asma es una enfermedad que afecta a los niños con factores de riesgo relacionados con problemas genéticos y ambientales [4] y como enfermedad crónica afecta la calidad de vida, el ausentismo escolar y produce elevados costos sanitarios. En México se estima que los costos directos anuales están entre 32 a 35 millones de dólares [5]. Esta enfermedad representa uno de los motivos de admisión a los servicios de salud con una alta hospitalización [6]. Un 70 % del costo total del asma está ocasionado por su mal control [7], que repercute en los gastos de hospitalización, visitas a urgencias, fallecimientos de los pacientes, etc. was $ 2,323 compared with $ 3,230 for the MON regimen. The SFC was the dominant strategy (both more effective and less expensive) using the SFC was associated with an incremental cost per additional asthmacontrolled of $ (5,467). Probabilistic sensitivity analysis tested numerous assumptions about the model cost and efficacy parameters and found that the results were robust to most changes. Conclusions: This analysis demonstrates that, compared with MON, SFC may be cost saving from the Mexican health care perspective for the treatment of pediatric patients with asthma. SFC provided a reduction in the number of severe exacerbations, frequent asthma symptoms and rescue medication use. Incremental cost-effectiveness analysis indicated the dominance of SFC because of both lower costs and greater efficacy. Palabras Claves: asma, costo/efectividad, montelukast, salmeterol/fluticasona. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. La farmacoterapia actual está dirigida a la broncodilatación para el alivio sintomático de los pacientes. Este tratamiento tiene como objetivo el control de síntomas, incluidos los nocturnos y los inducidos por el ejercicio, la prevención de crisis y una mejor función pulmonar, con los mínimos efectos adversos [8]. Para su control adecuado, las guías internacionales [9,10] proponen pautas terapéuticas basadas en la severidad del asma, que se estima por la frecuencia de las crisis, los episodios de despertar nocturno y los antecedentes de internaciones y consultas al servicio de emergencia. El tratamiento comprende el uso de fármacos que alivian los síntomas como los agonistas beta-adrenérgicos, los corticoides sistémicos y el bromuro de ipratropio y otros que previenen los síntomas como el cromoglicato y el nedocromilo, los corticoides inhalados, los agonistas beta de acción prolongada, la teofilina y los modificadores de los leucotrienos [11]. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Título corto: Costo-efectividad de salmeterol/fluticasona vs montelukast en asma infantil. * Autor de correspondencia: Kely Rely, CEAHealthTech, Repúblicas 1, Col. Banjidal, CP. 094500, México D.F.; Tel: ⫹ (52) 1 55-54351787; Fax: ⫹ (52) 50195598. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.033 S44 VALUE IN HEALTH 14 (2011) S43–S47 El objetivo del estudio es realizar un análisis costo-efectividad incremental para comparar los beneficios clínicos y económicos de la terapia con salmeterol y propionato de fluticasona (SFC) de la marca Seretide del Laboratorio Glaxo SmithKline y el leucotrieno montelukast (MON) en el tratamiento del asma en pacientes pediátricos del sector público sanitario de México. Métodos Se realizó un análisis costo-efectividad de las alternativas SFC y MON, para estimar la eficiencia de estos tratamientos. Se construyó un modelo para una cohorte hipotética de pacientes con asma leve a moderada persistente con características similares a los casos que fueron incluidos en el ensayo clínico “Pediatric Asma Clinical Evaluation” (PEACE) [12]. En este estudio se comparó los dos tratamientos en niños entre 6 y 14 años de edad con asma no controlada, los, que fueron controlados durante 12 semanas. Se consideraron 548 pacientes que fueron asignados aleatoriamente a los tratamientos. El grupo del SFC tuvo 281 pacientes y el grupo de MON 267 casos. Las características demográficas y los datos basales fueron similares en ambos grupos (edad media 9.3 años), con una media (DE) FEV1 de 1.49 (0.43 L) el SFC y de 1.48 (0,43 L) el MON. La media ajustada (SE), la variación en el FEM matutino fue de 45.88 (2.82) L/min con SFC y de 28.7 (2.86) L/min con MON (diferencia de 17.16 L/min, para un 95 % CI, 9.23–25.08, p ⬍0.001). En este estudio, se demostraron mejorías significativas en la función pulmonar, un porcentaje mayor de días libres sin síntomas, menor síntomas asmáticos y una reducción de medicamentos de rescate para los niños tratados con SFC y MON. Las dosis utilizadas fueron iguales a las del estudio PEACE. Se administraron el SFC (50/100 g) dos veces al día durante 12 semanas y el MON a dosis de 5 mg una vez al día. La información de la evolución de la enfermedad y los parámetros clínicos se obtuvieron a partir de este estudio [12]. Esta investigación tuvo una mayor proporción de pacientes que lograron una semana de buen control del asma con el SFC en relación al MON (83.3 % el SFC y 66.7 % el MON,). Existen varios estudios que evaluaron la eficacia y seguridad del SFC en un único inhalador en comparación con otras alternativas. Uno de ellos, es un estudio multicéntrico a doble ciego, que se realizó en los Estados Unidos, y el otro es un estudio prospectivo, aleatorio, doble ciego, realizado en América Latina y Turquía, donde se compararon el SFC y el MON en pacientes pediátricos (12–14). Los datos del estudio estuvieron basados en la eficacia y seguridad del ensayo PEACE y no aportaron otra información para la elaboración del modelo, siendo necesario los datos de fuentes secundarias mediante un panel de expertos, como la probabilidad y duración promedio de hospitalización y la asistencia a urgencias de los pacientes no controlados. El diseño del modelo y su validación fue realizado por un grupo de economistas y médicos neumólogos. El papel del comité de expertos consistió en validar la metodología y el modelo farmacoeconómico (tipo de modelo, población objetivo, criterio de efectividad, tipo de costos, datos clínicos, variables para el análisis de la sensibilidad), la definición de la hipótesis,, los datos de los recursos para cada estado de salud y la validación de los resultados. El panel de expertos evaluó los recursos asociados con los antiasmáticos. Esta técnica se utiliza con frecuencia para corregir los datos de insuficientes evidencias [15] y fue un instrumento para la utilización de determinados fármacos. Este panel trata de obtener un consenso de opinión a través de rondas de cuestionarios estructurados, intercalados con retroalimentación controlada a los participantes. El cuestionario inicial fue pilotado y validado en la práctica clínica y se emplearon entrevistas presenciales [16]. Este documento contenía preguntas sobre posología de los medicamentos, duración del tratamiento, terapia de rescate, estudios de laboratorio, la proporción de pacientes con asma no controlada en urgencias y/o hospitalización y la duración promedio de las mismas. Tabla 1 – Costo unitario del tratamiento con Salmeterol/ fluticasona vs Montelukast. Ítems Costo antiasmático - Salmeterol/fluticasona (50/100 mg/dıá) - Montelukast (5 mg/dıá) Atención médica - Consulta médica - Dıá cama - Dıá urgencia Costo ($ Mx) 8.41 12.60 821 4,769 1,105 Fuentes: Precio de compra del IMSS, segundo trimestre del 2009 y DOF. El horizonte temporal propuesto para el estudio fue menor a un año y se siguieron las recomendaciones de la Guía Nacional de Farmacoeconomía [17], de no aplicar tasa descuento para los costos y las efectividades. La perspectiva utilizada fue la del sistema público sanitario mexicano. Los costos incluidos en el análisis de costo efectividad (ACE) son los considerados por el Instituto Mexicano del Seguro Social (IMSS). Los costos directos asociados al tratamiento se estimaron a partir de los publicados en el IMSS [18] y los costos de la atención médica provienen del panel de expertos. En el ACE se consideró que no existieron diferencias en el número y gravedad de los acontecimientos adversos, por lo que no se contemplaron en los costos del tratamiento. Todos los costos sanitarios utilizados se presentaron al valor monetario del peso mexicano vigente año 2009. Los costos sanitarios estuvieron condicionados por dos factores: la utilización de recursos relacionados con la patología y tratamiento, y los costos unitarios de dichos recursos. Se determinaron tres grupos de costos: los costos del tratamiento, los costos relacionados con la exacerbación (hospitalizaciones y urgencias) y el tratamiento de rescate. Los costos en el modelo fueron los correspondientes a estos tres grupos: el primero incluyó los costos farmacológicos, el segundo incorporó los costos no farmacológicos, ya sea mediante manejo de la exacerbación “ambulatoria o urgencias” y el tercero correspondió al costo del tratamiento de rescate. Para estimar los costos del asma, se incluyeron en el modelo los siguientes supuestos: exacerbaciones que requirieron el uso de esteroides orales, exacerbaciones que requirieron el ingreso en urgencias y en el hospital, y los pacientes no controlados que requirieron consultas médicas adicionales. El consumo de fármacos se ajustó a las dosis disponibles en el Cuadro Básico del IMSS y su costo se calculó en función de las presentaciones farmacéuticas. Para el costo del tratamiento de rescate, su cálculo se basó en el costo ponderado obtenido por la multiplicación de la frecuencia de aparición de los síntomas por el costo unitario de adquisición del medicamento. Los costos no farmacológicos se estimaron de los asociados al manejo del paciente con asma y la información obtenida fue consensuada por el panel de expertos. Los costos de los antiasmáticos estándares fueron tomados de los precios vigentes del IMSS del 2009 [19]. Los costos unitarios se presentan en la Tabla 1. Las efectividades utilizadas provienen de la Iniciativa Global de Asma y del Instituto Nacional de Salud [8,19] que consideran el buen control del asma y los días libres de síntomas. Para el control del asma se definió el periodo de vigilancia total de 7 a 8 semanas sin síntomas durante el día; la no administración de medicamentos de rescate; un flujo espiratorio matinal de al menos el 80 % de lo esperado, que el paciente no se despierte en la noche con exacerbaciones y no tuviera una estancia de urgencia o eventos adversos de medicamentos. Se calcularon los cocientes costo/efectividad medios de cada alternativa, y el cociente costo/efectividad incremental, resultante S45 VALUE IN HEALTH 14 (2011) S43–S47 Tabla 2 – Probabilidades utilizadas en el modelo del análisis costo-efectividad de salmeterol/fluticasona. Determinıśtico Salmeterol/fluticasona % asma bien controlada * % dıás libres de sı´ntomas * % dıás libres de tratamiento de rescate * % noches sin despertar * Promedio de exacerbaciones * Porcentaje de ingreso urgencia † Porcentaje de ingreso hospitalización † Montelukast % asma bien controlada * % dıás libres de sı´ntomas * % dıás libres de tratamiento de rescate * % noches sin despertar * Promedio de exacerbaciones * Porcentaje de ingreso urgencia † Porcentaje de ingreso hospitalización † Atención médica Consulta asma controlada † Consulta asma no contralada † Porcentaje de ingreso urgencia † Porcentaje de ingreso hospitalización † Duración media de urgencia † Duración media hospitalización † Costo total Salmeterol/fluticasona ‡ Montelukast ‡ D.E. Distribución Alpha Beta N 0.833 0.090 0.090 0.360 0.120 0.048 0.010 0.022 0.017 0.017 0.029 0.019 0.021 0.010 beta beta beta beta beta beta beta 234 25 25 101 34 5 1 47 256 256 180 247 95 99 281 281 281 281 281 100 100 0.667 0.040 0.060 0.230 0.300 0.120 0.024 0.029 0.012 0.015 0.026 0.028 0.032 0.015 beta beta beta beta beta beta beta 178 11 16 61 80 12 2 89 256 251 206 187 88 98 267 267 267 267 267 100 100 1.5 2.5 0.400 0.200 0.25 6.00 0.56 1.56 0.05 0.04 0.39 3.13 beta beta beta beta beta beta 4 4 40 20 1 8 Gamma Gamma 25.00 25.00 2363 3283 473 657 0.375 0.625 60 80 0.625 0.625 100 100 94.51 131.32 Fuentes: * Maspero J et al. Clin Ther. 2008;30(8):1492–1504. † Expertos. ‡ Cálculo de modelo. de la opción con mejores resultados clínicos y un mayor costo frente a la de menores costos y peores consecuencias clínicas, para conocer el costo adicional por unidad extra de efectividad. Para estimar el costo-efectividad incremental por la utilización de SFC se utilizó la siguiente fórmula: [20] ⌬C⁄⌬E ⫽ 兺 n i⫽1 ⌬Ci⁄ 兺 n i⫽1 ⌬Ei Donde: ⌬C:es la diferencia entre el costo de tratamiento con SFC y MON. ⌬E:es la diferencia entre la eficacia de tratamiento con SFC y MON Para valorar la incertidumbre y la solidez de los resultados, se realizó un análisis de sensibilidad probabilístico mediante una simulación de Monte Carlo [21], simulando 1000 veces el costo-efectividad de cada comparación [22]. Se seleccionaron unas distribuciones fijas y se estimaron los parámetros de estas en función de los datos primarios del estudio [23]. Este tratamiento para la incertidumbre es aconsejable, cuando cabe la posibilidad de que los resultados puedan sufrir importantes cambios, porque no todos los pacientes se comportan como el caso “típico”, ya que el modelo reflejó la variabilidad que pueda existir entre los sujetos, como por ejemplo, los días de estancia en urgencia u hospitalización. Al simular los resultados, se observó con qué probabilidad estos se mantuvieron estables, ya que los parámetros son objeto de variación de forma simultánea. Se logró la validación interna de los resultados y se asignó una distribución gamma para los costos, una distribución normal para los recursos y una distribución beta para las probabilidades del modelo, según la variabilidad descrita en los estudios. En la Tabla 2 se muestran las distribuciones de los parámetros utilizados en el modelo probabilístico. Resultados En comparación con el MON, el SFC presentó una diferencia significativa con mayor cantidad de días libres de síntomas asmáticos (odds ratio [OR] ⫽ 1,74, IC 95 %, 1.07–2.82, p ⫽ 0,025), más días libres de rescate (OR, 3,24; IC del 95 % , 2.09-5.02, p ⬍0,001), más semanas de episodios controlados y una diferencia en las medianas del tratamiento durante 1-12 semanas (16,77 %, IC 95 %, 8.3– 16.77, p ⬍0,001). Las tasas de exacerbaciones reportadas a las 12 semanas fueron de 0,12 en SFC y de 0,30 en MON (razón de 0,40; para un 95 % IC, 0,29-0,57, p ⬍0,001). El cálculo de los costos del tratamiento, la efectividad incremental y el índice de costo efectividad incremental, reflejado en la Tabla 3. Tabla 3 – Resultado del análisis de costo efectividad SFC vs MON. Tratamiento Costo Costo incremental Efectividad Efectividad incremental ICER Montelukast Salmeterol/fluticasona $ 3,230 $ 2,323 $ (908) 0.67 0.83 0.17 $ (5,467) S46 VALUE IN HEALTH 14 (2011) S43–S47 Duración urgencia (0.10 a 0.50 h) -5,587 Costo SFC (+/- 10%) Costo MON (+/- 10%) No de consulta (1 a 3) % urgencia (+/- 10%) -5,395 -5,668 -5,263 -5,770 -5,162 -5,878 -5,057 -6,118 Duración hospitalización (4 a 6) -4,817 -6,295 % hospitalización (+/- 10%) -4,640 -6,708 Exacerbaciones (+/- 10%) -4,226 -6,913 -4,022 % asma control (+/- 10 - 12,507 %) -3,720 -13,500 -11,500 -9,500 -7,500 -5,500 -3,500 Razón de costo efectividad incremental ($/caso controlado del asma) Fig. 1 – Diagrama de tornado: análisis de la sensibilidad y la relación costo efectividad incremental de salmeterol/ fluticasona vs montelukast. El costo promedio del tratamiento por paciente con SFC y MON fue de $ 2,323 y $ 3,230 respectivamente. En el caso básico, el tratamiento con SFC ejerció un dominio con respecto al MON porque presentó un mayor beneficio al tener una efectividad incremental del 17 % y un costo incremental de $ 908 por paciente, con una razón de costo efectividad incremental de $ 5,467 por caso adicional tratado. En el diagrama de tornado se resume el análisis de sensibilidad de los pacientes controlados del asma, como muestra la Figura 1. Se observa el impacto de la diversificación de las variables en un rango determinado en el resultado del modelo. Las variables que más afectaron a los resultados de la razón de costo efectividad incremental por paciente fueron: porcentaje de paciente con un buen control del asma, porcentaje de exacerbaciones, porcentaje de hospitalización y su duración media. Se realizó un diagrama de dispersión de costos y efectividades incrementales en la simulación para 1000 muestras. Esta simulación de los casos individuales generó una nube de puntos que se distribuyeron según una tendencia de la tasa de costo efectividad incremental, como expresa el Gráfico 1. 4,000 3,000 Los puntos comprendidos en el cuadrante superior derecho se correspondieron a un incremento de la efectividad con un mayor costo. La mayor parte de los puntos se encontraron en el cuadrante superior derecho (82 %), lo que significa que el SFC presentó una mayor efectividad a un menor costo que el MON. Ese porcentaje indica que fue una estrategia costo-beneficiosa por ahorros económicos. La curva de aceptabilidad presentó una probabilidad para que la terapia SFC sea óptima, dados los datos generados por el análisis estocástico. Este resultado se interpreta como la probabilidad de que el SFC sea costo-efectivo, como muestra el Gráfico 2. Discusión La evidencia clínica demuestra los beneficios de adicionar un agonista 2 de larga acción, en pacientes con asma persistente moderada que no han respondido satisfactoriamente al tratamiento con corticosteroides. Los estudios clínicos controlados con salmeterol han demostrado que su adición mejora no sólo parámetros ventilatorios, como la tasa máxima del flujo y el VEF1, sino que recupera su calidad de vida [24]. Los resultados del estudio PEACE, evidenciaron que los niños tratados con SFC mejoraron su función pulmonar, ya que tuvieron 60 % menos de exacerbaciones, más semanas controladas y menor necesidad de medicación de rescate. Se demostró que el SFC logra un mayor control del asma en los parámetros evaluados. 2,000 1.00 1,000 0.90 0.80 -1.00 -0.80 -0.60 -0.40 -0.20 0 0.20 0.40 0.60 0.80 1.00 0.70 0.60 -1,000 0.50 0.40 -2,000 0.30 0.20 -3,000 0.10 0 -4,000 Efectividad incremental Gráfico. 1 – Resultados del análisis de sensibilidad multivariante en términos de costos y efectividad incrementales de SFC vs montelukast. $1,500 $3,000 $4,500 $6,000 $7,500 Disponibilidad a pagar MON SFC $9,000 Gráfico. 2 – Curva de aceptabilidad de análisis de costo efectividad de salmeterol/fluticasona versus montelukast en el tratamiento del asma. VALUE IN HEALTH 14 (2011) S43–S47 Se ha comprobado que la adición de un agonista 2 de larga acción al régimen con corticosteroides, como es el SFC disminuye la medicación de rescate, aumenta los días libre de síntomas y aún cuando no disminuye la prevalencia de exacerbaciones, sí ejerce una acción sobre la severidad de estas. Estos beneficios, como han sido reportados son relevantes para seleccionar una u otra opción de tratamiento [25]. Dado que el uso de agonistas 2 de larga acción no se asocia a largo plazo al deterioro en el control de los síntomas, el SFC representa un opción adecuada para el tratamiento de individuos de asma persistente moderada que no respondan adecuadamente a los corticosteroides, o en aquellos individuos que requieren aumentar las dosis de estos para lograr su control adecuado, como son los niños asmáticos [26]. El adecuado control del asma contribuye a disminuir los costos totales del tratamiento, específicamente las visitas a urgencias e ingresos hospitalarios. Los programas de atención farmacéutica, deben utilizar el fármaco más eficaz y seguro como el SFC, que contribuye a aumentar la calidad de vida y una reducción del costo de la atención sanitaria. Aunque es conocida la dificultad de extrapolar los resultados de estudios farmacoeconómicos de un país a otro, los efectos de esta investigación adaptada al entorno sanitario mexicano son consistentes con las evaluaciones realizadas para el SFC en el contexto internacional [27,28]. Se arribó a la conclusión de que SFC administrado en un mismo dispositivo, además de eficaz, resulta una intervención costo-beneficiosa por sus ventajas económicas [29,30]. Se deben tener en cuenta una serie de limitaciones y fortalezas del estudio. En primer lugar, se trata de un modelo teórico y en segundo lugar, los porcentajes de pacientes que logran un control total del asma y los días libres de síntomas se obtuvieron de ensayos clínicos no pragmáticos, por lo que sus resultados deben considerarse como estimaciones para un paciente típico. Por ello, fue necesario hacer varias asunciones consensuadas por el panel de expertos sobre el número de consultas médicas en pacientes controlados y no controlados, la duración de hospitalizaciones y las urgencias, y los porcentajes de pacientes que acuden a consultas porque requieren hospitalización. Para minimizar las limitaciones del modelo, se tomaron supuestos conservadores o valores promedio y se hizo un análisis de sensibilidad considerando varios escenarios con valores extremos, que confirmaron los resultados del caso básico. Como fortalezas del modelo, la estimación de la utilización de recursos se hizo por consensos de expertos y los costos unitarios se obtuvieron del IMSS. Además, los principales supuestos fueron sometidos a análisis de sensibilidad probabilístico, que confirmaron la estabilidad del estudio. Conclusiones El análisis demostró que en pacientes pediátricos con asma, la utilización de SFC es una medida costo-efectiva desde el punto de vista del sistema público sanitario mexicano. El tratamiento con SFC representó la mejor opción para el asma en comparación con MON, tanto en efectividad como en costos, por lo que resulto ser la opción dominante al presentar una mayor eficacia y al menor costo. Fuentes de financiamiento: Este estudio fue patrocinado para su realización por los Laboratorios Glaxo-SmithKline de México. REFERENCIAS [1] Braman SS. The global burden of asthma. Chest 2006;130(Suppl.):4S–12S. [2] Estudio Europeo del Asma. Prevalencia de hiperreactividad bronquial y asma en adultos jóvenes de cinco áreas españolas. Grupo Español del Estudio Europeo del Asma. Med Clin (Barc.) 1996;106:761–7. [3] Aguinaga OI, Arnedo PA, Bellido J, Guillen GF, et al. Prevalencia de síntomas relacionados con el asma en niños de 13–14 años de 9 poblaciones españolas. Grupo Español del Estudio ISAAC . Med Clin (Barc.) 1999;112:171–5. S47 [4] Callén M, Alústiza E, Solórzano C, et al. Prevalencia de asma y factores de riesgo de asma en Guipúzcoa: estudio multicéntrico caso-control. An Esp Pediatr 1995;43:347–50. [5] Grupo Español para el Manejo del Asma. Guía española para el manejo del asma. Arch Bronconeumol 2003;39(Suppl. 5):1– 42. [6] Rico FG, Barquera S, Cabrera DA, Escobedo S. Bronquial asthma healthcare costs in Mexico: analysis of trends from 1991-1996 with information from the Mexican Institute of Social Security. Invest Allergol Clin Immunol 2000;10:334 – 41. [7] Pan American Health Organization. Health in the Americas. Washington, DC: PAHO; 2002. [8] Global Initiative for Asthma. Global strategy for asthma management and prevention. Disponible en: http://www.ginasthma.com/. [Acceso julio 2004]. [9] Scottish Intercollegiate Guidelines Group. A guideline developers’ handbook. Publication No. 50. Edinburgh: SIGN; 2001. [10] British Guideline on the Management of Asthma. A national clinical guideline. Revisted edition April Edinburgh: BGMA; 2004. [11] Guía de Práctica Clínica de Asma. Comarca Gipuzkoa Este. Servicio Vasco de Salud editor. País Vasco: Osakidetza; 1999. [12] Maspero J, Guerra F, Cuevas F, Gutierrez JP, et al. Efficacy and tolerability of salmeterol/fluticasone propionate versus montelukast in childhood asthma: A prospective randomized, double-blind, double-lummy, paralel group study. Clin Ther 2008;30:1492–504. [13] Calhourn W, Nelson H, Nathan R, et al. Comparison of fluticasone propionate-salmeterol combination therapy and montelukast in patients who one symptomatic on short-acting 2-agonist alone. Am J Respir Crit Care Med 2001;164:759 – 63. [14] Pavord I, Woodcook A, Parker D, Rice L. Salmeterol plus flucasone propionate versus fluticasone propionate plus montelukast: a randomized controlled trial investigating the effects on among inflammation in asthma. Respir Res 2007;8:67. [15] Powell C. The Delphi technique: myths and realities. J Adv Nurs 2003; 41:376 – 82. [16] McKenna HP. The Delphi technique: a worthwile approach for nursing? J Adv Nurs 1994;19:1221–5. [17] Instituto Nacional de Salud Pública de México. Propuesta de guías para la conducción de estudios de evaluación económica como parte de la actualización de los cuadros básico de insumos para la salud del Consejo de Salubridad General.[Informe]Ciudad de México D. F:Consejo de Salubridad General; 2007. [18] Secretaría General del IMSS. Costos unitarios para la determinación de créditos fiscales derivados de capitales constitutivos, inscripciones improcedentes y atención a no derechohabientes. [Informe] Ciudad de México D. F: Diario Oficial de la Federación, 2009. [19] Global Initiative for Asthma. Global strategy for asthma management and prevention. Rev. NIH publication; no. 02-3659. Bethesda, Md: National Institutes of Health, National Heart, Lung, and Blood Institute, 2002. [20] Antoñanzas F. Evaluación económica aplicada a los medicamentos: características y metodología. In: Sacristán J, Badía X, Rovira J. (eds). Farmacoeconomía: evaluación económica de los medicamentos. Madrid: Editores Médicos, 1995. [21] Briggs AH, Goeree R, Blackhouse G, O’Brien B. Probabilistic analysis of cost-effectiveness models: choosing between treatment strategies for gastroesophageal reflux disease. Med Decis Making 2002;22:290 –308. [22] Briggs AH. Handling uncertainty in cost-effectiveness models. Pharmacoeconomics 2000;17:479 –500. [23] Claxton K, Schupher M, McCabe C, et al. Probabilistic sensitivity analysis for NICE technology assessment: not an optional extra. Health Econ 2005;14:339 – 47. [24] Van den Berg NJ, Ossip MS, Hederos CA, et al. Salmeterol/fluticasone propionate (50/100 microg) in combination in a Diskus inhaler (Seretide) is effective and safe in children with asthma. Pediatr Pulmonol 2000;30:97–105. [25] Price MJ, Briggs AH. Development of an economic model to assess the cost effectiveness of asthma management strategies Pharmacoeconomics 2002;20:183–94. [26] Barnes PJ. Scientific rationale for inhaled combination therapy with long-acting beta2-agonists and corticosteroids Eur Respir J 2002;19:182–91. [27] Lyseng-Williamson KA, Plosker GL. Inhaled salmeterol/fluticasone propionate combination: a pharmacoeconomic review of its use in the management of asthma. PharmacoEconomics 2003;21:951– 89. [28] Sheth K, Borker R, Emmett A, et al. Cost-effectiveness comparison of salmeterol/fluticasone propionate versus montelukast in the treatment of adults with persistent asthma. Pharmacoeconomics 2002;20:909–18. [29] O’Connor RD. Nelson H. Borker R, et al. Cost effectiveness of fluticasone propionate plus salmeterol versus fluticasone propionate plus montelukast in the treatment of persistent asthma. Pharmacoeconomics 2004;22:815–25. [30] Pieters WR, Wilson KK, Smith HC, et al. Salmeterol/fluticasone propionate versus fluticasone propionate plus montelukast: a costeffective comparison for asthma. Treat Respir Med 2005;4:129 –38. VALUE IN HEALTH 14 (2011) S48 –S50 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Costos de la Esclerosis Múltiple en Colombia Martin Romero, MD, MSc1,*, Carlos Arango, MD, MSc1, Nelson Alvis, MD, MSc, PhD2, Juan Camilo Suarez, MD3, Aristides Duque, MD, MEsp (Neurol)4 1 Fundación Salutia Centro de investigaciones económicas, de gestión y tecnologías en salud, Bogotá, Colombia; 2Departamento de Investigaciones Económicas y Sociales, Facultad de Ciencias Económicas, Universidad de Cartagena, Cartagena, Colombia; 3Instituto Neurológico de Colombia, Bogotá, Colombia; 4 Neurólogo – Fundación Santafé de Bogotá, Bogotá, Colombia A B S T R A C T Objective: To estimate, according to the states of disease (remission or relapse) and her level of progression (EDSS), the cost of treatment of Multiple Sclerosis (MS) in Colombia. Methods: From the perspective of the third payer, a cost study of MS was performed using two-way estimation techniques: a) “Top down” to estimate the costs during relapses from clinical registers of 304 patients; b) “bottom-up” to estimate the cost in remission from a questionnaire (Kobelt 2006) applied to 137 patients, located in different regions of Colombia. Results: The mean of patient’s age was 43,7 years and 73% of those were women. The mean annual cost per patient varied according to the disease phase, finding the highest value in Phase II (EDSS 3 – 5,5) with $ 50.581.216 COP (US$ 25.713) and the lowest one in Phase IV (EDSS 8 - 9,5) with $20.738.845 COP (US$ 10.543). The cost of Disease Modifier Drugs (DMD) represented 91.5% from the medial total annual cost of 1,2 and 3 Introducción La EM es una enfermedad del sistema nervioso central que produce una degeneración axonal [1]. Se estima que esta enfermedad afecta más de un millón de personas en el mundo [2]. Se ha demostrado que es menos frecuente en las regiones tropicales, sin que se hayan identificado los factores que lo explican [3,4]. En Colombia, se han encontrado prevalencias semejantes a otras poblaciones de bajo riesgo en el mundo [5,6], variando entre 1,48 en Antioquia, 4,98 en Risaralda [7] y 4,41 en Bogotá [8]. La EM es la segunda causa de discapacidad neurológica en adultos jóvenes [9]. En países donde la prevalencia es alta, el impacto económico de la enfermedad ha sido ampliamente estudiado [10 –29]. En Colombia, a pesar de su baja prevalencia, el impacto económico es marcado. Para el periodo 2002-2005, la EM fue el diagnóstico que representó el mayor valor recobrado al FOSYGA (FOSYGA Fondo de Solidaridad y Garantías. Fondo para el manejo de recursos del SGSSS) con un monto de cerca de $ 28 mil millones de pesos de 1998, cerca de 19,5 millones de phases. The participation of DMD was 58%.in the 4 phase. Indirect costs are minimal participation in all phases, except for 4 where increases at the expense of reduced consumption of DMD. Costs associated with the period of relapses of MS are low against the total cost, with an average cost of $ 2,433,182 COP ($ 1.237USD). Discussion: MS in Colombia is a disease with a behavioral pathology ”high cost ” to the social security system (SGSSS), generated mainly at the expense of their direct costs, which, even without including relapses, an aggregate amount of more than 75 times the annual premium cost of health insurance for Colombia ($ 430,488 COP) in the year under review (2008) Palabras Claves. costos, EDSS, esclerosis múltiple, recaídas, remisión. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. dólares del mismo año (cerca del 80% de lo que Colombia gastó en vacunas para el Programa Nacional de Inmunizaciones) [30,31]. No se conocen estudios del costo de la EM en Colombia aunque hay publicaciones asociadas: guías de manejo [32], artículos de revisión [33], un editorial sobre la enfermedad [34] y artículos sobre prevalencia [7,8]. Materiales y Métodos Se llevó a cabo un estudio de costos de la enfermedad desde la perspectiva del tercero pagador. La descripción de costos se hizo sobre pacientes ya diagnosticados considerando todos los consumos asociados al tratamiento (atenciones en salud como hospitalización, consultas, etc.) incluido el tratamiento de complicaciones y los medicamentos modificadores de la enfermedad (Disease Modifying Drugs, DMD) y los costos indirectos (incapacidades y pensiones, sin incluir las pérdidas de ingresos por reducción de la actividad laboral de pacientes y cuidadores) [35,36].Se estimaron los costos para las recaídas y los periodos de remisión. No se incluyó el costo de comorbilidades. Conflicts of interest: The authors declare total independence in this investigation. However, they state that they received an allowance from Bayer SA, which partially funded the research. Título corto: Costos de la esclerosis múltiple en Colombia * Autor de correspondencia: Martin Romero, Fundación Salutia Centro de investigaciones económicas, de gestión y tecnologías en salud y protección social, Carrera 71 B N 116 A 12, Bogotá – Colombia. Tel: ⫹57 1 613 4609. Fax: ⫹ 571 6179133. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.023 VALUE IN HEALTH 14 (2011) S48 –S50 Para la captura de la información se utilizó un diseño basado en la combinación de dos técnicas de costeo: “top-down” y “bottomup” [13,37,38]. En la primera, la información se obtiene de los registros de los hospitales de donde se tomaron los datos de atención durante las recaídas. En la segunda, la información se obtiene directamente de los pacientes durante los periodos de remisión mediante encuesta. Esta última, se ha utilizado por Kobelt para estimar costos de EM [12,13,15,19 –28], la cual se adaptó para Colombia con autorización de la autora. Los consumos encontrados en las encuestas fueron calculados a valores de mercado de 2008. Los precios de los DMD fueron estimados de acuerdo al valor autorizado por el Gobierno como pago en reembolsos. Con base en las prevalencias de la EM en Colombia [7], se calcularon 2.135 pacientes para 2008. Así se valoró, como tamaño apropiado de muestra (nivel de confianza 95% y error máximo del 5.5%), en 277 casos. Finalmente, se incorporaron al estudio historias clínicas de 304 pacientes (27 por encima de la muestra estimada) dentro del periodo 2003 a 2008, en 3 hospitales del país. Para la captura de datos de los periodos de remisión se entrevistaron 137 pacientes. Se incluyeron hospitales que prestan servicios de alto grado de complejidad, que cuentan con servicio de neurología y un sistema electrónico de registros clínicos. Las Historias Clínicas incluyeron los casos que correspondían al código G35 en el CIE 10 y cuyo diagnóstico era realizado por un médico neurólogo, con una fecha de diagnóstico no inferior a 2 años. Las entrevistas consideraron pacientes con diagnóstico confirmado de EM y que dieron su “consentimiento informado”. Para medir la discapacidad por EM se usó la escala Expanded Disability Status Scale (EDSS) (Kurtzke 1955-1983) [39 – 41]. Para efectos de análisis de los pacientes se agruparon en 4 fases: a) Fase 1 - Sin limitaciones: EDSS 0-2,5; b) Fase 2 - Con limitaciones moderadas: EDSS 3-5,5; c) Fase 3 - Con apoyo al caminar o en sillas de ruedas: EDSS 6-7,5 y d) Fase 4 - Restringido en Cama: EDSS 8-9,5. La información fue capturada por profesionales de la salud en formularios validados y cargada en una base de datos donde se cegaron los nombres de los pacientes. La información hospitalaria fue generada en archivos planos y analizada en hojas de cálculo utilizando Microsoft Excel®, dada su aplicabilidad para el nivel de análisis estadístico requerido de tipo descriptivo. Resultados La valoración de los costos de las recaídas se realizó con la data hospitalaria y para los periodos de remisión la obtenida de la encuesta. Las características básicas y distribución de los pacientes se muestran en la Tabla 1 en Materiales Complementarios en: doi: 10.1016/j.jval.2011.05.023. Los medicamentos clasificados como DMD, explican el 91,5% promedio de los costos de las fases 1, 2 y 3 de la EM (Tabla 2 en Materiales Complementarios en: doi:10.1016/j.jval.2011.05.023). En la fase 4, se reduce la presencia de los DMD y por tanto su costo desciende a una participación dentro de la estructura de costos de esta fase. La composición de la estructura de los costos directos de la EM en los periodos de remisión, sin incluir los DMD, corresponden a un valor relativamente homogéneo para las fases 1,2 y 3 que fluctúa entre un valor mínimo de 2,8% en la fase 4 y un valor mayor de 5,4% en la fase 2. La estructura de los costos indirectos fluctuó dentro de un rango estrecho entre un valor mínimo de 4% en la fase 1 y máximo de 4,8% en la fase 3. Sin embargo la fase 4 mostró un comportamiento elevado, con un valor de 39,1% por efecto de la reducción del costo directo de los DMD en esta fase (Tabla 2). En la Tabla 3 en Materiales Complementarios en: doi:10.1016/ j.jval.2011.05.023 se muestra la distribución de los costos de los periodos de recaídas de la EM por hospital participante y en la S49 tabla 4 se presenta la composición de los costos por tipo de atención prestada. Los costos de las estancias, los medicamentos y las imágenes diagnósticas explican el 80% de la estructura de costos de atención de las recaídas según se ve en la tabla 4. El costo total de las recaídas es mínimo frente al total del costo del periodo de remisión. Discusión Esta es una de las primeras aproximaciones a la descripción de los costos de la EM en países en “vías de desarrollo” y el único realizado en Colombia, utilizando métodos de estudio ya probados en otros países como ocurre con la identificación de los costos de la EM utilizando un mecanismo de “doble vía” [42,43], Los costos directos (la suma de costos atención en salud sin DMD y costos DMD) oscilan entre un 60,9 % en la fase 4 y un 96,0% en la fase 2. Como se aprecia en la tabla 2, la casi totalidad de estos costos, son explicados por los DMD, cifra mayor a la encontrada en países como Estados Unidos donde los costos Directos de atención fueron el 62,8% del total de costos y donde un 34% eran DMD [20] o en España (2006) en donde todos los medicamentos ocuparon más del 86% del total de costos directos y en donde un 52.4% correspondían a DMD [26]. Las diferencias en las estructuras de los costos de atención en salud en EM entre países se explica probablemente por los diferenciales de precios como ocurre en el Reino Unido, en donde se observa que el costo directo de los servicios de salud es cinco veces más elevado que en Colombia, mientras que por el contrario los costos de los medicamentos corresponden apenas a una tercera parte de los costos en Colombia [44 – 46]. A pesar de que las recaídas son la principal expresión de la EM, sus costos tienen un impacto relativo menor sobre el costo total, lo que se explica porque los DMD están indicados solo durante la remisión .Los principales componentes del costo de la recaída (Tabla 4 en Materiales Complementarios en: doi:10.1016/j. jval.2011.05.023) son las Imágenes Diagnósticas (32,41%) y los Medicamentos (25,38%), similar a lo reportado por O’Brien (2006) [47]. Los costos de las recaídas por fases EDSS no se obtuvieron, dadas las restricciones de las fuentes utilizadas, lo que constituye una limitación del estudio. La EM es una enfermedad clasificada de “alto costo” en Colombia [48], dado que alcanza un costo cuyos valores son al menos 75 veces superiores al valor de la prima de salud (UPC) per cápita: para 2008 el valor de cada UPC pagada fue de $430.488 COP ($218,9 USD). Agradecimientos A los hospitales participantes, a las Fundaciones FUNDEM y ALEM. Los autores agradecemos la participación de la Fundación de Esclerosis Múltiple y otras enfermedades- FUNDEM, Fundación Instituto Neurológico de Antioquia-INDEA y la Fundación Instituto Neurológico de Colombia pues sin su colaboración no hubiese sido posible obtener los resultados aquí presentados. Los autores actuaron con total independencia en todas las fases del estudio. Materiales Complementarios Material complementario que acompaña este artículo se puede encontrar en la versión en línea como un hipervínculo en doi: 10.1016/j.jval.2011.05.023 o si es un artículo impreso, estará en www.valueinhealthjournal.com/issues (seleccione el volumen, número y artículo). S50 VALUE IN HEALTH 14 (2011) S48 –S50 REFERENCIAS [1] Medina-Redondo F, Herrera-Carranza J, Sanabria C, et al. [The efficiency and cost-utility ratio of interferon beta in the treatment of multiple sclerosis in Andalusia]. Rev Neurol 2004;39:1– 6. [2] Miltenburger C, Kobelt G. Quality of life and cost of multiple sclerosis. Clin Neurol Neurosurg. 2002;104:272–5. [3] Noseworthy JH, Lucchinetti C, Rodriguez M, Weinshenker BG. Multiple sclerosis. N Engl J Med 2000;343:938 –52. [4] Kurtzke JF. Multiple sclerosis: changing times. Neuroepidemiology 1991;10:1– 8. [5] Cooper GS, Stroehla BC. The epidemiology of autoimmune diseases. Autoimmun Rev 2003;2:119 –25. [6] Jacobson DL, Gange SJ, Rose NR, Graham NM. Epidemiology and estimated population burden of selected autoimmune diseases in the United States. Clin Immunol Immunopathol 1997;84:223– 43. [7] Sanchez JL, Aguirre C, Arcos-Burgos OM, et al. [Prevalence of multiple sclerosis in Colombia]. Rev Neurol 2000;31:1101–3. [8] Toro J, Sarmiento OL, Diaz del Castillo A, et al. Prevalence of multiple sclerosis in Bogota, Colombia. Neuroepidemiology 2007;28:33– 8. [9] Casado V, Martinez-Yelamos S, Martinez-Yelamos A, et al. Direct and indirect costs of Multiple Sclerosis in Baix Llobregat (Catalonia, Spain), according to disability. BMC Health Serv Res 2006;6:143. [10] Phillips CJ. The cost of multiple sclerosis and the cost effectiveness of disease-modifying agents in its treatment. CNS Drugs 2004;18:561–74. [11] Taylor B, McDonald E, Fantino B, Sedal L, MacDonnell R, Pittas F, et al. The cost of multiple sclerosis in Australia. J Clin Neurosci 2007; 14:532–9. [12] Kobelt G, Pugliatti M. Cost of multiple sclerosis in Europe. Eur J Neurol 2005;12(Suppl. 1):63–7. [13] Kobelt G, Berg J, Lindgren P, Jonsson B. Costs and quality of life in multiple sclerosis in Europe: method of assessment and analysis. Eur J Health Econ 2006;7(Suppl. 2):S5–13. [14] Rieckmann P. [Early multiple sclerosis therapy in the effects of public health economics]. Med Klin (Munich) 2001;96(Suppl. 1):17–21. [15] Berg J, Lindgren P, Fredrikson S, Kobelt G. Costs and quality of life of multiple sclerosis in Sweden. Eur J Health Econ 2006;7(Suppl. 2): S75– 85. [16] De Judicibus MA, McCabe MP. The impact of the financial costs of multiple sclerosis on quality of life. Int J Behav Med 2007;14:3–11. [17] Ganzinger U, Badelt C, Vass K, et al. [Health care costs of multiple sclerosis in Austria. Cross-sectional study including consideration of quality of life]. Nervenarzt 2004;75:1000 – 6. [18] Henriksson F, Fredrikson S, Masterman T, Jonsson B. Costs, quality of life and disease severity in multiple sclerosis: a cross-sectional study in Sweden. Eur J Neurol 2001;8:27–35. [19] Kobelt G. Costs and quality of life for patients with multiple sclerosis in Belgium. Eur J Health Econ 2006;7(Suppl. 2):S24 –33. [20] Kobelt G, Berg J, Atherly D, Hadjimichael O. Costs and quality of life in multiple sclerosis: a cross-sectional study in the United States. Neurology 2006;66:1696 –702. [21] Kobelt G, Berg J, Lindgren P, et al. Costs and quality of life in multiple sclerosis in The Netherlands. Eur J Health Econ 2006;7(Suppl. 2):S55– 64. [22] Kobelt G, Berg J, Lindgren P, et al. Costs and quality of life of multiple sclerosis in Italy. Eur J Health Econ 2006;7(Suppl. 2):S45–54. [23] Kobelt G, Berg J, Lindgren P, et al. Costs and quality of life of multiple sclerosis in Germany. Eur J Health Econ 2006;7(Suppl. 2):S34 – 44. [24] Kobelt G, Berg J, Lindgren P, et al. Costs and quality of life of patients with multiple sclerosis in Europe. J Neurol Neurosurg Psychiatry 2006; 77:918 –26. [25] Kobelt G, Berg J, Lindgren P, et al. Costs and quality of life of multiple sclerosis in Switzerland. Eur J Health Econ 2006;7(Suppl. 2):S86 –95. [26] Kobelt G, Berg J, Lindgren P, et al. Costs and quality of life of multiple sclerosis in Spain. Eur J Health Econ 2006;7(Suppl. 2):S65–74. [27] Kobelt G, Berg J, Lindgren P, et al. Costs and quality of life of multiple sclerosis in the United Kingdom. Eur J Health Econ 2006;7(Suppl. 2): S96 –104. [28] Kobelt G, Berg J, Lindgren P, et al. Costs and quality of life of multiple sclerosis in Austria. Eur J Health Econ 2006;7(Suppl. 2):S14 –23. [29] McCrone P, Heslin M, Knapp M, et al. Multiple sclerosis in the UK: service use, costs, quality of life and disability. Pharmacoeconomics 2008;26:847– 60. [30] Cubillos L, Alfonso E. Análisis descriptivo preliminar de los recobros en el SGSSS 2002 a 2005 Bogotá D.C.: PARS. Ministerio de la Protección Social 2006. [31] Pinto Masis D, Castellanos M. Caracterización de los recobros por tutelas y medicamentos no incluidos en los POS. Gerencia y Políticas de Salud 2004;3:40 – 61. [32] Espinosa E, Pérez J C. Guías de manejo de esclerosis múltiple en niños. Acta Neurol Colomb 2002;18:52– 8. [33] Gutiérrez-Álvarez Á. Esclerosis múltiple: evidencias y controversias. Rev Cienc Salud 2006 enero-junio de 2006;4:52– 8. [34] Pradilla G, León-Sarmiento F. Esclerosis múltiple en Colombia: cerrando la brecha. Acta Neurol Colomb 2007;23:3–5. [35] Birnbaum HG, Ivanova JI, Samuels S, et al. Economic impact of multiple sclerosis disease-modifying drugs in an employed population: direct and indirect costs *. Curr Med Res Opin 2009;25(4): 869 – 877. [36] Trojano M, Paolicelli D, Fuiani A, et al. Postmarketing evidence of disease-modifying drugs in multiple sclerosis. Neurol Sci 2008;29(Suppl. 2):S225– 6. [37] Travis JF. Bottom-up cost accounting system a superior method for controlling healthcare costs. Mod Healthc 1987;17:50. [38] Chapko MK, Liu CF, Perkins M, et al. Equivalence of two healthcare costing methods: bottom-up and top-down. Health Econ 2008;18: 1188 –1201. [39] Kurtzke JF. Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS). Neurology 1983;33:1444 –52. [40] Gaspari M, Roveda G, Scandellari C, Stecchi S. An expert system for the evaluation of EDSS in multiple sclerosis. Artif Intell Med 2002;25: 187–210. [41] Kurtzke JF. A new scale for evaluating disability in multiple sclerosis. Neurology 1955;5:580 –3. [42] Durelli L, Verdun E, Barbero P, et al. Every-other-day interferon beta1b versus once-weekly interferon beta-1a for multiple sclerosis: results of a 2-year prospective randomised multicentre study (INCOMIN). Lancet 2002;359:1453– 60. [43] Prosser LA, Kuntz KM, Bar-Or A, Weinstein MC. Cost-effectiveness of interferon beta-1a, interferon beta-1b, and glatiramer acetate in newly diagnosed non-primary progressive multiple sclerosis. Value Health 2004;7:554 – 68. [44] PSSRU. Unit Costs of Health and Social Care 2008. London: Personal Social Services Research Unit; 2008 [updated 2008; cited 2009 May]; Available from: http://www.pssru.ac.uk/pdf/uc/uc2008/uc2008.pdf. [Accessed March 8, 2011]. [45] bnf.org. British National Formulary. London; 2009 [updated 2009; cited May 2009]; Available from: http://www.bnf.org/bnf/bnf/57/ 37092.htm?q⫽%22betaferon%22. [Accessed March 8, 2011]. [46] PLM FARMAPRECIOS. Colombia. Marzo- Abril de 2009. [47] O’Brien JA, Ward AJ, Patrick AR, Caro J. Cost of managing an episode of relapse in multiple sclerosis in the United States. BMC Health Serv Res 2003;3:17. [48] David I, Medina A, Martinez E. Enfermedades de alto costo en afiliado a un sistema institucional de aseguramiento y prestación de servicios de salud Rev Fac Nac Salud Pública 2006;24:98 –105. VALUE IN HEALTH 14 (2011) S51–S59 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Development and Validation of a Microsimulation Economic Model to Evaluate the Disease Burden Associated with Smoking and the CostEffectiveness of Tobacco Control Interventions in Latin America Andres Pichon-Riviere, MD, MSc, PhD1,2,*, Federico Augustovski, MD, MSc1,2, Ariel Bardach, MD, MSc1, Lisandro Colantonio, MD, MSc1, for the Latinclen Tobacco Research Group3 1 Institute for Clinical Effectiveness and Health Policy, Buenos Aires, Argentina; 2School of Public Health, University of Buenos Aires, Buenos Aires, Argentina; LatinCLEN Tobacco Research Group: Argentina (Institute for Clinical Effectiveness and Health Policy): Andres Pichon-Riviere, MD, MSc, PhD, Adolfo Rubinstein, MD, MSc, PhD, Federico Augustovski, MD, MSc, Ariel Bardach, MD, MSc, Lisandro Colantonio, MD, Andrea Alcaraz, MD, Sebastian Garcia-Marti, MD, Luz Gibbons, MSc; Argentina (Clinical Epidemiology Unit–University of Tucuman): Evelina Chapman, MD, MSc, Ramon Gonzalez, MD, Sara Aulet, MD; Bolivia (Clinical Epidemiology Unit–San Andres University): Maria del Pilar Navia Bueno, MD, MSc, Lijia Elizabeth Avilés Loayza, MD, MSc; Brazil (Universidade Federal do Rio de Janeiro): Antonio Jose Ledo Alves da Cunha, MD, MPH, PhD, Alberto José de Araujo, MD, MSc; Chile (Research and Training Center in Clinical Epidemiology–Frontera University): Sergio Muñoz, MS, PHD, Carlos Vallejos, MD, MS; Colombia (Clinical Epidemiology & Biostatistics Unit– Pontificia Universidad Javeriana): Juan Manuel Lozano Leon, MD, MSc, Esperanza Peña Torres, RN, MSc; Mexico (Clinical Epidemiology Unit–Instituto Mexicano del Seguro Social–Faculty of Medicine): Fernando Carlos Rivera, MSc, Araceli Camacho Chairez, MSc, Araceli Aguirre Granados, Patricia Clark Peralta, MD, MSc, PhD; Peru (Clinical Epidemiology Unit–Universidad Peruana Cayetano Heredia): Leandro Huayanay Falconi, MD, MSc, Cezar Loza Munárriz, MD 3 A B S T R A C T Objective: To describe the development and validation of a health economic model (HEM) to address the tobacco disease burden and the cost-effectiveness of smoking cessation interventions (SCI) in seven Latin American countries. Methods: The preparatory stage included the organization of the research network, analysis of availability of epidemiologic data, and a survey to health decision makers to explore country-specific information needs. The development stage involved the harmonization of a methodology to retrieve local relevant parameters and develop the model structure. Calibration and validation was performed using a selected country dataset (Argentina 2005). Predicted event rates were compared to the published rates used as model inputs. External validation was undertaken against epidemiologic studies that were not used to provide input data. Results: Sixty-eight decision makers were surveyed. A microsimulation HEM was built considering the availability and quality of epidemiologic data and relevant outcomes conceived to suit the identified information needs of Introduction Smoking is the single most preventable cause of disease and death worldwide, and this burden is increasingly shifting from upper to lower and middle-income countries. In the year 2000 there were 4.83 million premature tobacco-related deaths [1], and this number is expected to grow to 10 million per year by 2030 [2,3]. Currently half of the current tobacco-attributable deaths occur in high-income countries [1,3]; however, by the year 2030 7 out of 10 of these deaths are expected to occur in developing countries. This represents one out of six of all the deaths around the world [3]. decision makers. It considers all tobacco-related diseases (i.e., heart, cerebrovascular and chronic obstructive pulmonary disease, pneumonia/influenza, lung cancer, and nine other neoplasms) and can incorporate individual- and population-level interventions. The calibrated model showed all simulated event rates falling within ⫾ 10% of the sources (-9%–⫹5%). External validation showed a high correlation between published data and model results. Conclusions: This evidencebased, internally and externally valid HEM for the assessment of the effects of smoking and SCIs incorporates a broad spectrum of tobacco related diseases, SCI, and benefit measures. It could be a useful policymaking tool to estimate tobacco burden and cost-effectiveness of SCI. Keywords: cost-effectiveness, cost-utility, disease burden, economic model, Latin America, Monte Carlo microsimulation, smoking cessation interventions, tobacco, validation. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. Although the Framework Convention Tobacco Control from World Health Organization has been signed by almost every country in the Latin American region [4], tobacco control policies are still scarce in these countries. The lack of quality information related to the health and economic consequences of tobacco use in our region is an important barrier for the implementation of evidence-based tobacco control policies. This has led to a biased assessment by policy makers, resulting in a distorted prioritization of health policies where tobacco control interventions are considered less urgent than action on other diseases [5]. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Andres Pichon-Riviere, Institute for Clinical Effectiveness and Health Policy, University of Buenos Aires, Arevalo, 1457 (1414), Buenos Aires, Argentina. E-mail address: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.010 S52 VALUE IN HEALTH 14 (2011) S51–S59 Model-based health economic evaluations (HEEs) are widely accepted as decision-making tools [6] that can provide valuable information for the optimization of health resource allocation [7]. Although in many developed countries this “fourth hurdle” based on health economic evidence is required to shape health policies [8], there is still little experience in Latin America [9]. This project constitutes a collaboration among seven Latin American countries that aims to provide relevant evidence to inform tobacco control policies. The LatinCLEN Tobacco Research Group, an international and interdisciplinary network, is composed of eight research units from the Latin American chapter of the International Clinical Epidemiology Network in Argentina, Bolivia, Brazil, Chile, Colombia, Mexico, and Peru. The specific aims for this project were to select and develop the most suitable methodologic framework, as well as to elaborate a common health economic model to estimate the smoking-related disease burden and the cost-effectiveness of smoking cessation interventions (SCIs). In this article we present the details of the model’s development, structure, and validation using data inputs from one of the participating countries. Material and Methods The final model structure and its inputs were agreed after two stages; the preparatory stage and the development stage. Preparatory stage Organization of a research network to monitor the model building process and to ensure the generalizability to all participant countries. The LatinCLEN Tobacco Research Group was formed in 2004. The group designed two surveys that were completed in each country to 1) evaluate the availability, cost, and current coverage policies of SCIs; and 2) assess the availability and the quality of relevant information to be incorporated in each country-specific analysis (e.g., local epidemiology and cost of smoking-related diseases). Performance of a rapid systematic review of existing health economic evaluations regarding tobacco cessation strategies. Fortyfour individual studies and seven reviews published between 1984 and 2003 (search data up to November 2004) were critically assessed. Design and administration of a survey to health decision makers to explore country-specific information needs when deciding on the implementation and coverage of SCIs. As future users of the HEE, relevant decision makers from the different health sectors in the seven participant countries defined the key aspects to be considered (e.g., relevant time horizon and relevant perspective) and the outcomes to be reported (e.g., the number of cases prevented, life years gained, or quality-adjusted life years [QALYs]) in the HEE. Development stage This stage involved the following tasks: 1) definition of the methodology for the information source selection and parameter incorporation; 2) development of the model structure; and 3) calibration and validation of the model. The research group used Email and a Web-based platform to exchange documents, outlines, and ideas. The development of the model was completed in three phases: 1) based on information obtained in the preparatory stage, a first draft of the health economic model was sent to the participating countries for feedback, including the basic structure, disease states to be incorporated, and main assumptions; 2) three consultation rounds for refining the model description and structure; and 3) a face-to-face research meeting carried out in Buenos Aires, Argentina, during November 2006 where participants agreed on the final version of the HEE. Excel (Professional Edition 2003, then updated to version 2007, Microsoft Corp., Redmond, WA) with Visual Basic Macros (version 6.3, Microsoft Corp., Redmond, WA) was selected as the model platform to easily share the information. A software package was installed to improve the original Excel’s random number generator function [10,11]. Results Preparatory stage The surveys and data retrieved in each country showed that implementation of tobacco control interventions was a relevant issue in the region. Sixty-eight decision makers (9 –10 from each of the participating countries) completed the survey. The majority of decision makers belonged to the public (56%) or the social security (25%) health care sector and 80% considered that the lack of coverage for SCI adversely affected the prevalence of smoking in their institutions and countries. Ninety-three percent considered that this level of coverage should be increased and 83.3% believed that SCI should be included in the national lists or basket of mandatory coverage in their countries. When asked about what would be the most relevant information on the interventions needed when having to decide their incorporation into the health system, most of the decision makers identified the cost per QALY, cost per life-year saved, and budget impact information. The decision makers also mentioned a wide range of interventions that they considered should be evaluated for coverage, from population-wide interventions to pharmacological treatments, so the HEE had to be able to include all these aspects. On the other hand, the survey of epidemiologic data showed that the availability and quality of information in the region was very heterogeneous and poor, especially with regard to the incidence of events, thus making it necessary to design and harmonize a methodology to estimate locally relevant information in each country. All the results obtained during this first stage strongly influenced many of the decisions made later and shaped the type and structure of the model to be developed. Development stage Information source selection and parameter incorporation. We defined a decision rule that would establish a priority order among the possible data sources to populate the model: 1) use good quality local (country-specific) sources when available [12-14]; 2) use international sources when local data were unavailable or poor and when the parameter was considered transferable from other settings; or 3) derive or estimate the parameter from the best available local data when international sources were considered nontransferable. Special attention was paid to the estimation of baseline disease event incidence in nonsmokers because these data are keys to the generalizability of the model. Given the low availability of information encountered in the region, we defined a common methodology, anchored on national health statistics, to derive these parameters from mortality data. This methodologic assumption linking mortality to incidence data is a widely used assumption in epidemiologic and health economic models, used by the World Health Organization in tools such as DisModII or the WHO- VALUE IN HEALTH 14 (2011) S51–S59 CHOICE, and by GLOBOCAN [15-20]. Different approaches were taken for acute events or chronic conditions. For acute events, such as cardiac or cerebrovascular events, the first step was to obtain the age-, sex- and country-specific absolute risk of the event based on the specific mortality rate and the lethality of the event: Rpop.event ⫽ Rdeath (1) L where L is the lethality of the event and Rdeath is the age- and sex-specific mortality of the condition. Once this absolute risk is known, the baseline risk in nonsmokers was calculated based on the age-, sex- and country-specific smoking prevalence as well as disease-specific smoking relative risk: Rnosmk ⫽ Rpop.event (RRsmk ⫻ fsmk) ⫹ (RRformersmk ⫻ fformersmk) ⫹ fnosmk (2) where Rnosmk is the baseline event annual incidence in non-smokers, Rpop.event is the age- and sex-specific population risk (obtained with formula 1), RRsmk and RRformersmk are the relative risks of the event in smokers and former-smokers versus nonsmokers, and ƒsmk, ƒformersmk and ƒnosmk are the age- and sex-specific proportion of smokers, former-smokers, and nonsmokers. For cancer (as chronic conditions), the age and sex estimation of the probability of diagnosis was calculated using a more complex approach that considered both the annual mortality rate from national statistics as well as the estimated yearly survival rate since diagnosis. The age- and sex-specific risk of diagnosis for each cancer was calculated with the following formula: Rdxi ⫽ 冋兺 10 n⫽0 册 Rmi ⫻ Pn ⫻ 1 1 ⫺ S10 (3) where Rdxi is the risk of diagnosis at age i; Rmi⫹n is the population risk of death from the specific cancer at age i⫹n; Pn is the conditional probability of dying in year n after the diagnosis, conditional on dying within 10 years; and S10 is the proportion of survivors after 10 years. We assumed that those subjects surviving 10 years after a lung cancer diagnosis, or five for other cancers, return to the general population risk of death. Then, the formula (2) was applied to derive the baseline risk in nonsmokers. A special case was chronic obstructive pulmonary disease (COPD). Because national statistics are known to significantly underestimate COPD-related mortality, we estimated its incidence and prognosis based on international studies [21,22]. Model structure and operation. A first order Monte Carlo, or probabilistic microsimulation of individual subjects, was built. This model incorporates the natural history, costs, and quality of life of all the tobacco-related adult-specific diseases: coronary and noncoronary heart disease, cerebrovascular disease, COPD, pneumonia, influenza, lung cancer, and nine other neoplasms. This model allows the follow-up of the lives of thousands of individuals in hypothetical cohorts, calculating all outcomes for each patient in an annual basis, using the simulation of each individual’s history to ultimately obtain aggregated population results in terms of health and costs. Subjects can be assigned demographic and disease specific characteristics. The model updates the values of the various input parameters for each patient in a yearly basis and calculates event rates for outcomes on the basis of the variables and the underlying risk equations. The model runs on Visual Basic and captures the key parameters from four main spreadsheets: 1) sex- and age-specific epidemiologic data; 2) unit cost; 3) quality of life; and 4) interventions effects. It consists of two main modules. The first one is used for the analysis of the disease burden associated with smoking, in which age and sex detailed information of selected epidemiologic S53 and economic data is kept (e.g., the number of events suffered by the cohort throughout specific ages, the distribution of COPD stages or the prevalence of coronary heart disease). The second module is the one that performs the cost-effectiveness analysis, and it focuses on the comparison of the experience of two cohorts for whom different sets of interventions are defined. The cohorts are then followed during their lifetime, and the aggregated results are compared in terms of costs and benefits. Disease incidence, progression, and mortality. For each time period, the model estimates the individual risks of occurrence of each event, disease progression and death, based on the subject’s demographic attributes, smoking status, and clinical conditions. Table 1A shows a list of possible events a subject can suffer in each time period and the calculation method used to derive it. Table 1B shows the different health states considered in the model. The risk of death is calculated for each time unit as the age- and sex-specific general mortality, excluding the 16 disease-specific risks of death considered in the model, plus the risk of death of the events and conditions that the individual experiences during that time unit (Table 1C). For instance, in the case of myocardial infarction the model estimates its risk multiplying the age-and sex-specific risk in nonsmokers (baseline incidence) in each time period and for each subject, by the relative risk related to his smoking status. Then, a random number is generated; if this number is less tan or equal to the individual probability, the model assumes that the event is present. The risk of death, in this case, will be the general risk of death plus the myocardial infarction age- and sex-specific case fatality. When there is more than one simultaneous cause of death for a given subject in the same time period, a probabilistic approach is used to assign the final cause of death, weighted by the baseline risk of each competing cause. For COPD, besides the risk of acquiring this condition, the risk of progression to more severe stages in those already affected is estimated according to the individual’s smoking status. For oncologic conditions, the specific risk of death is estimated according to the number of years since diagnosis. This type of individual-based model allows for having multiple events in a given year, as they are not mutually exclusive. This was the main reason to choose this model instead of a state transition cohort model (i.e., Markov cohort). Smoking status and interventions. We considered three smoking status states: current smokers, former smokers, and never smokers. Smokers have a given probability of making a quit attempt in each time unit as well as a probability of succeeding in that attempt without any active intervention (background quitting rates). These probabilities are age- and sex-related, and in case of considering a population SCI, its effect could be modeled by directly influencing this background quitting rate (Table 1A). Similarly, former smokers have a given relapse probability related to the time elapsed since the successful quit attempt (background relapsing rates). The model was built to consider a wide range of intervention modalities: 1) interventions/policies with the objective of improving smoking cessation rates in smokers who have a quit attempt (e.g., nicotine replacement or behavioral interventions); 2) interventions/policies that increase the probability of smokers attempting to quit (e.g., media campaign) and; 3) mixed interventions/policies that influence both the probability of attempting to quit and the success rates (e.g., training primary care physicians in brief counseling interventions, including pharmacotherapy in benefit plans). Resource use, cost, and quality of life. The model is programmed to calculate the use of resources and the QALYs in each time unit as a summary of the events the subject experienced in that particular time unit with the active health conditions coming from S54 VALUE IN HEALTH 14 (2011) S51–S59 Table 1 – (A) Acute events, (B) chronic disease states, and (C) causes of death included in the model. A. Acute events Disease acute events MI Non MI CHD event Stroke COPD diagnosis COPD progression Pneumonia/Influenza Cancer diagnosis: lung, bladder, renal, lip/oral/pharynx, larynx, stomach, esophagus, pancreas, cervical cancer, leukemia Smoking behavior events Performing a quit attempt Succeeding in a quit attempt Relapsing after successful quit attempt Probability calculation Disease events: Baseline Risk in non-smokers (age/sex specific) x RRsmoking.status COPD progression: Baseline Risk in non-smokers (sex and years-inprevious-stage specific) x RRsmoking.status Performing or succeeding quit attempt: Baseline population probability (age/sex specific) x RR.INTERVENTION Relapsing: Risk based on years in former-smoker state (sex specific) B. Chronic disease states CHD patient Post - Stroke COPD Stage Lung cancer Bladder cancer Renal cancer Lip/oral/pharynx cancer Larynx cancer Stomach cancer Esophagus cancer Pancreas cancer Cervical cancer Leukemia Smoking status: Smoker Former smoker Never smoked C. Causes of death MI Non MI CHD event Stroke Pneumonia/Influenza Non-ischemic CV death COPD Lung cancer Bladder cancer Renal cancer Lip/oral/Pharynx cancer Larynx cancer Stomach cancer Esophageal cancer Pancreas cancer Cervical cancer Leukemia Mortality for all other causes Probability calculation Acute events deaths: Probability of the event x its lethality (age/sex specific) Non-ischemic CVD death: Baseline Risk in non-smokers (age/sex specific) x RRsmoking.status COPD: Stage specific mortality (sex specific) Cancer: Tumor annual specific mortality during the first five years after diagnosis (except lung cancer: 10 years) Mortality for all other causes General population mortality minus the mortality of the diseases included in the model (sex and age specific) CHD, coronary heart disease; COPD, chronic obstructive pulmonary disease; CV,cardiovascular; MI, Myocardial infarction; RR.INTERVENTION, relative risk of the intervention (either to improve the probability of performing a quit attempt or to improve the success rate of the quit attempt); RRsmoking.status, disease specific relative risk according to smoking status. prior time units. Costs and QALYs are simultaneously calculated in undiscounted and discounted fashions. Main assumptions. The occurrence of all the modeled events is independent and not mutually exclusive. In the case of coexistence of two or more events or conditions, the costs and the absolute risks of death are additive and the QALYs are multiplicative. Each event or condition has two sets of cost and QALYs estimates: one for the first year and the other for the subsequent years up to a customized specific time horizon. In the current version, this time horizon was set to a lifetime for the cardiovascular events and COPD, to 10 years in lung cancer, and to 5 years in all other neoplasms. Cancer survival was modeled as dependant of sex and the years since diagnosis, independent of age. COPD incidence was modeled as dependant of sex, age, and smoking status; its progression as dependant of sex, years in current stage, and smoking status. COPD deaths were modeled as dependant of sex and stage. Model outputs. Results can be presented according to age, sex, previous cardiovascular history, and other specific epidemiologic data. Different benefit measures that can be used to report the cost-effectiveness of the different smoking cessation strategies are cost per quitter; cost per year of life gained, cost per event averted, and cost per QALY. To reflect decision uncertainty, a graphical depiction of the results of several simulations (second-order uncertainty) each comparing a specific population of interest (first-order uncertainty) can be made. This can be shown in the cost-effectiveness plane as the incremental cost-effectiveness rate dispersion, as 95% confidence intervals, and also as cost-effectiveness acceptability curves to show the probability of each strategy being cost-effective according to locally relevant threshold values. To perform a probabilistic sensitivity analysis, the probability distribution of selected parameters are incorporated into the model. For each iteration, parameters values are recalculated and applied to the equations. In Figure 1 we present an example of the cost-effectiveness results of two hypothetical SCIs in 500 simulations of 5000 50-year old men. Different one-way or scenario sensitivity analyses can also be incorporated for the estimation of the effects of selected parameters. The detailed information provided by the model can also be used to perform budget impact analysis and to guide for locally relevant research priority setting [23,24]. Calibration and validation We applied the International Society for Pharmacoeconomics and Outcomes Research criteria for model development and reporting [25]. The model structure and the parameters’ calculation approach were validated and calibrated using a selected country dataset (Argentinean National Health Statistics year 2005) [26]. Internal validation. Internal testing and debugging were performed to ensure that the mathematical calculations were accurate and consistent with the specifications of the model. The model was checked and tested during the modeling process to identify any errors relating to data incorporation and modeling syntax. Null and extreme input values were used and the test of S55 VALUE IN HEALTH 14 (2011) S51–S59 (b) 2000 1500 1000 500 0 -0.2 -0.1 0 0.1 0.2 -500 0.3 0.4 probabilty new SCI is cost-effective Cost difference (Argentine pesos 2007) (a) 80% 70% 60% 50% 40% 30% 20% 10% 0% -1000 Effect difference (QALYs) 0K 10K 20K 30K 40K 50K 60K 70K maximun acceptable ceiling ratio in $'000 Argentine pesos Fig. 1 – (A) Cost effectiveness plane of cost- and effectrelated differences between two smoking cessation interventions (SCI). SCI-a: cost $472, effectiveness: 15% smoking cessation at 1 year; SCI-b: cost $1.254, effectiveness: 23% smoking cessation at 1 year (Argentine pesos 2007). Preliminary results for 500 simulated cohorts of 5000 50-year old male smokers after one quit attempt. Discount rate for costs and effects: 5%. (B) Cost-effectiveness acceptability curve showing the probability that SCI-b is cost-effective as compared to SCI-a over a range of values for the maximum acceptable ceiling ratio. replication using equivalent input values was applied. Inconsistencies were detected and programming errors corrected. Calibration. Calibration was performed to ensure that the model can reproduce the results of the sources used to run the model. General mortality and all age- and sex-specific death rates predicted by the model were compared with local health statistics, with a total of 16 parameters (excluding COPD mortality, universally agreed to be underestimated in national statistics) [21,22]. Sex- and age-specific model outputs were compared to the source rates and deviations from the expected values were analyzed. Mean simulated event rates within ⫾ 10% of the mean reference event rates were considered acceptable, and in cases of higher deviations, the risk equation for that particular event was modified to provide a better fit to the published data. The search stopped as soon as all the outputs were within 10% of the target results. As explained earlier, the disease event incidence was estimated from the age- and sex- specific mortality and the lethality of the event for acute conditions and COPD, and the yearly survival rate since diagnosis for cancers. These last two parameters (lethality and survival rate) were estimated from local and international studies and were allowed to vary ⫾ 15% to determine the best fitting parameter set. Table 2 describes the data sources, baseline risk parameters calculation process, and the allowed variation during the calibration process. Besides ensuring that the simulated results were within the prespecified range, the total number of events and the event incidence were graphed for each parameter according to age and sex. The resulting observed and expected curves were visually explored to confirm a good fit. Closeness of fit was additionally assessed by plotting predicted versus observed values outcomes, fitting a linear curve through the points with the intercept set at zero, and obtaining a squared linear correlation coefficient (R2). The final simulation set was composed of 20 cohorts (10 of men and 10 of women) of 25,000 continuing smokers, 25,000 smokers who quit smoking during follow-up and 25,000 lifelong nonsmokers followed through their lifetime. The sample size of the simulations was estimated on the basis of the standard error of the parameter with greater variability (lifelong risk of death from oral cavity cancer) to be able to obtain 95% confidence intervals within 10% in each cohort (smokers, former smokers, and nonsmokers). Incidence rates estimated from the simulated cohorts were transformed into absolute numbers of events in each age and sex strata following Argentina 2005 population distribution [26]. After calibration, the differences between published data and the model results ranged from -9% to ⫹5%. The curves’ shapes of the age- and sexspecific simulated number of events adequately overlapped with those of the expected values, showing, in all cases, an excellent internal validity. Figure 2 shows results in four conditions: myocardial infarction, kidney cancer, nonischemic cardiovascular disease, and oral cavity cancer. As expected, correlation between predicted and observed results was better among high incidence events (such as myocardial infarction, stroke, or lung cancer) and weaker for less frequent (thus with greater variability) events such as leukemia or oral cavity cancer. When predicted values were plotted against observed data to assess goodness of fit, the majority of values were on or close to the y ⫽ x line, indicative of perfect fit. Evaluation of the correlation between predicted and observed data produced R2 values that ranged from 0.758 to 0.999 (perfect fit ⫽ 1) indicating a very strong correlation. The regression lines obtained for the 16 parameters ranged from a gradient of 0.874 to 1.272, close to the Table 2 – Baseline annual risk in nonsmokers: data sources, calculation process and allowed variation during the calibration process. Parameter Calculation process Data inputs Baseline event incidence in nonsmokers – acute conditions (stroke, cardiac events, Pneumonia/Influenza) Baseline cancer incidence in nonsmokers Formula 1 and Formula 2 Baseline risk of COPD incidence Formula 2 – Age- and sex-specific mortality – Lethality – Smoking prevalence – Disease specific RR for smokers and former smokers – Age- and sex-specific mortality – Survival rate – Smoking prevalence – Disease specific RR for smokers and former smokers – Population COPD incidence (age and sex specific) – Smoking prevalence – Disease specific RR for smokers and former smokers Age- and sex-specific mortality Formula 3 and Formula 2 Risk of death from all other causes COPD, chronic obstructive pulmonary disease. Source Allowed variation [26], [36] [37], [38] [33] [39] None ⫾ 15% None None [26] [16], [17] [33] [39] None ⫾ 15% None None [21], [22] ⫾ 15% [33] [39] None None [26] ⫾ 15% S56 VALUE IN HEALTH 14 (2011) S51–S59 Fig. 2 – Calibration: Annual number of deaths predicted by the model in Argentina for each age strata compared to the 2005 Argentinean national health data [26] in four selected conditions: (A) myocardial infarction (women); (B) kidney cancer (women); (C) non-ischemic cardiovascular disease (men) and; (D) oral cavity cancer (men). perfect fit line (gradient ⫽ 1). These results are shown in Figure 3 for stroke, lung cancer, pancreatic cancer, and esophageal cancer. External validation. Model results were validated against selected published epidemiologic and clinical studies not used to provide input data. Stroke and myocardial infarction age- and sex- specific event rates predicted by model were compared with those from international and local available data: the World Health Organization MONICA Project data on stroke and myocardial infarction incidence (World Health Organization monitoring of trends and determinants in cardiovascular disease) [27,28] and the only population-based myocardial infarction incidence study performed in Argentina [29,30] (see Fig. 4A and Fig. 4C). Age- and sex-specific COPD predicted prevalence was compared with the results from the Latin American Project for the Investigation of Obstructive Lung Disease, a population-based study carried out in five Latin American cities [31] (see Fig. 4B). Lung cancer incidence and lung cancer mortality predicted by the model were compared to those of the global cancer estimations reported by the International Agency for Research on Cancer [16,17] (see Figs. 4D and 4E). As a final endpoint that is influenced by the others, the toll on life expectancy of smokers and former smokers predicted by the model was analyzed together with the effect of quitting smoking at age 55 years. Results were compared to the population-based study of male British doctors [32] (see Fig. 4F). In all cases a high correlation between published data and model results was observed. Discussion Smoking is the single most preventable cause of disease and death all around the world [1]. In Latin America tobacco control policies are still poor and access to SCI is very limited [33-35]. In a context of more limited resources local evidence from cost-effectiveness studies is essential to implement more efficient health policies. Although international evidence regarding the burden of tobacco-related diseases is extensive, it is widely known that the results of health economic evaluations cannot be directly transferred from one setting to another. Recently, countries such as the United Kingdom have changed tobacco intervention policies using cost-effectiveness data, suggesting that the presence of regional, accurate information in Latin America could lead to an increased availability of effective interventions and a better definition of local research priorities. Our study describes the development and validation of a HEE model to evaluate the disease burden associated with smoking and the cost-effectiveness of SCIs in Latin America. To ensure the local relevance of this model, decision makers and researchers from each participant country provided input from the beginning of the project. The main characteristics of the HEE were S57 VALUE IN HEALTH 14 (2011) S51–S59 (b) 250 500 450 400 350 300 250 200 150 100 50 0 Model predicted values Model predicted values (a) Stroke deaths y =0.9858x R² =0.9959 200 150 Lung cancer deaths y =0.9677x R² =0.9923 100 50 0 0 100 200 300 400 500 0 50 (c) 150 200 250 (d) 60 80 70 60 50 40 30 20 Pancreatic cancer deaths y =0.9734x R² =0.7709 10 Model predicted values 90 Model predicted values 100 Argentinean vital statistics data Argentinean vital statistics data 50 40 30 Esophagus cancer deaths y =1.0543x R² =0.8964 20 10 0 0 0 10 20 30 40 50 60 70 80 Argentinean vital statistics data 0 10 20 30 40 50 Argentinean vital statistics data Fig. 3 – Correlation plot of model predicted versus reported age specific deaths in four selected conditions: (A) stroke (women); (B) lung cancer (men); (C) pancreatic cancer (women); (D) esophageal cancer (men). The gradients of regression lines (y) and the correlation coefficients (R2) are reported in each graph. Reference population: Argentinean National Vital Statistics year 2005 [26]. defined taking into consideration the availability and quality of the required epidemiologic data in the region. The relevant outcomes of the HEE model were conceived to suit the different policy makers’ information needs identified in the preparatory stage of the study. The HEE model showed internal validity with all simulated event rates falling within ⫾ 10% of the source publications and it also showed an excellent external validity when model results were compared to selected published studies. This external validation considered different conditions analyzed from different perspectives: death incidence, disease prevalence, and survival experience of the simulated cohorts. This comprehensive validation process is reassuring regarding the adequate performance of the model. It showed to be a reliable tool that can be used to estimate the tobacco-related disease burden and the cost-effectiveness of different SCI in the participant countries. It is important to note that the validation that we are presenting here corresponds to a single country data set (Argentina), and that a similar validation process is required for each country in which the model is applied. This is the first multicountry collaborative project that, to our knowledge, developed and validated a microsimulation model capable to evaluate the cost-effectiveness of a wide range of SCIs in Latin America, from public health interventions to specific individual therapies. As opposed to most published tobacco economic models that are based on state transition structures, the main advantages of the microsimulation-based approach include the possibility of tracking each subject’s history (as opposed to the “memorylessness” of Markov models) and how it influences future events; the possibility of simultaneously experiencing different health states during follow-up (which would render a Markov model nearly unmanageable); and being intrinsically probabilistic, allowing to depict first order (person-level) uncertainty. Second order (parameter-level) uncertainty can be estimated in the probabilistic sensitivity analysis through the incorporation of defined distributions in selected parameters. In addition, it can be easily adapted to new information availability. Whereas other tobacco-related economic models are usually limited to cardiac disease and lung cancer, our model included most tobacco-related diseases and additionally encompassed cerebrovascular disease, COPD, pneumonia, and influenza as well as nine other neoplasms. It also allowed incorporation of background quitting and relapsing rates. Some considerations and possible limitations to our model include the following: 1) it is highly dependent upon local health statistics, which may, in some settings, misestimate the actual disease-specific toll; 2) it does not analyze the effect of secondhand smoking; and 3) it does not incorporate the effect of tobacco on mother and child health. This project was supported by grants from international agencies, allowing us to develop a generic tool not oriented to a particular SCI. In addition, because this model was designed taking into consideration the low availability and quality of information in the seven participant countries, it could be easily adapted to be used in other “information-poor” settings such as most low and middle income countries. Apart from being able to depict the incremental cost and effect of interventions for the economic evaluation, the model was conceived as a tool to provide burden of tobacco-related diseases and budget impact data. This is of utmost relevance to raise awareness of the health and economic consequences of smoking in the region. The need to more precisely estimate the burden of the smoking-associated diseases measured in terms of economics and clinical consequences (smoking related illnesses and quality-adjusted survival) still exists. Also, it is highly important to incorporate these tools in Latin America to inform decision makers about the cost-effectiveness of tobacco control policies. S58 VALUE IN HEALTH 14 (2011) S51–S59 (b) Annual MI attack rate per 10.000 250 70% 60% 200 Model results 150 Danish MONICA 100 COPD prevalence (a) Buenos Aires study 50 50% Model results 40% 30% PLATINO study 20% 10% 0% 0 35-39 45-49 55-59 65-69 35-39 75-79 45-49 55-59 700 (d) Model results 600 500 Finland MONICA 400 Russia MONICA 300 200 Lithuania MONICA 100 Lung Cancer incidence per 100,000 Annual stroke attack rate p/100.000 Age group (c) 0 45-54 Age group ≥ 85 300 250 Model results 200 Argentina 150 World 100 South America 50 55-64 15-44 300 (f) Model results 250 200 Argentina 150 World 100 South America 50 45-54 55-64 Age group ≥ 65 100 Percentage survival from age 35 Lung Cancer mortality per 100,000 75-79 0 35-44 (e) 65-69 Age group 89 (89) 80 (81) 82 60 74 (76) Non-smokers (58) 57 46 (50) Smokers 40 Stopped smoking at age 55 (26) 26 20 0 0 15-44 45-54 55-64 ≥ 65 40 50 60 70 Age (years) 80 90 100 Age group Fig. 4 – External validation against selected published epidemiologic studies. Results correspond to the male population. (A) Annual myocardial infarction event rates predicted by the model compared to population-based incidence studies: Danish WHO MONICA study register [27] and the myocardial infarction incidence study conducted in Argentina (Coronel Suarez-Province of Buenos Aires) [29,30]; (B) Model predicted COPD prevalence compared to a population-based prevalence study performed in Latin America (PLATINO Latin American Project for the Investigation of Obstructive Lung Disease) [31]; (C) Model predicted annual stroke event rates compared to WHO MONICA study register in selected countries with high stroke incidence rates (Finland WHO MONICA study register North Karelia province, Russia WHO MONICA study register Novosibirsk city, Lithuania WHO MONICA study register Kaunas city) [28]; (D) Model predicted lung cancer incidence compared to the International Agency for Research on Cancer estimations for Argentina, South America, and the world [16,17]; (E) model predicted lung cancer mortality compared to the International Agency for Research on Cancer estimations for Argentina, South America, and the world [16,17]; (F) Survival from age 35 for continuing smokers, lifelong non-smokers and effects on survival of stopping smoking at age 55 years. Model results for simulated cohorts of 250,000 subjects. In parenthesis, at age 60, 70, and 80 years model results are compared to the cohorts of male British doctors [32]. Sources of financial support: The study was made possible by unrestricted research grants awarded by the International Clinical Epidemiology Network, the Initiative for Cardiovascular Health Research in the Developing Countries; and the International Development Research Centre, Canadian Tobacco Control Research Initiative, American Cancer Society, Cancer Research UK, Institut National du Cancer France, and Department for International Development UK. The funding agreement ensured the authors’ independence in designing the study, interpreting the data, writing, and publishing the report. REFERENCES [1] Ezzati M, Lopez AD. Estimates of global mortality attributable to smoking in 2000. Lancet 2003;362:847–52. [2] Jha P, Chaloupka FJ, Moore J, et al. Tobacco addiction. In: Jamison DT, Measham AR, Breman JB et al. Disease Control Priorities in Developing Countries. (2nd ed.). New York: Oxford University Press, 2006. [3] Jha P, Chaloupka FJ. Curbing the epidemic: governments and the economics of tobacco control. Washington, DC: The World Bank, 1999. [4] Convenio Marco de la OMS para el control del tabaco. Organización Mundial de la Salud [Frame- work Convention on Tobacco Control VALUE IN HEALTH 14 (2011) S51–S59 [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] (FCTC). World Health Organization]. Available from: http://www. who.int/tobacco/framework/WHO_fctc_spanish.pdf. [Accessed March 9, 2011]. Abdullah AS, Husten CG. Promotion of smoking cessation in developing countries: a framework for urgent public health interventions. Thorax 2004;59:623–30. Sculpher MJ, Claxton K, Drummond MF, McCabe C. Whither trialbased economic evaluation for health care decision making. Health Econ 2006;15:677– 87. Hjelmgren J, Berggren F, Andersson F. Health economic guidelines— similarities, differences and some implications. Value Health 2001;4: 225–50. Taylor RS, Drummond MF, Salkeld G, Sullivan SD. Inclusion of cost effectiveness in licensing requirements of new drugs: the fourth hurdle. BMJ 2004;329;972–5. Banta D. Health technology assessment in Latin America and the Caribbean. Int J Technol Assess Health Care 2009,25(Suppl. 1):253– 4. Knusel L. On the accuracy of statistical distributions in Microsoft Excel 2003. Comput Stat Data Anal 2005;48:445–9. Keeling K, Pavur R. Numerical accuracy issues in using excel for simulation studies. Available from: http://portal.acm.org/citation. cfm?id⫽1162012. [Accessed March 9, 2011]. Silva LC, Ordúñez P, Paz Rodríguez M, Robles S. A tool for assessing the usefulness of prevalence studies done for surveillance purposes: the example of hypertension. Rev Panam Salud Publica 2001;10:152– 60. Moher D, Schulz KF, Altman DG. The CONSORT statement: revised recommendations for improving the quality of reports of parallelgroup randomised trials. Lancet 2001;357:1191– 4. Terwee CB, Bot SD, de Boer MR, et al. Quality criteria were proposed for measurement properties of health status questionnaires. J Clin Epidemiol 2007;60:34 – 42. Barendregt JJ, Van Oortmarssen GJ, Vos T, Murray CJ. A generic model for the assessment of disease epidemiology: the computational basis of DisMod II. Popul Health Metr 2003;1:4. Ferley J, Bray F, Pisani P, Parkin DM. GLOBOCAN 2002. Cancer incidence, mortality and prevalence worldwide. IARC CancerBase No. 5, version 2.0. Available from: http://www.emro.who.int/ncd/cancer_ globocan.htm. [Accessed 10 December 2010]. Parkin DM, Bray F, Ferlay J, Pisani P. Estimating the world cancer burden: GLOBOCAN 2000. Int J Cancer 2001;94:153– 6. Lauer JA, Rohrich K, Wirth H, Charette C, et al. PopMod: a longitudinal population model with two interacting disease states. Cost Eff Resour Alloc 2003;1:6. Gail MH, Kessler L, Midthune D, Scoppa S. Two approaches for estimating disease prevalence from population-based registries of incidence and total mortality. Biometrics 1999;55:1137– 44. Pisani P, Bray F, Parkin DM. Estimates of the world-wide prevalence of cancer for 25 sites in the adult population. Int J Cancer 2002;97:72– 81. Mannino DM. COPD: epidemiology, prevalence, morbidity and mortality, and disease heterogeneity. Chest 2002;121(5 Suppl.): 121S– 6S. Stang P, Lydick E, Silberman C, et al. The prevalence of COPD: using smoking rates to estimate disease frequency in the general population. Chest 2000;117(5 Suppl. 2):354S–9S. Levy DT, Bauer JE, Lee HR. Simulation modeling and tobacco control: creating more robust public health policies. Am J Public Health 2006; 96:494 – 8. Orme ME, Hogue SL, Kennedy LM, et al. Development of the health and economic consequences of smoking interactive model. Tob Control 2001;10:55– 61. Weinstein MC, O’Brien B, Hornberger J, et al. Principles of good practice of decision analytic modeling in health care evaluation: [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] S59 report of the ISPOR Task Force on Good Research Practices-Modeling Studies. Value Health 2003;6:9 –17. Dirección de estadísticas e información de salud. Argentina: Ministerio de Salud de la Nación, 2006. Estadísticas vitales – Información 2005, Serie 5, No. 29. [Directorate of statistics and health information. National Ministry of Health, 2006. National Health Statistics year 2005] Kirchhoff M, Davidsen M, Bronnum-Hansen H, et al. Incidence of myocardial infarction in the Danish MONICA population 1982–1991. Int J Epidimiol 1999;28:211– 8. Thorvaldsen P, Asplund K, Kuulasmaa K, et al. Stroke incidence, case fatality, and mortality in the WHO MONICA Project. Stroke 1995;26: 361–7. Caccavo A, Alvarez A, Bello F, et al. Incidencia poblacional del infarto con elevación del segmento ST o bloqueo de rama izquierda a lo largo de 11 años en una comunidad de la provincia de Buenos Aires. [Eleven Years Incidence of Infarction with ST Elevation or Left Bundle Branch Block on the Population of a Community in the Province of Buenos Aires] Rev Argent Cardiol 2007;75:185– 8. Ferrante D, Tajer C. ¿Cuantos infartos hay en la Argentina? [How many heart attacks there in Argentina?] Rev Argent Cardiol 2007;75: 161–2. Menezes AM, Perez-Padilla R, Jardim JR, et al. Chronic obstructive pulmonary disease in five Latin American cities (the PLATINO study): a prevalence study. Lancet 2005;366:1875– 81. Doll R, Peto R, Wheatley K, Gray R, Sutherland I. Mortality in relation to smoking: 40 years’ observations on male British doctors. BMJ 1994; 309:901–11. Programa VIGIA Argentina. Encuesta de tabaquismo de las 5 grandes ciudades de la Argentina. Buenos Aires, Argentina: Ministerio de Salud y Ambiente de la Nación, 2004. [VIGIA program Argentina. Survey of smoking in the big 5 cities of Argentina. Ministry of Health, 2004] Noble M, Pérez-Stable E, Casal E. El comportamiento médico en relación al tabaquismo. Boletín de la Acad Nac de Medicina 1996 [Physician behavior in relation to smoking. Bulletin of the National Medicine Academy 1996];74:413–25. Schoj V. Relevamiento regional de cobertura de tratamientos de la cesación tabáquica. Buenos Aires, Argentina: Ministerio de Salud y Ambiente de la Nación, 2004. [Regional survey on coverage of smoking cessation treatments. Buenos Aires, Argentina: National Ministry of Health, 2004] Sistema de Información Sanitaria del SNS. Estadísticas Sanitarias España 1991-2000: 5.8.6. Neumonía e Influenza. Número de defunciones por edad y sexo. [Health Information System of the SNS. Spain Health Statistics 1991–2000: 5.8.6. Pneumonia and influenza. Number of deaths by age and sex] Available from: http://www.msc.es/ estadEstudios/estadisticas/inforRecopilaciones/generales.htm. [Accessed April 28, 2008]. Egresos de establecimientos oficiales según variables seleccionadas. República Argentina - Año 2000. Buenos Aires, Argentina: Secretaría de Políticas, Regulación y Relaciones Sanitarias. Ministerio de Salud, 2003. [Discharges from public institutions according to selected variables. Argentina - Year 2000. Buenos Aires, Argentina: Secretariat of Policies, Regulations and Sanitation. Ministry of Health 2003] Gutierrez F, Masia M, Mirete C, et al. The influence of age and gender on the population-based incidence of community-acquired pneumonia caused by different microbial pathogens. J Infect 2006;53: 166 –74. Centers for Disease Control and Prevention. Smoking-attributable mortality, morbidity, and economic costs (SAMMEC). Adult SAMMEC, Relative Risk - CPS–II (82-88). Available from: http://apps.nccd.cdc.gov/ sammec/. [Accessed April 28, 2008]. VALUE IN HEALTH 14 (2011) S60 –S64 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Estimation and Comparison of EQ-5D Health States’ Utility Weights for Pneumoccocal and Human Papillomavirus Diseases in Argentina, Chile, and the United Kingdom Julieta Galante, MD, Msc(c)1, Federico Augustovski, MD, Msc, PhD1,2,*, Lisandro Colantonio, MD, Msc(c)1, Ariel Bardach, MD, Msc1, Joaquin Caporale, Msc1, Sebastian Garcia Marti, MD, Msc(c)1, Paul Kind, Mphil3 1 Institute of Clinical Effectiveness and Health Policy, Ciudad Autónoma de Buenos Aires, Argentina; 2Family and Community Medicine Division, Hospital Italiano de Buenos Aires, Universidad de Buenos Aires, Ciudad Autónoma de Buenos Aires, Argentina; 3Centre for Health Economics, University of York, York, UK A B S T R A C T Objectives: To estimate and compare EuroQol instrument (EQ-5D) health states’ values for pneumoccocal and human papillomavirus (HPV) diseases in Argentina, Chile, and the United Kingdom. Methods: Twelve vignettes were designed, pilot-tested, and administered to a convenience sample in a cross-sectional design to elicit descriptive EQ-5D state data. Country-specific EQ-5D time-trade-off-based weights were used to map these descriptive health states into local country preference weights. Descriptive analysis is reported and intercountry differences for each condition were compared using repeated measures analysis of variance. Results: Seventy-three subjects completed the survey. Pneumococcal disease-related health states mean values ranged from ⫺0.331 (sepsis, Chile) to 0.727 (auditive sequelae, Argentina). HPVrelated conditions ranged from 0.152 (cervical cancer, United Kingdom) to 0.848 (cervical intraepithelial neoplasia 1, Argentina). Chile had consistently the lowest mean values in pneumococcal states and in one HPV state, whereas those of the United Kingdom were the lowest in most HPV Introduction The measurement of health benefits is a critical activity associated with all aspects of the planning and delivery of health care, but the choice of unit of measure is not uniformly acknowledged. To help to guide health care-wide resource allocation decisions, it needs to be based on a generic system so that gains/losses can be compared across the widest possible range of interventions [1]. Quality-adjusted life years (QALYs) are a unit of measure which is made up of the product of quality of life and quantity of life. A QALY refers to 1 year of life in complete health. Health status, or quality of life, is measured on a scale in which full health has a value of 1.0 and dead has a value of zero [2–5]. QALYs led to much applied work based on cost-utility analysis, and approaches to prioritization based on incremental cost-per-QALY figures, both in upper- and lower-and-middle income countries [6]. states. Argentina had the highest mean values in both diseases. Differences in country-specific values for each health state were statistically (P ⬍ 0.001) significant except for six health states in which differences between Chilean and United Kingdom weights were nonsignificant. Conclusions: Utility values for most conditions differed statistically relevantly among analyzed countries, even though the same health states= descriptive set was valued for each. These results reflect the difference in social weights among different countries, which could be attributed to either different population values or valuation study methodologies. They stress the importance of using local preference weights for context-specific decision making. Keywords: Great Britain, Latin America, quality-adjusted life years, questionnaires. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. QALY weights are computed either by directly eliciting subjects preferences through direct methods (standard gamble, time-trade off, visual analog scale) or through a two-step approach: The first one involves classifying the health status with a preference-based, generic health-related quality of life measurement instrument; and the second is to translate this health state to the value that the general reference population have assigned to it in a previous valuation study. The EuroQol instrument (EQ-5D) is probably the most widely used standardized instrument for use as a measure of health outcome in economic evaluations [7]. Its descriptive system classifies a health state by a three-level Likert-type scale on five dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. The scales range from the best level (no limitations ⫽ 1) to the worst level (severe limitations ⫽ 3) and thus describe health states in a five-digit number. In addition, it has a visual analog scale item where the health state is valued in a single 0 to 100 scale. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Federico Augustovski, Director of the Economic Evaluations and HTA Department, Institute of Clinical Effectiveness and Health Policy, Ravignani 2024 (C1414CPV), Ciudad Autónoma de Buenos Aires, Argentina. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.007 VALUE IN HEALTH 14 (2011) S60 –S64 It has been shown that self-reported EQ-5D descriptive health status differ considerably between countries before weighting them through quality-of-life weighting methodologies [8]. One concern relates to the reported differences of preference weights for the same states from different countries [9 –19] Preference/utilities and preference weights studies are really scarce in the Latin American region, even though they are essential for locally relevant decision making. Local utilities are not generally used in our region, and this could be due not only to the fact that these values are not widely available, but also that they may seem to be transferable from utility weights from developed countries [20]. The objectives of our study were to estimate and compare EQ-5D health states’ preference values for pneumococcal and human papillomavirus (HPV) diseases in three different countries (Argentina, Chile, and the United Kingdom), using the same health states’ descriptive mix in the three countries. This work was part of a larger project in which we needed to obtain utility values for the selected health states as inputs for two vaccination cost-effectiveness models to be applied in different countries. Although this study was a substudy of a larger project, and due to the scarce research in this area in Latin America, we think that it makes a relevant regional contribution. Moreover, the conditions reported highly contribute to the burden of disease in Latin America [21,22]. This was the rationale behind the selection of the health states. S61 not all countries were similar, we run paired t tests with a Bonferroni-corrected alpha for multiple comparisons [24]. All statistical analyses were conducted using Stata/SE 8.0 (Stata Corp., College Station, TX). Results To obtain descriptive data regarding the different health states defined in the economic models, 12 health state vignettes (eight for pneumococcal diseases and four for HPV diseases) were designed, pilot-tested, and administered to a convenience sample of subjects in Argentina. The survey was confidential and anonymous. After describing each of the health states with a half-page vignette, they proceeded to complete one EQ-5D questionnaire for each. To evaluate and control sequence or order effects, three sets of questionnaires in which the health states were ordered in different ways were used and randomly administered to each third of the sample. Both as a primer and for use as descriptive data, the first health state for which it was asked to complete the EQ-5D questionnaire was, in all cases, their health status the day of the survey. Finally, some demographic descriptive data of the respondents was gathered. Between July and August 2009, 73 subjects completed the survey. Study sample characteristics are shown in Table 1 (available in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.007). Fifty-three percent of the respondents were women. Mean age was 31 years (range 22–58) and mean self-reported health status measured by EQ-5D’s visual analog scale was 86 out of 100. Utility values for the health states of the HPV vaccination model, obtained by pairing the five-digit number of each health state with its correspondent local weight, are shown in Table 2 and in Figure 1 (available in Supplemental Material found at: doi: 10.1016/j.jval.2011.05.007); whereas utility values for the health states of the pneumococcal vaccination model, obtained in the same way, are shown in Table 3, Figure 2, and Figure 3 (available in Supplemental Material found at: doi:10.1016/j.jval.2011.05.007). Because all values had an asymmetric distribution, we present both mean/confidence interval and median/interquartile range. In addition, in Table 4 (available in Supplemental Material found at: doi:10.1016/j.jval.2011.05.007) we show the visual analog scale summary values of each of the health states. For pneumococcal disease-related health states, means utility values ranged from ⫺0.331 (sepsis, Chile) to 0.727 (auditive sequelae, Argentina). Regarding HPV-related conditions, they ranged from 0.152 (cervical cancer, United Kingdom) to 0.848 (cervical intraepithelial neoplasia 1, Argentina). Chile had consistently the lowest coefficients in pneumococcal states and in one HPV state, whereas those of the United Kingdom were the lowest in most HPV states. Argentina had the highest coefficients in both disease groups. Mean differences between countries in pneumococcal health states were 0.256 (Argentina-Chile), 0.207 (Argentina-UK), and 0.048 (ChileUK); and those for HPV were 0.117 (Argentina-Chile), 0.133 (Argentina-UK), and 0.017 (Chile-UK). We found that the differences in country-specific values for each health state were statistically significant, and many of them of an important magnitude, except for six health states (cervical intraepithelial neoplasia 1, cervical intraepithelial neoplasia 2 and 3, cured cancer, meningitis, acute otitis media, and acute otitis media with myringotomy) in which differences between Chilean and English weights were nonsignificant. Argentinean weights resulted significantly different and higher for all the conditions. Analysis of questionnaire data and deriving of local preference weights Discussion The description of each health state consisted of an EQ-5D fivedigit number, as described above. On the other hand, time tradeoff-derived local weights for each of these health states for Argentinean [9], Chilean (Víctor Zárate, University of York, personal communication), and English populations [23] were available. By pairing the five-digit number of each health state with its correspondent local weight, utility values for the 12 health states could be obtained for each and all respondents. An initial descriptive analysis of the sample is presented, as is a descriptive report of the health states values for each country. To illustrate similarities and differences in country-specific values, the relationship between the Argentine, the Chilean, and the English values was graphically shown and assessed by analysis of variance repeated measures test. We examined the statistical significance of the differences in country-specific values for each health state. Furthermore, for the cases where a significant analysis of variance result was found, showing that Although it is not uncommon to assume that utility values to be used in economic evaluations are usually transferable from place to place, and many studies use for QALY calculations weights from other settings, there is growing evidence that utilities can be significantly and sometimes meaningfully different between settings [9 –19]. In our study we found that utility coefficients for each condition differed significantly between the three analyzed countries even considering that the same health states’ mix was valued in all three countries. This is why, even though our sample was a convenience sample, the fact that a health state can be descriptively different between countries (i.e., a typical pneumonia could be more severe), this could not account for the differences among countries’ utility values. Our study is a practical exercise that shows that in a real-life scenario and using the same set of health states for each disease state, the difference in country valuations introduce significant differences in results. This stress the importance of using local and not international weights in context-spe- Methods Descriptive data regarding the different disease-related EQ-5D health states S62 VALUE IN HEALTH 14 (2011) S60 –S64 Fig. 1 – Box plots representing EQ-5D translated coefficients for health states included in the human papillomavirus vaccination model. Central horizontal line of each box: median; upper hinge: 75th percentile; lowerhinge: 25th percentile; whiskers: upper (third quartile plus 1.5* interquartile range) and lower (first quartile minus 1.5* interquartile range) adjacent values; outside dots: outliers. cific decision-making processes such as cost-effectiveness analyses and economic evaluations. These differences we found reflect how systematic differences in social preference weights between countries can lead to differ- ent results in difference settings, and eventually to potentially different conclusions about effectiveness and cost-effectiveness of different alternative strategies. The reasons for the differences in the social value weightings in each country might be explained Fig. 2 – Box plots representing EQ-5D translated coefficients for four health states included in the pneumococcal vaccination model. Central horizontal line of each box: median; upper hinge: 75th percentile; lower hinge: 25th percentile; whiskers: upper (third quartile plus 1.5 · interquartile range) and lower (first quartile minus 1.5 · interquartile range) adjacent values; outside dots: outliers. VALUE IN HEALTH 14 (2011) S60 –S64 S63 Fig. 3 – Box plots representing EQ-5D translated coefficients for four health states included in the pneumococcal vaccination model. Central horizontal line of each box: median; upper hinge: 75th percentile; lower hinge: 25th percentile; whiskers: upper (third quartile plus 1.5 · interquartile range) and lower (first quartile minus 1.5 · interquartile range) adjacent values; outside dots: outliers. not only by the fact that they have different population characteristics and values, but they can also be at least partially explained by variations in the valuation protocol and method used for each country. Although all of the local valuation studies were done with the time-trade-off method, slightly different analysis procedures and different sampling methods were used to allow the valuations of all EQ-5D states to be interpolated from direct valuations, given that it is virtually impossible to generate direct valuations for all of the 243 possible EQ-5D health states. Chile and United Kingdom valuation studies used probabilistic sampling, whereas the Argentinean one used quota sampling. Secular trends in social preferences might also have some relevance. The United Kingdom valuation study was undertaken in 1993 [23], and the Chilean one 15 years later (Víctor Zárate, University of York, personal communication). Although studies exist that compared different valuation methods for similar health states and populations [25,26], there are fewer studies that attempted to compare potential differences in local utilities in a given set of identical mix of health states in different jurisdictions. As an example in our region, Augustovski et al. [9] published a study in 2009 where they developed a set of EQ-5D health states’ values for the Argentine general population and compared it with published values for the United States, finding meaningful and significant differences between them. Nevertheless, this study was based on all EQ-5D set of states and it was not related to any particular disease [9]. König et al. [8] published in 2009 a brief report where they compared general population health status measured by the EQ-5D in six European countries [8]. Even after adjusting for sociodemographic variables and with representative samples, self-reported EQ-5D health status differed considerably between countries, calling for caution when making international comparisons of disease burden and health care effectiveness and potential cost-effectiveness of different interventions. Jürges [27] decomposed in 2007 cross-national differences in self-reported general health into parts explained by differences in ’true’ health, measured by diagnosed conditions and measurements, and parts explained by cross-cultural differences in response styles, and concluded that failing to account for differences in reporting styles may yield misleading results. There are studies that reported that the use of anchoring vignettes successfully improved the comparability of self reported measures [28]. Newer studies also underline the problems of using value sets to weight profile data as EQ-5D derived health states and that caution should be taken when choosing a summary measure [29,30]. It would have been interesting to recruit different samples of subjects in Argentina, Chile, and the United Kingdom to evaluate if the same vignettes produced different responses regarding to which EQ-5D states they correspond, or even to make locally specific vignettes to reflect potential differences between disease states (i.e., ambulatory pneumonia) in three countries. Nevertheless, this was out of the scope and resources of our work and is an issue that could be addressed in the future. Acknowledgement The authors thank Jorge A. Gomez, from GlaxoSmithKline Biologicals, for helpful comments on a previous version. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi:10.1016/j.jval.2011.05.007, or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). S64 VALUE IN HEALTH 14 (2011) S60 –S64 REFERENCES [1] Coons SJ, Rao S, Keininger DL, Hays RD. A comparative review of generic quality-of-life instruments. Pharmacoeconomics 2000;17: 13–35. [2] Klarman HE, Francis JS, Rosenthal GD. Cost effectiveness analysis applied to the treatment of chronic renal disease. Med Care 1968;6: 48 –54. [3] Weinstein MC, Stason WB. Foundations of cost-effectiveness analysis for health and medical practices. N Engl J Med 1977;296:716 –21. [4] Williams A. Economics of coronary artery bypass grafting. Br Med J (Clin Res Ed) 1985;291:326 –9. [5] Drummond KF, Sculpher MJ, Torrance GW, et al. Methods for the Economic Evaluation of Health Care Programmes. Oxford, United Kingdom: OUP, 2005. [6] Shillcutt SD, Walker DG, Goodman CA, Mills AJ. Cost effectiveness in low- and middle-income countries: a review of the debates surrounding decision rules. Pharmacoeconomics 2009;27:903–17. [7] Rabin R, de Charro F. EQ-5D: a measure of health status from the EuroQol Group. Ann Med 2001;33:337– 43. [8] König HH, Bernert S, Angermeyer MC, et al. Comparison of population health status in six European countries: results of a representative survey using the EQ-5D questionnaire. Med Care 2009;47:255– 61. [9] Augustovski F, Irazola V, Velazquez A, et al. Argentine valuation of the EQ-5D health states. Value Health 2009;12:587–96. [10] Arnesen T, Trommald M. Are QALYs based on time trade-off comparable? A systematic review of TTO methodologies. Health Econ 2005;14:39 –53. [11] Huang IC, Willke RJ, Atkinson MJ, et al. US and UK versions of the EQ-5D preference weights: does choice of preference weights make a difference? Qual Life Res 2007;16:1065–72. [12] Johnson JA, Luo N, Shaw JW, et al. Valuations of EQ-5D health states: are the United States and United Kingdom different? Med Care 2005; 43: 221– 8. [13] Nan L, Johnson JA, Shaw JW, et al. A comparison of EQ-5D index scores derived from the US and UK population-based scoring functions. Med Decis Making 2007;27:321– 6. [14] Sakthong P, Charoenvisuthiwongs R, Shabunthom R. A comparison of EQ-5D index scores using the UK, US, and Japan preference weights in a Thai sample with type 2 diabetes. Health Qual Life Outcomes 2008;6:71. [15] Jelsma J, Hansen K, De Weerdt W, et al. How do Zimbabweans value health states? Popul Health Metr 2003;1:11. [16] Kharroubi SA, O’Hagan A, Brazier JE. A comparison of United States and United Kingdom EQ-5D health states valuations using a nonparametric Bayesian method. Stat Med 2010;29:1622–34. [17] Tsuchiya A, Ikeda S, Ikegami N, et al. Estimating an EQ-5D population value set: the case of Japan. Health Econ 2002;11:341–53. [18] Zarate V, Kind P, Chuang LH. Hispanic valuation of the EQ-5D health states: a social value set for Latin Americans. Value Health 2008;11: 1170 –7. [19] Badía X, Roset M, Herdman M, Kind P. A comparison of United Kingdom and Spanish general population time trade-off values for EQ-5D health states. Med Decis Making 2001;21:7–16. [20] Augustovski F, Iglesias C, Manca A, et al. Barriers to generalizability of health economic evaluations in Latin America and the Caribbean region. Pharmacoeconomics 2009;27:919 –29. [21] Yang BH, Bray FI, Parkin DM, et al. Cervical cancer as a priority for prevention in different world regions: an evaluation using years of life lost. Int J Cancer 2004;109:418 –24. [22] Constenla D, Gomez E, Pio de la Hoz F, et al. The Burden of Pneumococcal Disease and Cost-Effectiveness of a Pneumococcal Vaccine in Latin America and the Caribbean. Washington, DC: Albert B. Sabin Vaccine Institute, 2007. [23] Dolan P. Modeling valuations for EuroQol health states. Med Care 1997;35:1095–108. [24] Rosner BA. Fundamentals of Biostatistics (7th ed.). Belmont, CA: Thomson Brooks/Cole, 2010. [25] Bernert S, Fernandez A, Haro JM, et al. Comparison of different valuation methods for population health status measured by the EQ-5D in three European countries. Value Health 2009;12:750 – 8. [26] Krabbe PF, Essink-Bot ML, Bonsel GJ. The comparability and reliability of five health-state valuation methods. Soc Sci Med 1997;45:1641–52. [27] Jürges H. True health vs response styles: exploring cross-country differences in self-reported health. Health Econ 2007;16:163–78. [28] Salomon JA, Tandon A, Murray CJ. Comparability of self rated health: cross sectional multi-country survey using anchoring vignettes. BMJ 2004;328:258. [29] Parkin D, Rice N, Devlin N. Statistical analysis of EQ-5D profiles: does the use of value sets bias inference? Med Decis Making 2010; 30:556 – 65. [30] Wilke CT, Pickard AS, Walton SM, et al. Statistical implications of utility weighted and equally weighted HRQL measures: an empirical study. Health Econ 2010;19:101–10. VALUE IN HEALTH 14 (2011) S65–S70 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Evaluación Económica de un Programa de Inmunización Infantil en México Basado en la Vacuna Neumocócica Conjugada 13-Valente Emilio Muciño-Ortega, MSc1,*, Joaquín Federico Mould-Quevedo, PhD, MBA2, Raymond Farkouh, PhD3, David Strutton, PhD, MPH4 1 Coordinador de Farmacoeconomía, Pfizer S.A. de C.V., México D.F., México; 2Health Economics & Outcomes Research Director Latin America and Primary Care Emerging Market Business Unit, Pfizer Inc., New York, NY, USA; 3Associated Director, Global Healthcare Outcomes Assessment Vaccines, Pfizer Inc., Collegeville, PA, USA; 4Associated Vice President, Global Healthcare Outcomes Assessment Vaccines, Pfizer Inc., Collegeville, PA, USA A B S T R A C T Objectives: Vaccination is an effective intervention for reduce child morbidity and mortality associated to pneumococcus. The availability of new anti-pneumococcal vaccines makes it necessary to evaluate its potential impact on public health and costs related to their implementation. The aim of this study was to estimate the costeffectiveness and cost-utility of immunization strategies based on pneumococcal conjugated vaccines (PCV=s) currently available in Mexico from a third payer perspective. Material and Methods: A decision tree model was developed to assess both, economic and health impact, of anti-pneumococcal vaccination in children ⬍2 years (lifetime time horizon, discount rate: 5% annual). Comparators were: no-vaccination (reference) and strategies based on 7, 10 and 13-valent PCV=s. Effectiveness measures were: child deaths avoided, life-years gained (LYG) and quality adjusted life years (QALY=s) gained. Effectiveness, utility, local epidemiology and cost of treating pneumococcal diseases were extracted from published sources. Uni- Introducción En el año 2000, se estimó que las infecciones por Streptococcus pneumoniae (S. pneumoniae) o neumococo causaron alrededor de 14,5 millones de casos de enfermedad neumocócica severa y 825.000 muertes en niños menores de 5 años a nivel mundial [1]. Para Latinoamérica, se ha estimado que ocurren anualmente entre 12.000 y 28.000 muertes por enfermedad neumocócica en niños ⬍5 años [2]. Por otro lado, en los Estados Unidos de América (EUA) se presentaron más de cinco millones de eventos de otitis media (OM) al año [3] antes de la introducción de la vacuna neumocócica conjugada heptavalente (PCV-7, por sus siglas en inglés). El impacto social de las infecciones por S. pneumoniae en niños es mayor a la sola consideración de los años de vida perdidos por muerte prematura, ya que una proporción de los niños que las padecen desarrollará secuelas permanentes (daño neurológico, hipoacusia) que impactarán negativamente en su calidad de vida [4–7]. variate sensitivity analysis were performed. Results: Immunization dominates no-vaccination: strategy based on 13-valent vaccine prevented 16.205 deaths, gained 331.230 LY=s and 332.006 QALY=s and saved US$1.307/child vaccinated. Strategies based on 7 and 10-valent PCV=s prevented 13.806 and 5.589 deaths, gained 282.193 and 114.251 LY=s, 282.969 and 114.972 QALY=s and saved US$1.084 and US$731/child vaccinated, respectively. These results were robust to variations in herd immunity and lower immunogenicity of 10-valent vaccine. Conclusions: In Mexico, immunization strategies based on 7, 10 and 13-valent PCV=s would be cost-saving interventions, however, health outcomes and savings of the strategy based on 13-valent vaccine are greater than those estimated for 7 and 10-valent PCV=s. Palabras Claves: evaluación económica, neumococo, vacuna neumocócica conjugada 13 valente, México. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. La inmunización basada en la PCV-7 ha demostrado ser una intervención efectiva para la disminución de morbilidad y mortalidad por enfermedad neumocócica invasiva (ENI) en niños menores de 2 años [8,9] e incluso en otros grupos de edad [10,11]. La PCV-7 está formulada con los polisacáridos capsulares de los serotipos 4, 6B, 9V, 14, 18C, 19F y 23F de S. pneumoniae, presentes en alrededor del 78% de los casos de ENI en EUA [12]; sin embargo, debido a la variabilidad en la distribución geográfica de los serotipos de esta bacteria, en México, la cobertura serotípica de la PCV-7 en ENI se encuentra alrededor del 58% [12,13]. Actualmente se encuentra aprobada para su uso en México una vacuna formulada con tres serotipos adicionales a los presentes en la PCV-7 (1, 5 y 7F), denominada vacuna neumocócica conjugada 10-valente (PCV-10 por sus siglas en inglés ó PHIDCV, de Pneumococcal non-typeable Haemophilus Influenzae Protein D conjugate Vaccine), cuya cobertura serotípica en ENI en México es del 63% [12,13]. Se ha desarrollado una nueva va- Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Título corto: Costo-efectividad vacuna 13-valente en México. * Autor de correspondencia: Emilo Muciño-Ortega, Paseo de los Tamarindos N° 40, Col. Bosques de las Lomas; CP 05120, México D.F., México; Tel: ⫹52 (55) 50818625; Fax: ⫹52 (55) 50818600 ext. 8625. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.025 S66 VALUE IN HEALTH 14 (2011) S65–S70 Fig. 1 – Modelo de árbol de decisión en el que se incluyen las consecuencias de infección por S. pneumoniae en niños <2 años. cuna en la que se encuentran los serotipos de la PCV-10, más los serotipos 3, 6A y 19A (vacuna neumocócica conjugada 13-valente o PCV-13, por sus siglas en inglés), cuya cobertura serotípica en ENI en México es del 77% [13]. Para el contexto mexicano se ha generado evidencia del impacto económico y en salud pública de la inmunización basada en la PCV-7 vs no vacunar [14], de las consecuencias del uso de la PCV-13 vs la PCV-7 [15,16] y de la comparación de la PCV-13 vs la PCV-10 [16]. Con el fin de apoyar el proceso de toma de decisiones, es necesario contar con más información acerca del impacto en salud pública y los costos que tendrían México programas de inmunización universal basados en cada una de las vacunas neumocócicas conjugadas. El objetivo de este estudio fue estimar el costo-efectividad y costo-utilidad de la inmunización basada en la PCV-7, la PCV-10 y la PCV-13 respecto de no vacunar, en una cohorte de niños mexicanos ⬍2 años, en un horizonte temporal correspondiente a la esperanza de vida de la cohorte y desde la perspectiva del sistema público de salud en México. Material y métodos Descripción del modelo Se desarrolló un modelo tipo árbol de decisión para estimar el costo-efectividad y el costo-utilidad de estrategias de inmunización infantil contra S. pneumoniae. El modelo refleja los posibles desenlaces clínicos de la infección por S. pneumoniae en niños, tales como bacteriemia y meningitis, así como neumonía (con atención hospitalaria y con atención ambulatoria) y OM asociadas a todas las causas. Los pacientes que contraen bacteriemia o meningitis pueden desarrollar secuelas (deterioro neurológico o sordera), lo que reduce su calidad de vida (Fig. 1). Una cohorte de 3.793.867 niños (población ⬍2 años estimada para mediados de 2010 en México [17]) ingresa al modelo y enfrenta la probabilidad de sufrir infecciones por S. pneumoniae. La probabilidad de presentación de desenlaces clínicos producto de las infecciones por S. pneumoniae es una función de la vacuna empleada para la inmunización de una proporción de los niños ⬍1 año de la cohorte. El VALUE IN HEALTH 14 (2011) S65–S70 modelo estima el número de casos de cada enfermedad, muertes, pacientes con secuelas, así como el costo directo de la vacunación y del manejo de casos de enfermedad que se presentan con cada estrategia de vacunación al término de un año. Para estimar en el largo plazo el impacto de las muertes prematuras y la menor calidad de vida asociada a secuelas se consideró un horizonte temporal correspondiente a la expectativa de vida de la cohorte (74,3 años, basado en información del Consejo Nacional de Población [17]). Los resultados pueden incluir o no efectos de inmunidad de grupo (efectos indirectos) que ejerce la proporción vacunada de la cohorte sobre la proporción no vacunada. El modelo considera que no se presentan nuevos desenlaces clínicos (y en consecuencia tampoco costos) durante el resto del horizonte de análisis en la proporción de la cohorte que sobrevivió al primer año. Asimismo, no se consideran la probabilidad de la presentación eventos adversos asociados a las vacunas ni el costo de manejo de resistencia a antibióticos ni el de manejo de secuelas. Descripción de alternativas Las alternativas consideradas fueron las estrategias de vacunación basadas en la PCV-7, PCV-10 y PCV-13. A pesar de que sólo se trata de una opción hipotética (la PCV-7 se incluyó en la cédula de inmunización infantil en México desde 2008 [18]), se consideró a la estrategia de no vacunar como alternativa de referencia. Se administran 2 dosis y un refuerzo de la PCV-7 y la PCV-13 y 3 dosis y un refuerzo de la PCV-10. Perspectiva de análisis El modelo considera la perspectiva del sistema de salud al evaluar el costo médico directo en el que incurre una institución representativa del sistema de salud mexicano, como es el Instituto Mexicano del Seguro Social (IMSS) al atender casos de bacteremia, meningitis bacteriana, neumonía (con manejo ambulatorio y manejo hospitalario) y OM en niños ⬍2 años, así como también el costo de vacunación de una proporción de la cohorte. Medidas de efectividad En el análisis de costo efectividad (ACE), se consideraron como medidas de efectividad la reducción en el número de muertes y el número de Años de Vida Ganados (AVG=s) con cada vacuna (respecto de no vacunar). En el análisis de costo utilidad (ACU) se consideró como medida de resultado el número de Años de Vida Ajustados por Calidad (AVAC=s) ganados con cada vacuna (respecto de no vacunar). Los AVG=s y los AVAC=s se presentan descontados con una tasa anual del 5% [19]. Complementariamente, se estimó el costo por niño vacunado. Parámetros del modelo Para obtener los datos requeridos por el modelo, se realizó una revisión estructurada de la literatura publicada a la fecha, en bases de datos electrónicas (Medline, Health Star, Psycinfo, Embase, Cochrane Library Database e Inbiomed). Cuando no fue posible obtener datos específicos de México, se consideraron fuentes referidas al contexto latinoamericano y a la ausencia de estas, fuentes internacionales. La incidencia y tasa de fatalidad correspondientes a bacteremia, neumonía con manejo hospitalario debida a todas las causas y meningitis se extrajeron del trabajo de Valenzuela et al. [2]. La incidencia de OM se estimó a partir de los resultados de Nandí et al. [20]. La relación entre casos de neumonía con manejo hospitalario y ambulatorio en EUA es 1:7 [21,22]. Esta proporción fue usada para estimar la incidencia de neumonía con manejo ambulatorio en México. La tasa de secuelas neurológicas en meningitis se extrajo del estudio de Gómez et al. [23]. La tasa de hipoacusia por meningitis se extrajo de un trabajo realizado en el contexto de EUA [24]. S67 La cobertura serotípica de las vacunas en México en bacteremia, meningitis y neumonía fue estimada en base a resultados del Sistema Regional de Vacunas [13]. Se usó la cobertura serotípica en OM reportada por Rodgers et al. [25] y López et al. [26]. La eficacia clínica de la PCV-7 en ENI se extrajo del estudio de Black et al. [27], en neumonía debida a cualquier causa se estimó en base a Ray et al [24] y Grijalva et al [22] y en OM por cualquier causa se estimó en base a Ray et al. [24], Poehling et al. [28] y Fireman et al. [29]. Se ha estimado la eficacia clínica versus OM debida a todas las causas de una vacuna formulada con 11 serotipos de S. pneumoniae [30], suponiendo que la inmunogenicidad en 10 los serotipos comunes entre esta vacuna y la PCV-10 se mantiene, se estimó la eficacia clínica de esta versus OM por cualquier causa. Ante la ausencia de estudios clínicos de la eficacia de la PCV-10 y la PCV-13 en ENI y neumonía por todas las causas y OM por todas las causas (sólo la PCV-13), fue necesario estimar la eficacia de estas alternativas en base a su incremento en la cobertura serotípica respecto de la PCV-7 y los efectos directos e indirectos de esta última. Detalles de esta estimación se encuentran en disponibles en el anexo como material suplementario en: doi:10.1016/j.jval.2011.05.025. Oostenbrink et al. [31] estimaron un ponderador de utilidad de retraso mental leve. Se asume este dato como el ponderador de utilidad deterioro neurológico usado en el modelo. Se asumió que el ponderador de utilidad de niños sanos y de los enfermos que no presentan secuelas es 1,0, el ponderador para la muerte es 0,0. El modelo usa el costo médico directo del manejo de casos de bacteremia y neumonía estimado en un hospital mexicano de tercer nivel de atención [32]. El costo de manejo de caso de meningitis bacteriana, OM y atención ambulatoria de neumonía se obtuvieron de un reporte del IMSS [33]. Se asume que la práctica médica para el manejo de estos padecimientos ha sido consistente entre la fecha de realización de estos estudios y la actualidad, por lo que los costos correspondientes requerirían la actualización correspondiente al tiempo transcurrido. Los costos de la PCV-7 y la PCV-10 fueron extraídos del portal de compras del gobierno de México [34]. Se asumió que el costo de la PCV-13 es el mismo que el de la PCV-7 y que el costo de aplicación/ dosis de la vacuna es de US$1. Todos los costos fueron actualizados de acuerdo a la inflación acumulada en México del año de referencia en las fuentes a diciembre de 2009 [35] y se encuentran expresados en US$ de 2010 (MX$ 12,67⫽US$1) [36]. Una selección de los parámeros empleados en el análisis del caso base se presenta en la Tabla 1. Análisis de sensibilidad Se probó el impacto en las estimaciones del modelo por las siguientes variantes en las características de las vacunas: una menor inmunogenicidad de la PCV-10 (de 0,98 a 0,95), así como la presencia de efectos indirectos en la PCV-10 y la ausencia de efectos indirectos en la PCV-7 y la PCV-13. Resultados Bajo las consideraciones realizadas en el caso base, la estrategia basada en la PCV-7 presenta un costo incremental negativo (ahorros al sistema de salud) de US$ 2.014.526.391, la de la PCV-10 por US$ 1.359.669.841 y la de la PCV-13 por US$ 2.430.534.394. Se evitan 13.806, 5.589 y 16.205 muertes en las cohortes vacunadas con la PCV-7, PCV-10 y PCV-13, respectivamente, con el consiguiente impacto en los AVG=s y los AVAC=s ganados por cada una de ellas (Tabla 2). Las tres estrategias de vacunación dominan a la estrategia de no vacunar dado que resultaron ser más efectivas y menos costosas que ella. Para la estrategia basada en la PCV-13 se estimó S68 VALUE IN HEALTH 14 (2011) S65–S70 Tabla 1 – Parámetros empleados en el modelo. Tasa de fatalidad en niños ⬍ 2 años (%) Incidencia (Casos/100.000 niños ⬍ 2 años) Bacteremia neumocócica [2] Meningitis neumocócica [2] Neumonía manejo hospitalario [2] Neumonía manejo ambulatorio [2, 21, 22] Otitis meda [20] Tasa de secuelas por meningitis Deterioro neurológico [23] Hipoacusia [24] Cobertura serotípica de la PCV-7 12,0 12,0 7.173,0 49.780,6 36.000,0 18,0 136,0 % 58,3 61,9 58,5 59,0 61,5 94,0 Cobertura serotípica de la PCV-10 % Cobertura serotípica de la PCV-13 Bacteremia neumocócica [13] Meningitis neumocócica [13] Neumonía manejo hospitalario [13] Neumonía manejo ambulatorio [13] Otitis meda [25] Costos de manejo de casos 59,5 67,5 64,6 64,6 61,5 % 74,8 78,7 74,7 74,7 64,0 US$/evento Bacteremia neumocócica [32] Meningitis neumocócica [33] Neumonía manejo hospitalario [32] Neumonía manejo ambulatorio [33] Otitis meda [33] 14.946,9 5.651,6 16.541,1 1.560,9 831,6 Costos de vacunación US$/dosis PCV-7 [34] PCV-10 [33] PCV-13 (supuesto) 0,0 30,0 3,0 3,0 % Bacteremia neumocócica [13] Meningitis neumocócica [13] Neumonía manejo hospitalario [13] Neumonía manejo ambulatorio [13] Otitis meda [25, 26] Efectos directos de la PCV-7 en ENI [27] Bacteremia neumocócica [13] Meningitis neumocócica [13] Neumonía manejo hospitalario [13] Neumonía manejo ambulatorio [13] Otitis meda [25] Tasa de fatalidad en niños ⬍ 2 años (%) 19,8 17,1 19,8 Utilidades Sano o sin secuelas de ENI (supuesto) Deterioro neurológico [31] Pérdida auditiva [31] Muerte (supuesto) un ahorro de US$1.307/niño vacunado, siendo mayor en 20 y 80% al estimado para las PCV-7 y PCV-10, respectivamente. Análisis de sensibilidad El efecto económico y en salud de un menor factor de inmunogenicidad de la PCV-10 (0,95), así como la consideración de efectos indirectos en la PCV-10 y no consideración de efectos indirectos en las PCV-7 y PCV-13 se presenta en la Tabla 3. A pesar de la disminución de inmunogenicidad de la PCV-10 esta alternativa presenta un costo incremental negativo (US$ -1.315.335.152) y es más efectiva que no-vacunar 1 0,55 0,79 0,0 (dominando a esta última), su beneficio (ahorro neto /niño vacunado) disminuye en la misma proporción que el factor de inmunogenicidad. La consideración de la existencia de efectos indirectos en la PCV-10 incrementa el ahorro (US$2.198.540.406) y la efectividad de la estrategia basada en esta vacuna respecto de las estimaciones del caso base (PCV-10 domina no vacunación). El ahorro/niño vacunado correspondiente a la PCV-10 en este escenario es 62%mayor respecto de la estimación del caso base, sin embargo aún es US$ 124 menor, pero US$ 99 mayor a los estimados en el caso base para la PCV-13 y PCV-7, respectivamente. S69 VALUE IN HEALTH 14 (2011) S65–S70 Tabla 2 – Estimaciones para el caso base. Análisis incremental Estrategia de vacunación Muertes Costo total (US$) Muertes evitadas AVG=s AVAC=s ganados Costo incremental (US$) Razón de costo efectividad incremental* Ahorro por niño vacunado (US$) No vacunación PCV-7 PCV-10 PCV-13 64.930 51.124 59.340 48.725 8.599.074.176 6.584.547.785 7.239.404.335 6.168.539.782 — 13.806 5.589 16.205 — 282.193 114.251 331.230 — 282.860 114.871 331.897 — ⫺2.014.526.391 ⫺1.359.669.841 ⫺2.430.534.394 — Dominante Dominante Dominante — 1.084 731 1.307 * La razón de costo efectividad incremental se expresa en términos de $US/muerte evitada adicional, $US/AVG y $US/AVAC ganado. En el escenario sumamente conservador de la ausencia de efectos indirectos en las PCV-7 y PCV-13, la efectividad de ambas estrategias es mayor que la no vacunación y los costos incrementales son negativos (US$ ⫺1.245.428.617 y US$ ⫺1.601.390.273, respectivamente), dominando a la no vacunación. La PCV-13 presenta el mayor nivel de ahorro/niño vacunado, seguido por la PCV-10 y la PCV-7 (US$ 861, US$ 708 y US$ 670, respectivamente). Discusión De acuerdo a los resultados del presente estudio, un programa de inmunización basado en cualquiera de las tres vacunas analizadas tendría impacto favorable en la salud de la cohorte analizada respecto de no vacunar. Lo anterior se traduce en un importante ahorro de recursos, por lo que la pregunta que adquiere relevancia en este punto es cuál de las tres vacunas analizadas es la que presenta el perfil farmacoeconómico más favorable. En el análisis del caso base, la implementación de una estrategia de vacunación basada en la PCV-13 proveería los mayores beneficios en términos de muertes evitadas, AVG=s y AVAC=s ganados respecto de no vacunar. Este nivel de beneficios es alcanzado a un menor costo respecto del que representaría la vacunación con la PCV-10 y la PCV-7. Estos resultados fueron robustos a una disminución en el factor de inmunogenicidad de la PCV-10 (0,98 a 0,95), la consideración de efectos indirectos en la PCV-10 y la no consideración de efectos indirectos en la PCV-7 y la PCV-13. Ante la disponibilidad de la PCV-10 y la PCV-13, se han llevado a cabo evaluaciones económicas en diversos contextos nacionales, tales como Holanda [37], Grecia [38], Alemania [39], Canadá [40], Reino Unido [41] y México [16]. En general, los resultados de estas investigaciones sugieren que un programa de inmunización basado en la PCV-13 representa una opción más efectiva y menos costosa que programas basados en la PCV-7 y la PCV-10, tendencia a la que se suman los resultados de la presente investigación. Esta consistencia constituye un resultado muy importante para los clínicos y decisores mexicanos al brindarles certidumbre acerca de los beneficios farmacoeconómicos de cada una de las alternativas consideradas en el presente análisis. Esta investigación tiene limitaciones, entre las que se encuentra el hecho de que los parámetros epidemiológicos empleados fueron extraídos de literatura publicada y pueden no reflejar la carga actual de la enfermedad neumocócica en México. Se asume que cualquier diferencia respecto de la epidemiología no afecta de manera particular a alguna alternativa. La no consideración de eventos adversos y su costo de manejo en el las estimaciones del modelo tendría un impacto negativo en los ahorros generados por la vacunación, sin embargo muy probablemente este no sería significativo. Por otro lado, existe la posibilidad de que los beneficios de la PCV-7 y la PCV-13 se encuentren subestimados, ya que nuestro análisis no considera el impacto favorable de la PCV-7 en ⬎60 años [42]. Al no considerarse otros costos después del primer año de la inmunización (manejo de casos nuevos de enfermedades y secuelas) también se estaría subestimando el beneficio económico de la vacunación en el largo plazo. Asimismo, estimaciones que incluyesen costos por muerte prematura o pérdida de productividad muy probablemente constituirían un argumento que favorecerían aún más la vacunación. En conclusión, en México la inmunización con las vacunas PCV-7, PCV-10 y PCV-13 son intervenciones que redundarían en mayores beneficios en salud y menores costos respecto de la no vacunación. Sin embargo, al presentar la PCV-13 la mayor cobertura serotípica entre estas alternativas, el impacto de un programa de inmunización de niños ⬍ 1 año basado en ella llevaría al mayor beneficio en salud y en consecuencia a la mayor disminución en el consumo de recursos médicos y económicos destinados al manejo de casos de morbilidades asociadas a S. pneumoniae. Tabla 3 – Estimaciones para el análisis de sensibilidad. Estrategia de vacunación Muertes Costo total (US$) Muertes evitadas AVG=s AVAC=s ganados Costo incremental (US$) Razón de costo efectividad incremental* Ahorro por niño vacunado (US$) 5.507 112.556 113.158 ⫺1.315.335.152 Dominante 708 14.549 297.378 298.04 ⫺2.198.540.406 Dominante 1.183 Ausencia de efectos indirectos en la PCV-7 y la PCV-13 PCV-7 59.497 7.353.645.559 5.433 PCV-13 58.036 6.997.683.903 6.894 111.043 140.921 111.683 141.549 ⫺1.245.428.617 ⫺1.601.390.273 Dominante Dominante 670 861 Disminución de inmunogenicidad de la PCV-10 (0,95) PCV-10 59.423 7.283.739.024 Presencia de efectos indirectos en la PCV-10 PCV-10 50.381 6.400.533.770 * La razón de costo efectividad incremental se expresa en términos de $US/muerte evitada adicional, $US/AVG y $US/AVAC ganado. S70 VALUE IN HEALTH 14 (2011) S65–S70 Fuentes de financiamiento: Esta investigación se realizó con el apoyo financiero de Pfizer S.A. de C.V., sin que esto genere algún tipo de compromiso legal y/o sobre los resultados de la misma. Al momento de la realización del estudio, todos los autores eran empleados de Pfizer. Materiales Complementarios Material complementario que acompaña este artículo se puede encontrar en la versión en línea como un hipervínculo en doi:10.1016/j.jval.2011.05.025 o si es un artículo impreso, estará en www.valueinhealthjournal.com/issues (seleccione el volumen, número y artículo). REFERENCIAS [1] O=Brien KL, Wolfson L, Watt JP, et al. Burden of disease caused by Streptococcus pneumoniae in children younger than 5 years: global estimates. Lancet 2009;374:893–902. [2] Valenzuela MT, O’Loughlin R, de la Hoz F, et al. The burden of pneumococcal disease among Latin American and Caribbean children: review of the evidence. Rev Panam Salud Publica 2009;25:270 –9. [3] Marcy M, Takata G, Shekelle P, et al. Management of Acute Otitis Media. Evidence Report/Technology Assessment No. 15. AHRQ Publication No. 01-E010. Rockville, MD: Agency for Healthcare Research and Quality, 2001. Disponible en: http://www.ncbi.nlm.nih .gov/books/NBK33163/. [Consultado en agosto 8 de 2008]. [4] Pikis A, Kavaliotis J, Tsikoulas J, et al. Long-term sequelae of pneumococcal meningitis in children. Clin Pediatr 1996;35:72–7. [5] Pomeroy SL, Holmes SJ, Dodge PR, et al. Seizures and other neurologic sequelae of bacterial meningitis in children. N Engl J Med 1990;323:1651–7. [6] Weightman NC, Sajith J. Incidence and outcome of pneumococcal meningitis in northern England. Eur J Clin Microbiol Infect Dis 2005;24: 542– 4. [7] De Beer BA, Schilder AG, Zielhuis GA, et al. Natural course of tympanic membrane pathology related to otitis media and ventilation tubes between ages 8 and 18 years. Otol Neurotol 2005;26:1016 –21. [8] Kaplan SL, Mason EO, Wald ER, et al. Decrease of invasive pneumococcal infections in children among 8 children’s hospitals in the United States after the introduction of the 7-valent pneumococcal conjugate vaccine. Pediatrics 2004;113:443–9. [9] Black S, France EK, Isaacman D, et al. Surveillance for invasive pneumococcal disease during 2000 –2005 in a population of children who received 7-valent pneumococcal conjugate vaccine. Pediatr Infect Dis J 2007;26:771–7. [10] Pulido M, Sorvillo F. Declining invasive pneumococcal disease mortality in the United States, 1990 –2005. Vaccine 2010;28:889 –92. [11] Whitney CG, Farely MM, Hadler J, et al. Decline in invasive pneumococcal disease after the introduction of protein–polysaccharide conjugate vaccine. N Engl J Med 2003;348:1737–46. [12] Hausdorff WP, Bryant J, Paradiso PR, et al. Which pneumococcal serogroups cause the most invasive disease: implications for conjugate vaccine formulation and use, part I. Clin Infect Dis 2000;30:100–21. [13] Castañeda E, Agudelo CI, Regueira M, et al. Laboratory-based surveillance of Streptococcus pneumoniae invasive disease in children in 10 Latin American countries. Pediatr Infect Dis J 2009;28:e265–70. [14] Carlos F, Mercado G, Aguirre A, et al. Cost-effectiveness of heptavalent pneumococcal conjugate vaccine (PCV-7) in Mexico. Value Health 2009;12:A420. [15] Rivera R, Strutton D, Hwang S, et al. Eficacia en términos de costos de la vacunación conjugada contra el neumococo 13-valente comparada con la vacunación conjugada contra el neumococo 7-valente en México. Póster presentado en el XIII Congreso Latinoamericano de Infectología Pediátrica, Guayaquil, Ecuador, Agosto 12–15, 2009. [16] Strutton D, Hwang S, Earnshaw S, et al. Cost-effectiveness of 13-valent pneumococcal conjugate vaccination in Mexico. Póster presentado en el XIII Congreso Latinoamericano de Infectología Pediátrica, Guayaquil, Ecuador, Agosto 12–15, 2009. [17] Consejo Nacional de Población (México). Proyecciones de la población de México 2005–2050. Disponible en: http://www.conapo .gob.mx/index.php?option⫽com_content&view⫽article&id⫽36& Itemid⫽234. [Consultado en Febrero 20, 2010]. [18] Centers for Disease Control (Estados Unidos de America). Progress in introduction of pneumococcal conjugate vaccine--worldwide, 2000 – 2008. Morb Mortal Wkly Rep 2008;57(42):1148 –51. Disponible en: http:// www.cdc.gov/MMWR/preview/mmwrhtml/mm5742a2.htm#tab. [Consultado en febrero 20, 2010]. [19] Secretaría de Salud (México). Consejo de Salubridad General. Guía para la conducción de estudios de evaluación económica para la actualización del Cuadro Básico de Insumos del Sector Salud en México. 2008. Disponible en: http://www.csg.salud.gob.mx/descargas/ pdfs/cuadro_basico/GUxA_EVAL_ECON25082008_2_ech.pdf. [Consultado en febrero 20, 2010]. [20] Nandí-Lozano E, Espinosa LE, Viñas-Flores L, et al. Infección respiratoria aguda en niños que acuden a un centro de desarrollo infantil. Salud Publica Mex 2002;44:201– 6. [21] Grijalva CG, Poehling KA, Nuorti P, et al. National impact of universal childhood immunization with pneumococcal conjugate vaccine on outpatient medical care visits in the United States. Pediatrics 2006;118: 865–73. [22] Grijalva CG, Nuorti JP, Arbogast PG, et al. Decline in pneumonia admissions after routine childhood immunisation with pneumococcal conjugate vaccine in the USA: a time-series analysis. Lancet 2007;369:1179–86. [23] Gómez-Barreto D, Calderón-Jaimes E, Rodríguez RS, et al. Características clínico-microbiológicas de la meningitis por streptococcus pneumoniae resistente a la penicilina. Salud Pub Mexico 1999;41:397– 404. [24] Ray GT, Whitney C, Fireman B, et al. Cost-effectiveness of pneumococcal conjugate vaccine: evidence from the first 5 years of use in the United States incorporating herd effects. Pediatr Infect Dis J 2006;25:494 –501. [25] Rodgers G, Arguedas A, Cohen R, et al. Global serotype distribution among Streptococcus pneumoniae isolates causing otitis media in children: potential implications for pneumococcal conjugate vaccines. Vaccine 2009;27:3802–10. [26] López-Enríquez C, Blanco-Montero A, Espinosa-Monteros LE, et al. Middle-Ear Fluid Streptococcus pneumoniae susceptibility and serotype and distribution in Mexican children with acute otitis media. Pediatrics 2008;121(Suppl.):S129. [27] Black S, Shinefield H, Fireman B, et al. Efficacy, safety and immunogenicity of heptavalent pneumococcal conjugate vaccine in children. Pediatr Infect Dis J 2000;19:187–95. [28] Poehling KA, Talbot TR, Griffin MR, et al. Invasive pneumococcal disease among infants before and after introduction of pneumococcal conjugate vaccine. JAMA 2006;295:1668 –74. [29] Fireman B, Black SB, Shinefield HR et al. Impact of the pneumococcal conjugate vaccine on otitis media. Pediatr Infect Dis J 2003;22:10 – 6. [30] Prymula R, Peeters P, Chrobok V, et al. Pneumococcal capsular polysaccharides conjugated to protein D for prevention of acute otitis media caused by both Streptococcus pneumoniae and non-typable Haemophilus influenzae: a randomised double-blind efficacy study. Lancet 2006;367:740 – 8. [31] Oostenbrink R, Moll HA, Essink-Bot ML. The EQ-5D and the Health Utilities Index for permanent sequelae after meningitis A head-tohead comparison. J Clin Epidemiol 2002;55:791–9. [32] Navarrete-Navarro S, Armengol-Sánchez G. Costos secundarios por infecciones nosocomiales en dos unidades pediátricas de cuidados intensivos. Salud Pública Méx 1999;41(Suppl. 1):s51– 8. [33] Instituto Mexicano del Seguro Social. Costos por GRD. Dirección de Planeación y Finanzas. Coordinación de Presupuestos, Contabilidad y Evaluación Financiera. 2003. [34] Sistema electrónico de compras gubernamentales (México). Disponible en: www.compranet.gob.mx [Consultado en Febrero 20, 2010]. [35] Banco de México. Indices de precios al consumidor y UDIS. Inflación. Disponible en: http://www.banxico.org.mx/polmoneinflacion/ estadisticas/indicesPrecios/indicesPreciosConsumidor.html. [Consultado en Febrero 20, 2010]. [36] Banco de México. Estadísticas. Tipos de cambio. Tipo de cambio promedio del periodo. Disponible en: http://www.banxico.org.mx/ sistema-financiero/estadisticas/mercado-cambiario/tiposcambio.html. [Consultado en Agosto 1, 2010]. [37] Vemer P, de Greeff SCD, Schouls LM, et al. Seven, ten or thirtheen? The cost-utility of infant vaccination with a 7-, 10- or 13-valent pneumococcal conjugate vaccine in the Netherlands. Value Health 2009;12:A228. [38] Papanicolaou S, Kontodimas S, Syriopoulou V, et al. Clinical and economic benefits of national immunization with the 13-valent compared to 7- and 10-valent pneumococcal conjugate vaccine in Greece. Value Health 2009;12:A423. [39] Claes C, Mittendorf T, Kuchenbecker U, et al. Cost-effectiveness of switching strategies form a 7-valent to a 13-valent pneumococcal conjugate vaccine. Value Health 2009;12:A425. [40] Whillans F, Kwan H, Strutton DR, et al. Cost-effectiveness of 13-valent and 10-valent pneumococcal conjugate vaccination relative to 7valent pneumococcal conjugate vaccination in Canada. Value Health 2009;12:A425. [41] Patel R, Stoykova B, Lloyd AC, et al. A comparison of the costeffectiveness of the 13-valent (PCV13) and 10-valent pneumococcal conjugate vaccines in the UK. Value Health 2009;12:A428. [42] McBean AM, Park YT, Caldwell D, et al. Declining invasive pneumococcal disease in the U.S. elderly. Vaccine 2005;23:5641–5. VALUE IN HEALTH 14 (2011) S71–S77 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Gastos do Ministério da Saúde do Brasil com Medicamentos de Alto Custo: Uma Análise Centrada no Paciente Cristina Mariano Ruas Brandão, MSc1,*, Augusto Afonso Guerra Júnior, ScD2, Mariângela Leal Cherchiglia, PhD1, Eli Iola Gurgel Andrade, PhD1, Alessandra Maciel Almeida, ScD1, Grazielle Dias da Silva, MSc2, Odilon Vanni de Queiroz, MSc1, Daniel Resende Faleiros, Especialista2, Francisco de Assis Acurcio, ScD1 1 Programa de Pós-graduação em Saúde Pública da Universidade Federal de Minas Gerais, Belo Horizonte, Brazil; 2Secretaria de Estado de Saúde de Minas Gerais, Belo Horizonte, Brazil A B S T R A C T Objective: To describe the expenses of the Ministry of Health of Brazil with users of High-Cost Drug Program that began treatment between 2000 –2004, according to their demographic and clinical characteristics. Methods: We made a probabilistic-deterministic linkage of national databases of drugs and mortality, resulting in a historical cohort of patients using high-cost medications in 2000 –2004. The per capita spending on medicines were stratified by a follow-up period and described according to demographic, clinical and type of drug used. Results: The total population atended by the program was 611,419, being 63.5% female, average age 46 years. 41.9% of patients living in the Southeast and 29.7% in the Northeast of Brazil. 24.5% of patients began treatment in 2000, 12.4% in 2001, with increasing trend until 2004. The most prevalent diagnosis referred to the genitourinary system diseases Introdução No Brasil, o direito à saúde é positivado no ordenamento jurídico como um direito social, de acordo com a Constituição Federal. É garantido a todos os cidadãos mediante políticas sociais e econômicas que visem à redução do risco de doença e ao acesso universal e igualitário às ações e serviços para a sua promoção, proteção e recuperação [1]. A assistência farmacêutica é parte integrante do direito à saúde e provê o acesso gratuito aos medicamentos por meio de diferentes programas, dos quais se destacam: i. o componente da assistência farmacêutica básica, destinado ao atendimento dos agravos prevalentes e prioritários da Atenção Básica; ii. componente estratégico, utilizados para tratamento das doenças endêmicas, doenças sexualmente transmissíveis/Aids, hanseníase, tuberculose, hemoderivados, nutrição e controle do tabagismo; e iii. componente especializado, que são aqueles tratamentos cujo custo não pode ser suportado pela população [2]. O Programa de Medicamentos de Alto Custo (PMAC) do Ministério da Saúde, objeto desse estudo, provê tratamento farmacoterapêutico para pacientes com indicação de uso de medicamento de alto valor unitário ou que, em caso de uso crônico, seja um and the most common use of chemical groups were antianemic preparations. 40,941 deaths were detected (6.7% of total). The total expenditure per capita was R$4.794,34. Higher spending per capita was observed in males, aged 47, who lived in the Southeast of Brazil and began treatment in 2000, had diagnoses of infectious and parasitic diseases and used blood substitutes and perfusions solutions. Conclusion: The understanding of the expenses involved subsidizes restructuring actions and scheduling drug programs, also provides information for therapeutic groups which are priorities for analysis. Palavras Claves: drugs, expense, high cost. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. tratamento de custo elevado. Os usuários destes medicamentos são pacientes transplantados, portadores de insuficiência renal crônica, osteoporose, hepatites, doenças genéticas, dentre outras [3]. No período de realização do estudo este programa era denominado “Programa de Medicamentos de Dispensação em Caráter Excepcional/Alto Custo”. A partir de novembro de 2009, os medicamentos de alto custo integram o “Componente Especializado da Assistência Farmacêutica” [4]. Os gastos com medicamentos nesse programa têm apresentado crescimento contínuo. Em 2003, foram gastos aproximadamente R$1,05 bilhão e em 2005 estes recursos já somavam R$1,92 bilhão [5]. Apesar dos inúmeros esforços para organização e gerenciamento do PMAC, não se conhece o número de usuários cadastrados e, principalmente, de usuários atendidos por patologia [6]. Isto se deve ao processo de autorização de fornecimento dos medicamentos, realizado por meio do formulário de Autorização de Procedimento de Alto Custo (APAC), que serve como cadastramento do usuário no banco de dados nacional para fins gerenciais e de cobrança. A APAC apresenta caráter contábil e é focada no medicamento e não no indivíduo. Desta forma, indivíduos em uso contínuo de medicamentos podem ter inúmeras APAC’s, dificultando a análise de suas carac- Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Título resumido: Gastos do Ministério da Saúde do Brasil. * Autor de Correspondência: Cristina Mariano Ruas Brandão, Grupo de Pesquisas em Farmacoepidemiologia, Av. Antônio Carlos, 6627, Campus Pampulha, Belo Horizonte/MG, Brazil; Tel (31)3409-6861; Fax: (31)33574868. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.028 S72 VALUE IN HEALTH 14 (2011) S71–S77 Tabela 1 – Distribuição dos gastos individuais com medicamentos de Alto Custo, de acordo com o sexo e ano de seguimento. Ministério da Saúde, 2000 –2004. Seguimento Feminino (n⫽388.471) Masculino (n⫽222.948) Gasto mensal per capita (R$) 1° ano 2° ano 3° ano 4° ano 5° ano Geral Média Mín Máximo DP Média Mín Máximo DP P valor 325,74 314,91 318,01 321,87 381,47 312,39 0,06 0,06 0,06 0,19 0,17 0,06 48.684,16 49.406,78 48.143,88 46.382,28 47.313,40 48.684,16 1.062,86 1.077,40 1.212,59 1.243,55 1.406,78 1.021,18 474,44 431,71 426,29 417,27 447,28 458,47 0,10 0,23 0,22 0,22 0,22 0,10 47.001,71 49.547,71 39.729,24 41.070,18 47.765,26 47.001,71 1.214,74 1.099,40 1.157,68 1.092,03 1.080,30 1.184,07 * * * * * * Nota: * ⫽ p valor ⬍0,05 com teste t / Todos os registros de gastos foram atualizados para dezembro de 2006 / mín⫽ mínimo / DP ⫽ Desvio padrão. mínimo, dois registros de gastos no primeiro semestre de acompanhamento. terísticas individuais, assim como a identificação de sua trajetória no programa. O número de pacientes cadastrados no PMAC vem crescendo substancialmente nos últimos anos. Tendo em vista os crescentes aportes financeiros destinados ao programa e considerando a escassez de estudos sobre a utilização desses medicamentos, principalmente aqueles enfocando os gastos individuais, torna-se oportuna a realização de uma investigação que avalie esse programa e possa contribuir para uma melhor compreensão do perfil dos seus usuários, dos medicamentos dispensados e dos gastos a ele relacionados. O método de linkage, ou técnica de pareamento determinístico/probabilístico de bases de dados, tem se caracterizado como uma importante ferramenta para encontrar os registros referentes a um mesmo paciente nos arquivos e unificá-los em um único registro, permitindo a realização de estudos de acompanhamento do paciente nos sistemas de informação do Sistema Único de Saúde (SUS) [7]. O conhecimento do padrão de utilização de serviços e ações de saúde é essencial para sua estruturação e para que as decisões em relação aos custos sejam equânimes e efetivas [8]. O objetivo desse estudo é descrever os gastos do Ministério da Saúde com usuários do PMAC que iniciaram o tratamento entre 2000–2004, além de traçar o perfil epidemiológico desses pacientes. Gastos privados com medicamentos não foram incluídos na análise. Fonte de dados Foi construída uma Base Nacional de Usuários de Medicamentos de Alto Custo a partir dos registros existentes no banco de dados da APAC do Sistema de Informação Ambulatorial do SUS, utilizando-se o método de linkage determinístico-probabilístico. Consideraram-se somente os procedimentos com o código 36, relativos aos medicamentos de alto custo. O linkage teve como objetivo encontrar os registros para um mesmo paciente nos arquivos e unificá-los em um único registro. No processo de comparação, realizou-se a blocagem do arquivo unificado. Adicionalmente, compararam-se os registros com base em campos específicos através de escores, utilizando metodologia probabilística. Os campos usados foram município de residência, logradouro, UF de nascimento, dia, ano e mês de nascimento, sexo, CPF, nome, sobrenome(s) do meio, último sobrenome, nome da mãe, sobrenome(s) do meio da mãe e último sobrenome da mãe. Os procedimentos metodológicos adotados foram os mesmos aplicados na construção da “Base Nacional em Terapia Renal Substitutiva” e estão detalhadamente descritos em Cherchiglia et al. [7]. Adicionalmente, foi realizado o mesmo procedimento de linkage dos registros da Base Nacional de Usuários de Medicamentos de Alto Custo com o Sistema de Informação de Mortalidade (SIM). O objetivo deste relacionamento foi validar as informações de mortalidade dessa base de dados com as informações oriundas do SIM, também adotando procedimentos semelhantes aos aplicados na “Base Nacional em Terapia Renal Substitutiva”. A limpeza e padronização dos dados constituíram a etapa mais importante e trabalhosa desse processo, dada a grande freqüência de dados incon- Métodos Desenho e população do estudo Coorte histórica de todos os pacientes cobertos pelo PMAC que iniciaram tratamento no período de 2000 –2004 e apresentaram, no Tabela 2 – Distribuição da média dos gastos do Ministério da Saúde com medicamentos de Alto Custo de 2000 –2004, de acordo com a idade dos indivíduos no decorrer do período de seguimento. Seguimento Idade Até 47 anos (n⫽303.860) P valor Maiores de 48 anos (n⫽307.009) Gasto mensal per capita (R$) 1° ano 2° ano 3° ano 4° ano 5° ano Geral Média Mín Máximo DP Média Mín Máximo DP 479,85 493,39 493,74 481,30 521,49 462,16 0,06 0,06 0,22 0,22 0,17 0,06 48.684,16 49.547,71 47.011,69 46.382,28 47.765,26 48.684,16 1.306,64 1.350,91 1.452,46 1.402,67 1.489,91 1.254,12 281,07 229,24 212,46 210,19 247,52 270,10 0,06 0,23 0,06 0,19 0,19 0,06 47.624,31 45.961,76 48.143,88 44.523,72 40.313,09 47.624,31 894,10 713,45 768,44 785,48 810,63 877,55 Nota: * ⫽ p valor ⬍0,05 com teste t / Todos os registros de gastos foram atualizados para dezembro de 2006 / mín⫽ mínimo / DP ⫽ Desvio padrão. * * * * * * S73 CO*, NE* NE*, N* CO*, NE* CO*, NE* CO*, NE* NE* 1.106,79 1.025,18 1.049,33 972,80 1.020,46 1.044,29 NE* CO*, NE* NE* Nota: * ⫽ p valor ⬍0,05 com teste t para múltiplas comparações / Todos os registros de gastos foram atualizados para dezembro de 2006 / DP⫽Desvio padrão. 380,46 368,80 370,62 376,73 425,45 359,43 1.030,74 926,67 986,81 1.005,52 1.491,24 985,64 365,69 326,62 330,54 337,67 441,12 355,22 914,35 845,59 940,47 876,04 888,09 889,04 342,24 308,32 285,09 255,11 291,41 328,11 980,81 1.068,04 1.065,16 1.103,58 867,09 940,47 354,26 337,67 294,44 279,76 282,37 337,09 1° ano 2° ano 3° ano 4° ano 5° ano Geral Média DP Média DP NE* 414,52 400,74 417,77 424,30 471,98 401,46 DP Média Sudeste (SE) 1.292,12 1.250,08 1.377,47 1.360,05 1.459,26 1.255,53 CO*, NE*, N*, S* CO*, NE*, N*, S* CO*, NE*, N*, S* CO*, NE*, N*, S* CO*, NE*, S* CO*, NE*, N*, S* Média Sul (S) DP P valor Gasto mensal per capita (R$) P valor DP A população atendida pelo programa no período de 2000–2004 foi 611.419. O gasto total no período de 2000–2004 com medicamentos foi de R$2.931.351.490,21; gasto total per capita de R$4.794,34⫾20.992,21 (amplitude 2.115.922,46) e mediana de R$1.006,22. Mulheres representavam 63,5% da coorte. O gasto mensal per capita com medicamentos foi maior em indivíduos do sexo masculino durante o período de seguimento (Tabela 1). A idade média dos indivíduos foi de 46,46⫾20,41 e mediana 48 anos. O gasto médio mensal com medicamentos foi maior para indivíduos com idade até 47 anos, comparando-se aos maiores de 48 anos, durante todo o período de seguimento (Tabela 2). Em relação à região de residência, 41,9% dos indivíduos residiam na região sudeste, 29,7% na nordeste, 11,0% na sul, 10% na norte e 7,4% na região centro-oeste. A tabela 3 demonstra os gastos médios mensais com medicamentos nas regiões do Brasil. A região sudeste apresentou os maiores gastos médios com medicamentos e a região nordeste, os menores durante todo o período de seguimento. Outras regiões como norte e sul também apresentaram maiores gastos que a centro-oeste, nordeste e norte em determinados períodos de seguimentos (Tabela 3). Média Resultados Norte (N) O estudo foi submetido e aprovado pela Comissão de Ética em Pesquisa da Universidade Federal de Minas Gerais (Parecer n° ETIC0101/06). Nordeste (NE) Aspectos Éticos C-oeste (CO) A análise descritiva dos dados incluiu a distribuição de frequência para variáveis demográficas, clínicas e de medicamentos. Para a variável gasto individual realizaram-se medidas de tendência central e dispersão para cada variável e estratificado pelo período de seguimento. Teste-t e teste-t para múltiplas comparações foram utilizados para comparar diferenças entre médias. O nível de significância adotado foi 5%. Não foi realizada comparação estatística de médias para variáveis referentes ao diagnóstico e medicamentos devido ao elevado número de categorias apresentadas. Todas as análises foram realizadas com o software SPSS®17. Gasto mensal per capita (R$) Análise dos dados P valor Para o cálculo dos gastos foi realizada a soma dos gastos individuais por período de seguimento dividida pelo número de meses em que houve registro de gastos, obtendo-se desta forma o gasto médio per capita. Todos os gastos foram atualizados para dezembro de 2006, de acordo com o Índice de Preços ao Consumidor Amplo (IPCA) [10] e descritos na moeda brasileira (real – R$). Os gastos individuais foram estratificados por período de seguimento e descritos de acordo com as seguintes categorias de variáveis: 1) demográficas: sexo, idade (⬍47 e ⱖ48 anos, valor da mediana utilizado como ponto de corte), região de residência no início do tratamento, ano de início de tratamento (definido pela data em que o indivíduo recebeu a primeira medicação no programa: 2000 –2004); 2) clínicas: diagnóstico no início do tratamento segundo a Classificação Internacional de Doenças (CID-10) (capítulos e diagnósticos com maiores gastos), óbitos (sim/não); e 3) Medicamentos (terceira categoria da Anatomical Therapeutic Chemical Code (ATC) e princípios ativos com maiores gastos). Gasto mensal per capita (R$) Variáveis Seguimento sistentes, incompletos ou com erros de grafia. No SIM, cerca de 25% dos registros apresentam alguma informação inconsistente ou ausente, o que obriga que o software de relacionamento permita a identificação e tratamento, de forma diferenciada, de valores missing ou ausentes. Detalhes dos procedimentos metodológicos estão descritos em Queiroz et al. [9]. Tabela 3 – Distribuição da média dos gastos do Ministério da Saúde com medicamentos de Alto Custo, de 2000 –2004, de acordo com a região de residência dos indivíduos no primeiro registro na APAC no decorrer do período de seguimento. VALUE IN HEALTH 14 (2011) S71–S77 S74 VALUE IN HEALTH 14 (2011) S71–S77 Tabela 4 – Freqüência e distribuição da média dos gastos do Ministério da Saúde com medicamentos de Alto Custo no primeiro ano de acompanhamento dos indivíduos, de acordo com o diagnóstico agrupados em capítulos da CID-10. Brasil, 2000 –2004. Diagnóstico I II III IV V VI VII IX X XI XII XIII XIV XV XVI XVII XVIII XIX XX XXI N Algumas doenças infecciosas e parasitárias Neoplasias Doenças do sangue e dos órgãos hematopoiéticos e alguns transtornos imunitários Doenças endócrinas, nutricionais e metabólicas Transtornos mentais e comportamentais Doenças do sistema nervoso Doenças do olho e anexos Doenças do aparelho circulatório Doenças do aparelho respiratório Doenças do aparelho digestivo Doenças da pele e do tecido subcutâneo Doenças do sistema osteomuscular e do tecido conjuntivo Doenças do aparelho geniturinário Gravidez, parto e puerpério Algumas afecções originadas no período perinatal Malformações congênitas, deformidades e anomalias cromossômicas Sintomas, sinais e achados anormais de exames clínicos e de laboratório não classificados em outra parte Lesões, envenenamento e algumas outras consequências de causas externas Causas externas de morbidade e de mortalidade Fatores que influenciam o estado de saúde e o contato com os serviços de saúde N (%) Gasto mensal per capita (R$) Média DP 31.345 7380 7527 5,1 1,2 1,2 1.385,84 342,44 1.143,48 2.327,15 551,00 2.213,37 84.790 60.841 55.248 372 12.142 14.455 17.734 31.156 132.608 135.144 48 4 2.563 13,9 10,0 9,0 0,1 2,0 2,4 2,9 5,1 21,7 22,1 0,0 0,0 0,4 410,36 240,61 619,59 379,16 421,22 786,55 171,86 203,65 83,54 315,71 429,77 404,94 538,11 1.813,53 217,79 1.326,22 344,42 244,46 1.544,87 551,28 216,16 374,99 634,20 179,87 406,31 549,38 624 0,1 708,20 526,03 124 0,0 328,95 434,59 11 17.303 0,0 2,8 544,90 718,07 344,05 683,52 Nota: Todos os registros de gastos foram atualizados para dezembro de 2006 / DP ⫽ Desvio padrão. Em relação à data de início de tratamento, 24,5% dos indivíduos iniciaram o tratamento em 2000; 12,4% em 2001, 16,5% em 2002, 22,9% em 2003, e o restante (23,6%) em 2004. Em geral, observa-se que as maiores médias de gastos per capita ocorreram para indivíduos que iniciaram o tratamento em 2000, independentemente do período de seguimento. Para o primeiro ano de seguimento, a média dos gastos para indivíduos que iniciaram tratamento em 2000 foi de R$517,75⫾1.226,81 e para indivíduos que iniciaram em 2001 R$366,35⫾975,99, R$291,60⫾926,30 (2002), R$323,81⫾1.090,82 (2003) e R$360,19⫾1.220,97 (2004). Para o segundo ano de seguimento, a média dos gastos para indivíduos que iniciaram o tratamento em 2000 foi R$477,58⫾1.234,20 e para os anos subseqüentes foram de R$295,51⫾966,52 (2001), R$285,76⫾994,17 (2002) e R$286,92⫾976,12 (2003). Para o terceiro ano de seguimento, a média de gastos para indivíduos que iniciaram o tratamento em 2000 foi de R$400,67⫾1.277,64, maior que indivíduos que iniciaram em 2001 (R$327,64⫾1108,14) e 2002 (R$308,05⫾1059,83). Para o quarto ano de seguimento, a média dos gastos foi R$371,73⫾1.252,93 para indivíduos que iniciaram o tratamento em 2000 e menor para os que iniciaram em 2001 (R$337,51⫾963,09). Observou-se maior prevalência dos diagnósticos agrupados nos capítulos da CID-10 referentes às doenças do aparelho geniturinário, doenças do sistema osteomuscular e do tecido conjuntivo e doenças endócrinas, nutricionais e metabólicas, totalizando 57,7%. Em relação aos gastos, os maiores valores de médias foram observadas para indivíduos com diagnósticos de doenças infecciosas e parasitárias, doenças do sangue e dos órgãos hematopoiéticos e alguns transtornos imunitários e doenças do aparelho respiratório (Tabela 4). Os diagnósticos que apresentaram as maiores médias de gastos foram outras esfingolipidoses (R$21.867,36⫾ 10.397,31), síndrome Di George (R$6.806,28⫾11.668,42), anemia hemolítica auto-imune induzida por droga (R$5.609,17⫾6.765,84), outras anemias hemolíticas auto-imunes (R$5.179,72⫾4.486,30), púrpuras trombocitopênica idiopática (R$4.552,59⫾4.743,31), síndrome de Guillain-Barré (R$4.492,50⫾4.313,56), outras deficiências imunitárias combinadas (R$3.987,23⫾4.995,15), miastenia gravis (R$3.484,67⫾4.836,10) e esclerose múltipla (R$3.258,43⫾1.429,95). De acordo com as categorias da ATC, os grupos químicos de medicamentos que tiveram maior prevalência de uso foram as preparações antianêmicas, drogas que afetam a estrutura óssea e a mineralização e os antipsicóticos. As maiores médias de gastos foram para os componentes sanguíneos e soluções perfusionadas, hormônios hipotalâmicos e imunoglobulinas (Tabela 5). Considerando-se os princípios ativos, as maiores médias de gastos foram para indivíduos que utilizavam os seguintes medicamentos no início do tratamento: imiglucerase (R$22.211,41⫾10.270,03), infliximab (R$5.945,58⫾3.010,77), interferon ␣-peguilado (R$5.599,29⫾2.449,74), interferon -1a (R$3.473,07⫾1.579,15), octreotida (R$3.231,10⫾ 2.099,96), interferon -1b (R$3.166,57⫾1.147,00), salmeterol (R$3.124,22⫾2.451,61), octreotida Lar (R$3.119,92⫾2.189,26), dornase-␣ (R$2.572,13⫾1.338,26) e imunoglobulina (R$2.416,64⫾ 3.727,35). É interessante notar que imiglucerase, interferon ␣-peguilado e interferon -1a estão entre os medicamentos que geram os maiores gastos totais no Programa, pelo Ministério da Saúde. Detectou-se 40.941 óbitos (6,7% do total). Em relação aos gastos, no primeiro ano de seguimento dos indivíduos, não houve diferença estatisticamente significativa entre as médias dos gastos dos indivíduos que evoluíram para óbito ou não. Nos seguimentos subseqüentes, observou-se que as médias dos gastos foram maiores para indivíduos que sobreviveram (Tabela 6). S75 VALUE IN HEALTH 14 (2011) S71–S77 Tabela 5 – Freqüência e distribuição da média dos gastos do Ministério da Saúde com medicamentos de Alto Custo no primeiro ano de acompanhamento dos indivíduos, de acordo com seus grupos químicos (ATC). Brasil, 2000 –2004. Grupo químico Componente sanguíneo e soluções pefusionadas Hormônios Hipotalâmicos Imunoglobulinas Citocinas e Imunomoduladores Antivirais de ação direta Supressores da tosse, exceto combinações com expectorantes Relaxantes musculares, agentes de ação periférica Quaisquer outros produtos terapêuticos Adrenérgicos, Inalantes Digestivos, incluindo enzimas Outras drogas do sistema nervoso Hormônios do lobo pituitário anterior e análogos Digestivos, incluindo enzimas Agentes Imunossupressores Outras drogas sistêmicas para doenças obstrutivas das vias aéreas Hormônios do lobo pituitário posterior Antibacteriano Quinolona Agentes Alquilantes Antimetabólitos Outras preparações antianêmicas Contraceptivos Hormonais de uso sistêmico Drogas anti-demência Drogas usadas em desordens Aditivas Fórmulas infantis Antipsicóticos Vitaminas A e D, incluindo combinações das duas Antipsóricos de uso sistêmico Preparações antiacne de uso sistêmico Preparações de ferro Antiepléticos Hormônios anti-paratireóide Antidiarréico, Antiinflamatório intestinal/agentes Antiinfecciosos Agentes Anti-reumáticos Específicos Outros hormônios sexuais e moduladores do sistema genital Agentes Dopaminérgicos Corticosteróides de uso sistêmico Outros produtos ginecológicos Opióides Redutores de colesterol e triglicérides Agentes Anticolinérgicos Antimalárico Missings Drogas que afetam a estrutura óssea e a mineralização Preparações para tireóide Total N 466 707 3.504 28.549 8.178 1.179 5.078 1.388 10.889 1.221 1.133 10.576 1.812 44.346 3.506 2.752 61 54.157 118 72.497 2.562 13.674 142 863 60.886 44.451 12.748 17.235 10.583 17.506 28.651 15.091 1827 11.663 11.207 239 11.168 895 26.884 521 1839 564 66.614 1.489 611.419 N% 0,1 0,1 0,6 4,7 1,3 0,2 0,8 0,2 1,8 0,2 0,2 1,7 0,3 7,3 0,6 0,5 0,0 8,9 0,0 11,9 0,4 2,2 0,0 0,1 10,0 7,3 2,1 2,8 1,7 2,9 4,7 2,5 0,3 1,9 1,8 0,0 1,8 0,1 4,4 0,1 0,3 0,1 10,9 0,2 100,0 Gasto mensal per capita (R$) Média DP 22.211,41 3.173,69 2.416,47 1.779,16 1.390,37 1.076,40 1.010,89 974,46 902,64 877,08 849,81 780,61 490,34 479,01 432,72 408,29 386,84 380,03 349,54 302,98 285,00 281,84 278,06 254,33 240,42 229,79 208,86 189,93 175,81 173,32 131,49 128,18 116,09 112,21 111,75 104,15 96,46 92,55 64,45 57,48 42,14 37,08 26,50 14,06 379,96 10.270,03 2.143,92 3.725,79 2.238,16 2.336,40 1.490,97 1.878,57 1.000,39 1.657,31 1.135,54 184,55 593,43 693,89 790,67 1.060,01 1.028,89 895,19 691,18 1.075,00 234,21 668,34 835,56 607,31 217,51 214,52 239,06 271,14 143,76 214,58 169,20 213,58 233,99 223,44 189,41 139,32 353,17 206,20 206,36 110,81 68,49 324,12 112,49 60,56 95,80 1.122,91 Nota: Todos os registros de gastos foram atualizados para dezembro de 2006 / DP ⫽ Desvio padrão. Discussão No geral, os gastos per capita com medicamentos de Alto Custo foram elevados (média de R$4.794,34/ano) comparando-se ao gasto per capita em saúde (R$216,30 em 2002) [11]. Embora não diretamente comparáveis, apresentam-se também superiores aos gastos públicos per capita com medicamentos pelos países de alta renda, em 2003, estimados pela Organização Mundial da Saúde em US$73.6 (24.6, 204.3) [12]. Superam ainda os gastos per capita com medicamentos e outros produtos médicos não duráveis observados em 2004 nos países da Organização para a Cooperação e Desenvolvimento Econômico (OCDE) com economias de alta renda, que variaram entre US$212.00 (Nova Zelândia) e US$756.00 (Estados Unidos) (valores em dólares, paridade do poder de compra) [13]. Observaram-se maiores médias de gastos per capita para indivíduos do sexo masculino e maior utilização do programa por indivíduos do sexo feminino. Na literatura já está bem estabelecido o maior uso de serviços pelas mulheres [14 –16]. Mas geralmente elas apresentam mais doenças crônicas do que os homens e com menor gravidade [17,18], explicando, em parte, o maior gasto entre indivíduos do sexo masculino. Maiores gastos ocorreram em indivíduos mais jovens. Este achado está discrepante com a literatura, que mostra que idades mais avançadas requerem maior utilização de serviços, internações S76 VALUE IN HEALTH 14 (2011) S71–S77 Tabela 6 – Distribuição dos gastos de acordo com a ocorrência ou não de óbitos em indivíduos atendidos pelo Programa de Medicamentos de Alto Custo do Ministério da Saúde. Brasil, 2000 –2004. Período de seguimento Óbito P valor Não Sim Gasto mensal per capita (R$) 1° ano 2° ano 3° ano 4° ano 5° ano Geral Média Mín Máximo DP Média Mín Máximo DP 379,87 363,02 369,32 370,85 415,15 366,60 0,06 0,06 0,06 0,19 0,17 0,06 48.684,16 49.547,71 48.143,88 44.523,72 47.765,26 48.684,16 1.141,75 1.112,12 1.215,24 1.195,54 1.281,94 1.103,36 381,28 337,06 273,65 228,63 292,76 352,60 0,24 0,24 0,88 2,97 0,73 0,24 44.811,30 38.257,15 40.117,34 46.382,28 21.679,41 40.983,58 816,49 721,37 823,45 899,89 731,65 799,86 * * * * * Nota: * ⫽ p valor ⬍0,05 com teste t / Todos os registros de gastos foram atualizados para dezembro de 2006 / Mín⫽ mínimo / DP ⫽ Desvio padrão. hospitalares, maior uso de medicamentos [19], e consequentemente, maiores gastos. Todavia, para essa avaliação devem-se considerar os diferentes diagnósticos, suas características e custos, além de aspectos relacionados ao acesso ao programa. Uma avaliação em âmbito nacional do PMAC indicou a necessidade de definição, pelos Estados, de arranjos para melhorar o acesso, como a adoção de um grau de desconcentração na prestação do serviço que não tornasse demasiadamente oneroso o deslocamento dos doentes e que fosse compatível com a sua condição clínica. Pacientes que residem distante dos centros de dispensação geralmente são obrigados a superar obstáculos adicionais para terem acesso aos medicamentos, os quais se somam àqueles decorrentes de sua condição de enfermos [20]. É razoável supor que esses obstáculos apresentem-se com maior intensidade aos pacientes mais idosos, especialmente aqueles mais debilitados, portadores de condições de saúde que, em geral, requerem maiores gastos. Corrobora esta suposição uma pesquisa desenvolvida em cinco países que constatou que consultas a médicos no consultório, hospital ou telefone diminuem com a idade, aumentando as consultas no domicílio [21]. Análise de dados da Pesquisa Nacional por Amostra de Domicílios, realizada no Brasil em 1998, mostrou que a procura por atendimento médico entre homens e mulheres acima de 60 anos não aumentou com a idade [22]. Alguns estudos nacionais que analisaram medicamentos de alto custo para o tratamento das terapias renais substitutivas [23] e hepatite crônica B [24] não observaram diferenças nos gastos considerando-se a idade. Outras avaliações mais especificas seriam necessárias para se verificar possíveis diferenças relacionadas à idade. As regiões sudeste e nordeste apresentaram, respectivamente, as maiores e as menores médias de gastos com medicamentos. Esse fato pode estar relacionado à organização dos serviços de saúde – recursos disponíveis e características da oferta (disponibilidade de médicos, hospitais, ambulatórios) [25], que mostra condições mais favoráveis nas regiões sudeste e sul. Não obstante, constatou-se a inexistência de uma organização padrão para o programa no Brasil, sendo imprescindível um diagnóstico dos serviços prestados, como pressuposto para a discussão da estrutura organizacional necessária para a oferta adequada deste serviço [20]. Outra questão que pode explicar essas diferenças são as inequidades regionais, como observado em estudo que descreveu os gastos per capita com atenção ambulatorial e hospitalar. Foram detectados maiores gastos na região sudeste e sul comparando-se aos gastos na região nordeste, centro-oeste e norte [26]. Além disso, apesar de existirem protocolos clínicos, no quais são indicados diferentes medicamentos para o tratamento de doenças, a seleção da alternativa terapêutica depende da pactuação de cada estado e/ou fica a critério do médico prescritor, a exemplo do que ocorre com os medicamentos para tratamento da osteoporose [27]. Maiores médias de gastos per capita ocorreram para indivíduos que iniciaram o tratamento em 2000. Isto pode ser explicado em parte porque foi somente após a Política Nacional de Medicamentos (1998) que houve a criação ou ampliação de programas destinados a garantir o acesso da população a medicamentos [28]. Assim, é certo que os anos subseqüentes caracterizaram-se pelo rápido aumento da alocação de recursos com a finalidade de suprir a demanda por medicamentos [29]. Os indivíduos que apresentaram maiores médias de gastos tinham diagnósticos, geralmente, de doenças órfãs, raras e genéticas. O tratamento dessas doenças pelo SUS representa um grande avanço no acesso a medicamentos no sistema público de saúde, uma vez que a maioria da população não tem como arcar com os custos do tratamento e a não provisão de uma terapia farmacológica implica em aumento da morbi-mortalidade [30]. Ademais, o gasto com medicamentos pode diminuir outros gastos no cuidado com a saúde, sem afetar a saúde da população, sendo uma estratégia efetiva para diminuir a média geral dos gastos com a saúde [31]. No primeiro ano de seguimento dos indivíduos que evoluíram para óbito não houve diferença entre as médias dos gastos comparando-se aos que não morreram. Do segundo ao quinto ano, as médias dos gastos foram maiores para indivíduos que sobreviveram. Isto pode estar ocorrendo em virtude das diferenças de tecnologias utilizadas, sendo necessárias análises específicas por diagnóstico para avaliar essas diferenças, que não é o foco deste estudo. As avaliações econômicas baseiam-se no custo de oportunidade, isto é, a aplicação de recursos em determinadas tecnologias implica em não-provisão de outras [32]. Nesse sentido, deve-se dar atenção aos medicamentos que apresentam alto valor agregado, como a imiglucerase, de forma a evitar desperdício de recursos. Medicamentos que apresentam alto custo total em virtude do alto consumo também devem ter estratégias específicas de programação e aquisição de forma a permitir compras a um menor custo possível. A organização da Assistência Farmacêutica é um dos componentes principais nesse aspecto, que enfoca desde a programação de medicamentos até seu uso racional, permitindo uma alocação eficiente e obtendo-se o maior valor dos recursos empregados. Dado o alto grau de inovação tecnológica do setor saúde, novos procedimentos são incorporados, muitas vezes de forma acelerada e mesmo antes que evidências suficientes comprovem sua segurança, eficácia e efetividade [33]. E, frequentemente, estão associadas ao aumento dos custos em relação a tecnologias préexistentes. Acresce-se que os recursos da sociedade são sempre finitos frente à sua demanda. Como conseqüência, existe um per- VALUE IN HEALTH 14 (2011) S71–S77 manente conflito entre uso de recursos e necessidade de escolher entre alocações alternativas [33]. As avaliações econômicas em saúde, como as avaliações de custo-efetividade e custo-utilidade buscam analisar se os benefícios incorporados pelas novas tecnologias compensam seus custos adicionais. Nesse sentido, estudos de gastos, como o presente trabalho, são úteis para se definir o perfil dos gastos e estabelecer quais grupos de doenças deverão ser foco de avaliação. Doenças que apresentam elevado gasto médio individual, para as quais existem diferentes estratégias terapêuticas, devem ser alvos prioritários de avaliação. Ademais, os resultados apresentados poderão servir de subsídio na composição dos custos para diferentes avaliações econômicas. Uma das limitações do estudo consiste na utilização de base de dados administrativos, considerando os seus aspectos estruturais – como lacunas de informação clínica, dificuldades na codificação dos procedimentos e o caráter de faturamento, que restringem a possibilidade de desenvolvimento de avaliações a partir dessas informações. Outra limitação diz respeito à desatualização dos dados, dado que o programa a partir de 2006 sofreu alterações tanto na estrutura quanto na forma de financiamento. Apesar disso, destaca-se a escassez de estudos dessa natureza, a dificuldade em se trabalhar com pareamento de bases de dados tão extensas e com tantas peculiaridades e a grande potencialidade dos dados disponíveis em traçar a trajetória do usuário nos serviços de saúde, conforme descrito em estudos que utilizaram bases de dados administrativas [7,34]. Estima-se que o SUS seja responsável pela provisão da atenção à saúde a 75,5% dos cidadãos brasileiros [35] e os medicamentos representam parcela substancial dos gastos públicos em saúde. Uma melhor compreensão do perfil dos pacientes que utilizam o PMAC, assim como os gastos envolvidos, pode subsidiar ações que objetivem maior eficiência das políticas públicas de saúde, sem prejuízo de sua efetividade. Fontes de financiamento: Este artigo apresenta resultados do projeto de pesquisa “Avaliação Farmacoeconômica e Epidemiológica do Programa de Medicamentos Excepcionais do SUS- Brasil”. Foi realizado com o apoio financeiro do Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) da Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG) e parceria da Secretaria de Estado da Saúde de Minas Gerais (SES/MG) e Ministério da Saúde. [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] REFERENCIAS [26] [1] Brasil. Constituição da República Federativa do Brasil. Diário Oficial da União, 05 out 1988. [2] Brasil. Conselho Nacional de Secretários de Saúde. Documenta 20. Brasília: CONASS, 2010. 108p. [3] Brasil. Portaria GM n° 2.577 de 2006. Aprova o Componente de Medicamentos de Dispensação Excepcional. Diário Oficial da União, 27 out 2006. [4] Brasil. Portaria n°2981. Aprova o Componente Especializado da Assistência Farmacêutica. Diário Oficial da União, 01 dez 2009. [5] Brasil. Conselho Nacional de Secretários de Saúde. Assistência Farmacêutica no Sistema Único de Saúde. Brasília: CONASS, 2007. [6] Brasil. Conselho Nacional de Secretários de Saúde. Documenta 5. Brasília: CONASS, 2004. 64p. [7] Cherchiglia ML, Guerra Júnior AA, Andrade EIG, et al. A construção da base de dados nacional em terapia renal substitutiva (TRS) centrada no indivíduo: aplicação do método de linkage determinísticoprobabilístico. R Bras Est Pop 2007;24:163–7. [8] Sawyer DO, Leite IC, Alexandrino R. Perfis de utilização do serviços de saúde no Brasil. Cien Saude Colet 2002;7:757–76. [9] Queiroz OV, Guerra Júnior AA, Machado CJ, et al. A construção da Base Nacional de Dados em Terapia Renal Substitutiva (TRS) centrada no indivíduo: relacionamento dos registros de óbitos pelo subsistema de [27] [28] [29] [30] [31] [32] [33] [34] [35] S77 Autorização de Procedimentos de Alta Complexidade (Apac/SIA/SUS) e pelo Sistema de Informações sobre Mortalidade (SIM) – Brasil, 2000 –2004. Epidemiol Serv Saude 2009;18:193– 6. Instituto Brasileiro de Geografia e Estatística. Indicadores de Preços. Disponível em: http://www.ibge.gov.br/home/estatistica/ indicadores/precos/ipca15/defaultipca15.shtm. [Acesso em January 2008]. World Health Organization (WHO). National Health Accounts. Available from: http://www.who.int/nha/country/bra/en/. [Accessed October 16, 2009]. World Health Organization (WHO). Using indicators to measure country pharmaceutical situations. Fact Book on WHO Level I and Level II monitoring indicators. Geneve: WHO, 2006. OECD Health data 2010. Frequently requested data. Available from: http://www.oecd.org/document/16/0,3746,en_2649_37407_2085200_1_ 1_1_37407,00.html. [Accessed January 5, 2010]. Lima-Costa MF, Loyola Filho AI. Fatores associados ao uso e à satisfação com os serviços de saúde entre usuários do Sistema Único de Saúde na região metropolitana de Belo Horizonte, Estado de Minas Gerais, Brasil. Epidemiol Serv Saúde 2008;17:247–57. Verbrugge LM. The Twain meet: empirical explanations of sex differences in health and mortality. J Health Soc Behav 1989;30:282– 304. Capilheira MF, Santos IS. Fatores individuais associados à utilização de consultas médicas por adultos. Rev Saude Publica 2006;40:436 – 43. Travassos C, Viacava F, Pinheiro R, Brito A. Utilização dos serviços de saúde no Brasil: gênero, características familiares e condição social. Rev Panam Salud Publica 2002;11:365–73. Mendoza-Sassi R, Lberia JU. Utilización de los servicios de salud: uma revisión sistemática sobre los factores relacionados. Cad Saude Publica 2001;17:819 –32. Pinheiro RS, Viacava F, Travassos C, et al. Gênero, morbidade, acesso e utilização de serviços de saúde no Brasil. Cien Saude Colet 2002;7:687– 707. Blatt CR, Farias MR. Diagnóstico do Programa de Medicamentos Excepcionais do Estado de Santa Catarina - Brasil. Lat Am J Pharm 2007;26:776 – 83. Rowland D. A Five-nation perspective on the elderly. Health Aff 1992; 11:205–15. Lima-Costa MF, Barreto SM, Giatti L. Condições de saúde, capacidade funcional, uso de serviços de saúde e gastos com medicamentos da população idosa brasileira: um estudo descritivo baseado na Pesquisa Nacional por Amostra de Domicílios. Cad Saude Publica 2003;19: 735– 43. Guerra Junior AA, Acurcio FA, Andrade EIG, et al. Ciclosporina versus tacrolimus no transplante renal no Brasil: uma comparação de custos. Cad Saude Publica 2010;26:163–74. Almeida AM, Cherchiglia ML, Andrade EIG, et al. Epidemiologic profile and expenses of the Unified Health System (SUS) in the Program of High Cost Medicines (PHCM) to the treatment of chronic hepatitis B – 2000 to 2004, in Brazil. JBES 2009;1:80. Travassos C, Martins M. Uma revisão sobre os conceitos de acesso e utilização de serviços de saúde. Cad Saude Publica 2004;20(Suppl.):S190 – 8. Ugá MA, Piola FP, Porto SM, et al. Descentralização e alocação de recursos no âmbito do Sistema Único de Saúde (SUS). Cien Saude Colet 2003;8:417–37. Brasil. Portaria GM n°470. Aprova, na forma do Anexo desta Portaria, o Protocolo Clínico e Diretrizes Terapêuticas - Osteoporose -, Bifosfonados, Calcitonina, Carbonato de Cálcio, Vitamina D, Estrógenos e Raloxifeno. Diário Oficial da União 24 jul 2002. Brasil. Portaria n°3.916. Aprova a Política Nacional de Medicamentos. Diário Oficial da União 10 nov 1998. Vieira FS. Gasto do Ministério da Saúde com medicamentos: tendência dos programas de 2002 a 2007. Rev Saude Publica 2009;43:674 – 81. Araújo DN, Passos RBF, Souza CPR, et al. Financiamento do tratamento de doença de alto custo no Brasil. JBES 2009;1:44 –51. Crémieux PI, Ouellete P, Petit P. Do drug reduce utilization of other healthcare resources? Pharmacoeconomics 2007;25:209 –21. Palmer S, Raftery J. Opportunity cost. BMJ 1999;318:1551–2. Ministério da Saúde. Diretrizes Metodológicas: Estudos de Avaliação Econômica de Tecnologias em Saúde. Brasília, Ministério da Saúde: 2009.150p. Acurcio FA, Brandão CMR, Almeida AM, et al. Perfil demográfico e epidemiológico dos usuários de medicamentos de alto custo no Sistema Único de Saúde. REBEP 2009;26:263– 82. Bahia L, Costa AJL, Fernandes F, et al. Segmentação da demanda dos planos e seguros privados de saúde: uma análise das informações da PNAD/98. Cien Saude Colet 2002;7:671– 86. VALUE IN HEALTH 14 (2011) S78 –S81 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Health Care Resource Use and Costs in Opioid-Treated Patients with and without Constipation in Brazil Maira L. S. Takemoto, MSc, RN1,*, R. A. Fernandes, MD, MSc1, G. R. Almeida, PharmD2, R. D. C. Monteiro, PharmD2, M. Colombini-Neto, MD, MSc3, A. Bertola-Neto, BBA3 1 ANOVA - Knowledge Translation Consulting Group, Rio de Janeiro, Brazil; 2Pfizer Brazil, São Paulo, Brazil; 3AxisMed Gestão Preventiva da Saúde, São Paulo, Brazil A B S T R A C T Objective: To estimate the prevalence of constipation concomitant to opioid treatment and related resource use and costs from the private payer perspective. Methods: In this retrospective database analysis, patients receiving opioid therapy were identified from a longitudinal insurance claims database. An algorithm was used to identify patients receiving opioid therapy with coincident constipation-related claims according to ICD-10 codes, targeted procedures, and opioid use criteria. Resource use and costs were determined for these individuals and compared with patients receiving opioid therapy without constipation, without opioid therapy with constipation, and without both conditions. Results were compared using analysis of variance with a significance level of 0.05. Results: A total of 23,313 patients were classified as opioid-treated patients (2.2%) and 6678 of them had events related to constipation (29.0%). Compared with opioid-treated patients without constipation, incremental mean total costs per month per patient were 261.18 BRL (P ⬍ 0.001). The average cost per month for opioid-related Introduction Opioids are the mainstay therapy for patients with moderate to severe pain. According to the Brazilian Society for Study of Pain [1], the prevalence of chronic pain is about 30% within the country and noncancer pain is responsible for 60% to 70% of chronic pain cases. For these patients opioid treatment must be very carefully monitored and is generally reserved for refractory cases. Thus, it is estimated that about 10% of patients with chronic pain will eventually receive opioid treatment. Of these, 70% had moderate pain, indicating the use of a weak opioid and 30% strong or very strong pain with indication of strong opioids [2]. Pain is present in 30% of cancer patients undergoing chemotherapy and in 60% to 90% of those with advanced cancer [3]. Although effective in pain management [4], opioid therapy is frequently complicated by side effects [5]. With continued use, patients usually develop tolerance to those side effects, except constipation, which is the most common and usually the most constipation patients was 787.84 BRL, significantly higher than other patients (P ⬍ 0.001 for all comparisons). Among cancer patients, 24.4% was receiving opioids and 27.0% of those had constipation-related claims. As expected, the opioid therapy prevalence was significantly higher when compared to all patients (2.2% vs. 24.4%, P ⬍ 0.001). Cancer patients had, in average, higher costs than did noncancer patients in all four subgroups. Conclusions: Patients with constipation coincident with opioid treatment exhibited a significantly higher economic burden than did patients without the condition. These results indicate that reducing opioid-induced constipation could lead to potential cost savings for the health care system. Keywords: analgesics, constipation, costs and cost analysis, drug toxicity, opioid. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. debilitating side effect reported by patients, with a median frequency of 30% among noncancer patients (range 12%–52%) [6]. The prevalence in cancer patients is even higher [7], reaching 63% [8], and laxatives are required by 87% of terminally ill cancer patients taking oral strong opioids and by 74% of those receiving weak opioid therapy [9,10]. Constipation is also associated with a serious negative effects on patients’ health-related quality of life (HRQoL) and on society in terms of health care resource use and work productivity loss [11]. Patients with constipation have more hospital admissions, emergency room visits, home health services, nursing home care, physician visits, and laboratory tests, as well as higher mean all-cause costs for emergency, physician visits, nursing facilities, home health care, and prescription drug services compared to patients without constipation [12]. Further, it is known that constipation usually persists for as long as opioid therapy is administered [8]. In Brazil, opioid consumption was estimated at 1.1520 mg per capita, showing an average prescription below the world average Conflicts of interst: This study was funded in full by a grant from Pfizer, Brazil, to ANOVA. GR Almeida e RDC Monteiro are employees of Pfizer, Brazil. MLS Takemoto and RA Fernandes have served as consultants for different pharmaceutical companies in Brazil, including Pfizer. M Colombini-Neto and A Bertola-Neto have served as consultants for Pfizer, Brazil. * Address correspondence to: Maíra L. S. Takemoto, ANOVA - Knowledge Translation Consulting Group, General Polidoro 154/03, Botafogo, Rio de Janeiro-RJ, Brazil 22251-050. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.019 VALUE IN HEALTH 14 (2011) S78 –S81 consumption (5.5708 mg per capita) and indirect evidence of inadequate pain control in the country [13]. Despite these variations, a large number of patients are currently receiving opioid therapy for chronic pain worldwide. To date, no studies evaluating the prevalence of constipation in patients using opioids have been published in Brazil; neither their related management resource use nor costs. This study aimed to estimate the prevalence of constipation concomitant to opioid treatment and to compare resource use and costs in opioid-treated patients with and without constipation, from the private payer perspective in Brazil. Methods Data source Patients receiving opioid therapy were identified from a longitudinal insurance claims database (Axismed Database) consisting in 1,057,033 individuals observed during a 35-month period (December 2004 to December 2007). This database comprises about 3% of the population covered by health insurance plans across all five geographic regions in Brazil (national coverage). Its general data include patient demographics, medical claims, and enrollment date. Medical claims information includes date, type of procedure, provider specialty, amount paid, and diagnosis code using International Classification of Diseases, 10th revision (ICD-10) [14]. Average patient exposition time was 21.84 months. Study population An algorithm was developed through evidence-based clinical rules and expert opinion to identify patients receiving opioid therapy with coincident constipation-related claims according to ICD-10 codes and constipation-related procedures. Patients were initially segmented according to the presence of opioid therapy records. To be subsequently classified as a patient with constipation concomitant to opioid therapy, patients with evidence of opioid analgesics use had to meet at least one of the following eligibility criteria: at least one medical claim with an ICD-10 code potentially related to constipation (i.e., K59.0, K59.9, or R19.4) and/or at least one medical procedure potentially related to constipation (e.g., enemas or manual, endoscopic, or surgical fecal impaction removal). The individuals in the database were then classified in four groups: nonopioid-treated without constipation (NONC), nonopioid-treated with constipation (NOWC), opioid-treated without constipation (ONC), and opioid treated with constipation (OWC). To ensure more detailed analysis reflecting relevant clinical questions, individual with oncologic ICD-10 codes were separately analyzed, once opioid therapy and opioid-related constipation prevalence is usually higher among cancer patients. Outcomes measures Resource use and costs were collected for each individual during the 35-month follow-up period using a top-down approach [15]. Once individuals were followed for varying amounts of time, these variables (i.e., total costs, costs segmented by category, and resource use segmented by category) were then converted to per member/per month units dividing the results for all patients by the person-months of follow-up. Resource use results are presented as average consumption per month of six claims groups: outpatient procedures, consultations, tests and therapies, hospitalization, emergency department visits, and others. Hospitalizations are reported as hospital days. Costs were collected as reported in the administrative database, reflecting the amount effectively paid by each health insurance plan. Costs are presented in Brazilian reals (BRL), using reference values for 2009. S79 Statistical analyses Medical resource use and costs in patients with constipation coincident with opioid therapy (OWC) were compared with the outcomes observed in the other three groups (NONC, NOWC, and ONC). The per member per month outcomes were compared using analysis of variance, with corresponding P values reported with a significance level of 0.05. If the analysis of variance test indicated a statistically significant difference, further post hoc analyses were performed. SAS (version 9, 2002, SAS Institute Inc., Cary, NC) statistical software was used to perform all analyses. Results All patients In our study, 23,313 patients were classified as opioid-treated patients (2.2% of total population) and 6678 of them had events or ICD-10 codes related to constipation, resulting in a constipation prevalence of 29.0% among opioid-treated patients. The mean age of OWC patients was 51.58 ⫾ 19.35 years, 65.0% was women, 1.0% had cancer diagnosis, and the mean follow-up per individual ranged from 20.03 to 24.44 months in each group. Table 1 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.019 summarizes the baseline characteristics of the study sample, including average costs (mean ⫾ SD) and resource use (mean ⫾ SD) per patient during the follow-up. Mean cost per patient ranged from 2,122 BRL (NONC) to 17,206 BRL (OWC) and the average cost per patient for the entire sample was 2,486 BRL. Table 2 in Supplemental Materials found at: doi:10.1016/j. jval.2011.05.019 compares average cost and resource use per month of OWC patients with the other three subgroups. Compared to opioid-treated patients without constipation, average incremental costs per month per patients with the condition were 261.18 BRL (P ⬍ 0.001). The average cost per month for OWC patients was 787.84 BRL, significantly higher than ONC (526.66 BRL), NOWC (284.47 BRL), and NONC patients (90.17 BRL) (P ⬍ 0.001 for all comparisons). Patients with claims related to both conditions had significantly more days in hospital per month (0.25 vs. 0.497, P ⬍ 0.001), outpatient office visits (1.04 vs. 1.59, P ⬍ 0.001), outpatient procedures (4.69 vs. 14.05, P ⬍ 0.001) and tests and therapies (31.95 vs. 36.66, P ⬍ 0.001) than did patients without opioid-related constipation claims. Cancer patients Cancer patients were considered as a separate subgroup due to the expected higher prevalence of both conditions (opioid therapy and constipation) among those individuals. Oncology ICD-10 codes were identified for 9873 individuals, representing 1.0% of the total population. Among those, 24.4% was receiving opioid therapy and 27.0% of those had constipation-related claims. As expected, the opioid therapy prevalence was significantly higher among cancer patients when compared to all patients (2.21% vs. 24.4%, P ⬍ 0.001). Among opioid-treated patients the prevalence of constipation was similar in both groups (29.0% for all patients and 27.0% for cancer patients), regardless of cancer status. Table 3 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.019 presents the costs and resource use results for the cancer population, considering the same subgroups previously described. Cancer patients had, on average, higher costs than did noncancer patients in all four constipation and opioid status categories. The absolute difference between ONC and OWC patients, however, remains stable when compared to the observed difference in all patients’ analysis (263.21 BRL vs. 261.18 BRL, respectively). When segmented costs were analyzed, OWC patients resulted in higher costs due to tests and therapies and hospital days, but did not for other categories. S80 VALUE IN HEALTH 14 (2011) S78 –S81 Discussion Opioid-related constipation has multidimensional influences on a patient’s health status and consequently results in complex burdens for health care systems. To relieve constipation, patients often abandon their opioid medication, potentially impairing analgesia. Therefore, the burden of constipation can be not only from the direct influence of its symptoms on HRQoL and constipationrelated health care resources consumption, but also from medical resources to relieve the pain and the side effects of treatments taken to relieve the condition. The observed prevalence of constipation in our sample was quite similar to data reported by a meta-analysis focused in older patients without cancer (29.03% vs. 30.00%, respectively) [1]. The subgroup analysis of cancer patients showed a slightly lower prevalence (27.0%), which is different than results previously described, achieving 63% [8]. These published data refer to the higher rate reported by hospice cancer patients in the United States. A study published in 2001 [16] with 593 cancer patients treated by a pain service showed a 23% prevalence of constipation and this symptom was assessed as being frequently caused by the analgesic regimen. Furthermore, much of the variation in the frequency of constipation in patients treated with oipiods can be attributed to study design and population heterogeneity: age, sex, base pathology, type of opioid administered, its dose and duration, and subjective perception of constipation. Our findings indicated that, on average, opioid-treated patients were significantly more costly than patients without opioid-related events, and patients with constipation-related claims resulted in higher medical costs than those without constipation. The incremental costs observed for patients with constipationrelated claims coincident with opioid therapy, when compared to opioid-treated nonconstipated patients, were about 260 BRL per individual per month either for cancer patients or for the entire sample. The similar findings for both subgroups can indicate that the observed difference probably reflects the actual absolute difference in costs due to constipation related to opioid therapy. The effects of constipation in patients using opioids has already been studied in other countries showing similar results. A literature review was conducted to identify national and international cost-of-illness and prevalence studies addressing the burden of opioid-related constipation using a mix of controlled vocabulary and free text terms for constipation, opioid therapy, prevalence, and costs. PubMed and LILACS databases were searched and only US and European studies were found [7,9 – 12,17–22]. In the United States, Bell et al. [11] evaluated the effects of opioid-induced constipation on health care resource use, work productivity, and HRQoL with data from 2430 individuals, of whom 359 reported constipation. Opioid-induced constipation patients reported significantly more physician visits and alternative care visits. Significantly greater productivity loss and significantly lower HRQoL were observed in the constipated group, both signals of a negative influence on individuals’ health status. Iyer et al. [12] also compared the opioid use patterns, resource use, and costs of 39,485 US patients receiving opioid therapy who had constipation with those who did not. Patients with constipation had statistically significant higher resource use and all-cause costs compared to patients without constipation. Those studies showed that opioid-related constipation has a significant influence on costs and resource use in developing countries and it seems reasonable to believe that this association between higher health care resource consumption and opioid-related constipation could be observed in other similar countries, but there is still a lack of evidence concerning this issue in Brazil or Latin America. As stated before, average opioid prescription in Brazil is considered below the world average consumption and this is indirect evidence of inadequate pain control in the country [13]. In addition, resource use and medical costs are directly related to patients’ access to the health care system, reimbursement and coverage processes, local therapeutic patterns, and clinical guidelines, all of which are expected to be significantly different across countries, particularly if they have different health care system organizations. Study limitations include potential selection bias due to retrospective analysis of administrative database, misclassification of patients (ICD-10 codes are not homogeneously used in Brazilian clinical practice, particularly for general conditions such as constipation), and the lack of more detailed baseline clinical information in the original database to provide clinical and demographic variables that could be used to control for confounders. Trying to minimize those limitations, the opioid-treated patients were compared with nonopioid-treated patients and cancer patients were separately analyzed. We hypothesized that higher prevalence of cancer among OWC patients could lead to higher costs due to cancer treatments (e.g., chemotherapy, radiation therapy, and surgery) and not directly to opioid-related constipation. These hypotheses can be rejected once the same difference in costs was observed when only cancer patients were compared. In addition, opioid-treated patients are more likely to be in palliative care (i.e., not receiving high-cost cancer treatment). Retrospective claims database studies are still a novel field of research in Health Economics and Outcomes Research in Brazil and there is a recognizable absence of data concerning patients with private health care plans coverage. Our findings provided the first local overview of the burden associated to opioid-related constipation in Brazil. Further research is needed to validate those findings through primary data collection, preferably in a prospective fashion. Conclusions Patients with constipation coincident with opioid treatment exhibited a significantly higher economic burden than did patients without the condition. These results indicate that reducing opioidinduced constipation could lead to potential cost savings for the health care system. Source of financial support: Pfizer Brazil. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi:10.1016/j.jval.2011.05.019, or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). REFERENCES [1] Sociedade Brasileira para o Estudo da Dor (SBED). Capitulo Brasileiro da International Association for the Study of Pain (IASP). Projeto “Controle da Dor no Brasil”. Available from: http://www.dor.org.br/profissionais/s_projeto. asp. [Accessed June 27, 2010]. [2] Brasil Ministério da Saúde. Departamento Nacional de Auditoria do SUS. Coordenação de Sistemas de Informação - legislação federal. Portaria GM/MS n° 859, de 4 de Novembro de 2002. [3] Brasil Ministério da Saúde. Departamento Nacional de Auditoria do SUS. Coordenação de Sistemas de Informação - legislação federal. Portaria GM/MS n° 1.319, de 23 de Julho de 2002. [4] Kalso E, Edwards JE, Moore RA, McQuay HJ. Opioids in chronic noncancer pain: systematic review of efficacy and safety. Pain 2004;112: 372– 80. [5] Thomas J. Opioid-induced bowel dysfunction. J Pain Symptom Manage 2008;35:103–13. VALUE IN HEALTH 14 (2011) S78 –S81 [6] Papaleontiou M, Henderson CR, Turner BJ, et al. Outcomes associated with opioid use in the treatment of chronic noncancer pain in older adults: a systematic review and meta-analysis. J Am Geriatr Soc 2010; 58:1353– 69. [7] Panchal SJ, Müller-Schwefe P, Wurzelmann JI. Opioid-induced bowel dysfunction: prevalence, pathophysiology and burden. Int J Clin Practice 2007;61:1181–7. [8] McMillan SC. Assessing and managing opiate-induced constipation in adults with cancer. Cancer Control 2004;11(3 Suppl.):3–9. [9] Sykes NP. The relationship between opioid use and laxative use in terminally ill cancer patients. Palliat Med 1998;12:375– 82. [10] Pappagallo M. Incidence, prevalence, and management of opioid bowel dysfunction. Am J Surgery 2001;182(5A Suppl):11S–18S. [11] Bell T, Annunziata K, Leslie JB. Opioid-induced constipation negatively impacts pain management, productivity, and health-related quality of life: findings from the National Health and Wellness Survey. J Opioid Manage 2009;5:137– 44. [12] Iyer S, Davis KL, Candrilli S. Opioid use patterns and health care resource utilization in patients prescribed opioid therapy with and without constipation. Managed Care 2010;19:44 –51. [13] World Health Organization. 2005 Global Consumption of Morphine. 2005; Available from: http://www.painpolicy.wisc.edu/internat/global/ morphine05.pdf. [Accessed June 27, 2010]. [14] World Health Organization. International Classification of Diseases (ICD-10). Available from: http://www.who.int/classifications/icd/en/. [Accessed June 27, 2010]. S81 [15] Rice DP. Estimating the cost of illness. Am J Public Health 1967;57: 424 – 40. [16] Meuser T, Pietruck C, Radbruch L, et al. Symptoms during cancer pain treatment following WHO-guidelines: a longitudinal follow-up study of symptom prevalence, severity and etiology. Pain 2001;93: 247–57. [17] Penning-van Beest FJ, van den Haak P, Klok RM, et al. Quality of life in relation to constipation among opioid users. J Med Econ 2010;13: 129 –35. [18] Wee B, Adams A, Thompson K, et al. How much does it cost a specialist palliative care unit to manage constipation in patients receiving opioid therapy? J Pain Symptom Manage 2010;39:644 –54. [19] Kwong WJ, Diels J, Kavanagh S. Costs of gastrointestinal events after outpatient opioid treatment for non-cancer pain. Ann Pharmacother 2010;44:630 – 40. [20] Brown RT, Zuelsdorff M, Fleming M. Adverse effects and cognitive function among primary care patients taking opioids for chronic nonmalignant pain. J Opioid Manag 2006;2:137– 46. [21] Staats PS, Markowitz J, Schein J. Incidence of constipation associated with long-acting opioid therapy: a comparative study. South Med J 2004;97:129 –34. [22] Candrilli SD, Davis KL, Iyer S. Impact of constipation on opioid use patterns, health care resource utilization, and costs in cancer patients on opioid therapy. J Pain Palliat Care Pharmacother 2009; 23:231– 41. VALUE IN HEALTH 14 (2011) S82–S84 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Ideal Vial Size for Bortezomib: Real-World Data on Waste and Cost Reduction in Treatment of Multiple Myeloma in Brazil Luciana Clark, MD1,*, Ana Paula Castro, RN1, Anna Flávia Fortes, RN1, Fábio Santos1, Otávio Clark, MD, PhD1, Tobias Engel, MD1, Bruna Pegoretti, MD1, Vanessa Teich, BSIE2, Denizar Vianna, MD2, Fabíola Puty, MD1 1 Evidências, Campinas, Brazil; 2MedInsight, São Paulo, Brazil A B S T R A C T Objectives: Single-size vials of drugs may be a source of waste and increase in treatment costs. Bortezomib, indicated for multiple myeloma (MM) treatment, is available in 3.5-mg vials, a quantity higher than the average dose commonly prescribed. This analysis aimed to demonstrate, through real-world data, which would be the optimal vial presentation for bortezomib in Brazil and quantify the reduction in medication waste related to this option. Methods: From November 2007 to October 2009 all patients with MM treated with bortezomib were identified via the Evidências database. Analysis of prescribed, dispensed, and wasted doses, their costs and projections of the ideal vial size were performed. Results: Thirty-five patients (mean body surface area of 1.73 m2) received 509 infusions in 131 cycles of treatment (average of 3.77 cycles per patient). The average dose prescribed was 2.1 mg per infusion (95% confidence interval [CI] 1.97–2.26) with average Introduction Multiple myeloma (MM) is considered a chronic disease for which— despite many advances in therapy—there is currently no curative treatment available. The development of novel agents that target the tumor cell and the microenvironment, immunomodulators, proteasome inhibitors, and bisphosphonates has changed the standards of care for affected patients. Even with the achievement of complete responses in few cases, the ultimate goals for patients with MM are extended survival and quality of life [1–4]. The costs associated with current and emerging therapies, as well as supportive care, are significant and pose a tremendous financial burden to both patients and health care systems [5]. This is an important aspect to be considered, especially in underdeveloped countries and emerging economies in which the bulk of resources destined to health care is often reduced and unevenly distributed. Newer classes of cytotoxic agents are, for the most part, very expensive and there may be considerable resource savings in the judicious application of dose rounding without any negative clinical effect, given the significant interpatient pharmacokinetics and pharmacodynamics variability for most cytotoxic drugs [6]. Single-size vials of chemotherapy drugs may be an undernoticed source of waste and increase in treatment costs. There is, however, the possibility of customizing the dose of chemotherapy for waste of 39.5% of the vial content (95% CI 35.35– 43.76). The mean waste per patient per day was 1.38 mg (95% CI 1.24 –1.52). If a 3-mg vial were available, the average drug waste per patient per day would be 0.88 mg (95% CI 0.74 –1.03) or 36.2% less. With a 2.5-mg vial the waste would be 1.05 mg (95% CI 0.81–1.29) or 23.9% less. If two presentations were available (2.5 mg and 0.5 mg), the waste would be 0.52 mg (95% CI 0.4 – 0.63) or 62.5% less. Considering the price of the different vials to be proportional to the original 3.5-mg vial, the cost would be also reduced by the same rates described above. Conclusions: A simple adjustment in vial size may reduce the waste of bortezomib by 36% to 62% and can also reduce the cost of treatment. Keywords: bortezomib, cost-reduction, drug waste, myeloma. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. each patient not only by the body surface area (BSA) measurement, but also by adjusting the final dose to the nearest vial size. Previous published studies described this kind of adjustment as cost saving in several types of chemotherapy treatments [6,7]. Bortezomib (Velcade, Janssen-Cilag, Beerse, Antwerp, Belgium) is a drug frequently used in MM treatment that is available in Brazil and in many other countries only as a 3.5-mg vial. This presentation dose is higher than the average dose commonly prescribed and due to the lack of preservatives in the vial, it is mandatory that the drug be administered within 8 hours of preparation. This analysis aimed to demonstrate, through real-world data, which would be the optimal vial presentation for bortezomib in Brazil and to quantify the reduction in medication waste and costs related to it. Study design This study is a retrospective analysis of data extracted from Evidências Database (ED), an electronic system design to evaluate chemotherapy requests from health care providers and permit the approval or denial of coverage by auditors. The database is available as a secure Web site where patient data such as age, disease stage, BSA, and drugs requested among other information are prospectively stored. ED covers approximately 5% of the Brazilian pri- Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Luciana Clark, Rua Tranquillo Prosperi, 143, Bairro Santa Genebra II, Campinas – SP, Brazil 13084-778. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.013 VALUE IN HEALTH 14 (2011) S82–S84 vate health care market. For this analysis we used the payer’s perspective on the private health care system. During the period of November 2007 to October 2009, all patients diagnosed with MM (cases entered into ED with the International Statistical Classification of Diseases and Related Health Problems code C90.0), submitted to treatment containing bortezomib for at least one infusion were identified. All patients were treated in private care referral cancer centers, according to usual chemotherapy protocols and identified on the auditing medical system. Details regarding the prescribed dose of bortezomib, individually and collectively wasted doses, and the respective average doses per patient and per cycle were retrieved. The actual dose dispensed to perform the infusion was also calculated. That is important because it is often necessary to open multiple vials to fulfill the prescribed dose, thus accounting for increased waste. The costs of the wasted drug and the projection of costs saved with different vial sizes were calculated. The mean and average calculations were performed based on usual mathematical formulas and were represented as milligrams, US dollars, the Brazilian currency reals (R$), or square meters, according to the variable studied. The results are reported with the corresponding 95% confidence interval (CI). The amount of drug wasted was also expressed in percentage related to the total content of the 3.5-mg vial. The results of costs were expressed in US dollars with the cost of the 3.5-mg bortezomib vial being R$4218.33 or US$2343.51 (exchange rate R$1.80 to US$1 as of May 2010). The price list used was the Brazilian Official Price List [8], determined by the federal government, by the wholesale price plus taxes. Differently sized vials were priced proportionally to the existing vial for economic projections. No patient was identified by name or any particular characteristic during this study. All health care providers previously authorized the data extraction in signed contracts. Because the purpose of this study was to evaluate the optimal presentation for bortezomib, the clinical aspects and the outcomes regarding effectiveness and security of this drug were not taken into consideration. Results Thirty-five patients with MM treated with bortezomib (alone or in association with other cytotoxic drugs) were identified on ED. The mean body surface area was 1.73 m2. The average prescribed dose per infusion was 2.1 mg (95% CI 1.97–2.26) with average waste of 39.5% of the vial content (95% CI 35.35– 43.76) of the contents of a standard 3.5-mg vial per day or 5.39 mg per cycle. The patients received a total of 509 days of infusion distributed in 131 cycles of treatment (average of 3.77 cycles per patient). During the period of time analyzed, a total of 1781.5 mg were dispensed to cover 1075.7 mg prescribed, resulting in a gross loss of 705.7 mg (39.6%) of bortezomib. That means that of the US$1,192,841.50 spent on the drug, US$472,365.23 was wasted. The total amount of resource wasted per day, with all 35 patients, was US$121,420.95. Those data are described in Tables 1 and 2 in the Supplemental Materials found at: doi:10.1016/j.jval.2011.05.013. The average waste projections for every patient per day and per cycle of treatment, respectively, according with the different vial sizes proposed is described in Table 2 (in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.013). The percentage is defined in regard to the standard 3.5-mg vial. For a 3-mg vial, 0.88 mg/day (95% CI 0.74 –1.03) or 36.2% less and 3.44 per cycle (95% CI 2.85– 4.04). For a 2.5-mg vial, 1.05 mg/day (95% CI 0.81–1.29) or 23.9% less and 4.06 mg per cycle (24.6% more than with the 3.5-mg vial). If two different presentations were available, 2.5 mg and 0.5 mg, the waste per patient per day would be 0.52 mg (95% CI 0.4 – 0.63) or 62.5% less and 2.01 mg/per cycle (62.7% less than with 3.5-mg vial). S83 The cost reduction projections per day and per cycle, respectively, assuming a proportional price of the proposed vials to the 3.5-mg vial, are described in Table 3 in the Supplemental Materials found at: doi:10.1016/j.jval.2011.05.013. Using the currently available 3.5-mg vial, an average of US$926.88/day and US$3,607.33/cycle are lost. If a 3-mg vial were available, the average loss would be US$592.09/day (36.1% less than with the 3.5-mg vial size) and US$2306.51/cycle. With the 2.5-mg vial, the average waste per patient per day was US$704.54 (23.9% less than with the 3.5 mg vial size but 15.9% higher than with the 3-mg vial size) and $2717.97 per cycle. If considering the combination of 2.5-mg plus 0.5-mg vials sizes, the loss would be US$346.75/day (62.5% less than with the 3.5-mg vial size) and US$1348.15 per cycle (62.6% less). The total cost wasted per day for all 35 patients was US$45,424.01 (62.5% less than with the 3.5-mg vial size) or US$176,607.25 per cycle. Discussion Although MM accounts for only a small percentage of all cancers, the costs associated with treating and managing it are among the highest [9]. Recent developments in diagnosing, treating, and managing MM have led to novel treatment strategies. Immunomodulators, proteasome inhibitors, and bisphosphonates are improving response rates and preserving patients’ quality of life [10]; however, these agents are not replacing the older treatment modalities, but rather being used in addition to them [11]. The landscape of myeloma therapy has changed radically since 1999 with the introduction of new therapies and better prognostic indicators ushering in a new era of MM management [10]. Novel agents resulting in extended survival and better understanding of the biology of the disease have helped select patients most likely to benefit from stem cell transplantation. The costs associated with current and emerging therapies, as well as supportive care are significant and likely to increase further on, as patients begin to live longer [5]. Furthermore, the majority of cytotoxic chemotherapy protocols are based on dosage calculation from BSA. There is a growing body of evidence that demonstrates the large interpatient variability associated with dosing by BSA [12–16]. Despite this known variability, it is common practice for clinicians to calculate doses of chemotherapy to the nearest milligram based on BSA estimated to two decimal points [12]. In some instances dose rounding is employed by the clinician (e.g., capping) and in others by pharmacists preparing the drug, but this is still more the exception than the rule [12]. There are practical implications related with costs and unnecessary losses in preparing cytotoxic doses calculated to the exact milligram. Mertens et al. [11] conducted a study aimed to evaluate the effects of dose rounding on treatment cost. During the study period, 18 different anticancer drugs were administered 939 times. If dosage had been based strictly on BSA, drug costs would have been €509,664. Rounding off to whole ampoules with a dose margin of a maximum of 10% would have cost €465,619; a reduction of 8.6%. The rational application of the dose individualization principle based on body surface area may result in a substantial reduction in expenditure on anticancer drugs [11]. Dose rounding has been considered acceptable to within 5% of calculated dose because on the basis of pharmacokinetic and clinical issues this dose adjustment is not expected to have any significant effect on either response or toxicity [6,17–19]. Another possibility to help solve the waste problem would be to combine multiple infusions on the same day. In terms of our study, however, it would be highly unlikely, not only because of the rarity of the disease but also due to the fact that the 35 patients were located in 10 different states in Brazil. S84 VALUE IN HEALTH 14 (2011) S82–S84 The reason we assigned a proportional cost to the different bortezomib vials was because usually drug prices in Brazil are calculated to be proportional to their dosage; that is, if a medication is available in 50- and 100-mg vials, the latter will be twice as expensive as the former [8]. Drug waste may be defined as the consequence of an inappropriate disposal of unused or partially used ampoules, vials, or syringes of drugs [20]. It has been previously demonstrated that inefficiency of drug use and waste production may lead to a distinct economic loss, though experiences are limited and most studies are dated or focus on other therapeutic areas [9,20 –22]. Decreasing waste is an attractive cost-cutting strategy because it neither limits specific drug use nor affects quality of care [7]. One of the main reasons for drug waste was essentially the limited extent of chemotherapy medication shelf-life and the narrow availability of a range of vial sizes flexibly matching with possible drug dosages [7]. Adopted corrective measures were the logical consequence of these findings: if drug instability is a basis for drug waste, it is reasonable to use, whenever possible, multidose vials that retain a much longer microbial and chemical stability and to operate a per pathology/per drug distribution system of chemotherapy sessions over the week to allow the reuse of leftovers in other patients while respecting drug stability [7]. One of the limitations of this study was its small sample size. Although we used a database that covers 5% of the Brazilian private health care market, MM is much less frequent than other neoplasms such as breast or colon cancer and therefore it are important that these findings are confirmed in larger cohorts of patients. Finally, there has been much discussion on the rising prices of oncologic treatments and how much is too much [23]. We believe that this discussion is even more important in developing countries, which are plagued by a perennially insufficient health care budget. To keep the discussion active we made this analysis and intended to show how the simple adjustment of vial size in bortezomib could affect costs and minimize drug waste. Conclusions A simple adjustment in vial size from 3.5 to 3.0 mg reduces bortezomib waste by 36%. If presentations of 2.5 mg and 0.5 mg were available, the waste reduction could be as high as 62%. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi:10.1016/j.jval.2011.05.013, or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). REFERENCES [1] Durie BG. Role of new treatment approaches in defining treatment goals in multiple myeloma—the ultimate goal is extended survival. Cancer Treat Rev 2010;36(Suppl. 2):S18 –23. [2] Laubach JP, Richardson PG, Anderson KC. The evolution and impact of therapy in multiple myeloma. Med Oncol 2010;27(Suppl. 1):S1– 6. [3] Ludwig H, Beksac M, Blade J, et al. Current multiple myeloma treatment strategies with novel agents: a European perspective. Oncologist 2010;15:6 –25. [4] Richardson PG. Frontline multiple myeloma management: a clinical and cost update for managed care. In: Education NACfCM, ed., Continued Medical Education. Boston: Jerome Lipper Multiple Myeloma Center, Dana-Farber Cancer Institute, Harvard Medical School, 2010. [5] Cook R. Economic and clinical impact of multiple myeloma to managed care. J Manag Care Pharm 2008;14:19 –25. [6] Dooley MJ, Singh S, Michael M. Implications of dose rounding of chemotherapy to the nearest vial size. Support Care Cancer 2004;12: 653– 6. [7] Fasola G, Aita M, Marini L, et al. Drug waste minimisation and costcontainment in medical oncology: two-year results of a feasibility study. BMC Health Serv Res 2008;8:70. [8] Gobis O. Revista SIMPRO. São Paulo: SIMPRO publicações e tele processamento [SIMPRO Magazine-publications and tele processing]. 2009. [9] Diehl LD, Goo ED, Sumiye L, et al. Reducing waste of intravenous solutions. Am J Hosp Pharm 1992;49:106 – 8. [10] Cassidy J. Chemotherapy administration: doses, infusions and choice of schedule. Ann Oncol. 1994;5(Suppl. 4):25–9. [11] Mertens S, de Jongh FE. [Lower costs for anticancer drugs by safety margin around calculated dose and by fine-tuning on ampoule strength]. Ned Tijdschr Geneeskd 2009;153:B162. [12] Desoize B, Robert J. Individual dose adaptation of anticancer drugs. Eur J Cancer 1994;30A:844 –51. [13] Felici A, Verweij J, Sparreboom A. Dosing strategies for anticancer drugs: the good, the bad and body-surface area. Eur J Cancer 2002;38: 1677– 84. [14] Gurney H. Dose calculation of anticancer drugs: a review of the current practice and introduction of an alternative. J Clin Oncol 1996; 14:2590 – 611. [15] Ratain MJ. Body-surface area as a basis for dosing of anticancer agents: science, myth, or habit? J Clin Oncol 1998;16:2297– 8. [16] Reilly JJ, Workman P. Normalisation of anti-cancer drug dosage using body weight and surface area: is it worthwhile? A review of theoretical and practical considerations. Cancer Chemother Pharmacol. 1993;32: 411– 8. [17] Bajetta E, Di Bartolomeo M, Mariani L, et al. Randomized multicenter Phase II trial of two different schedules of irinotecan combined with capecitabine as first-line treatment in metastatic colorectal carcinoma. Cancer 2004;100:279 – 87. [18] Hempel G, Boos J. Flat-fixed dosing versus body surface area based dosing of anticancer drugs: there is a difference. Oncologist 2007;12: 924 – 6. [19] Moreno-Solorzano I, Ibeas-Rollan R, Monzo-Planella M, et al. Two doses of oxaliplatin with capecitabine (XELOX) in metastatic colorectal cancer. Clin Colorectal Cancer 2007;6:634 – 40. [20] Nava-Ocampo AA, Alarcon-Almanza JM, Moyao-Garcia D, et al. Undocumented drug utilization and drug waste increase costs of pediatric anesthesia care. Fundam Clin Pharmacol 2004;18:107–12. [21] Favier M, Fliche E, Bressolle F. Economic benefit of a centralized reconstitution unit of cytotoxic drugs in isolator. J Oncol Pharm Pract 1996;2:182– 85. [22] Gillerman RG, Browning RA. Drug use inefficiency: a hidden source of wasted health care dollars. Anesth Analg 2000;91:921– 4. [23] Meropol NJ, Schulman KA. Cost of cancer care: issues and implications. J Clin Oncol 2007;25:180 – 6. VALUE IN HEALTH 14 (2011) S85–S88 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Costos de la Diabetes en Ameŕica Latina: Evidencias del Caso Mexicano Armando Arredondo, PhD*, Esteban De Icaza, PhD A B S T R A C T The main objective was to identify economic burden from epidemiological changes and expected demand for health care services for diabetes in México. The cost evaluation method to estimate direct and indirect costs was based on instrumentation and consensus techniques. To estimate the epidemiological changes for 2009-2011, three probabilistic models were constructed according to the Box-Jenkins technique. Comparing the economic impact in 2009 versus 2011 (p⬍ 0.05), there is a 33% increase in financial requirements. The total amount for diabetes in 2010 (US dollars) will be $778,427,475. It includes $343,226,541 in direct costs and $435,200,934 in indirect costs. The total direct costs expected are: $40,787,547 for the Ministry of Health (SSA), serving to uninsured population; $113,664,454 for insured population (Mexican Institute for Social Security–IMSS-, and Institute for Social Introducción La diabetes es un problema de salud que requiere un abordaje integral, ya que su tendencia al incremento no ha sido impactada con los esfuerzos desarrollados y recursos económicos asignados para su resolución. Los altos costos en salud y el comportamiento demográfico de México, en el que se advierte un cambio en la pirámide poblacional, agregará mayores condiciones de riesgo para la población adulta. Tales tendencias incrementarán la demanda de servicios de atención para diabetes en el corto, mediano y largo plazo [1–3]. México ocupa actualmente el noveno lugar mundial en la prevalencia de diabetes. Este es un sitio alarmante, y más aún cuando las proyecciones de los especialistas internacionales refieren que para el año 2025, el país ocupará el sexto o séptimo lugar. Esta enfermedad se ha convertido en una epidemia mundial debido a los altos índices de muertes y la creciente demanda de servicios que se han registrado en los últimos 10 años [4]. Frente a ello, las diversas instituciones de salud en el país han comenzado a reforzar sus campañas preventivas para evitar altos costos, particularmente de las complicaciones asociadas. En efecto, para cuando se diagnostica la diabetes y sus complicaciones, los costos para su tratamiento son muy elevados y prácticamente el paciente va perdiendo sus años-vida productivos, con repercusiones importantes en términos de costos indirectos atribuibles a la diabetes [5– 6]. Además, los costos en la calidad de vida de estos pacientes son muy elevados y al mismo tiempo los montos financieros Security and Services for State Workers-ISSSTE-); $178,477,754 to users; and $10,296,786 to Private Health Insurance (PHI). Conclusions: If the risk factors and the different health care models remain as they are currently in the institutions analyzed, the financial consequences would be of major impact for the pockets of the users, following in order of importance, social security institutions and finally Ministry of Health. Allocate more resources to promotion and prevention of diabetes will decrease the cost increase by decreasing the demand for treatment of complications. Palabras Claves: cambios epidemiológicos, diabetes, Impacto económico, pacientes, sistema de salud. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. que emplea el sector salud para controlar problemas asociados a la diabetes se desconocen en la mayoría de los países. Lo anterior dificulta un uso y asignación eficiente de recursos para enfrentar el problema de demanda de servicios [7–9]. La perspectiva futura señala que se mantendrá el incremento en la cantidad de diabéticos [10]; Para el caso mexicano, dichas tendencias han sido ampliamente documentadas desde la Encuesta Nacional de Enfermedades Crónicas y corroboradas en los hallazgos recientes de la Encuesta Nacional de Salud y Nutrición [11]. Por ejemplo, la prevalencia de diabetes mellitus por diagnóstico médico previo en adultos mayores de 20 años en México, se ha incrementado de 4.6% en 1993,1 5.8% en 2000 a 17% en 2006. El impacto de esta enfermedad no sólo es en la mortalidad sino de manera muy importante en la morbilidad y en la calidad de vida, representando una enorme carga tanto para el individuo y su familia como para el sistema de salud y la sociedad en general [11]. En este contexto el propósito de este artículo es presentar los principales resultados sobre los costos directos, los costos indirectos y los requerimientos financieros esperados para atender la diabetes tipo 2 en México. Material Y Métodos Los resultados que se presentan provienen de un proyecto de investigación evaluativa basado en un diseño de tipo longitudinal Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Investigador del Instituto Nacional de Salud Pública, Cuernavaca Mor. México. Título corto: Impacto Económico de la Diabetes en México. * Autor de correspondencia: Armando Arredondo, Investigador del Instituto Nacional de Salud Pública, Av Universidad # 655, Cuernavaca Mor, México; Tel: (0052) 777 3-29-30-62; Fax: (0052) 777 3-29-30-00. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.022 S86 VALUE IN HEALTH 14 (2011) S85–S88 que consistió en el análisis de tendencias y series de tiempo de casos observados de problemas crónicos de salud para el período 1990-2008 y determinación de cambios epidemiológicos y de tendencias de casos esperados para el período 2009-2011. El principal objetivo del proyecto macro fue identificar el impacto económico de cabios epidemiológicos y tendencias de demanda de servicios para enfermedades crónico-degenerativas en México. Para el caso de la diabetes mellitus tipo 2, se hacen las siguientes precisiones metodológicas: La población de estudio se delimitó a toda la población adscrita por tipo de institución con diagnóstico de diabetes y que demandó servicios de atención para diabetes tipo 2. ● Los cambios en la demanda de servicios se determinaron mediante análisis de series de tiempo a partir de casos observados para el período 1990-2008. La determinación de modelos de pronóstico de la demanda esperada se refirió al periodo 2009-2011. El método de estimación se basó en la metodología de Box-Jenkins para pronósticos de salud [12–13]. El modelo de salida resultó en un modelo autoregresivo de orden 1 cuyos detalles y estadísticas se pueden ver en el anexo 1 de Materiales Complementarios en: doi: 10.1016/j.jval.2011.05.022) Definido el modelo, se estimaron los casos esperados de diabetes para el sistema de salud en su conjunto y para las 3 principales instituciones bajo estudio: SSA, IMSS e ISSSTE. Para la estimación de costos directos e indirectos esperados para a diabetes, se tomo como año de corte el 2010 por corresponder a la mitad del período de proyección. ● El manejo de casos promedio se determinó mediante la integración de un panel de expertos multidisciplinario con un mínimo de experiencia de 10 años en el manejo ambulatorio y hospitalario de pacientes diabéticos. El panel incluyó 4 expertos por tipo de institución, entre internistas, epidemiólogos, economistas y administradores. El panel de expertos basado en la técnica delphi, permitió identificar 7 instrumentos donde se indicaba la lista de insumos por función de producción en el manejo de casos promedio por tipo de institución. La validación de tales instrumentos fue realizada bajo la técnica de tiempos y movimientos. Esta técnica, que consistió en dar seguimiento y observación al 10 % de la demanda de pacientes en un periodo anual, permitió validar los 7 instrumentos guía donde fueron identificados los insumos requeridos para cada función de producción (consulta de primera vez, consulta subsecuente, estudios de diagnóstico, terapia medicamentosa, hospitalización, estudios de monitoreo y seguimiento, terapia intensiva). ● Los costos directos de manejo de caso se determinaron a partir de las funciones de producción, la combinación de insumos, los estándares de calidad y los costos de los insumos para cada subsector que fueron definidos por el panel de expertos de cada institución. Los costos al bolsillo de los usuarios se determinaron a partir de los costos atribuibles a diabetes calculados en base a los datos registrados en la encuesta nacional de ingreso gasto a partir del rubro de gastos en salud [14]. ● Los costos indirectos se determinaron usando el enfoque de capital humano desarrollado para América Latina, realizando los ajustes para el caso de México en función de tasas de mortalidad, discapacidad y esperanza de vida para la población mexicana. Este enfoque se basa en indicadores de mortalidad prematura y discapacidad temporal y discapacidad permanente atribuibles a diabetes [15]. ● Los costos por modelo de atención se determinaron a partir de los casos esperados para manejo ambulatorio y para manejo hospitalario definidos por el panel de expertos. ● Los requerimientos financieros para años futuros, se estimaron a partir de los casos esperados tomando como año base 2010, los costos de manejo de caso tomando como periodo de referencia el último año de casos observados y aplicando un factor de ajuste econométrico para controlar proceso inflacionario es- perado para servicios de salud en el corto, mediano y largo plazo. El factor de ajuste econométrico se basó en la tasa inflacionaria proyectada y acumulada a partir del índice inflacionario de precios a servicios de salud proyectado por el Banco de México. Período 2009-2011 [16]. Resultados Después de estimar los casos esperados para el período 2009-2011 y sus respectivos intervalos de confianza al 95% (ver tabla 1 anexo 2 en Materiales Complementarios en: doi:10.1016/j.jval.2011.05.022se decidió tomar como punto de corte 2010 para determinar los costos de atención a la diabetes para cada subsector. En el cuadro 1 se presenta la distribución de los costos totales anuales (directos e indirectos) para todos los casos que demandarían servicios en el manejo de la diabetes para las diferentes instituciones bajo análisis. Los resultados de este estudio mostraron evidencia de que el menor costo promedio fué en instituciones para no asegurados, seguidos del subsector público para asegurados. El mayor costo promedio corerspondió a instituciones privadas. Es importante resaltar que las diferencias observadas respecto al costo promedio de manejo de caso por tipo de institución se explican principalmente por 4 factores: el protocolo de manejo de caso ( tipo y combinación de insumos), los estándares de productividad, los estándares de calidad y del costo de los insumos para cada subsector. En relación al costo directo de las diferentes funciones de producción, llama la atención que los insumos de mayor impacto se refieren a los medicamentos, seguidos de costos de servicios de consulta y diagnóstico y en menor grado de costo de hospitalización por descompensación sin considerar manejo de complicaciones. . Para el peso relativo del costo en el manejo integral de las principales complicaciones de la diabetes, en todas las instituciones, el mayor impacto está en los costos para el manejo de nefropatía diabética, siguiéndole de mayor a menor impacto retinopatía, enfermedad cardiovascular, neuropatía diabética y finalmente enfermedad vascular periférica. En relación a los costos directos el mayor impacto está en el bolsillo de los usuarios con el 51% del gasto total para diabetes; le siguen en orden de importancia las instituciones de seguridad social y finalmente, instituciones para no asegurados (ver cuadro 1). Sobre los costos indirectos de la atención, en esta dimensión de costos se pudieron determinar los costos indirectos para usuarios que se atienden en las tres principales instituciones del sector público y costos atribuíbles a usuarios que se atienden en instituciones privadas. Estos costos representan el 43% del costo total de la diabetes en México. Se distribuyen en 3 categorías de estimación: costos por mortalidad prematura (5%), costos por discapacidad permanente ( 93%) y costos por discapacidad temporal ( 2%), Respecto al peso relativo de costos directos versus costos indirectos, los costos indirectos representan el 56% y 44% respectivamente sobre los costos totales de la diabetes en México (ver cuadro 1). Discusion Y Conclusiones La relevancia de incorporar aspectos epidemiológicos y económicos a la perspectiva clínica, constituye una propuesta integral para el análisis y evaluación del desempeño y los costos de la atención para el sistema de salud. En efecto, el desarrollo de estudios de investigación evaluativa que integren una valoración económica con valoraciones clínica y epidemiológica, se torna relevante para un abordaje de mayor efectividad en el abatimiento de los costos asociados al problema de la diabetes o de cualquier otra enfermedad. Las evidencias generadas sobre costos directos e indirectos, plantean la necesidad de diseñar e implementar nuevos mecanis- VALUE IN HEALTH 14 (2011) S85–S88 S87 CUADRO 1 – Costos Directos, Costos Indirectos Y Costos Totales Anuales Atribuibles A La Diabetes En México, 2010: SSA. IMSS, ISSSTE, USUARIOS, Seguros Privados De Salud. (DLS.DE EUA). mos de planeación estratégica que permitan controlar, contener y reducir los costos atribuibles a la diabetes, particularmente los costos asociados a sus complicaciones por tipo de institución, tal como se resaltó con los resultados de este estudio. Respecto a los costos al bolsillo de los usuarios, no deja de llamar la atención el alto peso relativo del origen de los gastos para diabetes desde el ingreso familiar y sus implicaciones en materia de equidad y acceso a la salud en México. En efecto, decir que de cada 100 pesos que se gastan en diabetes en México, aproximadamente 51 pesos proviene de los hogares/ingresos familiares, representa una carga social de muy alto impacto que evidentemente tendrá un efecto considerable en la medición del gasto catastrófico en salud del país, sobretodo tratándose de un padecimiento de alto costo y alta prioridad como problema de salud pública en México. S88 VALUE IN HEALTH 14 (2011) S85–S88 Dicho de otra forma, la atención a la salud en México se distribuye mas o menos de la siguiente manera: población que se atiende en instituciones para la seguridad social 48%, población que se atiende en instituciones para no asegurados 42% y población que se atiende en instituciones privadas 10%.De acuerdo a nuestros resultados quiere decir que, respecto a los costos directos, de cada 100 pesos que se gastan en diabetes en México, 52 se gastan en el 10% de la población, 33 en el 48% (asegurados) de la población y 15 pesos en el 42% restante de la población (no asegurados). Lo que sin duda queda también en evidencia es que el problema de la diabetes en México representa un alto impacto económico por todas las consideraciones epidemiológicas, económicas y organizacionales. En este sentido la valoración económica de la diabetes plantea de manera pertinente que el abordaje para su resolución, en términos de asignación y flujo de recursos económicos, presenta problemas de inequidad y de acceso a la salud dependiendo del grupo social al que pertenecen los pacientes con diabetes y sus familias. Respecto a los costos indirectos, es importante resaltar que el peso relativo de costos directos vs. Indirectos resultó muy similar a los costos de diabetes en otros estudios realizados en Canadá y EUA [5–7]. Por otra parte, aunque no constituyen un impacto directo sobre el presupuesto en salud, en términos de costo e impacto social, representan una alta carga que la sociedad deberá asumir, sobre todo en términos de productividad perdida por muerte prematura y discapacidad, sea temporal o permanente. Los requerimientos financieros calculados constituyen una base de información fundamental para la planeación estratégica. En efecto, dadas las consecuencias financieras del cambio epidemiológico esperado, no solo se fundamenta y justifica la necesidad de invertir mayores recursos financieros para las actividades de promoción y prevención, de manera que se pueda minimizar y controlar el daño a la salud y así evitar la carga económica para los sistemas de salud. En resumen la ganancia económica en la productividad y en la eficiencia, se podrán dar en la medida que se conocen los costos unitarios por funciones de producción para las diferentes etapas del proceso de atención médica. De esta manera se pueden establecer tanto patrones de equipamiento, como patrones de productividad y eficiencia de los recursos utilizados, justificándolos en relación a los ahorros que generarán. Desde una perspectiva sistémica, puede decirse que de acuerdo a los resultados presentados, si los factores de riesgo y los diferentes modelos de atención ambulatoria y hospitalaria permanecieran más o menos como están actualmente, ,las consecuencias financieras de la diabetes serían de mayor impacto para el IMSS, siguiendo en orden de importancia la SSA y finalmente el ISSSTE. Por otra parte, también es importante resaltar que los requerimientos financieros para el tratamiento de la diabetes, tanto de la demanda de servicios ambulatorios como hospitalarios, representaran aproximadamente el 6% del presupuesto total asignado para la población no asegurada y aproximadamente el 9.5% del presupuesto total asignado para población asegurada. Finalmente, es necesario señalar que para mayor confiabilidad, validez y pertinencia de estos resultados, se recomienda implementar sistemas de monitoreo de costos y análisis posteriores que permitan actualizar la medición anual o bianualmente de los costos de manejo promedio de la diabetes. De esta manera se podrán ajustar permanentemente de acuerdo con cambios inflacionarios, cambios en los costos de los insumos y sobre todo, cambios en las tendencias de la demanda de servicios de salud para diabetes por tipo de institución. Ciertamente para valorar los cambios en la demanda a partir de cambios epidemiológicos esperados, también se recomienda que el modelo probabilístico desarrollado se actualice anualmente con los datos observados hasta el año de actualización. En este sentido, se tendrá mayor impacto en la toma de decisiones en materia de asignación y uso eficiente de los recursos destinados a la resolución del problema de la diabetes en México. Fuentes de financiamiento: El estudio base de este proyecto recibió financiamiento del International Development Research Centre-Canadá y del CONACYT-México. Materiales Complementarios Material complementario que acompaña este artículo se puede encontrar en la versión en línea como un hipervínculo en doi: 10.1016/j.jval.2011.05.022 si es un artículo impreso, estará en www.valueinhealthjournal.com/issues (seleccione el volumen, número y artículo). REFERENCES [1] Arredondo A, Barceló A, The Economic Burden of Out-of-Pocket Medical Expenditures for Patients Seeking Diabetes Care in Mexico. Diabetologia 2007;50:435–36. [2] American Diabetes Association. Economic Costs of Diabetes in the US in 2002. Diabetes Care 2003;26:917–32. [3] Frenk J, Ruelas E, Lozano R, et al. Demanda y oferta de servicios médicos: Obstáculos a la mejoría del sistema de salud en México. (Ed.). FUNSALUD; México D.F., 1994. [4] Secretaría de Salud. Información básica sobre recursos y servicios del Sistema Nacional de Salud. Informe Técnico.México D.F. 2009:68 –76. [5] Caro J, O’Brien JA, Shomphe LA, et al. Lifetime Costs of Complications Resulting From Type 2 Diabetes in the U.S. Diabetes Care, 2002, vol 25: 476 – 481. [6] International Diabetes Federation, Direct cost to the health care sector.Diabetes Health Economics. International Diabetes Federation. Brussels, Belgium. 1999:13-15. [7] Dawson KG, Gomes D, GersteinH, et al. The economic cost of diabetes in Canada, Diabetes Care 2002;25:1303– 07. [8] Barcelo A, Daroca MC, Ribera R, et al. Diabetes in Bolivia. Pan American Juo of Public Health 2001;10:318 –23. [9] Arredondo A, Zuñiga A. Economic burden of diabetes in middleincome countries: The Mexican Case. Diabetes Care 2004;29:104 –9. [10] SSA, IMSS, ISSSTE. Boletín de Información Estadística. Casos de morbilidad hospitalaria por demanda específica, 1993-2009. SSA (Ed.). México, D.F., 2010. [11] Instituto Nacional de Salud Pública. Encuesta Nacional de Salud y Nutrición-2006. Daños y servicios de la salud. Cuernavaca, Mex. INSPSSA- 2008:74-81. [12] Murray A. Statistical Modelling and Statistical Inference: Measurement error in the explanatory variables. Box-Jenkins technique in Statistical Modelling in GLIM (3rd ed.). New York, Oxford, 2005. [13] Instituto Nacional de Salud Pública. Métodos de estimación sobre demanda esperada para enfermedades crónico-degenerativas. Informe Técnico de Memoria Metodológica. Cuernavaca, México. Febrero del 2010:63-78. [14] Instituto Nacional de Estadística Geografía e Informática. Encuesta Nacional de Ingreso Gasto. Resultados sobre gastos en servicios de salud. México DF., 2008:214 –26. [15] Barceló A, Aedo C, Rajpathak S, Robles S. The cost of diabetes in Latin America and the Caribbean. Bull World Health Organ 2003;81:19 –27. [16] Banco de México. Índice Nacional de Precios por Servicios Médicos en México. Cuadernos Mensuales, Base 1980⫽100. La Actividad Económica en México. 1983-2002. Gerencia de Investigación Económica. Banco de México Ed. México DF, México: 2009:46-68. VALUE IN HEALTH 14 (2011) S89 –S92 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Short-Term Therapy with Enoxaparin or Unfractionated Heparin for Venous Thromboembolism in Hospitalized Patients: Utilization Study and Cost-Minimization Analysis Catia Argenta, MSc,1 Maria Angélica Pires Ferreira, MSc,2 Guilherme Becker Sander, PhD,2 Leila Beltrami Moreira, PhD1,2,3,* 1 Graduate Program in Medical Sciences, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil; 2Pharmacy and Therapeutics Committee, Hospital de Clínicas de Porto Alegre, Porto Alegre, Brazil; 3Pharmacology Department, Universidade Federal do Rio Grande do Sul; National Institute of Science and Technology for Health Technology Assessment–CNPq, Porto Alegre, Rio Grande do Sul, Brazil A B S T R A C T Objectives: To evaluate the direct costs of venous thromboembolism (VTE) treatment with unfractionated heparin (UFH) and low-molecular weight heparin, from the institutional perspective. Methods: This is a real-world cohort study that included inpatients treated with UFH or enoxaparin for deep venous thromboembolism or pulmonary embolism in a tertiary public hospital. To estimate medical costs we computed the acquisition costs of drugs, supplies for administration, laboratory tests, and hospitalization cost according to the patient ward. Results: One hundred sixty-seven patients aged 18 to 92 years were studied (50 treated with UFH and 117 with enoxaparin). The median of days in use of heparin was the same in both groups. Activated partial thromboplastin time was monitored in 98% of patients using UFH and 56.4% using enoxaparin. Nonstatistically significant differences were Introduction Deep venous thrombosis (DVT) in the lower extremities is the most frequent manifestation of venous thromboembolism (VTE), with an incidence of 0.48 to 1.6 cases/1000 persons-year among community residents [1–3]. It is a common condition affecting mainly inpatients and rates increase with age. The most lifethreatening manifestation is pulmonary embolism (PE), affecting 0.23 to 0.69 cases/1000 person-years [4]. The treatment of VTE recommended by the American College of Chest Physicians [5] involves short-term low-molecular-weight-heparin (LMWH) or unfractionated heparin (UFH) therapy plus long-term oral warfarin therapy. Anticoagulant therapy with UFH followed by warfarin prevents thrombus extension, reduces the risk of recurrent thrombosis, and prevents death in patients with VTE [6,7]. Subcutaneous LMWH is as effective and safe as conventional UFH therapy, but does not require laboratory monitoring and is less likely to cause bleeding, immune thrombocytopenia, and osteoporosis [8 – 10]. LMWH preparartions differ considerably in composition, observed between groups in the number of bleeding events (10.0% and 9.4%; P ⫽ 1.00); blood transfusion (2.0% and 2.6%; P ⫽ 1.00); death (8.0% and 3.4%; P ⫽ 0.24); and recurrent VTE, bleeding, or death (20.0% and 14.5%; P ⫽ 0.38). Daily mean cost per patient was US$12.63 ⫾ $4.01 for UFH and US$9.87 ⫾ $2.44 for enoxaparin (P ⬍ 0.001). The total costs considering the mean time of use were US$88.39 and US$69.11. Conclusion: The treatment of VTE with enoxaparin provided cost savings in a large teaching hospital located in southern Brazil. Keywords: heparin, deep venous thrombosis, utilization study, cost analysis. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. which could result in different antithrombotic effects, but there is no evidence that any LMWH preparation is better or worse than another in terms of efficacy or safety outcomes [11]. Enoxaparin is among the most widely studied treatments for VTE [12,13]. Despite LMWH having a greater acquisition cost, previous pharmacoeconomic analyses have shown that LMWH is more cost-effective than UFH [14 –16]. It has been calculated that outpatient treatment with LMWH may save $1641 per patient in comparison to UFH hospital treatment [17]. This economic benefit of outpatient treatment of VTE seems to be present in different health systems of developed countries, although the same can not be extrapolated to developing nations. Economic evaluations in developing countries are desirable to estimate VTE hospital treatment cost with UFH and LMWH, because people with low income do not have access to outpatient treatment with LMWH. We conducted a cohort study to evaluate the direct costs of short-term heparin anticoagulation treatment for VTE in a large teaching hospital located in southern Brazil, from the institutional perspective. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Leila Beltrami Moreira, Farmacologia Clínica sala 947, Hospital de Clínicas de Porto Alegre, Ramiro Barcelos, 2350, 90.035-903, Porto Alegre, RS, Brazil. E-mail : [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.017 S90 VALUE IN HEALTH 14 (2011) S89 –S92 Methods This was a cohort study that prospectively included patients hospitalized from March 2005 through January 2007 in a university-affiliated, general tertiary teaching hospital with 749 beds in southern Brazil. These patients had been treated with intravenous UFH or subcutaneous LWMH for suspected or confirmed DVT or PE. The study was approved by the institutional review board. Patients receiving UFH or LWMH were identified through the institution’s computerized prescription system and had their clinical records revised to be included in the study. All patients receiving an anticoagulation dose of heparin to treat VTE were potentially eligible. The LMWH included on the hospital formulary was enoxaparin, based in the lowest price acquisition policy of the institution. Patients identified with DVT or PE were prospectively followed-up until the end of the heparin anticoagulation period. The only exclusion criterion was age younger than 18 years. Data about diagnosis confirmation, anticoagulation regimen, duration of treatment, laboratory monitoring with activated partial thromboplastin time (aPTT), bleeding, thrombocytopenia, blood transfusion, and protamine prescription were recorded for all included patients. Adverse events potentially related to DVT or its treatment were defined as combined endpoint and included in-hospital death, bleeding, or recurrent VTE. Costs were assessed directly from the hospital’s records on prices paid for each one of the elements involved in patients’ assistance, considering the prices during their period of hospitalization. This information was provided directly by the hospital, so retrospective costs did not have to be estimated based on any other indirect information. Total medication costs were calculated for each patient, considering the prospectively collected data on medicines, supplies, and laboratory tests. To estimate direct medical costs we computed the acquisition costs of drugs and supplies associated with UFH treatment, laboratory tests, and hospitalization cost according to the patient ward. The cost of use of automatic pump for UFH administration was not computed because they were not hospital property and its charge was covered by the cost of pump-specific infusion equipment required. The costs were converted into US dollars considering the mean exchange rate from April to May 2007. Data were analyzed with PASW Statistics version 18.0 (2009, SPSS Inc., Chicago, IL) and a level of significance of 0.05 was set. Chi-square statistics were used in the comparison of categorical variables and Student t or Mann-Whitney tests were applied to compare continuous variables. Although costs and time of treatment had small skewness deviation, t test and Mann-Whitney results were similar and t test value was provided to compare means. Logistic regression modeling was applied to analyze the association of heparin form and composite clinical outcomes, to take into account potential confounders identified in crude analyses. Sensitivity analyses were performed during June 2010, considering the actual cost of drugs to account for acquisition prices variation. Propensity score for LWMH prescription was computed to adjust for indication bias. Results From the 200 patients included, 33 were excluded because they had been treated for arterial thrombosis. Twelve patients received both heparins and were classified in the group of UFH (n ⫽ 7) or LMWH (n ⫽ 5) according to the first drug prescribed. The change from the first prescribed drug to the other was more frequent in the UFH group (14.0% vs. 4.3% P ⫽ 0.044). The characteristics of the groups are similar when excluding the 12 pa- tients that received both heparins (Table 1 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.017). Patients were 18 to 92 years old; UFH was first prescribed for 50 (29.9%) patients and LMWH for 117 (70.1%) patients. DVT was treated in 107 cases, PE in 43 cases, and both conditions in 17 cases. Diabetes, surgery, infectious disease, and renal failure were significantly more frequent in the group that began anticoagulation with UFH. Previous PE was more frequent in LMWH group (P ⫽ 0.015). Only general surgery had crude association with combined endpoint (death, bleeding, or recurrent VTE). Among nine critically ill patients, six were treated with UFH (12%) and three with LMWH (2,6%) (chi-square P ⫽ 0.036). The DVT and PE diagnoses were confirmed in 93.0% and in 51.7% of patients, respectively. The median of days in use of heparin was the same in both groups and warfarin was initiated at the first day in 28.1% of 32 patients on the UFH group and in 24.2% of 89 patients on LMWH group treated with oral anticoagulant. aPTT was monitored in 98% of patients receiving UFH and in 56.4% of patients receiving LWMH. Half of the UFH group had aPTT measured less than once a day (Table 2 in Supplemental Materials found at: doi: 10.1016/j.jval.2011.05.017). Nonstatistically significant differences were observed in the number of bleeding events (10.0% in UFH and 9.4% in LMWH; chi-square P ⫽ 1.00), blood transfusion (2.0% and 2.6%; chi-square P ⫽ 1.00), and death (8.0% and 3.4%; chi-square P ⫽ 0.242). Only one patient in the UFH group evolved to PE. Combined endpoint consisting of recurrent VTE, bleeding, or in-hospital death occurred in 20.0% and 14.5% of patients in the UFH and LMWH groups, respectively (chi-square P ⫽ 0.379). In a logistic regression model, adjusted for propensity score to receive UFH and for the hospital ward (clinical or obstetric unit, surgery unit, or intensive care unit), LMWH group members showed no significant risk reduction of combined outcome (odds ratio 0.87; 95% confidence interval 0.32–2.41; P ⫽ 0.79). We calculated the drug cost per day of treatment taking into account the actual price for the institution (Table 3 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.017). We excluded patients who were treated initially with one form of heparin and then changed to other on the follow-up (12 individuals). Daily mean cost per patient was US$12.63 ⫾ $4.01 for UFH and US$9.87 ⫾ $2.44 for LWMH (t test P ⬍ 0,001). The total cost of short-term heparin treatment considering the mean length of use was US$88.39 and US$69.11, respectively, representing a cost saving of US$19.28 per heparin treatment. The medication itself is the main component of the LMWH group costs (92.88%), whereas it represents only 5.25% of costs in the UFH group (Fig. 1 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.017). Sensitivity analysis was performed changing the drug cost according to the actual price of acquisition in 2010 converted to US dollars (exchange rate in June 26, 2010). The daily mean cost became US$81.48 ⫾ $55.52 for UFH and US$14.23 ⫾ $8.42 for LWMH (t test P ⬍ 0,001), and the total unadjusted cost of short-term heparin anticoagulation was respectively US$757.31 ⫾ $359.64 and US$570.94 ⫾ $201.76 (t test P ⫽ 0.002), representing a cost saving of US$186.37 per heparin treatment. Discussion This was a pharmacoeconomic analysis of heparin use to treat VTE conducted in a developing country, through cost analysis from a public health perspective. This kind of analysis is justified based on the therapeutic equivalence between initial treatment of DVT and PE with UFH and LMWH [5]. One characteristic that distinguishes our study is the evaluation of a real cohort rather then a hypothetical one, concomitantly including groups of patients receiving different treatments. This reflects the usual care of these patients and makes it possible to estimate the actual costs of treatment in a defined setting. VALUE IN HEALTH 14 (2011) S89 –S92 Patients included in the study were younger then those of hypothetical cohorts [17,18] and similar to those of other pharmacoeconomic analyses that collected data from individual patients [14,19]. Sex is not a risk factor for VTE and is in accordance with the nonsignificant predominance of women in the sample [20]. The 18.6% prevalence of confirmed PE in the studied sample is similar to the 19% in a clinical trial of VTE [12]. In the same trial the major or minor bleeding rates were lower then those recorded in our cohort but they were not different between the treatment groups either. There was no difference on clinical outcomes between the two groups. This is in accordance with others studies [12,13], but clinical trials are not all consistent about equivalence between enoxaparin and UFH. A meta-analysis [21] demonstrated a statistically significant reduction in clinical outcomes with LMWH when combining all trials, but benefit persisted only for reduction of the thrombus size when the 11 studies with adequate concealment of allocation before randomization were considered. Comparison with literature is difficult because the conditions affecting enrolled patients in clinical trials are quite different and the small sample size is a limitation that precludes a definitive conclusion of our study. We did not expect that LMWH prescription was greater than UFH prescription because the institutional policy during the study period reinforced the use of UFH for VTE treatment based on equal efficacy and lower acquisition costs. The preferred use of UFH in renal failure is recommended in view of the plasmatic accumulation due to delayed elimination and risk of excessive anticoagulant effect [22]. In surgical patients, the preference could be related to the potential for reversion of anticoagulant effect with protamine and the shorter half-life then LMWH in case of bleeding. On the other hand, the preference for using LMWH in face of previous PE is not evidence-based. The crossover from one drug to another was not expected either, but we did not investigate the reasons for it. The greater proportion of infected, critically ill, and diabetic patients in the UFH group suggests they were at higher risk and may be responsible for the statistically nonsignificant increase in death rate in this group. The propensity score was constructed to adjust for these risk factors and was included in the logistic regression model that showed no independent association of heparin group with the composite endpoint. But the nonsignificant relative risk reduction of 17% favoring LMWH may be explained by the small sample size and lack of power. The total costs of initial treatment for VTE were less expensive with enoxaparin than with UFH, providing an economic advantage of 21.0%. It provided cost savings regardless of the abuse in ordering monitoring laboratory tests when LMWH was prescribed. If we had computed the costs of automatic infusion pump use this economic advantage would be even greater. The economic advantages are in agreement with those found in studies that compared the costs of UFH and LMWH for VTE treatment [14,15,16,18,23,24] despite the differences in cost components like home care [14,23], outpatient treatment [14], and the health care providers’ perspective [18]. It must be emphasized that the institution is a tertiary university-affiliated hospital that acquires great quantities of the drug. Therefore, it is a favored buyer that can purchase LMWH at prices significantly lower than the average wholesale price. The costs saving associated to LMWH may be explained mainly by the supplies associated with UFH treatment. Besides of the same duration of UFH and LMWH use, computing the costs of hospitalization must implicate in greater economic differences because UFH seems to be mainly indicated for more severely ill patients. It must be noted that total cost, including hospitalization cost was not adjusted for the clinical disadvantages of patients with UFH. The sensitivity analysis was performed computing actual drugs’ prices in June 2010. The economic LMWH advantage became even S91 greater as a result. Because of evidence of a drug contaminant associated with greater incidence of adverse reactions UFH had its commercialization stopped [25]. When commercialization restarted, there was a considerable price increase for UFH, inflating costs still further. Conclusions Treatment of VTE with enoxaparin provided costs savings in a large teaching hospital located in southern Brazil, from the institutional perspective. Despite differences in the settings where studies of economic evaluation were conducted, the findings agree in regard to the economic advantages of use of LMWH. Implementation of a critical pathway for anticoagulation is desirable to promote rational use of heparins and to save costs. Source of financial support: This study was funded by Fundo de Incentivo à Pesquisa/Hospital de Clínicas de Porto Alegre. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi:10.1016/j.jval.2011.05.017, or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). REFERENCES [1] Anderson FA Jr, Wheeler HB, Goldberg RJ, et al. A population-based perspective of the incidence and case-fatality rates of deep vein thrombosis and pulmonary embolism. The Worcester DVT Study. Arch Intern Med 1991;151:933– 8. [2] Nordström M, Lindblad B, Bergqvist D, et al. A prospective study of the incidence of deep-vein thrombosis within a defined urban population. J Intern Med 1992;232:155– 60. [3] Naess IA, Christiansen SC, Romundstad P, et al. Incidence and mortality of venous thrombosis: a population-based study. Thromb Haemost 2007;5:692–9. [4] Silvestein MD, Heit JA, Mohr DN, et al. Trends in the incidence of deep vein thrombosis and pulmonary embolism: a 25-year populationbased study. Arch Intern Med 1998;158:585–93. [5] Hirsh J, Raschke R. Heparin and low-molecular-weight heparin: the seventh ACCP conference on antithrombotic and thrombolytic therapy. Chest 2004;126:188s–203s. [6] Brandjes DBM, Heijboer H, Buller HR, et al. Acenocoumarol and heparin compared with acenocoumarol alone in the initial treatment of proximal-vein thrombosis. N Engl J Med 1992;327:1485–59. [7] McRae S, Eikelboom JM. Latest medical treatment strategies for venous thromboembolism. Expert opinion on pharmacotherapy 2007;8:1221–33. [8] Majerus PW, Tollefsen DM. Bood coagulation and anticoagulant, thrombolitic, and antiplatelet drugs. In: Brunton LL, Lazo JS, Parker KL, (ed.). The Pharmacological Basis of Therapeutics. (11th ed.). New York: Brunton LL, Lazo JS, Parker KL, 2005. [9] Snow V, Qassem A, Barry P, et al. Management of venous thromboembolism: a clinical practice guideline from the american college of physicians and the american academy of famy physicians. Ann Inten Med 2007;146:204 –10. [10] Quinlan DJ, McQuillan A, Eikelboom JW. Low-molecular-weight heparin compared with intravenous unfractioned heparin for treatment of pulmonary embolism: a meta-analysis of randomized, controlled trials. Ann Intern Med 2004;140:175– 83. [11] Kearson C, Ginsberg JS, Julian JA, et al. Comparison of fixed-dose weight-ajusted unfractionated heparin and low-molecular-weight heparin for acute treatment of venous thromboembolism. JAMA 2006; 296:23–30. [12] Merli G, Spiro TE, Olsson CG, et al. Subcutaneous enoxaparin once or twice daily compared with intravenous unfractionated heparin for treatment of venous thromboembolic disease. Ann Intern Med 2001; 134:191–202. [13] Mismetti P, Quenet S, Levine M, et al. Enoxaparin in the treatment of deep vein thrombosis with or without pulmonary embolism. Chest 2005;128:2203–10. S92 VALUE IN HEALTH 14 (2011) S89 –S92 [14] Boucher M, Rodger M, Johnson JA, et al. Shifting from inpatient to outpatient treatment of deep vein thrombosis in a tertiary care center. A cost-minimization analysis. Pharmacoterapy 2003; 23:301–9. [15] Spyropoulos AC, Hurley JS, Ciesla GN, et al. Management of acute proximal deep vein thrombosis: phamacoeconomic evaluation of outpatient treatment with enoxaparin vs inpatient treatment with unfractioned heparin. Chest 2002;122:108 –14. [16] Gould MK, Dembitzer AD, Sanders GD, et al. Low-molecular-weight heparins compared with unfractioned heparin for treatment of acute deep venous thrombosis: a cost-effectiveness analysis. Ann Intern Med 1999;130:789 –99. [17] Holzheimer RG. Low-molecular-weight heparin (LMWH) in the treatment of thrombosis. Eur J Med Res 2004;9:225–39. [18] Aujesky D, Smith K, Cornuz J, et al. Cost-effectiveness of lowmolecular-weight heparin for treatment of pulmonary embolism. Chest 2005;128:1601–10. [19] Huse DM, Cummins G, Taylor DC, et al. Outpatient treatment of venous thromboembolism with low-molecular-weight heparin: an economic evaluation. Am J Manag Care 2002;(89 Suppl.):S10 – 6. [20] Hureta CJS, Wallander MA, Garcia RLA. Risk factors an short-term mortality of venous thromboembolism diagnosed in the primary care setting in the United Kingdom. Arch Intern Med 2007;167:935– 43. [21] Dongen CJJ, Belt AGM, Prins MH, et al. Fixed dose subcutaneous low molecular weight heparins versus adjusted dose unfractionated heparin for venous thromboembolism. Cochrane Database Syst Rev 2004;18:CD001100. [22] Lim W, Dental F, Eikelboom JW, et al. meta-analysis: low-molecularweight heparin and bleeding in patients with severe renal insufficiency. Ann Intern Med 2006;144:673– 84. [23] Spyropoulos AC, Frost FJ, Hurley JS, et al. Costs and clinical outcomes associated uit low-molecular-weight heparin vs unfractioned heparin for perioperative bridging in patient receiving long-term oral anticoagulat therapy. Chest 2004;125:1642–50. [24] Knight KK, Wong J, Hauch O, et al. Economic and utilization outcomes associated with choice of treatment for venous thromboembolism in hospitalized patients. Value Health 2005;8:191–200. [25] Kishimoto TK, Viswanathan K, Ganguly T, et al. Contaminated heparin associated with adverse clinical events and activation of the contact system. N Engl J Med 2008;358:2457– 67. VALUE IN HEALTH 14 (2011) S93–S95 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval The Burden of Moderate/Severe Premenstrual Syndrome and Premenstrual Dysphoric Disorder in a Cohort of Latin American Women Alexandre Schiola, MD1, Julia Lowin, MSc2,*, Marion Lindemann, RPh3, Renu Patel, MSc2, Jean Endicott, PhD4 1 Bayer HealthCare Pharmaceuticals, São Paolo, Brazil; 2IMS Health, London, UK; 3Bayer HealthCare Pharmaceuticals, Berlin, Germany; 4Columbia University College of Physicians and Surgeons, New York, NY, USA A B S T R A C T Objectives: The aim of this study was to investigate the relationship between symptom severity, cost, and impairment in women with moderate/severe premenstrual syndrome (PMS) or premenstrual dysphoric disorder (PMDD) in a Latin American setting. Methods: A model was constructed based on analysis of an observational dataset. Data were included from four Latin American countries. Responder-level data were analysed according to four categories of symptom severity: Category 1 comprised Daily Record of Severity of Problems score 21 to 41.9, Category 2 score was 42 to 62.9, Category 3 score was 63 to 83.9, and Category 4 was a score of 84 or higher. Burden was estimated in terms of impact on job and activities using the modified work productivity and impairment questionnaire and affect on quality of life using the SF-12 questionnaire. Costs were estimated in Brazilian reals from a Brazilian private health care and societal perspective. The outputs of the analysis were estimates of burden, mean annual cost and affect on quality of life (as measured by quality adjusted life years) by symptom severity. Confidence intervals around key Introduction Premenstrual dysphoric disorder (PMDD) is a severe form of premenstrual syndrome (PMS). Key features include depressed mood, anxiety, affective liability, persistent anger or irritability, and change in appetite or sleep. By Diagnostic Statistical Manual, 4th edition, definition, in women with PMDD symptoms are severe enough to markedly affect usual daily activities. More than 80% of women of reproductive age may experience some emotional and/or physical premenstrual symptoms [1-3] and approximately 3% to 8% of menstruating women have symptoms that are severe enough to meet the specific diagnostic criteria for PMDD [4,5]. Expert review of country-specific data suggests that the prevalence of PMS/PMDD may be higher in Latin American countries than in North American or European cohorts [6]. PMDD and moderate/severe PMS are associated with significant impairment as measured by a number of scales [7,8]. How- outcomes were generated through nonparametric bootstrapping. Results: Analysis suggests a significant cost burden associated with moderate/severe PMS and PMDD with mean per patient annual costs estimated at 1618 BRL (95% confidence interval 957–2,481). Although the relationship between cost, quality of life, and severity was not clear, analysis showed a consistent relationship between disease severity and measures of disease burden (job and daily activity). Burden on activities increased with disease severity. Conclusions: Our analysis, conducted from a Latin American perspective, suggests a significant burden and an increasing impairment associated with moderate/ severe PMS and PMDD. Keywords: burden, daily record of severity of problems, premenstrual dysphoric disorder, premenstrual syndrome. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. ever, there are only a limited number of studies available that attempt to quantify the influence of these impairments [9,10]. To the author’s knowledge, no studies have been conducted from a Latin American perspective. In the absence of such data, a comprehensive awareness of the impact of PMDD is lacking. Within Latin America, PMDD is not fully accepted as a discrete disease entity. Providing more evidence on the burden of PMDD is likely to increase the awareness of PMDD and more clearly define the disease. A multicountry observational study was recently conducted in women to assess the impact of pre-menstrual symptoms (see Box 1 in Supplemental Material found at doi:10.1016/j.jval. 2011.05.008). The impact was measured through assessment of direct and indirect outcomes via the modified work productivity and impairment questionnaire (m-WPAI) and the impact on quality of life through use of the SF-12 questionnaire. The study has generated new information about resource use and quality Conflicts of interest: Bayer HealthCare Pharmaceuticals has commissioned IMS Health HQ Ltd to conduct a model to investigate the relationship between symptom severity, cost, and impairment in women with moderate/severe premenstrual syndrome (PMS) or premenstrual dysphoric disorder (PMDD) in a Latin American setting. The preparation of the work and the development of this manuscript was based on the results of the model development which was conducted as a collaboration between IMS Health HQ Ltd, Jean Endicott as external expert, and Bayer Health Economics and Outcomes Research team, i.e. by the authors listed. * Address correspondence to: Julia Lowin, IMS Health, 7 Harewood Avenue, London NW1 6JB, UK. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.008 S94 VALUE IN HEALTH 14 (2011) S93–S95 of life among women with PMS [11]. This article reports an analysis of these data designed to quantify the burden associated with PMDD and moderate/severe PMS in women from a Latin American subset of the dataset. Methods The aim of our study was to investigate the relationship between symptoms, burden, and impairment in women with moderate/ severe PMS or PMDD, based on the findings of a multicenter observational study (the IMPACT study) and assignment of countryspecific cost data. The IMPACT study [11] reported symptom severity, measured by the Daily Record of Severity of Problems (DRSP) form, alongside impairment measured by a modified WPAI scale, resource use, productivity, and quality of life. The DRSP is an accepted measure of the severity of symptoms and impairment at various phases of the menstrual cycle [12] (see Box 2 in Supplemental Materials found at; doi:10.1016/j.jval.2011.05.008). The WPAI is a validated scale for measurement of the effects of disease on daily living activities and includes assessment of affect on productivity. Versions of the scale have been validated in a Latin American setting [13]. The original WPAI was modified to include additional questions related to resource use. In our analysis, impairment (as measured by the m-WPAI), annual direct and indirect costs (estimated from reported resource use and lost productivity), and quality-adjusted life years (QALYs) (estimated from reported quality of life) were estimated according to symptom severity for a subset of women included in the observational study. The observational study was pan-global with data collected from women in 19 countries. Our analysis focused on Latin America, and was limited to consideration of a subset of the cohort of women from the four Latin American countries included in the observational study (Brazil, Venezuela, Colombia, and Mexico). We additionally considered only those women experiencing moderate/severe PMS or PMDD. Data were pooled to increase the power of the analysis and retain the statistical integrity of the dataset. We explored the health statistics of the four included countries to determine if they constituted an economically similar group suitable for pooled analysis [14]. The aim was to develop a dataset that might serve as a proxy for individual country analyses in a Latin American setting. The country perspective selected was that of Brazil and the cost analysis was conducted from a Brazilian perspective. A societal perspective based on private health care costs was chosen. The Brazilian health care system is a two-tier system with a national health system (Sistema Único de Saúde) as well as a supplementary private medical system funded by private funds or through work insurance schemes. The burden of disease calculated is most relevant to those women who used the private system in that these women would be most likely to seek treatment for PMDD. We therefore analyzed the burden of moderate to severe PMS and PMDD from a private health perspective, including the costs to the government from lost productivity. Severity was defined by DRSP total luteal score, and costs and QALYs were reported by the defined categories of severity. The approach to categorization followed previously reported analyses [15]. Five severity categories were initially defined on the basis of previous analysis. Initial review of the data indicated very limited numbers in the top two severity categories; these categories were therefore combined to improve the sample size of the analysis. The reported analyses are based on four categories: Category 1 comprised women with DRSP scores 21 to 41.9, Category 2 was made up of women with scores 42 to 62.9, Category 3 was made up of women with scores of 63 to 83.9, Category 4 included women with scores of 84 or more. Category 1 includes those women with minimal symptoms; Category 4 includes women at the most se- vere end of the symptom score spectrum. These women are expected to show significant impairment. To quantify the costs, available per cycle data on lost productivity and medical resource use were extracted from the dataset for each responder and unit costs were applied. These data were then extrapolated to estimate an annual expected cost. We conducted this analysis assuming a Brazilian perspective. The number of hours of lost productivity as reported in the m-WPAI data were multiplied by an hourly wage for Brazil to estimate the total cost of lost productivity per respondent per cycle and were extrapolated to estimate annual costs. Costs for respondents not in employment (and therefore reporting lost productivity relating to household chores) were calculated in the same manner, as the cost of their time is represented by the opportunity cost, which is to be in paid employment (following the human capital approach). We obtained a wage of 7.23 BRL per hour for female manual workers in the private sector [16]. Data related to reduced productivity (i.e., presenteeism) and affect on daily living activities were also analysed by DRSP category. Data were analysed according to the degree of affect on work or daily living activities reported in the m-WPAI. Responses were collapsed into two categories, moderate/severe affect, or mild/not relevant affect. To better understand the study findings, reported treatment use was also estimated and analysed according to country. Results Table 1 (Table 1 in Supplemental Materials found at doi:10.1016/ j.jval.2011.05.008) shows expenditure on health by country using a number of indicators from the World Health Organization database [14]. Because our analysis relies on having a sufficient sample size to assess effects of disease severity by category, we chose to include all countries in the analysis despite some discrepancies in the similarity of their health expenditures (the implications of this are reviewed in the Discussion). The analysis dataset comprised a total of 292 women classified as having moderate/severe PMS or PMDD. Data on quality of life were reported for a subset of these women and 63 women are included in the analysis of utility outcomes. Women were assigned to categories according to their DRSP total luteal symptom score. Unit costs (Table 2 in Supplemental Materials at doi:10.1016/ j.jval.2011.05.008) were applied to resource estimates to estimate costs. Mean per cycle values for lost productivity and medical resource use were calculated for each DRSP category and extrapolated to estimate the expected mean annual cost and utility influence as measured by QALY (Table 3 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.008). Confidence intervals (CIs) are reported to address the issue of uncertainty. The mean annual cost per woman with moderate/severe PMS or PMDD (independent of severity) was estimated at 1618 BRL (95% CI 957 BRL–2481 BRL). The level of influence on work and daily living activities was estimated. Figure 1 (in Supplemental Materials found at doi: 10.1016/j.jval.2011.05.008) shows the proportion of women who rated the burden of their condition on job activities as “not relevant/mild” or “moderate/severe” by DRSP category. Figure 2 (in Supplemental Materials at doi:10.1016/j.jval.2011.05.008) shows the proportion of women who rated the burden on daily living activities as “not relevant/mild” or “moderate/severe” by DRSP category. The profile of individual country resource use was explored. Figure 3 (in Supplemental Materials found at: doi:10.1016/j.jval.2011.05. 008) reports the percentage of women who reported treatment use across all resource categories. VALUE IN HEALTH 14 (2011) S93–S95 Discussion and Implications Categorical analysis of the observational data according to DRSP score found that work and daily activity impairment increased with DRSP severity. Our analysis of the data suggests a significant cost burden associated with moderate/severe PMS and PMDD although the relationship between severity and cost is not clear. In this example, costs are estimated from a Brazilian societal perspective. Analysis of the individual country resource use data suggested differences in resource profiles. We would suggest that difference in resource use may in part be explained by the different distribution of disease severity across the individual country datasets (Figure 4 in Supplemental Materials found at: doi:10.1016/j.jval. 2011.05.008). For example, Brazil has a much lower proportion of women in the lowest DRSP category (13% compared to 34%, 29%, and 22% for Colombia, Venezuela, and Mexico, respectively), whereas the Venezuelan and Mexican cohorts had proportionally smaller numbers of women in the highest DRSP category. This is in itself an interesting finding that may suggest a difference in perception of severity of symptoms across the Latin American countries included in the analysis. Further investigation of this is beyond the scope of the current article. Analysis of the costs associated with medical resource use and lost productivity and the effects on quality of life did not show the expected relationship with disease severity. Our analysis showed a relationship between disease severity and measures of disease burden (e.g., job burden, activity burden, and number of treatments used). This analysis showed that, as expected, the proportion of women who experience burden on their activities/use of treatments increases with disease severity. However, results suggest that this did not translate into increasing costs associated with resource use and productivity loss. This is contrary to findings of previous analyses [15]. It is possible that the reason for this outcome is that women in Latin American countries are both less likely to present for costly treatment and less likely to be absent from work (despite severity of symptoms) then their Organisation for Economic Co-operation and Development counterparts. Presenteeism and affect on daily activities are reported consistently across the datasets and show a clear relationship with increasing severity as measured by DRSP score. It would be interesting to explore the reasons for the lack of a stable relationship between more expensive resources used and DRSP category to more fully understand if region-specific difference in health care availability drives the differences between this and the Organisation for Economic Co-operation and Development analysis. Despite this, the estimated cost burden associated with women with moderate/severe PMD or PMDD was found to be high with the mean annual burden estimated in the region of 1500 BRL. Further quantification of the monetary effects of PMDD and research into potential geographical interactions is needed to better understand the burden of moderate to severe PMS and PMDD. The dataset reported here has a wide applicability and the approach described could be used to generate such evidence for other countries. Acknowledgments The dataset used in this research was provided by ZEG, courtesy of Professor Lothar Heinemann. S95 Source of financial support: Marion Lindemann and Alexandre Schiola are employees of Bayer HealthCare Pharmaceuticals. Jean Endicott is an employee of New York State Departement of Mental Health, New York, NY, USA and received a consultancy fee. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi:10.1016/j.jval.2011.05.008, or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). REFERENCES [1] Johnson SR. The epidemiology and social impact of premenstrual symptoms. Clin Obstet Gynecol 1987;30:367–76. [2] Rodrigues IC, de Oliveira E. Prevalência e convivência de mulheres com síndrome pré-menstrual [Prevalence and sociability of women with premenstrual syndrome]. Arq Ciênc Saúde 2006;13:146 –52. [3] Petta CA, Duarte Osis MJ, de Padua KS, et al. Premenstrual syndrome as reported by Brazilian women, Int J Gynecol Obstet 2010;108:40 –3. [4] Cohen LS, Soares CN, Otto MW, et al. Prevalence and predictors of premenstrual dysphoric disorder (PMDD) in older premenopausal women. J Affect Disord 2002;70:125–32. [5] Wittchen HU, Becker E, Lieb R, Krause P. Prevalence, incidence and stability of premenstrual dysphoric disorder in the community. Psychol Med 2002;32:119 –32. [6] Bahamondes L, C’ordova-Eguez S, Pons JE, Shulman L on behalf of the Latin America Experts Group. Perspectives on premenstrual syndrome/premenstrual dysphoric disorder. Dis Manage Health Outcomes 2007;15:263–77. [7] Pearlstein T, Steiner M. Premenstrual dysphoric disorder: burden of illness and treatment update. J Psychiatry Neurosci 2008;33:291–301. [8] Halbreich U, Borenstein J, Pearlstein T, Kahn LS. The prevalence, impairment, impact, and burden of premenstrual dysphoric disorder (PMS/PMDD). Psychoneuroendocrinology 2003;28(Suppl. 3):1–23. [9] Borenstein JE, Dean BB, Leifke E, et al. Differences in symptom scores and health outcomes in premenstrual syndrome. J Womens Health (Larchmt) 2007;16:1139 – 44. [10] Yonkers KA, Holthausen GA, Poschman K, Howell HB. Symptom-onset treatment for women with premenstrual dysphoric disorder. J Clin Psychopharmacol 2006;26:198 –202. [11] Heinemann LA, Minh TD, Filonenko A, Uhl-Hochgräber K. Explorative evaluation of the impact of severe premenstrual disorders on work absenteeism and productivity. Womens Health Issues 2010;20:58 – 65. [12] Endicott J, Nee J, Harrison W. Daily record of severity of problems (DRSP): reliability and validity. Arch Womens Ment Health 2006;9:41–9. [13] Ciconelli RM, Soárez PC, Kowalski CC, Ferraz MB. The Brazilian Portuguese version of the work productivity and activity impairment: general health (WPAI-GH) questionnaire. Sao Paulo Med J 2006;124:325–32. [14] World Health Organisation. WHO statistical Information System (WHOSIS). Available from: http://www.who.int/whosis/en. [Accessed March 8, 2011]. [15] Lowin J, Endicott J, Patel R, et al. Estimating the Burden of Women Suffering from PMS/PMDD: Analysis of a Cross-Sectional Dataset. October 2009, ISPOR 12th Annual European Congress, Paris, France. Available from: http://ispor.org. [Accessed March 8, 2011]. [16] Ministry of Labour (Relação Anual de Informações Sociais - 2008 Ministerio do trabalho). Annual Social Information. Available from: http://www.mte.gov.br/rais/default.asp. [Accessed March 8, 2011]. VALUE IN HEALTH 14 (2011) S96 –S99 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval HEALTH OUTCOMES ANALYSIS Análisis de la Satisfacción con los Cuidados en Salud a Través del Cuestionario EORTC IN-PATSAT32 en Pacientes con Cáncer de Mama, Linfoma no Hodgkin y Cáncer Colo-Rectal en Diferentes Etapas Clínicas. Relación con las Características Socio-Demográficas, Estados Co-Mórbidos y Variables del Proceso de Atención en el Instituto Mexicano del Seguro Social Luz-Ma-Adriana Balderas-Peña, MD, PhD1,*, Daniel Sat-Muñoz, MD2, Iris Contreras-Hernández, MD, MSc3, Pedro Solano-Murillo, MD4, Guillermo-Allan Hernández-Chávez, MD, MSc4, Ignacio Mariscal-Ramírez, MD4, Martha Lomelí-García, Chemist, Pharm4, Margarita-Arimatea Díaz-Cortés, Eng5, Joaquín-Federico Mould-Quevedo, PhD6, Juan-Manuel Castro-Cervantes4, Oscar-Miguel Garcés-Ruiz4, Gilberto Morgan-Villela, MD, MSc4 1 Unidad de Investigación Médica en Epidemiología Clínica, UMAE Hospital de Especialidades del Centro Médico Nacional de Occidente, Instituto Mexicano del Seguro Social, Departamento de Farmacobiología del Centro Universitario de Ciencias Exactas e Ingenierías de la Universidad de Guadalajara, Guadalajara, Jalisco, México; 2Servicio de Oncología Quirúrgica, División de Cirugía, Hospital General Regional No. 46, Guadalajara, Jalisco, Instituto Mexicano del Seguro Social, Departamento de Morfología del Centro Universitario de Ciencias de la Salud de la Universidad de Guadalajara, Guadalajara, Jalisco, México; 3Unidad de Investigación en Economía de la Salud del Centro Médico Nacional Siglo XXI, Instituto Mexicano del Seguro Social, México, D. F., México; 4División de Oncología, UMAE Hospital de Especialidades del Centro Médico Nacional de Occidente, Instituto Mexicano del Seguro Social, Guadalajara, Jalisco, México; 5 Unidad de Investigación y Estudios de Posgrado, División de Electrónica y Computación, Centro Universitario de Ciencias Exactas e Ingenierías, Universidad de Guadalajara, Guadalajara, Jalisco, México; 6Instituto Tecnológico de Estudios Superiores de Monterrey, Campus Ciudad de México, Guadalajara, Jalisco, México A B S T R A C T Introduction. In Mexico cancer is a public health burden. Nowadays the health care systems pay special attention to patient’s perception and satisfaction of the health care received. Satisfaction with quality of health care has an impact in the adherence to the treatment. Objective. To evaluate the satisfaction with the quality of health care received at the IMSS in a group of cancer patients [non Hodgkin lymphoma (NHL), breast and colorectal cancer]. Socio-demographic features, co-morbid diseases, and attendance processes impact on satisfaction are also evaluated. Results. 476 cancer patients were studied: 314 with breast cancer, 92 with NHL and 70 with colorectal cancer. In women with breast cancer the mean score to nurses’ interpersonal skills in non-classified disease group and clinical stage III group were: 73.64⫾32.53, 90.00⫾18.25 respectively (p⫽0.005), nurses’ availability in non-classified and clinical stage III group were: 69.71⫾30.25, 89.21⫾19.00 respectively (p⫽0.003). In subjects with NHL the mean scores for doctors’ technical skills in Introducción El cáncer es un problema de Salud Pública en México [1]. La Agencia Internacional para la Investigación en Cáncer (IARC: International Agency for Research on Cancer) publicó: En 2008, el 56% de clinical stage I and III groups, were: 63.69⫾37.78, 80.30⫾18.46 respectively (p⫽0.017), doctors’ information provision scores in subject in clinical stage I and IV were: 49.40⫾40.75, 79.49⫾24.63 respectively (p⫽0.043). In the group of colorectal cancer patients the mean of the score to exchange of information between clinical stage II and clinical stage III group were 50.00⫾41.83, 84.21⫾22.37 respectively (p⫽0.036). Were not observed association between attendance processes features and general satisfaction. Conclusions. In Mexico 50% of cancer patients are attended at the IMSS. The continued evaluation of the satisfaction with health care received by the health care service users is important to enhance attention’s quality. Palabras Claves. breast cancer, colorectal cancer, IN-PATSAT32, nonHodgkin lymphoma, satisfaction with health care, process of care. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. los 12.7 millones de nuevos casos de cáncer y 63% de los 7.6 millones de muertes por cáncer se presentan en países subdesarrollados [2]. En México durante 2008, hubo 127,604 casos nuevos de cáncer (tasa de incidencia 117.54/100,000 habitantes) [2]. El Programa Nacional de Salud 2007–2012, reportó 40,000 defunciones Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Título corto: Satisfaction with health in cancer patients at the IMSS. * Autor de correspondencia: Balderas-Peña Luz-Ma-Adriana, 1000 Belisario Domínguez, Colonia Independencia, Guadalajara, Jalisco, México 44340; Tel: 52 33 3618 2661; Fax: 52 33 3663 1834. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.026 VALUE IN HEALTH 14 (2011) S96 –S99 por cáncer en México durante 2005 [3,4]. En Jalisco la incidencia de cáncer durante 2008 fue 91.10/100,000 habitantes; en mujeres la tasa fue 105.6/100,000 mujeres y se reportaron 9,278 casos nuevos de cáncer, de los cuales el Instituto Mexicano del Seguro Social (IMSS) reportó el diagnóstico inicial en el 28.58% (2,652 casos). La neoplasia más frecuente en Jalisco fue cáncer de mama (n⫽1,251; 18.6%), seguido por cáncer del aparato genital masculino (n⫽920; 13.7%) y gastrointestinal (n⫽921; 13.7%). Seis tumores aparecen en los primeros 20 lugares de mortalidad general, destacan: mama (16°) y colo-rectal (20°). Los casos incidentes de linfoma no Hodgkin (LNH) fueron 169 y son la neoplasia linfática más frecuente; datos que destacan la importancia de estos, dentro del Sistema Nacional de Salud Mexicano y de otros países con sistemas de salud y economías similares [5]. Cada vez se presta mayor atención a como los pacientes perciben la calidad de los cuidados médicos que reciben (PRO; siglas en inglés de patient-reported outcomes), en aspectos como: calidad de vida y satisfacción, en su asociación con variables socio-demográficas, médicas y psicológicas [6-8]. En los PRO se contemplan las expectativas de los pacientes respecto al apoyo emocional y cognitivo recibido de los prestadores de servicios de salud [9]. Los resultados previos muestran cómo la satisfacción, influye en el apego al tratamiento [10]. Existe la necesidad generalizada de identificar las prioridades de mejora en los servicios de salud que se brindan a los pacientes con cáncer. Dentro de las ya identificadas encontramos: emocionales, de información a pacientes y familia, reducción en tiempos de espera [11] y la existencia de un cuidado coordinado entre el médico y el equipo de salud. En este contexto los PRO se consideran complemento y factor pronóstico independiente de los resultados biomédicos tradicionales (supervivencia global, supervivencia libre de enfermedad) que aportan información para la toma de las decisiones en el cuidado de los pacientes oncológicos y la implementación de políticas de salud eficientes [12-14]. Estos aspectos reflejan el compromiso y la solidaridad social, que forman parte de la filosofía en la prestación de los servicios de salud dentro del IMSS, al que tienen derecho 43’583,112 de usuarios y los pacientes oncológicos son atendidos de forma coordinada entre sus diferentes unidades médicas y niveles de atención, con énfasis en la satisfacción de su población usuaria. El objetivo de este estudio fue evaluar la satisfacción de pacientes con cáncer de mama, colo-rectal y LNH en diferentes etapas clínicas, respecto a la calidad de los cuidados médicos recibidos en el IMSS y el impacto de las variables socio-demográficas, estados co-mórbidos y variables del proceso de atención sobre los puntajes de satisfacción. Métodos Diseño Estudio transversal, 476 casos incidentes del 1° de enero del 2008 al 31 de enero del 2009: 314 (65.97%) con cáncer de mama, 70 (14.71%) con colo-rectal y 92 (19.33%) con LNH, atendidos en dos hospitales de segundo y uno de tercer nivel de atención del IMSS, en Guadalajara, Jalisco, México. Se aplicó el cuestionario EORTC (European Organisation for Research and Treatment of Cancer) IN-PATSAT32 en español, validado para población mexicana [15,16], y un cuestionario de variables socio-demográficas y del proceso de atención. Etapa clínica (EC), tiempos de atención y tratamientos se tomaron del expediente clínico. El proyecto fue aprobado por la Comisión Nacional de Investigación Científica del IMSS. Recolección de datos Los pacientes fueron reclutados en unidades de quimioterapia ambulatoria y consulta externa de oncología de los hospitales par- S97 ticipantes. Se estudiaron: género, edad, EC, escolaridad, estado civil, co-mórbidos (diabetes mellitus, hipertensión arterial, enfermedades cerebro-vasculares, enfermedades reumáticas y otros), proceso de atención (tiempo entre el diagnóstico y la realización de un procedimiento quirúrgico indicado, entre la indicación y la administración de la quimioterapia y/o radioterapia), realización o no de cirugía, administración o no de quimioterapia y/o radioterapia, número de ciclos de quimioterapia, número de sesiones de radioterapia, apoyo psicológico individual o grupal, orientación nutricional y terapia física-rehabilitación. Análisis estadístico Se determinó confiabilidad y consistencia interna en escalas multi-ítem del cuestionario IN-PATSAT32 mediante alfa de Cronbach; éstas incluyeron: satisfacción con el cuidado médico y de enfermería (habilidades de comunicación interpersonal, habilidades técnicas, disposición para proporcionar información y disponibilidad) y satisfacción con los cuidados de salud (amabilidad y utilidad de la información proporcionada por otro personal del hospital, tiempo de espera y accesibilidad). Se contemplaron tres escalas de un ítem: intercambio de información, comodidad-limpieza y satisfacción general. Todas las escalas (multi-ítem y un ítem) están construidas de manera similar. El promedio bruto (raw score) de cada ítem individual es sumado y en las escala multiítem dividido entre el número de ítems que integran la escala; esos puntajes de las escalas son transformados linealmente para obtener un puntaje de 0 a 100 (puntaje más alto representa un mayor nivel de satisfacción); acorde a las fórmulas e instrucciones proporcionadas en el Scoring Procedure for the EORTC INPATSAT 32 [16]. Los tiempos de atención se cuantificaron en días; se calcularon medianas (Md) y percentiles 25 y 75, los puntajes del cuestionario de satisfacción se analizaron con promedios y desviaciones estándar. Las diferencias entre EC se analizaron con ANOVA y pos hoc con estadístico de Bonferroni. Las diferencias por género, estado marital, escolaridad, antecedentes personales de cáncer y co-mórbidos se analizaron con T de student. Las variables relacionadas con administración de tratamiento u obtención de apoyo, se categorizaron de forma dicotómica. Se cuantificaron ciclos de quimioterapia y sesiones de radioterapia. La asociación se analizó con coeficiente de correlación de Pearson (rP). Un valor p menor o igual a 0.05 se consideró significativo. Los datos fueron analizados en Excel 2007 y en SPSS V16.0 (SPSS. Chicago, IL, USA). Resultados Se estudiaron 476 pacientes: 314 con cáncer de mama, 92 con LNH y 70 con cáncer colo-rectal, con edades promedio de 52.29 (⫾10.43), 53.1 (⫾18.09) y 56.03 (⫾12.97) respectivamente. Predominó el género femenino con 84.7% y 15.3% masculino (cuadro 1 en Material Complementario en: doi:10.1016/j.jval.2011.05.026). El estado marital se clasificó en: no unidos (n⫽136; 28.57%), y unidos (70.17%; n⫽334), el resto (n⫽6) no lo especificaron. La escolaridad predominante fue educación básica y media básica (57.77%; n⫽275), seguida por educación media superior en pacientes con cáncer de mama (18.5%; n⫽58) y LNH (23.9%; n⫽22) y por educación superior en cáncer colo-rectal (25%; n⫽17) (cuadro 2 en Material Complementario en: doi:10.1016/j.jval.2011.05.026). Variables del proceso de atención (en días) Cáncer de mama: tiempo entre primera consulta y confirmación de cáncer: Md 30 (1 a 124.5); tiempo de atención en unidad de medicina familiar (UMF): Md 1 (1 a 30); tiempo de atención en hospital de segundo nivel: Md 120 (30 a 392.5); tiempo de atención en hospital de tercer nivel: Md 91 (30 a 270); tiempo entre confirmación diagnóstica y cirugía: Md 21 (14 a 34.5 días); tiempo entre S98 VALUE IN HEALTH 14 (2011) S96 –S99 prescripción y administración de quimioterapia (QT): Md 6 (3 a 14); tiempo entre prescripción y aplicación de radioterapia (RT): Md 10 (2 a 29.5). La mediana de ciclos de QT fue 6 (6 a 8), y de sesiones de RT 25 (25 a 25; un ciclo mamario completo). LNH LNH: tiempo entre primera consulta y confirmación de cáncer: Md 22.5 (12.25 a 136); tiempo de atención en UMF: Md 40 (1 a 143.75); tiempo de atención en hospital de segundo nivel: Md 5.5 (1 a 97.5); tiempo de atención en hospital de tercer nivel: Md 210 (56.26 a 395); tiempo entre confirmación y cirugía: Md 17 (0.75 a 67.5); tiempo entre prescripción y administración de QT: Md 11.5 (5.75 a 15); tiempo entre prescripción y aplicación de RT: Md: 8.5 (0 a 49.25). Mediana de ciclos QT: 11.5 (5.75 a 11), y de sesiones RT: 8 (6 a 15). Cáncer colo-rectal: tiempo entre primera consulta y confirmación de cáncer: Md 30.00 (3.00 a 90); tiempo de atención en UMF: Md 1 (1 a 60); tiempo de atención en hospital de segundo nivel de atención: Md 8 (2 a 250); tiempo de atención en hospital de tercer nivel: Md 60 (30 a 365); tiempo entre confirmación y cirugía: Md 0 (0.00 a 12.00) días; tiempo entre prescripción y administración de QT: Md 10 (2 a 24); tiempo entre la prescripción y aplicación de RT: Md 10 (10 a 21). Mediana de ciclos de quimioterapia administrados: 6 (4 a 10), y sesiones de radioterapia 25 (22 a 25). La mayor parte de pacientes no recibieron apoyo psicológico (90.54%), en grupos de auto-ayuda (94.12%), orientación nutricional (90.13%), terapia física y rehabilitación (99.16%) u otros servicios de apoyo (cuadro 2 en Material Complementario en: doi: 10.1016/j.jval.2011.05.026). El análisis de confiabilidad y consistencia interna mostró valores de alfa de Cronbach mayores a 0.700 en todas las escalas multi-ítem. Los puntajes se analizaron por EC y por patología con los siguientes hallazgos. Cáncer de mama: percepción de las habilidades de comunicación interpersonal en enfermería (p⫽0.005), en el grupo de pacientes con enfermedad no clasificable y en EC-III (73.64⫾32.53, 90.00⫾18.25 respectivamente), disponibilidad del personal enfermería (69.71⫾30.25, 89.21⫾19.00 respectivamente; p⫽0.003) percibida por estos grupos de pacientes (cuadro 3 en Material Complementario en: doi:10.1016/j.jval.2011.05.026). LNH: percepción de habilidades técnicas de los médicos; percibidas por pacientes en EC-I y EC-III (63.69⫾37.78, 80.30⫾18.46 respectivamente; p⫽0.017), disposición para brindar información por parte de los médicos; percibida por los pacientes en EC-I y IV (49.40⫾40.75, 79.49⫾24.63 respectivamente; p⫽0.043) (cuadro 4 en Material Complementario en: doi:10.1016/j.jval.2011.05.026). Cáncer colo-rectal: Intercambio de información percibido por pacientes en EC-II versus (vs.) EC-III (50.00⫾41.83, 84.21⫾22.37 respectivamente; p⫽0.036) (cuadro 4 en Material Complementario en: doi:10.1016/j.jval.2011.05.026). Se analizó también la satisfacción al agrupar a los pacientes por género, estado civil, escolaridad, antecedentes personales de cáncer y co-mórbidos. Se encontró una menor satisfacción en mujeres con cáncer de mama sin co-mórbidos que en aquellas con alguna patología en: habilidades de comunicación interpersonal (73.63⫾27.34; 81.42⫾21.49 respectivamente; p⫽0.007), habilidades técnicas (77.83⫾22.97, 83.61⫾18.50 respectivamente; p⫽0.018), disposición para proporcionar información (72.29⫾28.17, 78.65⫾24.97; respectivamente p⫽0.044) en el personal médico; amabilidad y utilidad de la información proporcionada por otro personal del hospital (67.16⫾25.77, 73.19⫾25.77 respectivamente; p 0.047), tiempo de espera (70.58⫾26.30, 78.25⫾23.18; respectivamente p⫽0.009) y satisfacción general (70.57⫾25.56, 78.88⫾20.44; respectivamente p⫽0.002) (cuadro 5 en Material Complementario en: doi:10.1016/j.jval. 2011.05.026). En LNH se encontraron diferencias en: habilidades de comunicación interpersonal de enfermería (87.71⫾18.78, 95.83⫾91.0; p⫽0.040) y satisfacción general (83.75⫾23.03, 94.64⫾10.64; p⫽0.023), los puntajes de pacientes sin antecedente de co-mórbidos fueron más altas (cuadro 6 en Material Complementario en: doi:10.1016/j.jval.2011.05.026). En cáncer colo-rectal se observaron diferencias en: disposición del médico para proporcionar información (75.31⫾31.22; 79.82⫾24.17; p⫽0.032); el puntaje más bajo fue de pacientes con educación media superior, seguidos por aquellos con educación básica y media básica. La calificación más alta fue la otorgada por pacientes con educación superior. Pacientes con antecedentes de un cáncer previo mostraron mayor satisfacción con enfermería en: habilidades de comunicación interpersonal (100.00⫾0.00, 75.13⫾28.59; p⫽0.000), habilidades técnicas (100.00⫾00, 76.39⫾27.37; p⫽0.000), disposición para proporcionar información (100.00⫾0.00, 71.97⫾28.03; p⫽0.000) y disponibilidad (100.00⫾0.00, 74.81.28.80; p⫽0.000). En tiempo de espera (100.00⫾0.00, 70.26 versus 28.23; p⫽0.000) también se observaron diferencias. Los pacientes no unidos calificaron mejor la amabilidad y utilidad de la información proporcionada por otro personal del hospital (83.33⫾23.17; 64.94⫾28.42; p⫽0.021), y la satisfacción general (82.81⫾23.66; 66.98⫾26.30; p0.035). No se observó asociación entre variables del proceso de atención y satisfacción general (cuadro 7 en Material Complementario en: doi:10.1016/j.jval.2011.05.026). Conclusiones En México el IMSS atiende aproximadamente al 50% de los pacientes con cáncer, por ello, la evaluación de la satisfacción de los derechohabientes con los cuidados en salud es fundamental para elevar la calidad de la atención. Los casos incidentes estudiados, representaron el 25% de los casos de cáncer de mama reportados en 2008 en Jalisco (n⫽1251), 17.72% de cáncer colo-rectal (n⫽395) y el 54.47% de LNH (n⫽169), por lo que resultados son representativos de estas tres patología atendidas en el IMSS Jalisco y pueden extrapolarse al entorno del Sistema de Salud de México y Latinoamérica [5]. La EORTC publicó en el año 2005 la validación del instrumento IN-PATSAT32 [17], las publicaciones posteriores lo han mostrado como un instrumento útil para el análisis de aspectos relevantes en el cuidado, comparación de tratamientos, intervenciones psicosociales o de servicios de salud en pacientes de diversos países y entornos socio-culturales [18]; incluidos los hispanoparlantes, para quienes ya fue validado [19]. Los datos publicados [17-20] concuerdan con lo observado en nuestro estudio; la satisfacción no mostró asociación con las variables del proceso de atención, pero sí, con la EC, el estado marital, el nivel de escolaridad y la presencia de co-mórbidos [18]. La satisfacción con el cuidado médico se ha evaluado a través de IN-PATSAT32 en el Reino Unido, en pacientes con cáncer esofágico y gástrico. Las calificaciones obtenidas fueron en promedio: 74.2, satisfacción con personal médico, 71.9 enfermería, 61.4 en el intercambio de información y 72.5 en satisfacción general. Lo cual concuerda con nuestros resultados. También se observó que la satisfacción es independiente de la enfermedad y la modalidad de tratamiento. En conjunto esta información que puede ser utilizada por el proveedor de la atención y mejorar la percepción del paciente en cuanto a la calidad del cuidado médico que brindan las instituciones de salud [21]. Agradecimientos Los autores agradecemos la participación del grupo de encuestadores (Rosa-Emilia Ramírez-Conchas, Martha-Cristina BalderasPeña, Miguel-Ángel Martínez-López, Adolfo-Leonardo GómezBalderas, Alfredo Prieto-Moreno y Febe-Eréndira Balderas-Peña) VALUE IN HEALTH 14 (2011) S96 –S99 en la realización de este proyecto, pues sin su capacidad laboral no hubiese sido posible obtener los resultados aquí presentados. Fuentes de financiamiento: Fundación IMSS A.C. Materiales Complementarios Material complementario que acompaña este artículo se puede encontrar en la versión en línea como un hipervínculo en doi: 10.1016/j.jval.2011.05.026o si es un artículo impreso, estará en www.valueinhealthjournal.com/issues (seleccione el volumen, número y artículo). REFERENCES [1] Sistema de Indicadores de Género Morbilidad y Mortalidad. Instituto Nacional de las mujeres http://estadistica.inmujeres.gob.mx/formas/ tarjetas/Morbilidad_y_mortalidad1.pdf. [Último acceso 07 de diciembre de 2010]. [2] Ferlay J, Shin HR, Bray F, Forman D, et al. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. International Journal of Cancer. 2010;127(12):2893–917. Media Centre - IARC News 2010. GLOBOCAN 2008: Cancer Incidence and Mortality Worldwide. http:// globocan.iarc.fr. [Último acceso 29 de diciembre de 2010]. [3] Programa Nacional de Salud 2007–2012. Disponible en: http://alianza .salud.gob.mx/descargas/pdf/pns_version_completa.pdf. [Último acceso 01 de diciembre de 2010]. [4] Plan Nacional de Desarrollo 2007 – 2012. Numeral 3.2 Salud. Disponible en: http://pnd.calderon.presidencia.gob.mx/index.php?page⫽salud. [Último acceso 01 de diciembre de 2010]. [5] Gutiérrez-Carranza A, Carranco-Ortiz BG, Sandoval-Urban E, Hernández-Sánchez JE, et al. Registro Estatal de Cáncer. Jalisco 2008. Secretaría de Salud Jalisco. Dirección General de Salud Pública. Disponible en: http://www.jalisco.gob.mx/wps/wcm/connect/ f347130040963031b746b79c8da0b43f/Cancer2008.pdf?MOD⫽AJPERES& CACHEID⫽f347130040963031b746b79c8da0b43f. [Último acceso 13 de diciembre de 2010]. [6] Lis CG, Rodeghier M, Grutsch JF, Gupta D. Distribution and determinants of patient satisfaction in oncology with a focus on health related quality of life. BMC Health Services Research 2009;9:190. [7] Kleeberg UR, Feyer P, Günther W, Behrens M. Patient satisfaction in outpatient cancer care: a prospective survey using The PASQOC questionnaire. Support Care Cancer 2008;16:947–54. S99 [8] Gotay CC, Kawamoto CT, Bottomley A, Efficace F. The prognostic significance of patient-reported outcomes in cancer clinical trials. J Clin Oncol 2008;26:1355– 63. [9] Feyer P, Kleeberg UR, Steingräber M, et al. Frequency of side effects in outpatient cancer care and their influence on patient satisfaction--a prospective survey using the PASQOC questionnaire. Support Care Cancer 2008;16:567–75. [10] Goldzweig G, Meirowitz A, Hubert A, et al. Meeting expectations of patients with cancer: relationship between patient satisfaction, depression, and coping. J Clin Oncol 2010;28:1560 –5. [11] Landercasper J, Linebarger JH, Ellis RL, et al. A quality review of the timeliness of breast cancer diagnosis and treatment in an integrated breast center. J Am Coll Surg 2010;210:449 –55. [12] Lipscomb J, Gotay CC, Snyder CF. Patient-reported outcomes in cancer: a review of recent research and policy initiatives. CA Cancer J Clin 2007;57:278 –300. [13] Forbat L, Hubbard G, Kearney N. Patient and public involvement: models and muddles. J Clin Nurs 200918:2547-54. [14] Hausdorf K, Rogers C, Whiteman D, et al. Rating access to health care: Are there differences according to geographical region? Aust N Z J Public Health 2008;32:246 –9. [15] EORTC IN-PATSAT32, versión validada para población mexicana. http://www.eortc.be/home/qol/downloads/f/PATSAT/IN-PATSAT32% 20Spanish%20(Mexico).pdf. [Último acceso 13 de diciembre de 2010]. [16] http://www.eortc.be/home/qol/downloads/f/SCManualIN-PATSAT32 .pdf. [Ultimo acceso 13 de diciembre de 2010]. [17] Brédart A, Bottomley A, Blazeby JM, et al. European Organisation for Research and Treatment of Cancer Quality of Life Group and Quality of Life Unit. An international prospective study of the EORTC cancer in-patient satisfaction with care measure (EORTC IN-PATSAT32). Eur J Cancer 2005;41:2120 –31. [18] Brédart A, Coens C, Aaronson N, et al. EORTC Quality of Life Group and EORTC Quality of Life Unit. Determinants of patient satisfaction in oncology settings from European and Asian countries: preliminary results based on the EORTC IN-PATSAT32 questionnaire. Eur J Cancer 2007;43:323–30. [19] Arraras JI, Vera R, Martínez M, et al. The EORTC cancer in-patient satisfaction with care questionnaire: EORTC IN-PATSAT32 Validation study for Spanish patients. Clin Transl Oncol 2009;11:237– 42. [20] Jayasekara H, Rajapaksa L, Bredart A. Psychoetric evaluation of the European Organization for Research and Treatment of Cancer inpatient satisfaction with care questionnaire (‘Sinhala’ version) for use in a South-Asian setting. Int J Qual Health Care 2008;20:221– 6. [21] Avery KN, Metcalfe C, Nicklin J, et al. Satisfaction with care: an independent outcome measure in surgical oncology. Ann Surg Oncol 2006;13:817–22. VALUE IN HEALTH 14 (2011) S100 –S107 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Cost-Effectiveness of Supervised Exercise Therapy in Heart Failure Patients Eduardo M. Kühr, MD, MSc1, Rodrigo A. Ribeiro, MD, MSc2, Luis Eduardo P. Rohde, MD, ScD1, Carisi A. Polanczyk, MD, ScD1,2,* 1 Cardiology Division, Hospital de Clínicas de Porto Alegre, Department of Medicine and Graduate Program of Cardiology, Federal University of Rio Grande do Sul, Porto Alegre, Brazil, and National Institute for Health Technology Assessment, CNPq, Brazil; 2Graduate Program in Epidemiology, Federal University of Rio Grande do Sul, Porto Alegre, Brazil, and National Institute for Health Technology Assessment, CNPq, Brazil A B S T R A C T Objective: Exercise therapy in heart failure (HF) patients is considered safe and has demonstrated modest reduction in hospitalization rates and death in recent trials. Previous cost-effectiveness analysis described favorable results considering long-term supervised exercise intervention and significant effectiveness of exercise therapy; however, these evidences are now no longer supported. To evaluate the costeffectiveness of supervised exercise therapy in HF patients under the perspective of the Brazilian Public Healthcare System. Methods: We developed a Markov model to evaluate the incremental cost-effectiveness ratio of supervised exercise therapy compared to standard treatment in patients with New York Heart Association HF class II and III. Effectiveness was evaluated in quality-adjusted life years in a 10-year time horizon. We searched PUBMED for published clinical trials to estimate effectiveness, mortality, hospitalization, and utilities data. Treatment costs were obtained from published cohort updated to 2008 values. Exercise therapy intervention costs were obtained from a reha- Introduction Heart failure (HF) is a common health care problem worldwide, with elevated costs associated to its treatment [1]. During the past 20 years several effective therapies have changed HF management and clinical outcomes and these have been formally evaluated through economic analyses [2-4]. The decrease in HF mortality was followed by an increase in its prevalence, with direct effect on health care budgets resulting from the rising number of hospitalizations and therapeutic procedures [5]. HF is a complex syndrome characterized by reduced exercise tolerance and the involvement of multiple physiopathologic mechanisms [6]. In the past patients were often advised to limit their efforts in daily activities; however, several studies suggest that exercise training may reduce mortality and morbidity in HF patients [7,8]. These studies also demonstrated that exercise training could be performed safely in appropriately evaluated cases of patients who present in clinically compensated New York Heart bilitation center. Model robustness was assessed through Monte Carlo simulation and sensitivity analysis. Cost were expressed as international dollars, applying the purchasing-power-parity conversion rate. Results: Exercise therapy showed small reduction in hospitalization and mortality at a low cost, an incremental cost-effectiveness ratio of Int$26,462/quality-adjusted life year. Results were more sensitive to exercise therapy costs, standard treatment total costs, exercise therapy effectiveness, and medications costs. Considering a willingness-to-pay of Int$27,500, 55% of the trials fell below this value in the Monte Carlo simulation. Conclusions: In a Brazilian scenario, exercise therapy shows reasonable cost-effectiveness ratio, despite current evidence of limited benefit of this intervention. Keywords: costs, health economics, heart failure, physical therapy. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. Association (NYHA) functional class II and III, as endorsed by current guidelines [9,10]. For health care managers, the decision to incorporate exercise therapy in treatment of patients with HF should be based in several perspectives, including cost-effectiveness studies of the intervention. In 2001, Georgiou et al. [11] published a cost-effectiveness analysis of supervised exercise intervention in HF patients showing a very favorable cost-effectiveness ratio of $1773 per life-year saved, considering a 14-month period of supervised exercise intervention in a time horizon of 10 years applied to a North-American setting. Recently a multicenter randomized controlled trial of 2331 HF outpatients [12] described an exercise-based intervention being compared with standard treatment. After 2.5 years of follow-up, including a short training period in a facility followed by homebased exercise sessions, a benefit was observed only after adjustment for other prognostic predictors of the primary endpoint. The authors concluded that exercise training is a safe intervention associated with a modest reduction in hospitalization and mortality, Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: C. A. Polanczyk, National Institute for Health Technology Assessment, Hospital de Clínicas de Porto Alegre, Ramiro Barcelos 2350, Building 21, 5th Floor, 90035-007 - Porto Alegre, RS, Brazil. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.006 VALUE IN HEALTH 14 (2011) S100 –S107 S101 Fig. 1 – Schematic representation of the decision model. far from the assumed estimations in previous cost-effectiveness analysis [12]. In this study we evaluated the economic impact of a supervised exercise intervention in a hypothetic stable outpatient HF cohort, considering current evidence of effectiveness and costs, offering health care professionals an updated assessment on the role of exercise in the management of HF. computed all-cause mortality in our model, considering evidence of exercise intervention studies. A schematic representation of our decision tree is shown in Figure 1. The discount rate for both cost and effectiveness was 5% per year. We used the public third-party payer perspective and a 10year time horizon. Survival data Methods Target population The target population was composed of 60-year old patients at baseline, with clinically stable NYHA class II or III HF, intended to reproduce the population in exercise interventions studies in HF. Decision model structure We developed a model based on two competing strategies: 1) standard HF care; and 2) standard HF care plus an exercisebased intervention [13]. We constructed our decision tree model with Markov transitional states using TreeAge Pro 2009 Suite software (release 1.0.1, TreeAge Software Inc., Williamstown, MA), tracking a hypothetical cohort of HF patients over time receiving one of the strategies. During each 1-year cycle, patients could remain alive or die; patients alive could also remain stable or be hospitalized. After having been hospitalized, these patients could die or remain alive, with a lower survival rate, simulating the natural history of HF. In the intervention arm we assumed that exercise could reduce mortality and hospitalization rates, according to expected rates of effectiveness. We Survival rates were based in data from a specialized HF outpatient clinic from a university hospital in Brazil whose patients’ characteristics are similar to the exercise intervention studies’ populations. This cohort was composed of 318 patients (68% men), with a median age of 61 years (interquartile range 50 –71 years). Thirty-seven percent of patients had ischemic heart disease as the HF etiology; 87% were currently taking beta blockers, and 91% were taking angiotensin converting enzyme inhibitors. The prevalence of diabetes mellitus in this cohort was 30%, 41% of these patients had hypertension, and 11% were tobacco users. The annual rate of hospital admission in this cohort was 16%, and patients who had been hospitalized had a diminished survival rate compared with those who had not been hospitalized [14]. Median follow-up of this cohort was 75 months (95% confidence interval [CI] 68 – 81). To project survival during the 10-year time horizon, we built a survival curve (Fig. 2) using a Weilbull function. Two different curves were built based on hospitalization status. The final equations for the survival functions were Exp(⫺((0.0004*(_stage)^1.0715)) in nonhospitalized patients and Exp(⫺((0.00018*(_stage)^1.3627)) in hospitalized patients. S102 VALUE IN HEALTH 14 (2011) S100 –S107 were included, as described in the Consort statement (Fig. 3). Thirteen trials met the eligibility criteria, resulting in a total of 3458 patients being included: 1746 in the exercise group and 1712 in the control group, as shown in Table 1 [7,12,16 –26]. When one arm of a study contained no events, we added 0.5 to each cell to allow effect size calculation. Combined events (death or hospitalization) showed a pooled risk ratio of 0.878 (95% CI 0.805– 0.957) using the Mantel Haenszel fixed-effects model. Pooled risk reduction for mortality was 0.957 (95% CI 0.865–1.058) and for hospitalization was 0.90 (95% CI 0.831– 0.973). HF treatment costs Fig. 2 – Undiscounted survival curve projections for standard treatment and exercise therapy plus standard treatment, compared with actual survival curve of heart failure (HF) patients. Utilities To estimate quality-adjusted life years (QALYs), we assumed a utility index of 0.80 for HF (95% CI 0.78 – 0.82), using recommended values proposed by Göhler et al. [15] in their decision-analytic model, which was weighted by NYHA classes from actual prevalence on the Hospital de Clinicas de Porto Alegre cohort (66% and 33% of NYHA class II and III, respectively). We used a discount rate of 5% (range 3%–7%). Exercise therapy intervention The model assumed a supervised exercise therapy session occurring in a facility center, with direct supervision of a qualified exercise training professional (eg, a physiotherapist). These sessions consist of an initial warm-up, followed by 15 to 30 minutes of aerobic training on a treadmill or performing stationary cycling in individually prescribed program. All patients should be monitored during the exercise session and each physiotherapist could supervise up to four patients each session. In initial phases of rehabilitation (until 12 weeks), every patient in the intervention group should participate in at least three weekly 1-hour supervised group-based exercise sessions. After 36 sessions we determined that patients should return once a week to the rehabilitation center for additional supervised sessions during a 9-month period, totaling 72 sessions per year in the first intervention stage. In subsequent stages we considered a maintenance program of 50 yearly sessions (weekly supervised sessions). Annual HF treatment costs in the Public Healthcare System (PHS) were extracted from a Brazilian cohort described by Araujo et al. [27]. These authors described treatment costs categorized in outpatients and hospital costs, in 2002 values. We updated published values to 2008 costs according to the official Brazilian inflation index [28]. Hospital costs were adapted from our university hospital billing system, which includes in standard admission payments some high-complexity diagnostic procedures, including special drugs and devices. For outpatient cost estimates for stable conditions we included drug costs, complementary exams, and ambulatory costs (Table 2). All costs are expressed in international dollars (Int$), using the purchasing power parity conversion rate. According to a 2008 report of the World Bank regarding conversion rates [29], Int$1 ⫽ R$1.357. We used an annual discount rate of 5% in all described costs, expressed in internation dollars. Intervention costs (Supervised Exercise) In Brazil outpatient cardiac rehabilitation procedures are not reimbursed by the PHS. To establish yearly intervention costs we used data from a rehabilitation center in a private facility of an insurance plan, located in the Brazilian state of Santa Catarina. Salaries of fitness center personnel accounted for 47% of the total annual cost of rehabilitation. Maintenance costs (including rental costs, licenses, and repairs) and equipment costs (e.g., cycle ergometers and pulse oximeters) accounted for 33% and 20%, respectively. This structure is able to provide 2000 annual Intervention effectiveness We performed a literature search in PUBMED using “Heart Failure” and “Exercise Therapy” MeSH terms, selecting randomized controlled trials published from 1980 to 2009. We included trials with the following characteristics: 1) randomized parallel group controlled trials; 2) exercise programs during at least 8 weeks or more initially based in facility centers; and 3) clinically relevant endpoints (death and/or hospitalization) described in the results. Initial search identified 169 articles that were subsequently scrutinized and those that fulfilled inclusion criteria Fig. 3 – Flow diagram of study selection for exercise effectiveness. S103 VALUE IN HEALTH 14 (2011) S100 –S107 Table 1 – Studies meeting inclusion criteria to estimate exercise effectiveness. Author, year (ref) Austin, 2005 [16] Belardinelli, 1999 [7] Gianuzzi, 2003 [17] Jolly, 2009 [18] Jonsdottir, 2006 [19] Kiilavuori, 2000 [20] Kulcu, 2007 [21] Hambrecht, 2000 [22] McKelvie, 2002 [23] Nilsson, 2008 [24] O’ Connor, 2009 [13] Wielenga, 2000 [25] Willenheimer, 1998 [26] Total Sample size (exercise controls) Events (exercise, controls) Combined Deaths Hospitalization 13 23 14 34 2 2 23 25 9 13 0 1 0 2 5 4 11 10 2 1 918 958 2 4 0 3 999 1082 5 4 9 20 0 1 7 5 2 2 0 0 0 1 3 2 11 10 2 1 189 198 1 4 0 0 229 248 9 19 5 14 2 1 16 20 7 11 0 1 0 2 2 2 0 0 0 0 729 760 1 1 0 3 771 834 Exercise: 86 Controls: 98 Exercise: 50 Controls: 49 Exercise: 45 Controls: 45 Exercise: 84 Controls: 85 Exercise: 21 Controls: 22 Exercise: 12 Controls: 15 Exercise: 27 Controls: 26 Exercise: 35 Controls: 34 Exercise: 90 Controls: 91 Exercise: 40 Controls: 40 Exercise: 1159 Controls: 1172 Exercise: 41 Controls: 39 Exercise: 22 Controls: 30 Exercise: 1712 Controls: 1746 training hours to four patient groups, with a total cost of Int$66,633, resulting in a single session cost of Int$8.33 per patient. The proposed cost of each session is similar to the estimated costs per session used by a previously published model [11], which would cost Int$7.14 in the Brazilian scenario. Values used in the model are lower than reimbursement values in Brazil (Int$17.65 in the first 12-week program and Int$5.88 in the maintenance program). Table 2 – Model variables, values, and sources. Variable Median age of cohort (y) Annual risk of hospitalization (%) Exercise variables Annual hospitalization risk reduction (%) Annual mortality risk reduction (%) Costs Annual exercise intervention costs in the first year (Int$) Base 60 16 0.90 0.957 Sensitivity analysis variation 50–70 10–22 0.831–0.973 0.865–1.058 375 187–562 Annual exercise intervention cost in 2- to 10-y interval (Int$) 100 50–150 Total annual conventional heart failure treatment costs (Int$) Annual hospitalization costs (Int$) Annual ambulatory costs (Int$) Annual complementary exams costs (Int$) Annual medication costs (Int$) Total annual exercise-group heart failure treatment costs (conventional heart failure treatment cost plus exercise intervention costs) (Int$) Utilities Heart failure patient utility (New York Heart Association Class II-III) Discount rate (%) 3752 1798 54 363 1538 4187 1,876–5,628 899–2,697 27–81 182–545 769–2,306 2,093–6,280 0.80 5 0.78–0.82 3–7 Source 15 15 Meta-analysis Meta-analysis Health care insurance plan, primary data Health care insurance plan, Primary data 17 17 17 17 17 Estimated 16 Estimated S104 VALUE IN HEALTH 14 (2011) S100 –S107 Table 3 – Model predicted cost, effectiveness, and cost-effectiveness of competing strategies. Total cost (Int$) Conventional treatment Conventional treatment ⫹ Exercise therapy 12,720 15,331 Effectiveness Incremental cost-effectiveness Mean life years Mean QALYs Int$/LYS Int$/QALY 5.45* 5.58* 4.36* 4.46* – 21,169 – 26,461 Int$, international dollars; LYS, life year saved; QALY, quality-adjusted life year. * All values are discounted. Sensitivity analysis We performed one-way sensitivity analysis in all model parameters described in Table 2. Model effectiveness variables (mortality and hospitalization reduction with exercise) were varied between the boundaries of meta-analysis confidence intervals. Costs were varied ⫾50% of their original values; utilities varied according to previously described values. Discount varied between 3% and 7%. Two-way sensitivity analysis was performed in the most important model variables. Model robustness was tested in a Monte Carlo simulation, with a generation of 1000 trials and variation in the range described above. Gamma distribution was chosen for cost variables, lognormal distribution for effectiveness, and beta distribution for utility variables. Results The model predicted a mean survival of 5.58 years in the exercise group and 5.45 years in the control group. When adjusted for quality of life, we found a mean survival of 4.46 and 4.36 QALYs, respectively, as shown in Table 3. The exercise intervention increased 0.13 life-years and 0.10 QALYs. The total costs in the intervention group was Int$15,331 and Int$12,720 in the standard care group. The incremental cost-effectiveness ratio (ICER) was Int$21,169 per life-year and Int$26,462 per QALY. Monte Carlo simulation with 1000 trials is represented in Figure 4. Results from one-way sensitivity analyses are summarized in Table 4, and our evaluation was adequate to most one-way analyses. At the lowest values of established intervals for variables related to HF standard costs of treatment—such as ambulatory costs, complementary exam costs, hospitalization costs, and drugs costs—there was small influence on the ICERs. Utili- ties and discount rates variations also produce discrete modifications on the ICERs, varying between Int$25,816 and Int$27,140 and Int$25,155 and Int$27,832, respectively. More pronounced effect occurred when varying probability of hospitalization, with an ICER of Int$24,499 at a 22% annual rate and of Int$30,350 at a 10% annual hospitalization rate. Three variables had a more expressive effect on the base case estimates: exercise intervention costs, estimated mortality reduction, and estimated hospitalization reduction with rehabilitation program. When the relative risk (RR) for hospitalization with exercise was 0.83, the ICER would be Int$20,856; with a RR of 0.97, the ICER increased to Int$36,245. Regarding exercise costs, a cost of Int$217 per year of supervised exercise produces an ICER of Int$13,501, and a cost of Int$899 per year increased the ICER to Int$52,056 (Table 4). The variable with greater impact on the ICER in the sensitivity analysis was mortality reduction with exercise: a RR of 0.86 increased 0.24 in total QALY, resulting in an ICER of Int$12,738/ QALY; a RR of 0.96 yielded 0.09 additional QALY and an ICER of Int$28,141/QALY. Finally, considering no effect on mortality, exercise intervention results in additional 0.04 QALY and an ICER of Int$66,576/QALY. Two-way sensitivity analysis altering hospitalization reduction effect and exercise cost is illustrated in Figure 5. Assuming a mortality RR of 0.96 with exercise and a willingness-to-pay of Int$27,495, the exercise intervention should cost below Int$440 yearly to become cost-effective, assuming a RR of 0.90 in hospitalization with exercise. In the Monte Carlo simulation, we established a threshold of three times the Brazilian gross domestic product, which represents Int$27,495 (R$37,311) in 2008 and evaluated how many simulations fell below this value. As shown in Figure 6, 55% of trials were below this value. In 34% of trials exercise was more costly Table 4 – One-way sensitivity analysis. Variables Mortality reduction with exercise Hospitalization reduction with exercise Exercise intervention costs Annual rate of hospitalization Utility of heart failure Discount rate Ambulatory costs Hospitalization costs Complementary exam costs Medication costs Fig. 4 – Monte Carlo 1000 trials scatter plot. The number of points below wiliness-to-pay threshold line are 55.2% (Int$27,500/quality-adjusted life year—three times Brazil’s gross domestic product per capita). Lower ICER (Int$/QALY) Higher ICER (Int$/QALY) 12,738 20,856 Dominated 36,245 13,501 24,499 25,816 25,155 26,430 25,945 26,251 25,571 52,056 30,350 27,140 27,832 26,493 26,977 26,672 27,353 Note: Range values are those presented in Table 2. ICER, international cost-effectiveness ratio; Int$, international dollars; QALY, quality-adjusted life year. VALUE IN HEALTH 14 (2011) S100 –S107 Fig. 5 – Three-way sensitivity analysis considering exercise costs (Int$) and hospitalization risk reduction, assuming mortality risk reduction with exercise = 0.96 and a willingness-to-pay of Int$ 27,500. and above the wiliness-to-pay and in 10.8% the exercise intervention was inferior (dominated). Discussion The results of this model show that exercise therapy in HF patients has a modest but favorable incremental cost-effectiveness ratio of Int$26,462 per QALY and Int$21,169 per life-year in a Brazilian PHS scenario. The results were consistent considering sensitivity analyses performed and assumptions described. Our results show that this intervention has a reasonable costeffectiveness ratio when compared to other incoming therapies, such as implantable cardioverter defibrillator devices [30], but closer to proposed willingness-to-pay for Brazil, as demonstrated in the acceptability curve. The ICER of this intervention is also higher than the estimated Brazilian hemodialysis cost per life-year gained (US$10,065) [31]. During the past three decades the exercise training approach concerning HF patients moved from absolute restraint to enthusiastically prescribed, based on a growing body of evidence suggesting physiologic and clinical benefit from exercise. Randomized controlled trials showed significant reductions in the composite endpoint of hospitalization and mortality in HF patients submitted to supervised sessions of exercise interventions, reaching almost 50% in one report [7], and averaging a 35% reduction in mortality and 28% in hospitalization rates in a meta-analysis published in 2004 [8]. Nonetheless, a recently published trial failed to prove superiority of an exercise therapy intervention in a multicenter-based strategy [12]. The HF-Action trial [12] was designed to test the hypothesis that exercise training prescribed to HF patients would reduce mortality, and although overall intention-to-treat results were of small magnitude, protocol-driven results and those adjusted for prognostic variables indeed suggest some benefit on mortality beyond reduction of hospitalization and improvement on quality of life, functional capacity, and other markers of well-being [32,33]. The true effect of exercise on mortality in this population is a matter of debate, although exercise is still considered a key aspect in the management of HF [9,34]. Our model was sensitive to this parameter; assuming even a small effect of 4% reduction in mortality, modest reduction of combined risk of death or HF-related hospitalization, exercise would still have a favorable cost-effectiveness ratio. Unfortunately we do not have large Brazilian studies describing the actual result of such interven- S105 tions. There are several short-term randomized studies evaluating the effect of exercise among Brazilian HF patients, mostly limited to physiologic parameters, functional class, or quality of life [35,36]. We could not identify studies that have evaluated hard clinical endpoints such as death or hospitalization. Indirect evidence from these studies; however, support the concept that the results obtained locally are similar to the ones described by other international research groups [35,36]. Based on these assumptions we believe our findings could be a reasonable estimate of the cost-effectiveness ratio for the Brazilian PHS scenario, although the effect of the health care system itself and cultural and socioeconomic characteristics on the main result might be unpredictable. In another published cost-effectiveness analysis [11], the impressive absolute reduction of 19% fewer hospitalizations in the exercise group compared to standard treatment group and additional survival of 1.82 years was the determinant factor in effectiveness estimation, with a cost-effectiveness ratio of $1773 per life-year saved. In our model, we considered a less pronounced increase in survival (0.13 years) produced by exercise intervention, reflecting a more realistic ICER taking into account current data. Prospective economic evaluation from HF-Action [33] demonstrated that along the trial the exercise training program was of little systematic benefit in terms of overall resource use, but individual data indicate that most estimates were consistent with a decrease in costs (89.9%) and an increase in QALYs (76.5%), and that most of the bootstrap replications were either associated with cost-saving result or with ICER below $50,000 per QALY. Usual exercise intervention in HF requires at least 12 weeks of training, because during a period of this length patients can receive adequate care and perceive functional improvement, increasing adherence to the intervention. In our model, we determined that the 36 sessions should be supplemented with additional weekly supervised sessions to reinforce compliance to the intervention in an attempt to reach the effectiveness demonstrated in clinical trials. Perhaps these additional costs could be counterbalanced with increased effectiveness, but we opted to consider a conservative strategy in light of current evidence. Assuming that including different professionals and activities in every program is proportional to increasing costs of the intervention, reducing the number of sessions in a facility center is a valuable attempt to reduce treatment costs and increase cost-effectiveness ratio. In the real world exercise therapy is not considered the core of the standard care for HF patients. Commonly HF patients are Fig. 6 – Acceptability curve with a willingness-to-pay of Int$27,500/quality-adjusted life year. S106 VALUE IN HEALTH 14 (2011) S100 –S107 aged and have other diseases (e.g., osteoarticular limitations) that restrict their mobility; plus, successful exercise therapy demands motivation and both family and patient engagement time. In today’s health care system model, which is based in hospital procedures, there are few examples of rehabilitation centers to promote significant changes in patients’ habits and minimize usual risk factors associated with cardiovascular diseases, such as lack of exercise, tobacco use, inappropriate food ingestion, and depression. Despite all these points, there is evidence—including that produced in our study—that committed patients with cardiac disease who have access to facilities with trained professionals and appropriate equipment could benefit from exercise therapy with reasonable cost-effectiveness ratio [37]. Common to other interventions that also rely on human behavior, long-term adherence to exercise in patients with HF remains a challenge and requires additional research to determine strategies aimed at improving compliance, which might be associated with increased effectiveness. Areas of needed research include identifying the subgroups of patients who benefit the most from this intervention, as well as the determining the optimal intensity, duration, and frequency of exercise needed to maximize clinical benefits. Some limitations in our modeling should be mentioned. First, we established a constant effectiveness. In the real world a considerable number of patients oscillate their attendance in rehabilitation programs. Another limitation is that true cost of HF treatment could be higher than assumed costs, although we used primary data from a cohort of HF patients, with values updated to 2008. As we assumed a third-party payer (i.e., government) perspective, we did not considered patient or family displacement cost in regard to the rehabilitation facility; indirect cost associated with productivity losses were also not included, although HF in Brazil is a retirement cause insured by the PHS. The main caveat regarding the results of our study is the broad variance in exercise effectiveness, even though in two-way sensitivity analysis we found that every scenario with effectiveness greater than 5% (considering mortality reduction) the exercise intervention brought benefits to HF patients. Exercise therapy seems to be safe and should be considered in stable HF patients as part of treatment regardless of individual beliefs. Although effectiveness of exercise therapy in HF patients seems to be lower than initially expected, its cost-effectiveness ratio remains acceptable for health policy decision makers to incorporate this intervention in the care of patients with HF. Source of financial support: Dr. Rohde and Dr. Polanczyk received research sponsorship from Conselho Nacional de Desenvolvimento Científico e Tecnológico, Brazil. This study was supported by the Brazilian Institute for Health Technology Assessment. REFERENCES [1] Lee WC, Chavez YE, Baker T, Luce BR. Economic burden of heart failure: a summary of recent literature. Heart Lung 2004;33:362–71. [2] Gregory D, Udelson JE, Konstam MA. Economic impact of beta blockade in heart failure. Am J Med 2001;110(Suppl. 7A):74S– 80S. [3] Andersson F, Cline C, Ryden-Bergsten T, Erhardt L. Angiotensin converting enzyme (ACE) inhibitors and heart failure. The consequences of underprescribing. Pharmacoeconomics 1999;15:535–50. [4] McAlister FA, Ezekowitz J, Dryden DM, et al. Cardiac resynchronization therapy and implantable cardiac defibrillators in left ventricular systolic dysfunction. Evid Rep Technol Assess (Full Rep) 2007):1–199. [5] Mosterd A, Hoes AW. Clinical epidemiology of heart failure. Heart 2007;93:1137– 46. [6] Executive summary: HFSA 2006 Comprehensive Heart Failure Practice Guideline. J Card Fail 2006;12:10 –38. [7] Belardinelli R, Georgiou D, Cianci G, Purcaro A. Randomized, controlled trial of long-term moderate exercise training in chronic heart failure: effects on functional capacity, quality of life, and clinical outcome. Circulation 1999;99:1173– 82. [8] Piepoli MF, Davos C, Francis DP, Coats AJ. Exercise training metaanalysis of trials in patients with chronic heart failure (ExTraMATCH). BMJ 2004;328:189. [9] Dickstein K, Cohen–Solal A, Filippatos G, et al. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure 2008: The Task Force for the Diagnosis and Treatment of Acute and Chronic Heart Failure 2008 of the European Society of Cardiology. Developed in collaboration with the Heart Failure Association of the ESC (HFA) and endorsed by the European Society of Intensive Care Medicine (ESICM). Eur J Heart Fail 2008;10:933– 89. [10] Wenger NK. Current status of cardiac rehabilitation. J Am Coll Cardiol 2008;51:1619 –31. [11] Georgiou D, Chen Y, Appadoo S, et al. Cost– effectiveness analysis of long–term moderate exercise training in chronic heart failure. Am J Cardiol 2001;87:984 – 8, A4. [12] O’Connor CM, Whellan DJ, Lee KL, et al. Efficacy and safety of exercise training in patients with chronic heart failure: HF-ACTION randomized controlled trial. JAMA 2009;301:1439 –50. [13] Sonnenberg FA, Beck JR. Markov models in medical decision making: a practical guide. Med Decis Making 1993;13:322–38. [14] Beck-da Silva L, Goldraich L, Bonzanini L, et al. Pulse pressure and QRS width evaluation as an inexpensive tool for heart failure assessment. Congest Heart Fail 2009;15:222–7. [15] Gohler A, Geisler BP, Manne JM, et al. Utility estimates for decisionanalytic modeling in chronic heart failure-health states based on New York Heart Association classes and number of rehospitalizations. Value Health 2009;12:185–7. [16] Austin J, Williams R, Ross L, et al. Randomised controlled trial of cardiac rehabilitation in elderly patients with heart failure. Eur J Heart Fail 2005;7:411–7. [17] Gianuzzi P, Temporelli PL, Corra U, Tavazzi L. Antiremodeling effect of long-term exercise training in patients with stable chronic heart failure: results of the Exercise in Left Ventricular Dysfunction and Chronic Heart Failure (ELVD-CHF) Trial. Circulation 2003;108: 554 –9. [18] Jolly K, Taylor RS, Lip GY, et al. A randomized trial of the addition of home-based exercise to specialist heart failure nurse care: the Birmingham Rehabilitation Uptake Maximisation study for patients with Congestive Heart Failure (BRUM–CHF) study. Eur J Heart Fail 2009;11: 205–13. [19] Jónsdóttir S, Andersen KK, Sigurosson AF, Sigurosson SB. The effect of physical training in chronic heart failure. Eur Heart J 2006;8:97– 101. [20] Kiilavuori K, Naveri H, Salmi T, Harkonen M. The effect of physical training on skeletal muscle in patients with chronic heart failure. Eur J Heart Fail 2000;2:53– 63. [21] Kulcu DG, Kurtais Y, Tur BS, Gulec S, Seckin B. The effect of cardiac rehabilitation on quality of life, anxiety and depression in patients with congestive heart failure. A randomized controlled trial, shortterm results. Eura Medicophys 2007;43:489 –97. [22] Hambrecht R, Gielen S, Linke A, et al. Effects of exercise training on left ventricular function and peripheral resistance in patients with chronic heart failure: a randomized trial. JAMA 2000;283: 3095–101. [23] McKelvie RS, Teo KK, Roberts R, et al. Effects of exercise training in patients with heart failure: the Exercise Rehabilitation Trial (EXERT). Am Heart J 2002;144:23–30. [24] Nilsson BB, Westheim A, Risberg MA. Long-Term effects of a groupbased high-intensity aerobic interval-training program in patients with chronic heart failure. Am J Cardiol 2008;102:1220 –24. [25] Wielenga RP, Huisveld IA, Bol E, et al. Safety and effects of physical training in chronic heart failure. Results of the Chronic Heart Failure and Graded Exercise study (CHANGE). Eur Heart J 1999;20: 872–9. [26] Willenheimer R, Erhardt L, Cline C, Rydberg E, Israelsson B. Exercise training in heart failure improves quality of life and exercise capacity. Eur Heart J 1998;19:774 – 81. [27] Araujo DV, Tavares LR, Verissimo R, et al. Cost of heart failure in the Unified Health System. Arq Bras Cardiol 2005;84:422–7. [28] Instituto Brasileiro de Geografia e Estatística. Available at: http://www .ibge.gov.br/home/estatistica/indicadores/precos/inpc_ipca/ipca-inpc_ 200908_3.shtm. [Accessed September 9, 2009]. [29] Word development indicators. Available at: http://data.un.org/Data .aspx?q⫽purchase⫹power⫹parity&d⫽CDB&f⫽srID%3a29947. [Accessed April 7, 2009]. VALUE IN HEALTH 14 (2011) S100 –S107 [30] Ribeiro RA, Stella SF, Camey SA, et al. Cost-effectiveness of implantable cardioverter-defibrillators in Brazil: primary prevention analysis in the public sector. Value Health 2010;13:160 – 8. [31] Sesso R, da Silva CB, Kowalski SC, et al. Dialysis care, cardiovascular disease, and costs in end-stage renal disease in Brazil. Int J Technol Assess Health Care 2007;23:126 –30, 132. [32] Flynn KE, Piña IL, Whellan DJ, et al. Effects of exercise training on health status in patients with chronic heart failure: HF-ACTION randomized controlled trial. JAMA 2009;301:1451–9, 1533. [33] Reed SD, Whellan DJ, Li Y, et al; HF-ACTION Investigators. Economic evaluation of the HF-ACTION (Heart Failure: A Controlled Trial Investigating Outcomes of Exercise Training) randomized controlled trial: an exercise training study of patients with chronic heart failure. Circ Cardiovasc Qual Outcomes 2010;3:374 – 81. S107 [34] Keteyian SJ, Piña IL, Hibner BA, Fleg JL. Clinical role of exercise training in the management of patients with chronic heart failure. J Cardiopulm Rehabil Prev 2010;30:67–76. [35] Winkelmann ER, Chiappa GR, Lima CO, et al. Addition of inspiratory muscle training to aerobic training improves cardiorespiratory responses to exercise in patients with heart failure and inspiratory muscle weakness. Am Heart J 2009;158:e1–7. [36] Lima MM, Rocha MO, Nunes MC, et al. A randomized trial of the effects of exercise training in Chagas cardiomyopathy. Eur J Heart Fail 2010;12:866 –73. [37] Papadakis S, Reid RD, Coyle D, et al. Cost-effectiveness of cardiac rehabilitation program delivery models in patients at varying cardiac risk, reason for referral, and sex. Eur J Cardiovasc Prev Rehabil 2008;15: 347–53. VALUE IN HEALTH 14 (2011) S108 –S114 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Estimating the SF-6D Value Set for a Population-Based Sample of Brazilians Luciane N. Cruz, ScD1,*, Suzi A. Camey, ScD2, Juliana F. Hoffmann, MSc1, Donna Rowen, PhD3, John E. Brazier, PhD3, Marcelo P. Fleck, ScD4, Carisi A. Polanczyk, ScD1,5 1 Graduate Studies Program in Epidemiology, Health Technology Assessment Institute, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; 2Statistics Department, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; 3School of Health and Related Research, University of Sheffield, UK; 4Graduate Studies Program in Psychiatry, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; 5Cardiology Division, Hospital de Clinicas, Porto Alegre, Brazil A B S T R A C T Objectives: SF-6D is a preference-based measure of health developed to estimate utility values from the SF-36. The aim of this study was to estimate a weighting system for the SF-6D health states representing the preferences of a sample of the Southern Brazilian general population. Methods: A sample of 248 health states defined by the SF-6D was valued by a sample of the southern Brazilian population using the standard gamble. Mean and individual level multivariate regression models were fitted to the standard gamble valuation data to estimate preference weights for all SF-6D health states. The models were compared with those estimated in the UK study. Results: Five hundred twenty-eight participants were interviewed, but 58 (11%) were excluded for failing to value the worst state. Data from 469 subjects producing 2696 health states valuations were used in the regression analysis. In contrast to the best performing model for the UK data, the best performing model for the Brazilian data was a random effects Introduction Increasingly, decision makers, providers, patients and the public require that expenditures on health be justified according to expected outcomes. In this context, the decision-making process in health and health care policy has never been more important to reduce inefficiencies and eliminate ineffective medical procedures [1]. Cost-effectiveness analysis is increasingly used in the decision-making process for resource allocation of health care resources. An important tool in this analysis is the quality-adjusted life year (QALY), an index that combines quantity and quality of life [2]. International guidelines for studies of cost-effectiveness [3] and institutions of health technology assessment such as the National Institute of Health and Clinical Excellence [4] in the United Kingdom have recommended that the QALYs are the reference outcome for economic evaluation. Health-related quality of life measures suitable for calculating QALYs are those that incorporate preferences into their scoring system. Examples of widely used generic instruments are the EQ-5D [5], Health Utilities Index [6], and SF-6D [7]. The latter was developed to model using only the main effects variables, highlighting the importance of adopting a country-specific algorithm to derive SF-6D health states values. Inconsistent coefficients were merged to produce the final recommended model, which has all significant coefficients and a mean absolute difference between observed and predicted standard gamble values of 0.07. Conclusions: The results provide the first population-based value set for Brazil for SF-6D health states, making it possible to generate quality-adjusted life years for cost-utility studies using regional data. Besides, utility weights derived using the preferences of a sample from a southern Brazilian population can be derived from existing SF-36 data sets. Keywords: cross-cultural, preference-based measures, SF-6D, utility. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. obtain a preference-based index from the SF-36 for use in economic evaluation. The growing number of studies conducting health states valuation surveys around the world has contributed to cross-cultural comparisons. Studies evaluating the EQ-5D [e.g., 8 –12] and SF-6D [13–15] in diverse cultures have shown that health state preferences are different from the preferences derived in the country where the measure was originally created. The aim of this study was to estimate preference weights for SF-6D health states that represent the preferences of a sample of the Southern Brazilian general population. This article presents results of the valuation survey and the modeling of the valuation data to produce utility values for all SF-6D health states, comparing results with those from the original UK study [7]. Methods This study follows the same protocol as the original UK SF-6D valuation study [7]. First, the SF-36 was reduced in size to generate health states to be valued by respondents. Second, a valuation Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Luciane Nascimento Cruz, Rua Ramiro Barcelos, 2350/Building 21, Porto Alegre, RS, Brazil 90035-903. E-mail: [email protected] 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.012 S109 VALUE IN HEALTH 14 (2011) S108 –S114 survey was undertaken in the Southern Brazilian general population. Finally, the results of the survey were modeled to predict utility values for all possible health states generated by the SF-6D. Deriving the SF-6D Health State Classification With the aim of deriving the descriptive system of the Brazilian SF-6D, we selected the same SF-36 items used in the original study in the United Kingdom [7], using the version of the SF-36 translated and validated in Brazil by Ciconelli et al. [16]. The result was a six-dimension descriptive system, each with multiple levels: 1) physical functioning, six levels; 2) role limitations, four levels; 3) social functioning, five levels; 4) pain, six levels; 5) mental health, five levels; and 6) vitality, five levels. This descriptive system is identical to the UK classification system except the language. Selection of Health States A health state was defined by selecting one level of each dimension. Considering the number of dimensions and levels, the SF-6D generates 18,000 health states. We used the same 248 health states valued in the original UK study to enable comparisons. Severity was assessed by summing the levels of each dimension levels for each health state. States were ranked using this sum score and divided into quintiles to form five groups according to severity. Each individual rated five intermediate states (one state of each group of severity) plus the worst health state generated by SF-6D (the “pits” state) [7]. Sampling The sample consisted of individuals aged from 20 to 64 years selected from the general population of Porto Alegre, a capital city in southern Brazil. The sample size was calculated such that each of the 247 health states (excluding pits) was valued at least 10 times. Because each respondent would value five intermediate states the sample size estimation was 494 respondents. A two-stage cluster random selection design was used. Primary units were represented by census sectors of the municipality according to division made by the Brazilian Geography & Statistics Institute [17]. In a second stage, households were systematically selected in each census sector. The interviewers visited seven domiciles in each of the 108 selected sectors, inviting all residents who met the following inclusion criteria to take part in the study: aged 20 to 64 years and literate and with no physical or mental incapacity that would prevent reading and understanding required in the valuation tasks. Study Protocol A team of interviewers made up of graduate students from healthrelated courses received intensive training in the application of the interview protocol. Interviews were conducted in the respondents’ own home. The interview followed the protocol used in the UK SF-6D valuation study [7]: 1) respondents completed the SF-6D; 2) ranking exercise and performed using a group of eight cards containing the five intermediate states along with the best state defined by the SF-6D, the worst state, and immediate death; and 3) valuation of five intermediate health states plus the worst state using the standard gamble (SG) technique. As a visual aid for the SG task, a visual prop developed by a team from MacMaster University [18] was used. Following the valuation of the five intermediate states, respondents valued the worst SF-6D state (pits) using a modified version of the SG. The SG choice depended on if the respondent had ordered the worst state better or worse than immediate death in the ranking exercise: 1) if better than death, the respondent must choose between the certain prospect of the worst state and the uncertain prospect of full health or immediate death; and 2) if worse than death, the choice was between the certain prospect of death and the uncertain prospect of full health or the worst state. The SG value produced from this SG task is measured on the full health-dead one to zero scale used to produce QALYs. All values derived for the intermediate states are then adjusted onto the full health-dead one to zero scale using the pits value for each individual. Taking P to represent the value given to the worst state for individual i and SG to represent the values of an intermediate state j, the formula used to generate adjusted values SGADJ for all intermediate health state valuations is: SGADJij ⫽ SGij ⫹ (1 ⫺ SGij) * Pi[7] These values were used in the regression analysis. In the final part of the interview respondents answered a range of sociodemographic questions. All respondents signed the Informed Consent Form approved by the University’s Institutional Review Board. Variables The sociodemographic variables obtained were: sex, age, marital status (married and not married), level of education (during years of study), job status (employed, unemployed, informal job, housewife, student, retired, and other), and economic class. Economic class was evaluated using an index called the Brazil Criterion that classifies the population into seven classes according to the purchasing power and level of education of the head of the family [19]. The division and its equivalence in terms of monthly family income in dollars are, approximately: Class A1: $3800, Class A2: $2300, Class B1: $1400, Class B2: $800, Class C: $460, Class D: $212, and Class E: $103. Table 1 – Sociodemographic characteristics of the included and excluded subjects. Variables Age (mean ⫾ SD) Sex Man Woman Marital status Married Not married Economic class A1 A2 B1 B2 C D Years of study Up to 4 5 to 8 years 9 to 11 12 or more Job status Employed Informal job Unemployed Housewife Student Retired Other Included (n ⫽ 469) % Excluded (n ⫽ 58) % P 41 ⫾ 13 42 ⫾ 14 0.65 41 59 29 71 0.07 63 37 60 40 1 14 18 25 36 6 2 9 9 17 53 10 4 16 35 45 8 31 32 29 54 8 7 14 4 7 5 44 9 5 21 6 10 5 0.42 0.05 0.07 0.22 S110 VALUE IN HEALTH 14 (2011) S108 –S114 Descriptive statistics of health state values were generated and compared to UK values. Modeling Regression analysis was used to estimate preference weights for each level and dimension of the classification system to enable preference weights to be estimated for all states rather than simply those included in the valuation study. This analysis followed the same protocol as the UK study [7], using linear models to estimate the relationship between the SF-6D classification system and SG adjusted values obtained in the valuation survey. Models were estimated at both the individual and aggregate (mean values for the 248 health states) levels. Several models were constructed to predict the health states values, but the general individual level model can be defined by the formula: yij ⫽ g共ⱊXij ⫹ rij兲 ⫹ ij (1) where i ⫽ 1,2, . . . . m represents the valued health state and j ⫽ 1, 2, . . . n represents respondents. The dependent variable, yij, is the SG adjusted score for the health state i valued by respondent j (SGADJ), x is a vector of dummy independent variables (X␦) for each level of dimension ␦ do SF-6D. For example, x31means that the variable refers to the dimension 3 at level 1. For any health state, X␦ is defined as: X␦ ⫽ 1, if, for this state, dimension ␦ is at level ( ⱖ 2) X␦ ⫽ 0, if, for this state, dimension ␦ is at level Assuming a simple linear model, the intercept represents the state of perfect health (111111). The level ⫽ 1 is a baseline for each dimension. The coefficient of the dummy variables represents the main effect of a move from level 1 to the other levels in each dimension. The final value of each health state can be estimated by summing the coefficients of the levels of each dimension present in a given state [7]. Table 2 – Descriptive statistics for 40 health state valuations comparing Brazil and the United Kingdom [7]. State 111112 412152 211111 423343 213323 224223 131542 111215 221432 122233 523634 112221 214535 342353 312552 443215 345122 141653 134322 344145 241531 431623 321455 432623 112521 341123 535645 315515 532124 541432 323333 241635 323644 534644 423433 124314 334254 434654 432255 645655 N 11 8 8 10 9 8 10 9 9 9 10 9 9 12 6 11 10 9 12 10 11 8 9 10 13 8 11 10 9 12 14 10 8 11 9 5 8 10 9 469 Brazil United Kingdom Min Max Mean Median SD Min Max Mean Median SD 0,42 0,25 0,30 0,19 0,06 0,44 0,26 0,10 0,05 ⫺0,40 0,19 0,00 0,05 0,05 0,19 0,05 0,10 0,15 ⫺0,23 ⫺0,28 0,05 ⫺0,14 0,05 0,06 ⫺0,48 0,15 0,05 0,02 0,05 0,00 ⫺0,88 0,10 0,00 0,05 0,00 0,10 0,10 ⫺0,28 ⫺0,38 ⫺1,00 1,00 0,98 1,00 1,00 0,92 0,95 0,96 1,00 0,96 0,94 0,91 0,96 0,94 1,00 0,99 0,98 0,88 0,98 0,97 0,98 0,97 0,80 1,00 0,98 0,94 0,80 0,75 0,86 0,85 0,75 0,99 0,75 0,75 1,00 0,91 0,70 0,93 0,81 0,97 1,00 0,82 0,70 0,69 0,69 0,67 0,64 0,60 0,60 0,58 0,57 0,57 0,57 0,56 0,55 0,55 0,55 0,53 0,53 0,51 0,50 0,50 0,48 0,46 0,46 0,46 0,43 0,39 0,39 0,38 0,38 0,37 0,36 0,35 0,35 0,33 0,31 0,29 0,28 0,21 0,14 0,95 0,79 0,73 0,81 0,80 0,58 0,62 0,96 0,53 0,78 0,60 0,53 0,75 0,67 0,51 0,66 0,62 0,65 0,59 0,66 0,59 0,69 0,42 0,47 0,55 0,37 0,38 0,34 0,43 0,36 0,51 0,34 0,26 0,20 0,10 0,28 0,20 0,14 0,15 0,50 0,21 0,29 0,28 0,29 0,28 0,20 0,25 0,44 0,29 0,45 0,23 0,37 0,38 0,39 0,37 0,31 0,29 0,30 0,38 0,41 0,37 0,38 0,34 0,26 0,35 0,26 0,28 0,28 0,27 0,26 0,53 0,23 0,29 0,32 0,36 0,24 0,28 0,34 0,46 0,38 0.10 0.19 0.00 0.12 0.53 ⫺0.66 0.53 0.53 0.14 0.05 0.51 0.00 0.29 0.10 ⫺0.06 0.29 0.00 0.10 ⫺0.57 0.28 ⫺0.88 0.10 0.07 0.19 0.10 ⫺0.56 0.19 0.29 0.10 0.05 ⫺0.09 0.10 ⫺0.28 ⫺0.15 0.06 ⫺0.66 ⫺0.85 0.00 0.93 1.00 1.00 0.98 1.00 0.96 1.00 0.98 1.00 0.99 0.98 0.99 0.98 0.95 1.00 1.00 0.91 1.00 0.98 0.99 0.99 0.99 1.00 0.94 0.99 0.76 0.97 1.00 1.00 0.98 0.99 0.99 0.98 1.00 0.99 0.98 1.00 1.00 0.50 0.78 0.44 0.74 0.80 0.42 0.90 0.81 0.83 0.61 0.82 0.63 0.66 0.53 0.67 0.64 0.39 0.57 0.51 0.75 0.45 0.60 0.55 0.71 0.76 0.10 0.56 0.79 0.69 0.68 0.54 0.40 0.35 0.58 0.70 0.58 0.38 0.48 0.59 0.90 0.38 0.79 0.85 0.45 0.97 0.84 0.91 0.57 0.89 0.78 0.79 0.64 0.81 0.67 0.36 0.59 0.63 0.88 0.67 0.65 0.56 0.73 0.92 0.10 0.55 0.84 0.75 0.76 0.57 0.29 0.32 0.60 0.94 0.80 0.55 0.48 0.29 0.27 0.31 0.25 0.17 0.41 0.14 0.15 0.23 0.33 0.17 0.37 0.23 0.35 0.35 0.25 0.34 0.27 0.48 0.24 0.47 0.33 0.30 0.21 0.31 0.39 0.25 0.21 0.29 0.32 0.37 0.31 0.32 0.36 0.35 0.46 0.61 0.42 Min, minimun; Max, maximum. VALUE IN HEALTH 14 (2011) S108 –S114 S111 The first model was an ordinary least square regression model, considering g as a linear function [7]. Each individual health state valuation was considered as an independent observation, regardless of if it was valued by the same individual. The second model was a random effects model, which takes account of variation both within and between respondents. For this model, the error term of formula (1) is subdivided so that: ij ⫽ uj ⫹ eij (2) where uj is the respondent variation and eij is an error term for the ith valuation of the jth individual. A random variation is assumed for both terms. Estimation was through restricted maximum likelihood. Additional strategies were undertaken to deal with the possible effects of interaction between the levels of different dimensions of the SF-6D, as described in the original UK study [7] The models were evaluated considering the following criteria: 1) inconsistencies in the estimated coefficients, because the coefficients of dummy variables representing each level of SF-6D are expected to be negative and increasing in absolute size as the level of severity increases (amongst coefficients with statistical significance); 2) mean absolute error and the proportion of predictions outside 0.05 (% absolute error ⬎ 0.05) and 0.10 (% absolute error ⬎ 0.10) ranges on either side of the observed value; 3) goodness of fit measured using Akaike=s information criterion (AIC) and Bayes information criterion (BIC). Predictions were further tested in terms of bias (t test), normality of residuals (Jarque-Bera [JB]) and the presence of autocorrelation in the prediction errors (Ljung-Box [LB]). Analysis was performed using SPSS version 16.0 (SPSS Inc., Chicago, IL) [20], R 2.9.1 (R Development Core Team, Vienna, Austria) [21] and STATA 9.0 (Stata Corp., College Station, TX) [22]. Fig. 1 – Histogram for adjusted health state valuation. Figure 1 presents a histogram for the 2696 individual adjusted health state valuations with the skewness coefficient. As found in the UK study, negative values (states considered worse than death) were comparatively rare (Brazil 6.6% and United Kingdom 6.9%) and the proportion of valuations at the maximum value (1.0) was small (Brazil 0.4% vs. United Kingdom 0.5%). Regarding the worst health state, 63% of respondents valued it as better than death, while in the United Kingdom the rate was 73%. Modeling Results Study Population A total of 889 households were visited from July 2007 to March 2008. Interviews were conducted in 354 (40%) of them, because for the others it was impossible to contact residents after three consecutive visits or residents did not wish to participate in the study. Out of 846 eligible individuals identified, 537 accepted to take part, obtaining a response rate of 64%, similar to that of 65% found in the population of the UK study. Out of 537 respondents, 10 (1.9%) were excluded because they did not complete the ranking or SG tasks, leaving a sample of 527 respondents. A total of 58 (11%) respondents were excluded as they failed to value the worst state, because without this value it was not possible to adjust the values of all other intermediate states. Eighty-eight respondents (17%) gave the same value to the five intermediate states values but were not excluded from the analysis. The comparison between excluded and included subjects according to socio-demographic characteristics is presented in Table 1. There were no important differences regarding those variables between the two groups, except for economic class (P ⫽ 0.05). There are a higher percentage of people from lower economic classes (C and D) in the excluded group. The final sample contains 2696 observed standard gamble valuations across 248 health states from 469 respondents. Health State Values Descriptive statistics for 40 of the 248 health states are shown in Table 2, comparing the values observed in the Brazilian population with those obtained in the original UK study. Overall, the Brazilian mean health state values were lower and range from 0.14 to 0.82 with large standard deviations. Several models were estimated following the strategy proposed by Brazier et al. [7], but only the best models in terms of predictive ability are described in this article. Therefore, the results are presented for the random effects models with and without the intercept restricted to unity (Table 3). In the random effects model without the intercept forced to unity (model 1 in Table 3) 15 of 25 coefficients were significant with two inconsistencies, where the estimated effect decreases from level 4 to level 5 for the physical functioning dimension and level 2 of the vitality dimension did not show the expected negative sign. In terms of predictive ability, the proportion of prediction errors under 0.1 and 0.05 was 88% and 61%, respectively. The predictions are unbiased (P ⬎ 0.05), but prediction errors are not normally distributed (JB test). Moreover, there is autocorrelation in the prediction errors (LB statistics), as can be seen in Figure 2, which shows the observed and predicted values for model 1. There is a tendency to overpredict at low health state values and under-predict at high health state values. To be used to generate QALYs, the best health state generated by the SF-6D (111111) should be equal to 1 and death must be equal to zero. The best way to ensure that the best state has the value 1 is to restrict the intercept to unity [7]. Table 3 shows the random effects model (model 3) with the constant forced to unity. There was an increase in the number of significant coefficients compared to the previous model 1, but a higher number of inconsistencies. All coefficients had the expected negative sign. Regarding the number of prediction errors model 3 performed worse than model 1, with a higher proportion of absolute errors greater than 0.05 and 0.1. As in model 1, the predictions were unbiased and the residuals were not normally distributed. An important advantage of model 3 was the absence of autocorrelation in the prediction errors (LB test not significant). Figure 3 shows the curves of observed and predicted values for the 248 health states valued using model 3. S112 VALUE IN HEALTH 14 (2011) S108 –S114 Table 3 – Main effects models and consistent models. Constant forced to unity RE (Model 1) RE Consistent model (Model 2) C C PF2 PF3 PF4 PF5 PF6 RL2 RL3 RL4 SF2 SF3 SF4 SF5 PAIN2 PAIN3 PAIN4 PAIN5 PAIN6 MH2 MH3 MH4 MH5 VIT2 VIT3 VIT4 VIT5 N Inconsistencies MAE %AE ⬎ 0.05 %AE⬎ 0.10 t (mean⫽0) JBPRED LB AIC BIC 0.628 ⫺0.018 ⫺0.026 ⫺0.055 ⫺0.036 ⫺0.130 ⫺0.022 ⫺0.030 ⫺0.050 0.007 ⫺0.015 ⫺0.049 ⫺0.068 ⫺0.015 ⫺0.014 ⫺0.016 ⫺0.038 ⫺0.095 ⫺0.037 ⫺0.037 ⫺0.044 ⫺0.087 0.037 0.012 0.022 ⫺0.031 2696 2 0.050 39 12 1.149 57.021 3.172 281.08 446.26 p SE 0.00 0.30 0.11 0.00 0.03 0.00 0.11 0.03 0.00 0.63 0.32 0.00 0.00 0.35 0.39 0.33 0.02 0.00 0.02 0.02 0.00 0.00 0.02 0.45 0.17 0.06 0.02 0.02 0.02 0.02 0.01 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.02 0.02 0.02 0.02 0.01 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 C P SE PF2 PF3 0.652 ⫺0.021 ⫺0.023 0.00 0.24 0.15 0.02 0.02 0.02 PF45 PF6 RL2 RL3 RL4 ⫺0.043 ⫺0.133 ⫺0.022 ⫺0.029 ⫺0.047 0.00 0.00 0.10 0.03 0.00 0.01 0.02 0.01 0.01 0.01 SF3 SF4 SF5 ⫺0.022 ⫺0.051 ⫺0.069 0.08 0.00 0.00 0.01 0.01 0.01 PAIN234 PAIN5 PAIN6 MH2 MH3 MH4 MH5 ⫺0.019 ⫺0.043 ⫺0.097 ⫺0.035 ⫺0.037 ⫺0.041 ⫺0.086 0.14 0.00 0.00 0.03 0.02 0.01 0.00 0.01 0.02 0.01 0.02 0.02 0.02 0.02 0.051 0.00 0.01 VIT5 2696 0.051 41 12 1.094 60.111 4.554 407.394 531.139 RE (Model 3) C 1.000 ⫺0.088 ⫺0.070 ⫺0.104 ⫺0.064 ⫺0.177 ⫺0.076 ⫺0.065 ⫺0.089 0.063 ⫺0.067 ⫺0.102 ⫺0.114 ⫺0.105 ⫺0.068 ⫺0.075 ⫺0.110 ⫺0.140 ⫺0.091 ⫺0.088 ⫺0.090 ⫺0.133 ⫺0.049 ⫺0.043 ⫺0.040 ⫺0.083 2696 7 0.064 50 21 1.661 15.774 0.193 637.257 796.28 RE consistent model (Model 4) P SE 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.137 0.00 0.02 0.02 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.02 0.02 0.02 0.02 0.01 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 C P SE 1.000 PF2345* PF6 ⫺0.080 ⫺0.183 0.00 0.00 0.01 0.02 RL23 RL4 ⫺0.073 ⫺0.087 0.00 0.00 0.01 0.01 SF3 SF4 SF5 ⫺0.033 ⫺0.066 ⫺0.078 0.00 0.00 0.00 0.01 0.01 0.01 PAIN234† PAIN5 PAIN6 ⫺0.089 ⫺0.116 ⫺0.146 0.00 0.00 0.00 0.01 0.01 0.01 MH23 MH4 MH5 ⫺0.087 ⫺0.088 ⫺0.131 0.00 0.00 0.00 0.02 0.01 0.02 VIT234 VIT5 ⫺0.050 ⫺0.086 0.00 0.00 0.01 0.02 2696 0.065 50 20 1.916 15.599 6e-04 583.14 683.346 AIC, Akaike=s information criterion; BIC, Bayes information criterion; C, coefficients; JB, Jarque-Bera; LB, Ljung-Box; MAE, Mean absolute error; MH, mental health; PAIN, bodily pain; PF, physical functioning; RE, random effects; RL, role limitations; SE, standard error; SF, social functioning; VIT⫽ Vitality. * Coefficients PF23 and PF34 were inconsistent, so they were merged in PF2345 coefficient. † Coefficient PAIN23 and PAIN4 were inconsistent, so they were merged in PAIN234. Consistent Models Models without inconsistencies were estimated following an approach used in the UK SF-6D [23]. Consistent models were constructed from models 1 and 3 by aggregating levels of each dimension if inconsistencies occurred; that is if the coefficients on each level did not represent an additional decrease in the health state value. The results are shown in Table 3. Model 2 is a consistent version of model 1. From level 4 to level 5 of the physical dimension there was an inconsistency, as the coefficient of PF4 was higher than PF5 in absolute value. In the consistent model these two levels were merged. Social functioning level 2 was merged with the level 1 reference point because the coefficient was insignificant and inconsistent (positive sign), as were the coefficients of levels 2, 3, and 4 of the vitality dimension and hence these were also merged with the level 1 reference point. In model 3, all coefficients were significant but inconsistent levels were merged in model 4 (PF2 and PF3; PF4 and PF4; RL2 and RL3; PAIN2 and PAIN3; MH2 and MH3; and VIT2, VIT3, and VIT4). Using this methodology, models 2 and 4 are consistent versions of models 1 and 2 respectively. Both are similar in terms of predictive ability compared to their original models. Model 4 has a higher mean absolute error and higher proportion of predictions outside 0.05 and 0.10 ranges in comparison to model 2 and lower AIC and BIC when compared to models 1 and 2, but has the advantage of having the intercept fixed at 1 according to the conventional utility scale. Furthermore, the model does not suffer from autocorrelation in the errors of prediction. For these reasons, this model appears to be the most appropriate. Figure 4 shows the observed and predicted values for model 4. Table 4 shows a comparison between the Brazilian consistent model (model 4) and the United Kingdom consistent model used to derive SF-6D [7]. Statistically, these are not comparable models, since the UK model was estimated by ordinary least squares (OLS) using data at a mean level, including the interaction term VALUE IN HEALTH 14 (2011) S108 –S114 Fig. 2 – Observed and predicted health state valuation for the random effects model (1). (“MOST”). The OLS models with interaction terms applied to the study data performed worse than all other models estimated in terms of predictive ability. Mean absolute error increased to 0.102 and the percentage of predictions errors greater than 0.05 and 0.10 was 70% and 42%, respectively, with problems of autocorrelation in the errors. In addition, our OLS models performed worse than random effects models using AIC and BIC. The coefficients in the models can be interpreted as representing any decrement in utility associated with health deteriorating from full health. In the recommended model 4, the value of full health is equal to one, since the intercept was forced to unity. Utility values for all possible SF-6D health states can be obtained by using the coefficients estimates by subtracting from 1 the utility decrement associated with each level in a given health state. For Fig. 3 – Observed and predicted health states valuation for the random effects model with constant forced to unity (3). S113 Fig. 4 – Observed and predicted health state valuation for the consistent random effects model with constant forced to unity (4). example, taking the state 245633, the estimated value would be: 1 ⫺ 0.080 ⫺ 0.087 ⫺ 0.078 ⫺ 0.146 ⫺ 0.087-0 ⫽ 0.52 Discussion The results of this study provide the first Brazilian populationbased value set for health states, making it possible to generate QALYs for cost-utility studies using regional data and preferences. We acknowledge that the population of Porto Alegre is not representative of the whole Brazilian population. On the other hand, this kind of survey is complex and expensive, so it could be difficult to have algorithms for all regions in such a large country as Brazil. We believe that using SF-6D health states preferences values from the Porto Alegre population to conduct cost-effectiveness studies in Brazil is a more suitable alternative than using values obtained in other countries. This is a relatively new research area in Latin America, with only one study valuing EQ-5D health states in the population of Argentina, recently published [11]. The Brazilian SF-6D preference weights estimated here offer a method for producing utility values from existing SF-36 data. Because the overall aim was to construct a model to predict values for all possible health states generated by the SF-6D, the main criterion used to choose the most appropriate model was predictive ability in terms of mean absolute errors between observed and predicted health state values and percentages of prediction errors greater than 0.05 and 0.01. On this basis, the random effects models performed best among all models tested. On theoretical grounds it is argued that the constant term in the models should be set equal to one to conform to the conventional utility scale. This would suggest model 4 as the preferred model. One concern is the existence of inconsistencies between coefficients. This finding can be related to the difficulty in attempting to value a comparatively large classification system describing 18,000 health states. Due to sample size issues, some values are not very stable. This result is similar to that found by the authors of the original study and other researchers in different countries [7,13–15]. It is likely that increasing the number of health states to be valued, which in turn requires a larger number of individuals, S114 VALUE IN HEALTH 14 (2011) S108 –S114 may overcome some of the inconsistencies and insignificant coefficients for some levels in some dimensions [23]. Moreover, the inconsistencies may be expected due to the classification system of the SF-6D, with so many levels in each dimension. Furthermore, SG is not an easy instrument to complete, and some people can have difficulty understanding the exercise, leading to illogical responses. The percentage of respondents who were not able to give a valid value to the worst SF-6D health state was higher in lower economic classes. The criterion to classify economic class is based on income and purchasing power. In Brazil, a higher economic class guarantees greater access to education, information, health care, and even greater social inclusion. Consequently, the higher classes have greater opportunity of attaining conditions required to make decisions on their own health, and perform more complex tasks. Because conducting a population survey is very costly, in terms of time and resources, an alternative for dealing with inconsistent coefficients was proposed by Brazier et al. [23]. Following this strategy, it was possible to construct consistent models by merging inconsistent coefficients. Consistent models were estimated from models 1 and 3, containing only the levels that contribute significantly to the final health state value. Therefore, model 4 is recommended for calculating utility values for this Brazilian version of SF-6D health states. The differences between Brazilian and UK models reinforce the importance of using a country-specific algorithm to calculate utility values in the national context. The best fitting model in the Brazilian data was a random effects consistent model that takes into account variations in two levels, inter- and intraindividual, whereas for the UK data the best model was an OLS consistent model using data at a mean level. In general, the coefficients of the Brazilian model were larger than those found in the United Kingdom, leading to a greater decrement in utility values from full health. Although the bodily pain dimension appears to be the most important dimension in determining the health state value for both cultures, the physical functioning dimension seems to have greater importance for the Brazilians. Differences between the weights of different dimensions among diverse cultures was also found in other countries for SF-6D [13–15], and similar instruments such as EQ-5D [8 –12,24], emphasizing the need for more studies analyzing the association between cross-cultural variables and preference measurement. Although we recognize some caveats regarding interpreting our study, it is important to emphasize the difficulty of conducting a population survey in Brazil. Due to high rates of urban violence, in many census sectors people live in buildings with security systems that greatly hinder access to residents. For these reasons it was necessary to adopt the strategy of replacement of losses and refusals, visiting a larger number of households than planned to obtain the required number of interviews. The availability of a regional algorithm for calculation of utility scores represents an opportunity to undertake local health economics research in Brazil. To date, studies of cost-utility in Brazil have been conducted using secondary data derived from other countries, generally developed countries. This study estimated preference weights using a random sample of the southern Brazilian general population, suitable for incorporation in the decisionmaking process for resource allocation and public health policies in Brazil. Sources of financial support: This study was funded by CNPQ/ Brazil (Edital MCT-CNPq/MS-SCTIE-DECIT – No. 36/2005). Dr. Luciane Cruz received a graduate research scholarship from CAPES, Brazil. Professors Polanczyk and Fleck received a research scholarship from CNPq/Brazil. REFERENCES [1] Patrick DL, Erickson P. Applications of health status assessment to health policy. In: Spilker B, ed., Quality of Life and Pharmacoeconomics in Clinical Trials (2nd ed.). Philadelphia: Lippincott-Raven, 1996. [2] Kind P, Lafata JE, Matuszewski K, Raisch D. The use of QALYs in clinical and patient decision-making: issues and prospects. Value Health 2009;12(Suppl. 1):S27–30. [3] Weinstein MC, Siegel JE, Gold MR, et al. Recommendations of the Panel on Cost-effectiveness in Health and Medicine. JAMA 1996;276:1253– 8. [4] National Institute for Health and Clinical Excellence. Available from: www.nice.org.uk. [Accessed March 7, 2010]. [5] EuroQol—a new facility for the measurement of health-related quality of life. The EuroQol Group. Health Policy 1990;16:199 –208. [6] Furlong WJ, Feeny DH, Torrance GW, Barr RD. The Health Utilities Index (HUI) system for assessing health-related quality of life in clinical studies. Ann Med 2001;33:375– 84. [7] Brazier J, Roberts J, Deverill M. The estimation of a preference-based measure of health from the SF-36. J Health Econ 2002;21:271–92. [8] Greiner W, Claes C, Busschbach JJ, von der Schulenburg JM. Validating the EQ-5D with time trade off for the German population. Eur J Health Econ 2005;6:124 –30. [9] Tsuchiya A, Ikeda S, Ikegami N, et al. Estimating an EQ-5D population value set: the case of Japan. Health Econ 2002;11:341–53. [10] Johnson JA, Luo N, Shaw JW, et al. Valuations of EQ-5D health states: are the United States and United Kingdom different? Med Care 2005; 43:221– 8. [11] Augustovski FA, Irazola VE, Velazquez AP, et al. Argentine valuation of the EQ-5D health states. Value Health 2009;12:587–96. [12] Zarate V, Kind P, Chuang LH. Hispanic Valuation of the EQ-5D health states: a social value set for Latin Americans. Value Health 2008;11: 1170 –7. [13] Brazier JE, Fukuhara S, Roberts J, et al. Estimating a preference-based index from the Japanese SF-36. J Clin Epidemiol 2009;62:1323–31. [14] Lam CL, Brazier J, McGhee SM. Valuation of the SF-6D Health states is feasible, acceptable, reliable, and valid in a Chinese population. Value Health 2008;11:295–303. [15] Ferreira LN, Ferreira PL, Pereira LN, et al. A Portuguese value set for the SF-6D. Value Health 2010;13:624 –30. [16] Ciconelli RM, Ferraz MB, Santos W, et al. Brazilian-Portuguese version of the SF-36. A reliable and valid quality of life outcome measure. Rev Bras Reumatol 1999;39:143–50. [17] Censo Populacional 2000. Instituto Brasileiro de Geografia e Estatística [Brazilian Institute of Geography and Statistics]. Available from: www.ibge.gov.br. [Accessed March 2010]. [18] Furlong W, Feeny D, Torrance G, Barr R, Horsman J. Guide to Design and Development of Health-State Utility Instrumentation. Hamilton, Ontário, Canada: McMaster University, 1990. Working Paper Series No. 90-0. [19] Critério Brasil 2003. Associação Brasileira de Empresas de Pesquisa [Brazilian Association of Research Companies] 2010 March 7. Available from: www.abep.org. [Accessed March 07, 2010]. [20] Statistical Package for Social Sciences. Available from: http://www.spss.com/software/. [Accessed March 6, 2010]. [21] R: a language and environment for statistical computing. R Foundation for Statistical Computing. Available from: www. R-project.org. [Accessed March 6, 2010]. [22] Data analysis and statistical software. Available from: www.stata.com. [Accessed March 6, 2010]. [23] Brazier JE, Roberts J. The estimation of a preference-based measure of health from the SF-12. Med Care 2004;42:851–9. [24] Badia X, Roset M, Herdman M, Kind P. A comparison of United Kingdom and Spanish general population time trade-off values for EQ-5D health states. Med Decis Making 2001;21:7–16. VALUE IN HEALTH 14 (2011) S115–S118 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Factores Asociados al Incumplimiento de los Tratamientos con Antidepresivos en Santiago, Chile Marcela Jirón, PhD, PharmD, MSc1,*, Leslie Escobar, PharmD, PhD Student1, Leonardo Arriagada, PharmD2, Sebastián Orellana, PharmD1, Ariel Castro, PharmD, MSc1,2 1 Químico Farmacéutico, Departamento de Ciencias y Tecnología Farmacéuticas, Facultad de Ciencias Químicas y Farmacéuticas, Universidad de Chile, Santiago, Chile; 2Químico Farmacéutico, Hospital Clínico de la Universidad de Chile, Santiago, Chile A B S T R A C T Objective: To identify factors associated with non-compliance to antidepressant’s (AD) treatment in Santiago, Chile. Methods: A cross-sectional study was carried out in a household randomized and representative sample of 1000 individuals aged 15 years and older. Treatment adherence was studied in AD consumers using logistic regression to estimate factors associated with non-compliance in doses or time of treatment. Results: Antidepressant non-compliance was 52.8% and their main associated factor was income. Gender and educational level were also associated with AD non-compliance. Conclusions: Antidepressant Introducción En países desarrollados, se ha descrito un aumento en el consumo de antidepresivos (AD) debido a diferentes situaciones, tales como mayor acceso a atención médica y a AD, aumento en la prevalencia de depresión, mejor reconocimiento de las enfermedades asociadas a su uso tanto psiquiátrico como no psiquiátrico, nuevos usos descritos como fibromialgia y migraña, entre otras razones [1]. No obstante, este aumento en el consumo de AD ha sido explicado principalmente por un mayor número de prescripciones por trastornos depresivos [2]. Se han realizado pocos estudios en países en vías de desarrollo, como Chile, para determinar la prevalencia de trastornos psiquiátricos en la población. Vicente et al. [3] reportaron que un 31,5% de la población chilena ha tenido al menos una vez en su vida un trastorno mental, siendo la depresión una de las más prevalentes (9,2%). Asimismo, se sabe que el consumo de AD aumentó en más de 470% en Chile entre 1992 y 2004 [4]. Sin embargo, existe limitada información sobre el cumplimiento a terapias con AD y sus factores asociados. Los antecedentes disponibles respecto a incumplimiento de las terapias con AD muestran que más del 40% de los pacientes discontinúan el tratamiento dentro de los 3 primeros meses [5]. En pacientes con depresión el incumplimiento varía entre 4070% [6] y tienen 3 veces más riesgo de no cumplir las terapias [7], principalmente por suspensión anticipada del tratamiento, obs- non-compliance was associated with income and sociodemographic characteristics in Chilean patients. Therefore, health strategies to improve AD compliance should consider inequities in access to medications and characteristics of the AD consumers. Palabras Claves: antidepressants treatment, compliance, risk factors, Santiago de Chile. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. taculizando el logro de los objetivos terapéuticos y aumentando el riesgo de relapso o la reaparición de los síntomas [8]. Se ha visto en estudios controlados que la tasa de abandono al tratamiento AD varía entre 20-40% debido principalmente a eventos adversos [9]. En cambio, en estudios naturalísticos, la frecuencia de incumplimiento alcanza el 50-60% debido a que los pacientes refieren “sentirse mejor” [9,10]. Entre las razones que explicarían el incumplimiento se encuentran el temor a los posibles efectos adversos antes de iniciar o durante el tratamiento AD, problemas económicos para adquirirlos o desacuerdo del paciente con la terapia [11–13]. Por lo tanto, el incumplimiento estaría determinado por condiciones multifactoriales. Dado el aumento en el consumo de AD en Chile, las consecuencias negativas del incumplimiento en los tratamientos, tanto para el paciente, la sociedad y el sistema de salud [13] y la falta de antecedentes sobre adherencia a AD en países de Latinoamérica, la presente investigación tuvo por objetivo identificar los factores asociados al incumplimiento de los tratamientos con AD en la población de Santiago de Chile. Método Mediante un estudio transversal se entrevistó en hogares a una muestra aleatoria y representativa de adultos de 15 o más años, en Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Título corto: Factores asociados al incumplimiento con antidepresivos. * Autor de correspondencia: Marcela Jirón, Sergio Livingstone 1007 (ex Olivos), Independencia, Santiago, Chile, Casilla 233-1; Tel: 56-2 - 978 2838; Fax: 56-2 - 978 2990. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.035 S116 VALUE IN HEALTH 14 (2011) S115–S118 Santiago de Chile. Para esto 25 encuestadores aplicaron un cuestionario sobre el consumo y cumplimiento del tratamiento con AD, independiente de la razón de uso. Para asegurar la representatividad de la muestra respecto a la población de Santiago, los datos fueron ponderados por género según Censo Nacional 2002 y nivel socioeconómico según AIM Chile [14]. Chile tiene una población aproximada de 15 millones de personas y en Santiago, su capital, viven cerca de 6 millones de habitantes, principalmente en zonas urbanas. Santiago se encuentra dividido en 34 comunas, constituidas por manzanas, las cuales fueron aleatorizadas usando mapas cartográficos. Posteriormente, para la selección de cada hogar se contó el número de viviendas en cada manzana y se seleccionó arbitrariamente un hogar cada 8 viviendas, siguiendo el sentido del reloj, partiendo por la primera en la esquina noreste de cada manzana. En cada hogar se seleccionó una persona a entrevistar utilizando la tabla Kish [15]. No se incluyó a personas en cárceles, hospitales, pacientes institucionalizados o sin hogar. Un total de 1543 viviendas fueron seleccionadas para conseguir 1000 personas dispuestas a participar en la investigación (tasa de respuesta 64,8%). Esta investigación fue conducida por Profesores de la Universidad de Chile y contó con aprobación del Comité de Ética de la Facultad de Ciencias Químicas y Farmacéuticas de dicha Universidad. Todos los participantes dieron su consentimiento para participar y los datos fueron codificados para asegurar la confidencialidad de la información. Instrumento La encuesta utilizada fue especialmente diseñada y validada a través de pruebas iterativas hasta lograr reproducibilidad de respuesta. Contaba con preguntas sobre consumo y cumplimiento del tratamiento AD, tipos de AD utilizados (Inhibidores Selectivos de la Recaptación de Serotonina (ISRS), Inhibidores de la Recaptación de Norepinefrina (IRNE), productos naturales y otros), e información sociodemográfica del encuestado (edad, género, años de escolaridad, estado civil (con o sin pareja), nivel socioeconómico (alto, medio-alto, medio, medio-bajo, bajo), seguro de salud (privado o público), ocupación (con o sin trabajo) y hábitos (tabaco y alcohol). La identificación del AD utilizado se realizó usando listados de medicamentos disponibles en el mercado farmacéutico chileno. El tiempo transcurrido entre el uso del AD y la entrevista fue en el último mes, último año o alguna vez en su vida. Los encuestadores tenían experiencia y entrenamiento en la aplicación de encuestas, además recibieron capacitación específica sobre el cuestionario y los objetivos de la investigación. Para asegurar la calidad de la información recolectada, dos supervisores re-contactaron al 30% de la muestra para confirmar las respuestas obtenidas. El 100% de los datos digitados fue revisado y cualquier diferencia de información fue resuelta con una nueva visita. Sólo datos consistentes y cuestionarios completos se consideraron exitosos. Análisis Estadístico Entre los entrevistados se seleccionó aquellos que alguna vez en su vida usaron AD, para usos psiquiátricos y no psiquiátricos, incluyendo aquellos que habían suspendido sus terapias con AD al momento de la entrevista. Se clasificó como incumplidores a quienes modificaron las indicaciones de uso del AD por iniciativa propia o no médica, y en ellos se estudiaron los factores asociados al incumplimiento de la dosis o duración/tiempo de la terapia AD. Se definió como incumplidor de dosis aquellos pacientes que declararon modificar la cantidad diaria de AD indicada; mientras que incumplidor de tiempo se definió como aquel que modificó la frecuencia diaria de administración del AD o la duración del tratamiento según lo indicado. Los análisis se realizaron utilizando STATA 8.0. A través de regresiones logísticas se evaluó los posibles factores asociados al incumplimiento de los tratamientos con AD, con sus correspondientes Odds Ratios (OR) e intervalos de confianza al 95% (IC95%). Para variables con más de 2 estratos se utilizaron variables dummy o imaginarias para los análisis [16]. Resultados De los 171 sujetos que usaron AD alguna vez en su vida, el 59,1% los utilizó el último año y 36,3% el último mes. El 97,1%(166) contestó las preguntas sobre cumplimiento del tratamiento AD y el 52,8% resultó incumplidor. El 30,6% de los 166 modificó la dosis y el 45,6% varió la duración del tratamiento. El 85,9% declaró tener prescripción médica del AD. La caracterización sociodemográfica de los consumidores de AD se encuentra en la Tabla 1 de material complementario en: doi:10.1016/j.jval.2011.05.035. Las principales razones de incumplimiento de la dosis del AD fueron sentirse mejor (32,8%), temor a los efectos adversos (30,8%), sentirse mal cuando lo usaba (26,2%), olvido (7,5%) y falta de dinero para adquirirlo (5,9%); mientras que para quienes variaban la duración del tratamiento AD, estos valores fueron 33,7%, 24,8%, 30,7%, 10,1% y 6,5%, respectivamente. No se especificó si el temor a efectos adversos se debía a experiencias propias o a creencias. Entre los AD utilizados se encontraban los ISRS (88,2%), principalmente fluoxetina (65,1%); IRNE (39,1%); productos naturales (11,6%) y otros (9,6%). Algunos sujetos utilizaron más de un AD. Factores asociados al incumplimiento del tratamiento con antidepresivos Se encontraron asociaciones significativas entre el incumplimiento del tratamiento con AD y el ingreso económico, género y nivel educacional. Sin embargo, la influencia de estos factores varió según tipo de incumplimiento (dosis o tiempo) (Tabla 2 de material complementario en: doi:10.1016/j.jval.2011.05.035). Al estudiar el incumplimiento en general a AD, se encontró que tener la condición de ingresos medio-bajo presentó 3,6 veces más riesgo de modificar la dosis o la duración de la terapia que aquellos de ingresos altos (IC95% 1,19-10,71). Específicamente, el incumplimiento en la dosis AD se asoció a ingresos medio-bajo y escolaridad de hasta 8 años, presentando 3,3 veces más riesgo de modificar la dosis indicada que aquellos de ingresos altos (IC 95% 1,04-10,28) y 2,5 veces más riesgo de variar la dosis que aquellos con más de 12 años de estudios (IC95% 1,016,02), respectivamente. En hombres de ingresos medio-bajo este riesgo aumentó a 11 veces (IC95% 1,11-109,39). Asimismo, el incumplimiento de la duración de la terapia con AD, se asoció con tener ingresos medio-bajo y bajo, presentando 5,3 (IC95% 1,51-18,40) y 3,3 (IC95% 1,09-9,88) veces más riesgo de abandonar la terapia que aquellos de ingresos altos, especialmente en mujeres de ingresos bajos (OR 4,3; IC95% 1,11-16,52). El análisis de regresión múltiple mostró que el incumplimiento al tratamiento con AD estuvo asociado a ingresos medios y bajo (OR 3,1; IC95% 1,4-6,8; p⬍0,05) y a niveles de escolaridad de hasta 8 años (OR 2,4; IC95% 1,1-5,3; p⬍0,05), respecto a niveles de ingresos altos y más de 12 años de escolaridad, respectivamente. No se encontraron asociaciones entre el incumplimiento al tratamiento AD y la edad, seguro de salud, ocupación, estado civil y hábitos. Discusión y Conclusiones Aún cuando el consumo de AD en Chile ha tenido un crecimiento importante, hasta la fecha no hay publicaciones sobre el incumpli- VALUE IN HEALTH 14 (2011) S115–S118 miento de tratamientos con AD y sus posibles causas. La presente investigación aporta antecedentes a la escasa información disponible en Latinoamérica. Nuestros resultados muestran que en Santiago el incumplimiento con AD supera el 52%, mayor a lo publicado en algunos trabajos [13,17–19]. Las razones de abandono más frecuentemente mencionadas, como sentirse mejor y temor a efectos adversos, son coincidentes con otras publicaciones [9,10]. Aunque otras razones descritas como baja tolerabilidad [5], creencias y actitudes frente a los medicamentos y enfermedades mentales [12,20], no fueron explícitamente mencionadas por los entrevistados. El temor a los eventos adversos del AD podría, en algunos casos, determinar que el paciente no utilice el medicamento [12]. Se ha reportado que el temor a los efectos adversos es la principal razón de abandono de las terapias antidepresivas, sobre todo si se trata de ISRS [11], grupo terapéutico más utilizado por esta muestra y por otros trabajos con AD [21]. Dado el buen perfil de seguridad de los ISRS comparado con otros AD, parece más importante reforzar la comunicación médico-paciente cuando se decida utilizar estos medicamentos. Los factores asociados al incumplimiento con AD encontrados fueron similares a los reportados por otros autores, como nivel educacional, ingresos y género [9,12]. Sin embargo, su influencia varía según lugar y diseño del estudio. En Santiago, el abandono del tratamiento se asoció a bajos ingresos, especialmente en mujeres, lo que podría explicarse por inequidad y/o mayor vulnerabilidad a eventos afectivos en la mujer. En cambio, el mayor riesgo de incumplimiento en la dosis fue en bajos niveles de escolaridad, especialmente en hombres de ingresos medio-bajo, posiblemente debido a que los sujetos con baja escolaridad requieren instrucciones especiales sobre sus terapias [22], y a que los hombres que mejoran su funcionalidad tienen más riesgo de incumplimiento [23]. Entonces, ¿qué tipo de acciones se debieran realizar para mejorar el cumplimiento? A simple vista pareciera que una estrategia para disminuir las tasas de abandono y mejorar el conocimiento sobre los tratamientos es una adecuada monitorización e información a los usuarios sobre potenciales beneficios y eventos adversos de la terapia [5]. Sin embargo, se han analizado variadas acciones relacionadas con intervenciones educacionales y de recursos sanitarios, pero los resultados son controversiales, ya que no mejoran de manera satisfactoria la adherencia terapéutica a largo plazo [24]. Según la OMS, la capacidad de que los pacientes sigan adecuadamenteme las instrucciones del tratamiento está afectada por más de una barrera, como aspectos sociales y económicos, el tipo de sistema de salud disponible, las características de la enfermedad, etc. [13]. Por consiguiente, a juicio de estos autores los esfuerzos para mejorar la adherencia a AD en Santiago debieran orientarse a resolver posibles inequidades en el acceso a AD y a diseñar estrategias ajustadas a las características sociodemográficas del paciente, las cuales de acuerdo a la presente investigación, debieran enfocarse a grupos de pacientes de bajos ingresos económicos y con niveles de escolaridad inferiores a 8 años. Además, habría que reforzar especialmente el cumplimiento de dosis en hombres con bajo nivel educacional y bajos ingresos, y prevenir el abandono de la terapia especialmente en mujeres de ingresos bajo y medio-bajo; poniendo especial énfasis en reforzar la relación médico-paciente para abordar temores y creencias sobre los eventos adversos del tratamiento. Cabe destacar que a la fecha de recopilación de datos de la presente investigación (2004) aún no se implementaba el Régimen de Garantías Explícitas en Salud (GES) en Chile, sistema que asegura las condiciones de acceso, oportunidad y resguardo financiero para algunas enfermedades, por lo que las condiciones actuales pueden ser diferentes. Sería interesante evaluar el impacto del GES en el acceso S117 a tratamientos con antidepresivos y su posible efecto sobre el cumplimiento. Algunas de las limitaciones en la interpretación de estos resultados tienen relación con la presencia de sesgos propios de la metodología utilizada durante la recolección de la información, la cual estuvo basada en las declaraciones de los entrevistados y pudiera ser menos exacta por sesgos de memoria, como también la posible estigmatización sobre trastornos mentales y el uso de medicamentos, entre otros. Estos sesgos podrían subestimar el uso de AD y/o variar el perfil de cumplimiento. No obstante y aunque con menor poder, se hizo un análisis adicional para los grupos de pacientes que declararon usar AD en el último mes y último año, posteriormente se compararon con el análisis original y se encontró que las tendencias para los factores asociados al incumplimiento con AD fueron similares entre los grupos. Asimismo, por la metodología utilizada no fue posible confirmar si el uso de los AD fue apropiado o si los pacientes que recibieron AD cumplían con los criterios diagnósticos que avalaran su uso, especialmente si cerca del 74% de la población chilena accede a medicamentos en farmacias sin exigírseles receta médica [25], reafirmando la utilidad de obtener información directamente del paciente y en lugares diferentes de donde se entrega habitualmente la atención médica. Tampoco fue posible especificar el momento en que ocurrió el incumplimiento. Por otro lado, las dificultades en conseguir financiamiento para desarrollar estas investigaciones determina largos periodos de tiempo en su ejecución y análisis, por lo tanto, las condiciones pueden variar en el tiempo, afectando los resultados obtenidos. Es necesario tener presente que algunos de los factores estudiados varían a lo largo del tiempo y podrían existir condiciones que no estaban presentes al momento del consumo de AD, por lo tanto, habría que ser prudente con los resultados obtenidos. Por otro lado, existen grupos de pacientes no representados en la muestra seleccionada. Finalmente, aunque el uso de ponderadores es una técnica ampliamente utilizada para ajustar muestras a una distribución poblacional esperada, estos no corresponden a un valor real y por lo tanto, son una estimación realizada bajo supuestos del investigador. Esperamos que estos resultados estimulen la ejecución de futuras investigaciones a nivel nacional y en Latinoamérica, que permitan evaluar las consecuencias del incumplimiento, así como también, desarrollar intervenciones para mejorar el uso de los medicamentos y el manejo clínico de pacientes, que ayuden a prevenir resultados negativos en salud y promover el desarrollo de estrategias que mejoren la adherencia a antidepresivos. Dichas estrategias debieran incluir aspectos sociodemográficos y económicos en la población chilena. Fuentes de financiamiento: Proyecto financiado por el DI Salud 02/7-2 y DI I2 05/05-2 de la Universidad de Chile. Los autores declaran no tener conflictos de intereses y afirman que este artículo no ha sido sometido en forma simultánea a otra publicación. Materiales Complementarios Material complementario que acompaña este artículo se puede encontrar en la versión en línea como un hipervínculo en doi: 10.1016/j.jval.2011.05.035o si es un artículo impreso, estará en www.valueinhealthjournal.com/issues (seleccione el volumen, número y artículo). REFERENCIAS [1] Cascade EF, Kalali AH, Thase ME. Use of antidepressants. expansion beyond depression and anxiety. Psychiatry 2007;4:25– 8. S118 VALUE IN HEALTH 14 (2011) S115–S118 [2] Pincus HA, Tanielian TL, Steven C, et al. Prescribing trends in psychotropic medications: primary care, psychiatry, and other medical specialties. JAMA 1998;279:526 –31. [3] Vicente B, Kohn R, Rioseco P, et al. Lifetime and 12-month prevalence of DSM-III-R disorders in the Chile psychiatric prevalence study. Am J Psychiatry 2006;163:1362–70. [4] Jirón M, Machado M, Ruiz I. Consumo de Antidepresivos en Chile: 1992-2004. Rev Méd Chile 2008;136:1147–54. [5] Lin EH, Von Korff M, Katon W, et al. The role of the primary care physician in patients’ adherence to antidepressant therapy. Med Care 1995;33:67–74. [6] Demyttenaere K. Noncompliance with antidepressants: who’s to blame? Int Clin Psycopharmacol 1998;13(Suppl.2):S19 –25. [7] DiMatteo MR, Lepper H, Croghan TW. Depression is a risk factor for noncompliance with medical treatment: meta-analysis of the effects of anxiety and depression on patient adherence. Arch Intern Med 2000;160:2101–7. [8] Melfi CA, Chawla AJ, Croghan TW, et al. The effects of adherence to antidepressant treatment guidelines on relapse and recurrence of depression. Arch Gen Psychiatry 1998;55:1128 –32. [9] Demyttenaere, K. Risk factors and predictors of compliance in depression. Eur Neuropsychopharmacol 2003;13(Suppl.3):S69 –75. [10] Demyttenaere K, Enzlin P, Dewé W, et al. Compliance with antidepressant in primary care setting, 1: beyond lack of efficacy and adverse events. J Clin Psychiatry 2001;62(Suppl.22):S30 –3. [11] van Geffen EC, van Hulten R, Bouvy ML, et al. Characteristics and reasons associated with nonacceptance of selective serotoninreuptake inhibitor treatment. Ann Pharmacother 2008;42:218 –25. [12] Jorm AF, Christensen H, Griffiths KM. Belief in the harmfulness of antidepressants: results from a national survey of the Australian public. J Affect Disord 2005;88:47–53. [13] Sabaté E. Adherence to log-term therapies: evidence for action. WHO Library Cataloguing-in-Publication Data, 2003. [14] Asociación Chilena de Empresas de Investigación de Mercado. Grupos socioeconómicos. Chile 2008. [15] Kish L. Sampling organizations and groups of unequal sizes. Am Sociol Rev 1965;30:564 –72. [16] Daniel W. Regresión y correlación múltiple. En: Daniel W, Bioestadística: base para el análisis de las ciencias de la salud.Ciudad de México: Noriega Editores, 1995. [17] Cantrell CR, Eaddy MT, Shah MB, et al. Methods for evaluating patient adherence to antidepressant therapy. Med Care 2006:44;300 –3. [18] Robinson RL, Long SR, Chang S, et al. Higher cost and therapeutic factors associated with adherence to NCQA HEDIS antidepressant medication management measures: analysis of administrative claims. J Manag Care Pharm 2006;12:43–54. [19] Bulloch AG, Patten SB. Non-adherence with psychotropic medications in the general population. Soc Psychiatry Psychiatr Epidemiol 2010;45: 47–56. [20] Aikens JE, Kroenke K, Swindle RW, et al. Nine-month predictors and outcomes of SSRI antidepressant continuation in primary care. Gen Hosp Psychiatry 2005;27:229 –36. [21] Olfson M, Marcus SC. National patterns in antidepressant medication treatment. Arch Gen Psychiatry 2009;66:848 –56. [22] ten Doesschate MC, Bockting CL, Koeter MW, et al. Predictors of nonadherence to continuation and maintenance antidepressant medication in patients with remitted recurrent depression. J Clin Psychiatry 2009;70:63–9. [23] Demyttenaere K, Enzlin P, Dewé W, et al. Compliance with antidepressants in a primary care setting, 2: the influence of gender and type of impairment. J Clin Psychiatry 2001;62(Suppl.22): S34 –7. [24] McDonald HP, Garg AX, Haynes RB. Interventions to enhance patient adherence to medication prescription: scientific review. JAMA 2002; 288:2868 –79. [25] Situación de Salud en Chile 1999. Tarjeta de presentación. Ministerio de Salud, Chile, 2000. VALUE IN HEALTH 14 (2011) S119 –S121 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Health-Related Quality of Life of Patients Recieving Hemodialysis and Peritoneal Dialysis in São Paulo, Brazil: A Longitudinal Study Mirhelen Mendes de Abreu, MD,1,* David R. Walker, PhD,2,† Ricardo C. Sesso, MD, PhD,3 Marcos B. Ferraz, MD, PhD4 1 Universidade Federal de São Carlos, São Carlos, Brazil; 2Baxter Healthcare Corp., McGaw Park, IL, USA, at the time of study; 3Nephrology Division, Universidade Federal de São Paulo, São Paulo, Brazil; 4São Paulo Centre for Health Economics, Universidade Federal de São Paulo, São Paulo, Brazil A B S T R A C T Objectives: The aim of this study was to evaluate quality of life in patients undergoing hemodialysis (HD) or peritoneal dialysis (PD) in São Paulo, Brazil. Methods: Inclusion criteria for this is a 1-year prospective study included being 18 years of age or older and clinically stable receiving chronic dialysis. Quality of life was measured using the SF-12 and the Kidney Disease Quality of Life questionnaires at baseline, 6 months, and 12 months. Patients who completed the surveys for all three periods were evaluated. Differences in quality of life scores were measured using univariate and multivariate regression analyses. Results: One hundred eighty-nine of 249 (76%) HD patients and 161 of 228 (71%) PD patients completed all three surveys. The PD group was older and a larger number had diabetes. PD patients consistently had higher scores than HD patients at all three measurement periods for patient satisfaction (P ⫽ 0.002, P ⫽ 0.005, and P ⫽ 0.005, respectively), encouragement/support from staff (P ⫽ 0.003, P ⫽ 0.017, and P ⫽ 0.029, Introduction Health status and health-related quality of life (HRQoL) are core components of health outcomes. HRQoL measures can be assessed with both generic and disease-specific instruments [1]. Renal replacement therapy (RRT) is a life-saving treatment for patients with end-stage renal disease. The two main treatment modalities are transplantation and dialysis (i.e., hemodialysis [HD] or peritoneal dialysis [PD]) [2,3]. Because of an increase in survival rates for patients with endstage renal disease, HRQoL has become increasingly important as an outcome measure in the evaluation of dialysis treatments. [4] It has also been suggested that for patients receiving RRT, quality of life measures can be used to predict future morbidity and mortality [5]. Studies examining the difference in quality of life outcomes associated with HD and PD have mixed results [4,5]. More than 77,000 patients in Brazil were receiving chronic dialysis during January 2009 [6]; ranking the country among the top three in the world in terms of absolute number of dialysis patients. However, in Brazil, as in most developing countries, there have been no known prospective studies completed comparing HRQoL in HD and PD populations. This study is part of a larger one eval† respectively), and burden of kidney disease (P ⫽ 0.003, P ⫽ 0.017, and P ⫽ 0.057, respectively). The HD group had a greater percent of patients who clinically improved from baseline to 12 months compared to PD patients for sleep quality, social support, encouragement/support from staff, and overall health. Scores for other dimensions of the Kidney Disease Quality of Life and SF-12 questionnaires were not significantly different between the PD and HD groups. Conclusions: The results provide evidence that PD and HD patients have equivalent health-related quality of life in several domains, although the former performed better in some quality of life domains despite being older and having more comorbidities. Keywords: chronic disease, hemodialysis, peritoneal dialysis, quality of life. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. uating economic aspects of dialysis therapy, and aims to investigate the impact of PD and HD on HRQoL. Patients and Methods Study Population This study recruited PD and HD patients older than age 18 years who had been on the same dialysis modality for at least 1 month. Patients were identified from six dialysis centers in São Paulo, Brazil. Exclusion criteria included hospitalized patients at time of study initiation and patients who planned to transfer from their modality within six months. A similar number of PD and HD patients were selected. HD patients were randomly selected and matched to PD counterparts who started dialysis during the 9-month enrollment period, by sex, age group (⬍ 45 years, 45–59 years, and ⱖ 60 years), and dialysis center. At the beginning of the enrollment period, we assessed all patients receiving dialysis and selected new patients meeting the inclusion criteria who started dialysis during the 9-month enrollment period. All PD patients at each center meeting the inclusion criteria were se- Employee at Baxter Healthcare Corporation at the time of the study. Conflict of interest: The authors indicated that this study was sponsored by Baxter Healthcare Corporation. * Address Correspondence to: Mirhelen Mendes de Abreu, Federal University of São Carlos, Medicine Department, Av. Washington Luis, Km 235, SP 310, São Carlos, Brazil 13565-905. E-mail: [email protected] 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.016 S120 VALUE IN HEALTH 14 (2011) S119 –S121 lected, then, using a systematic random sampling method within the prespecified strata of sex and age, the HD patients were subsequently selected. This allowed us to reach the desired sample size of approximately 150 patients per group, which provided 80% power to detect a 10% or greater between-group difference in the mean scores of the quality of life dimensions. This was a prospective multicenter study with 1 year of followup. For the purpose of this article, only patients who completed 12 months of follow-up were included in the analysis. Assessment of HRQoL The SF-12 (mental component summary and physical component summary scores) and the Kidney Disease Quality of Life Short Form (KDQoL-SF) questionnaires were used to assess HRQoL. Originally, the KDQOL-SF is a self-reported measure that assesses the functioning and well-being of people with kidney disease and receiving dialysis [7]. The questionnaire consists of 80 items divided into 19 dimensions: SF-36 (eight dimensions/36 items): physical functioning (10 items), role limitations caused by physical problems (four items), role limitations caused by emotional problems (three items), pain (two items), general health perceptions (five items), social functioning (two items), emotional well-being (five items), energy/fatigue (four items), and one item about health status compared to 1 year ago; kidney-disease-targeted items (11 dimensions/43 items): symptom/problem list (12 items), effects of kidney disease (eight items), burden of kidney disease (four items), cognitive function (three items), quality of social interaction (three items), sexual function (two items), sleep (four items), social support (two items), work status (two items), overall health rating (one item scored separately), patient satisfaction (one item), and dialysis staff encouragement (two items). In this study, we used the SF-12, instead of SF-36. The SF-12 health survey is designed to be quick to use while retaining the validity of the parent SF-36 and the capacity to distinguish between the health of groups of subjects of different age and sex, and with different conditions. The loss of reliability associated with fewer defined health levels was regarded as an acceptable trade-off with practicality and length in the context of large group studies [8]. The scales range from 0 to 100, with a higher score representing better HRQoL [8,9]. From the specific KDQoL domains, we also derived a kidney disease component summary score [8,9]. Other Data Sources Medical, clinical, and laboratory data were collected at baseline by direct interview with the patient and their charts. All clinical and HRQoL data were collected at baseline, and at 6 and 12 months of follow-up. Statistical Analysis To compare differences in the means between HD and PD for the three time periods, t tests were used. Benjamini and Hochberg procedures were applied to evaluate changes in quality of life [10]. Clinically significant changes in quality of life for the individual domains were defined as a difference of ⫾ 5 points [10]. Clinically significant changes for the physical component scores and mental component scores are defined as ⫾ 5.7 points and ⫾ 6.3 points, respectively [10]. Potential independent variables included age category, sex, race, comorbities, lab values, comorbities (0 and 1 were dummy variables), dialysis modality (0 ⫽ PD, 1 ⫽ HD) and years receiving dialysis. A P value less than 0.05 was considered statistically significant and was retained in the final regression model. We used one multivariate regression to compare the influence of dialysis modalities on the quality of life domains for all three time periods and another to examine the impact of PD and HD on the change in quality of life scores from baseline to 12 months [10,11]. Adjusters used in the analysis included demographics, comorbidities, lab values (albumin and hemoglobin), time receiving dialysis (in years), and type of health insur- ance (public or private). The baseline score was included as an adjustor in the regression analyses evaluating the change in score from baseline to 12 months. Statistical analyses were conducted using SAS software, version 9.1.3, 2009 (SAS Institute Inc., Cary, NC). The study was approved by the Committee for Ethics in Research at the Federal University of São Paulo. Results Between April 2007 and February 2009, 249 HD patients and 228 PD patients were interviewed. 189 (76%) HD patients and 161 (71%) PD patients completed the 12-month study. Of the whole sample, 60 HD patients did not participate in the study due to the following reasons: 14 died, three received a transplant, six changed dialysis modality, 28 had less than 12 months of follow-up, four moved to a different dialysis center, and five were lost to follow-up. Among PD patients nonparticipation reasons were 12 died, three received a transplant, 17 changed dialysis modality, 26 had less than 12 months of follow-up, three moved to a different dialysis center, and six were lost to follow-up. Demographic and socioeconomic profiles of patients completing all three quality of life surveys are shown in Table 1 in Supplemental Materials found at: doi:10.1016/ j.jval.2011.05.016. PD patients were older (P ⫽ 0.009), had fewer years on dialysis (P ⫽ 0.003) and have higher rates of diabetes (P ⫽ 0.04). Quality of life scores for PD patients at baseline were significantly higher than for HD patients for several domains, including effects and burden of kidney disease, work status, encouragement from dialysis staff, patient satisfaction with care, and kidney disease component score. Analysis after 6 months showed that burden of kidney disease, encouragement/support from staff, and patient satisfaction with care scores remained significantly in favor of PD and, at the end of 12 months, encouragement/support from staff and patient satisfaction with care domains continued to be significantly higher in PD compared to HD patients (Table 2 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.016). A greater percent of patients receiving HD compared to PD had clinically significant improvements in HRQoL from baseline to 12 months (P ⫽ 0.004). Regression analyses showed that at baseline, burden of kidney disease, effects of kidney disease, dialysis staff encouragement, overall health rating, patient satisfaction with care, work status, and kidney disease component summary score were significantly better for the PD group. At 6 and 12 months, burden of kidney disease, staff encouragement/support, and patient satisfaction continued to be better in PD patients (Table 3 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.016). Finally, a further multivariate regression analysis regarding the change in quality of life from baseline to 12 months showed that HD significantly improved by 4.86 points (P ⫽ 0.0285) compared to PD on quality of social interaction. For patient satisfaction, PD patients improved by 4.85 points (P ⫽ 0.0275) compared to HD patients from baseline to 12 months. There was no other significant influence of dialysis modality on the other HRQoL domains. Discussion Dialysis care has increased in importance throughout the world because of the growing prevalence of patients receiving RRT and its related morbidity and mortality and high social and financial costs [12]. There is limited prospective evidence analyzing the burden of both dialysis modalities on HRQoL. This study aimed at describing HRQoL within the context of dialysis care. At baseline, both study groups were similar in sex, race, and several comorbidities, but PD patients were older, had higher rates of diabetes and hypertension, and were on dialysis for a shorter time period than HD patients. Although this could be considered a limitation of our study, these differences correspond to the reality of the VALUE IN HEALTH 14 (2011) S119 –S121 patients allocated to these modalities in Brazil [12]. The selection of the patients in this study intended to preserve their main characteristics observed in the general dialysis population in Brazil, namely the greater age and comorbidity in the PD group [12]. Therefore, patients’ matching between the groups at the enrollment was not too strict. In addition, an extensive multivariate linear regression analysis was performed, adjusting for a number of baseline characteristics. Patients receiving PD had better scores in several quality of life domains throughout the study period, even after adjusting for several factors, including age and comorbidity. However, considering that dialysis treatment is chronic, maybe these statistically significant results could need more time of observation to have better comprehension regarding the clinically significance results. Previous research has found little difference in the HRQoL between patients using HD and PD [5,12]. Liem et al. [5] conducted a systematic review on HRQoL of HD and PD, as well as transplant patients. They did not find statistically or clinically significant differences between dialysis modalities. So far, we know of no prospective study that looked at HRQoL in both dialysis modalities in a developing country. Because only 10% of all chronic dialysis patients in Brazil are receiving PD, we increased the representativeness of this group to better evaluate it [12]. HD is the most commonly used dialysis modality in many developing countries [12]. In Brazil, the reasons for the predominant allocation of patients on HD involve clinical, political, social, and economic factors [12]. Our findings show that several QOL dimensions were systematically better for PD patients during the follow-up, particularly burden of kidney disease, encouragement from staff, and satisfaction with care. Considering the clinically significant changes over time, HD had greater improvement in sleep, social support, and health status and did worse in cognitive status compared to PD. However, these results should be interpreted with caution because HD patients started with an overall lower quality of life evaluation. Literature on the responsiveness of different quality of life measures is growing, but this evidence is not available for all quality of life instruments [13]. To minimize potential floor and ceiling effects, the literature recommends the inclusion of the baseline quality of life measure as a covariate in the selected statistical model (as we did in this study), in which the contribution of the baseline to the follow-up scores is directly estimated. In general, this will result in a more powerful test of the treatment effect and an unbiased estimate of change [13]. In addition, we applied the Benjamin and Hochberg method, which is suitable for multiple comparisons [10]. From the methodologic perspective, we used the absolute change to analyze significant clinical changes between the groups over time [10]. Another limitation is that this is an observational study and some other confounding factors may not have been completely adjusted in the comparative analysis between PD and HD. Finally, the cases analyzed with complete follow-up corresponded to about 75% of the whole sample; nonetheless the patients not included in this report did not have significant differences in clinical and demographic parameters compared with those analyzed. A strength of this study is the prospective design, which has been rarely used in previous studies on this issue. As such, it shows how quality of life is subject to variation over time due to ageing, development of complications, changes in comorbidity, or due to a patient’s adjustment to the treatment. The HRQoL tools were applied at baseline, 6 months, and 12 months. For the purpose of this article, we analyzed patients who participated during the full study period. This study is a first step in the development of more comprehensive evidence of the HRQoL benefits of PD compared to HD in developing countries. Finally, this study was carried out in São Paulo, which is the main region for RRT in the country. Census data indicate that the distribution of demographic characteristics of São Paulo, com- S121 pared with other major metropolitan areas in Brazil, are similar [14]. Despite the mentioned limitations, this study contributes to the emergent investigation of quality of life and its consequences to decisions in clinical practice and heath policy in RRT. This prospective study provides evidence that quality of life is, in general, similar for PD and HD patients with some domains favoring PD. These findings could be used in the planning of health care services and patient management. Conclusions There was no difference between PD and HD patients’ scores on the SF-12 instrument. PD consistently performed better than HD on burden of kidney disease, patient satisfaction, and staff encouragement/ support domains. HD had greater clinical improvements on several variables but they had started lower at baseline and thus had more room for improvement. However, these differences are not clinically significant. Future analyses should compare incident HD and PD patients and use more sophisticated matching techniques. Source of financial support: Baxter Healthcare Corporation. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi:10.1016/j.jval.2011.05.016, or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). REFERENCES [1] Garratt A, Schmidt L, Mackintosh A, et al. Quality of life measurement: bibliographic study of patient assessed health outcome measures. BMJ 2002;324:1417–21. [2] Smith KW, Avis NE, Assmann SF. Distinguishing between quality of life and health status in quality of life research: a meta-analysis. Qual Life Res 1999;8:447–59. [3] Evans RW, Manninen DL, Garrison LP Jr, et al. The quality of life of patients with end-stage renal disease. N Engl J Med 1985;312:553–9. [4] Cameron JI, Whiteside C, Katz J, Devins GM. Differences in quality of life across renal replacement therapies: a meta-analytic comparison. Am J Kidney Dis 2000;35:629 –37. [5] Liem YS, Bosch JL, Arends LR, et al. Quality of life assessed with the Medical Outcomes Study Short-Form 36-Item Health Survey of patients on renal replacement therapy: a systematic review and metaanalysis. Value Health 2007;10:390 –97. [6] Sesso R, Lopes AA, Thomé FS, et al. Relatório do censo brasileiro de diálise [Brazilian dialysis survey report], 2009. J Bras Nefrol 2010;30:233–8. [7] Duarte PS, Ciconelli RM, Sesso R. Cultural adaptation and validation of the Kidney Disease and Quality of Life—Short Form (KDQOL-SF 1.3) in Brazil. Braz J Med Biol Res 2005;38:261–70. [8] Ware JE, Kosinski M, Keller SD. A 12-item short-form health survey. Med Care 1996;34:220 –3. [9] Ciconelli RM, Ferraz MB, Santos W, et al. Brazilian-Portuguese version of the SF-36: a reliable and valid quality of life outcome measure. J Rheumatol 1999;39:143–50. [10] Thissen D, Steinberg L, Kuang D. Quick and easy implementation of the Benjamini-Hochberg procedure for controlling the false positive rate in multiple comparisons. J Educ Behav Stat 2002;27:77– 83. [11] Zhang J, Quan H, Ng J, Stepanavage ME. Some statistical methods for multiple endpoints in clinical trials. Control Clin Trials 1997;18:204 –21. [12] Andrade MV, Junoy JP, Andrade EI, et al. Allocation of initial modality for renal replacement therapy in Brazil. Clin J Am Soc Nephrol 2010;5:637–44. [13] Guyatt GH, Osoba D, Wu AW, et al. Clinical Significance Consensus Meeting Group. Methods to explain the clinical significance of health status measures. Mayo Clin Proc 2002;77:371– 83. [14] Brazil travel. The history of Sao Paulo. Available from: http://lanic.utexas.edu/project/etext/llilas/outreach/fulbright07/kintz.pdf. [Assessed March 31, 2011]. VALUE IN HEALTH 14 (2011) S122–S125 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Health Status and Health Behaviors in Venezuelan Pharmacy Students Yajaira M. Bastardo, PhD* Facultad de Farmacia, Universidad Central de Venezuela, Caracas, Venezuela A B S T R A C T Objectives: The goals of this study were to assess the self-reported health status of Venezuelan pharmacy students, and to examine the association between self-reported health status and health behaviors in these individuals. Methods: A random sample 171 of pharmacy students, ranging in age from 18 to 35 years were surveyed using a written questionnaire. Health status was assessed using the Medical Outcomes Study Health Survey Short-Form 36 (SF-36). The sample consisted of 127 women and 44 men. The sample had a mean age of 22.3 ⫾ 2.71 years. The associations between health status and health behaviors were examined using both bivariate and multivariate models. The bivariate association was examined by t tests. Multiple regression analysis was used to model each SF-36 score separately using as independent variables sex, lack of regular exercise, regular smoking, and alcohol consumption. Results: The regression model explained between 6% and 12% of the variance in perceived health status. Controlling for other variables in the model, male students had significantly Introduction Perceived health is an integrated indicator of the subjective assessment of heath. Several demographic factors have been found to be associated with self-reported health status, including age, sex, and socioeconomic status [1– 4], which suggest that it could be a good surrogate marker for individual health. A perception of poor health functioning is an independent predictor of loss of functional ability and mortality [5]. Many of the studies on the predictive value of self-reported health status have been limited to vulnerable populations, such as the elderly [5]. It is only recently that attention has been paid to relatively young and healthy populations, such as university students [6 –9]. Years at university might be associated with considerable demands that could affect health and health status of individuals. While attending college, students face financial constraints, heavy study workloads, and adjustments to new social networks. In addition, during these years negative habits such as tobacco and alcohol consumption could be reinforced in this population. Although association between health behaviors and health status of university students has been reported in the literature, most of the studies are from developed countries; therefore, their findings are not necessarily applicable to Venezuela and other Latin American countries where students’ services; living arrangements; and social, cul- higher scores in bodily pain, general health, vitality, and social functioning than female students. Controlling for other variables in the model, lack of regular exercise was associated with significantly lower scores in physical functioning, bodily pain, and vitality; and regular smoking with significantly lower scores in physical functioning and general health. Controlling for other variables in the model, students who reported consuming alcohol had significantly higher scores in role-physical, bodily pain, and social functioning than students who did not report to consuming alcohol. Conclusions: This exploratory study demonstrates sex differences in health behaviors and perceived health status in pharmacy students. Health status is associated with several health behaviors in this sample of pharmacy students. Keywords: health behaviors, health status, SF-36, pharmacy students, Venezuela. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. tural, and political particularities are very different. To our knowledge, health status of Venezuelan pharmacy students has not been examined. The goals of this study were to 1) assess the self-reported health status in Venezuelan pharmacy students; and 2) examine the association between self-reported health status and health behaviors in these individuals. Methods Design This article describes an exploratory cross-sectional study. The study method was a survey using a written questionnaire. Participants The study population consisted of all undergraduate students enrolled full-time in the School of Pharmacy at the Central University of Venezuela, in Caracas. This is the largest of the three pharmacy schools in Venezuela. The other two schools of pharmacy in Venezuela are a small school in a private university in Caracas and a public school in the city of Mérida. Based on previous studies with the Medical Outcomes Study Health Survey Short-Form 36 (SF-36) Conflicts of interest: The author has indicated that she has no conflicts of interest with regard to the content of this article. * Address correspondence to: Yajaira M. Bastardo, Universidad Central de Venezuela, Facultad de Farmacia, Apto postal 40109 Nueva Granada, Caracas 1040, Venezuela. E-mail: [email protected] 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.020 VALUE IN HEALTH 14 (2011) S122–S125 in Venezuela [10], a targeted sample size of 150 was estimated. We assumed a 25% nonresponsive rate. A random sample of 200 students in the School of Pharmacy at the Central University of Venezuela was invited to participate in a study that consisted of completing a written health survey. Participation was voluntary and anonymous. No incentives were given for participation. Consent was assumed if the questionnaire was completed and returned. The survey was completed by 171 students, which represented a response rate of approximately 85.5%. Measures A survey instrument was used to assess the variables of interest in this study. The instrument, which was designed to take 10 minutes to complete, included the several scales described below. Demographics Demographics examined in this study included age, sex, and income. Age was measured as a continuous variable. Sex was coded as a dummy variable. Income was measured in three categories: less than two times the minimum wage, two to three times the minimum wage, and more than three times the minimum wage. S123 were asked an additional question about which medicines they were taking. Health Status Health status was assessed using the SF-36 [11]. The SF-36 is a generic health status questionnaire designed to assess eight health domains: physical functioning (10 items), role limitations due to physical (four items) or emotional (three items) health problems, bodily pain (two items), general health perceptions (five items), vitality (energy/fatigue) (four items), social functioning (two items), and mental health (five items). One additional item that is not scored asks respondents to compare current health with that of 1 year ago. The SF-36 is scored by summing item responses after reversing some items to ensure that a higher score always indicates better health-related quality of life. Raw scores are then linearly transformed to a 0 to 100 scale with 0 representing the lowest possible score and 100 the highest possible score [11]. The SF-36 has a high degree of acceptability and data quality. For this study a Spanish version of the SF-36 obtained from the Medical Outcomes Trust developed and validated by International Quality of Life Assessment project was used [12]. Data Analysis Reported Health Behaviors Three health behaviors were investigated in this analysis. Cigarette Smoking Cigarette smoking was assessed by asking the participant, “How frequently did you smoke cigarettes during the last 4 weeks?” The response categories were “never,” “twice or thrice a month,” “once or twice a week,” “almost daily,” and “daily.” Participants were defined as regular smokers if they reported smoking almost daily or daily. Alcohol Consumption Alcohol consumption was assessed by asking the participant, “How frequently did you have an alcoholic beverage during the last 4 weeks?” The response categories were “never,” “once a month,” “twice or thrice a month,” “once or twice a week,” and “almost daily.” Participants were defined as alcohol consumers if they reported having an alcohol beverage during the previous month. Descriptive statistics are presented for all the variables. Categorical variables are presented as percentages. Range, mean, and standard deviation were computed for continuous variables. Chi-square was used to map out major differences in male and female students’ health behaviors. The internal consistency of each multi-item SF-36 scale was evaluated using Cronbach’s coefficient alpha. According to Nunnally [13], alpha levels between 0.65 and 0.80 are acceptable for group comparisons. A coefficient alpha of 0.70 was set a priori as the goal for all scales of our study. The extent to which SF-36 scores differed in participants grouped by sex was examined. The association between health status and health behaviors was examined using both bivariate and multivariate models. The bivariate association was examined by t tests. To determine which variables were associated with different concepts of health status controlling for other variables in the model, multiple regression analysis was used to model each SF-36 score separately using as independent variables sex, lack of regular exercise, regular smoking, and alcohol consumption. All data analyses were performed using SSPS for Windows version 10.0 (1999, SPSS Inc., Chicago, IL). A priori, for all tests the significance level was set at P ⱕ 0.05. Lack of Regular Exercise Lack of regular exercise was assessed by asking the participant, “How frequently did you exercise or carried out any sport over the last 4 weeks?” The response categories were “never,” “once a month,” “twice or thrice a month,” “once or twice a week,” and “almost daily.” Lack of regular exercise was defined as exercising less than once a week during the past month. Illness Illness was assessed by asking whether or not the participant was experiencing an illness. Respondents were coded as one if they reported having an illness. They were coded as zero if they answered otherwise. Respondents who reported having an illness were asked an additional question about their illness. Medicines Use Medicines use was assessed by asking whether or not the participant was taking medicines. Respondents were coded as one if they reported taking medicines. They were coded as zero if they answered otherwise. Respondents who reported taking medicines Results Description of the Sample The sample consisted of 127 women and 44 men. The sample had a mean age of 22.3 ⫾ 2.71 years, (range 18 –35 years). The mean age did not differ among the men and women (t test t ⫽ 1.67; P ⫽ 0.097). The sample had relatively high income: 47.8% reported an income more than three times the minimum wage, 23.9% reported an income between two to three times the minimum wage, and 28.3% reported an income less than two times the minimum wage. As expected, the sample was healthy; only 30 subjects (17.5%) reported having an illness. Forty subjects (23.4%) reported current medication use; of those, 11 (27.5%) were taking oral contraceptives and eight (20%) were taking vitamins. Health Behaviors Of the students surveyed, 113 (66.1%) had at least one drink in the previous month, 106 (62%) did not exercise regularly, and 18 S124 VALUE IN HEALTH 14 (2011) S122–S125 (10.5%) smoked regularly during the last month. There were differences in certain health behaviors between genders. Male students were more likely to consume alcohol (84.1% vs. 59.8%; chisquare 19.568; P ⫽ 0.001) and to exercise regularly (65.9% vs. 28.3%; chi-square 8.573; P ⫽ 0.002) than female students. There was no difference in smoking behavior (13.6% vs. 9.4%; chi-square 0.608; P ⫽ 0.301) between male and female students. Health Status The internal consistency reliability of the SF-36 was high. The Cronbach’s alpha coefficients ranged from 0.72-0.89. Means of each SF-36 score grouped by sex and health behaviors are shown in Tables 1 and 2 in Supplemental Materials found at: doi:10.1016/ j.jval.2011.05.020. Bivariate Analysis In bivariate analyses, sex was significantly associated with all SF-36 domains. Lack of regular exercise was significantly negatively associated with physical functioning, vitality, and mental health. Regular smoking was significantly negatively associated with physical functioning and general health. Alcohol consumption was significantly positively associated with role-physical, bodily pain, vitality, and social functioning. Multiple Regression Analysis Multiple regression analysis was used to model each SF-36 score separately using as independent variables sex, lack of regular exercise, regular smoking, and alcohol consumption. Table 3 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.020 shows the  coefficients obtained for each independent variable and the adjusted R2 values obtained for each regression model, which is a measure of the proportion of variance of the dependent variable accounted for by the regression model. Except role-emotional, all the models were significant (F values from 3.51– 6.71). The regression models explained between 6% and 12% of the variance in the different SF-36 domains. Controlling for other variables in the model, male students had significantly higher scores in bodily pain, general health, vitality, and social functioning than female students. Controlling for other variables in the model, lack of regular exercise was associated with significantly lower scores in physical functioning, bodily pain, and vitality; and regular smoking with significantly lower scores in physical functioning and general health. Controlling for other variables in the model, students who reported consuming alcohol had significantly higher scores in role-physical, bodily pain, and social functioning than students who did not report to consuming alcohol. Discussion The aim of this study was to describe health status of Venezuelan pharmacy students, and to examine the association between health behaviors and health status in this population. Alhough the measure of lack of regular exercise used herein is slightly different in term of time frame from other studies, our results are comparable to those reported previously by Steptoe and Wardle [6] for university students in Western Europe and Martins Bion et al. [14] in Brazil. As found in previous studies, male students were more likely to exercise regularly than female students in our sample [8,15]. Drinking and smoking behaviors found in this study do not agree with findings from other studies. The students in our sample tended to report lower levels of drinking (66.1%) and smoking (10.5%) than university students in Europe [6,8] and the United States [7]. Compared to other studies in South America, the alcohol use in our sample was very similar to that reported by Passos et al. [16] for medical students in four public universities in Rio de Janeiro and lower than reported by Pillon et al. [17] for first-year university students in Brazil. As found in other studies [8,16,17], male students were more frequent drinkers than female students. Tobacco consumption in our sample was considerably lower than reported in other studies in South America [18,19]. Contrary to other studies [8], there was not a difference in smoking behavior between male and female students in our sample. The overall health status of pharmacy students in Venezuela was good. In the bivariate analysis, for all health concepts, male students scored significantly higher than female students. Even controlling for other variables in the model, male students had significantly higher scores in bodily pain, general health, vitality, and social functioning than female students. These findings agree with previous literature showing sex differences in health status [2,8]. The data from this exploratory study do not shed light on the factors that may contribute to female students’ lower ratings in self-reported health, but warrant further investigation to explain why, even in a relatively privileged group of the society, sex differences of such magnitude persist. Representatives from both sexes scored higher in health concepts closely related with physical health, which was expected in this relatively young sample. The scores in vitality and mental health were considerably low, which suggest some burnout in our sample that should be further investigated. University students are under the influence of several stressors with unknown consequences to their health [8]. In one study that examined stress in health professions students, pharmacy students were found to have more stress and distress than medical and dental students [20]. This exploratory study has identified an important health issue that requires attention from the Student Well-Being Center and deserves consideration by health professionals who provide care to this population. The health of the students must be an important public health concern for the society. Because of this, it is important to understand better the relation of self-reported health status and health behaviors in Venezuelan university students to ensure health promotion strategies well aligned to their needs. In addition, we found significant associations between some health concepts and health behaviors in bivariate and multivariate models. Not surprisingly, in both models, regular exercise was positively associated with physical functioning and vitality. As expected, regular smoking was negatively associated with physical functioning and general health. Surprisingly, some SF-36 health concepts were significantly positively associated with alcohol consumption. These findings suggest an optimism bias in our sample. In addition, alcohol consumption was associated with sex, which could confound the results in our sample. There are limitations of this study that should be considered when evaluating the results. First, the study used a random sample of pharmacy students attending the Central University of Venezuela in Caracas. Thus, the results of this study may not be generalized to pharmacy students from the other two universities in Venezuela that offer a pharmacy program. Second, although the study showed significant associations between health status and some health behaviors studied, its cross-sectional design does not allow us to draw conclusions regarding the direction of the relationships or causality. For example, we do not know if regular smokers in this sample truly have poorer health or if they rate their health poorer because they are aware that smoking is a bad health habit that could affect their health. Despite these limitations, this exploratory study is the first in Venezuela that used the SF-36 in examining health status in a relatively young and healthy Venezuelan population. VALUE IN HEALTH 14 (2011) S122–S125 Conclusions This exploratory study demonstrates sex differences in health behaviors and perceived health status in pharmacy students. Health status is associated with several health behaviors in this sample of pharmacy students. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi:10.1016/j.jval.2011.05.020, or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). REFERENCES [1] Idler EL. Age differences in self-assessment of health: age changes, cohort differences, or survivorship? J Gerontol 1993;48:S289 –300. [2] Dunn JR, Walker JD, Graham J, et al. Gender differences in the relationship between housing, socioeconomic status, and selfreported health status. Rev Environ Health 2004;19:177–95. [3] Hemingway H, Nicholson A, Stafford M, et al. The impact of socioeconomic status on health functioning as assessed by the SF-36 questionnaire: the Whitehall II Study. Am J Public Health 1997;87:1484 –90. [4] Damian J, Ruigomez A, Pastor V, Martin-Moreno JM. Determinants of self assessed health among Spanish older people living at home. J Epidemiol Comm Health 1999;53:412–16. [5] Mossey JM, Shapiro E. Self-rated health: a predictor of mortality among the elderly. Am J Public Health 1982;72:800 – 8. [6] Steptoe A, Wardle J. Health behaviour, risk awareness and emotional well-being in students from Eastern Europe and Western Europe. Soc Sci Med 2001;53:1621–30. [7] Prokhorov AV, Warneke C, de Moor C, et al. Self-reported health status, health vulnerability, and smoking behavior in college students: Implications for intervention. Nicotine Tob Res 2003;5:545–52. [8] Vaez M, Laflamme L. Health Behaviors, self-rated health and quality of life at university: a study among Swedish first-year university students. J Am Coll Health 2003;51:156 – 62. S125 [9] Saupe R, Nietche EA, Cestari ME, et al. Qualidade de vida dos acadêmicos de enfermagem. Rev Latino-am Enfermagem 2004;12: 636 – 42. [10] Bastardo YM, Kimberlin CL. Relationship between quality of life and social support in HIV-infected persons in Venezuela. AIDS Care 2000; 12:673– 84. [11] Ware JE, Snow KK, Kosinski M, et al. SF-36 Health Survey. Manual & Interpretation Guide. Boston: New England Medical Center Health Institute, 1993. [12] Alonso J, Prieto L, Antó JM. La versión española del SF-36 Health Survey (Cuestionario de Salud SF-36): Un instrumento para la medida de los resultados clínicos [The Spanish version of the SF-36 Health Survey (the SF-36 health questionnaire): an instrument for measuring clinical results]. Med Clín (Barcelona) 1995;104:771– 6. [13] Nunnally JC. Psychometric Theory (2nd ed.). New York: McGraw-Hill, 1978. [14] Martins Bion F, de Castro Chagas MH, de Santana Muniz G, et al. Estado nutricional, medidas antropométricas, nivel socioeconómico y actividad física en universitarios brasileños [Nutritional status, anthropometrical measurements, socio-economic status, and physical activity in Brazilian university students]. Nutr Hosp 2008; 23:234 – 41. [15] Dos Santos Ferreira da Silva G, Bergamaschine R, Rosa M, et al. Avaliação do nível de atividade física de estudantes de graduação das áreas saúde/biológica [Evaluation of the level of physical activity of graduate students of health/biological areas]. Rev Bras Med Esporte 2007:13:39 – 42. [16] Passos S. Brasil P, Santos M, Aquino MT. Prevalence of psychoactive drug use among medical students in Rio de Janeiro. Soc Psychiatry Epidemiol 2006;41:989 –96. [17] Pillon S, O’Brein B, Chavez K. The relationship between drug use and risk behaviors in Brazilian university students. Rev Latino-am Enfermagem 2005;13(Spec):1169 –76. [18] Prat-Marin A, Fuentes-Almendras MM, Sanz-Gallen R, et al. Epidemiología del tabaquismo en los estudiantes de ciencias de la salud [Epidemiology of tobacco use in health-science students]. Rev Saude Publica 1994:28:100 – 6. [19] Danjoy León D, Ferreira P, Pillon SC. Conocimientos y prácticas sobre el consumo de tabaco en estudiantes de pregrado de farmacia, Lima, Perú [Knowledge and practice regarding tobacco use among pharmacy undergraduate students in Lima, Peru]. Rev Latino-am Enfermagem 2010;18(Spec):582– 8. [20] Henning K, Ey S, Shaw D. Perfectionism, the imposter phenomenon and psychological adjustment in medical, dental, nursing, and pharmacy students. Med Educ 1998;32:456 – 64. VALUE IN HEALTH 14 (2011) S126 –S129 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Influence of Organic and Functional Dyspepsia on Work Productivity: The HEROES-DIP Study Guilherme Becker Sander, MD, MsC1,*, Luiz Edmundo Mazzoleni, MD, PhD2, Carlos Ferrnando de Magalhães Francesconi, MD, PhD2, Giácomo Balbinotto, PhD3, Felipe Mazzoleni, MD4, Andre Castagna Wortmann, MD, MsC5, Israel de Quadros Cardoso6, Alexandre Luis Klamt, MD7, Tobias Cancian Milbradt8 on behalf of the Helicobacter Eradication Relief of Dyspetic Symptoms Trial Investigators 1 Universidade Federal do Rio Grande do Sul, School of Medicine, Programa de Pós-graduação: Ciências em Gastroenterologia e Hepatologia, Porto Alegre, Brazil; Universidade Federal do Rio Grande do Sul, School of Medicine, and Hospital de Clinicas de Porto Alegre, Department of Gastroenterology, Porto Alegre, Brazil; 3 Universidade Federal do Rio Grande do Sul, Faculty of Economics, Porto Alegre, Brazil; 4Hospital de Clinicas de Porto Alegre, Department of Internal Medicine, Porto Alegre, Brazil; 5Universidade de Caxias do Sul, School of Medicine, Porto Alegre, Brazil; 6Universidade Federal do Rio Grande do Sul, School of Medicine, Porto Alegre, Brazil; 7Hospital de Clinicas de Porto Alegre, Department of Gastroenterology, Porto Alegre, Brazil; 8Universidade Federal do Rio Grande do Sul, School of Medicine, Programa de Pós-graduação: Ciências Médicas, Porto Alegre, Brazil 2 A B S T R A C T Objectives: Dyspepsia is defined as persistent or recurrent abdominal pain or discomfort centered in the upper abdomen. Dyspepsia represents up to 8.3% of all primary care physician visits and causes huge economic costs to patients and to the economy as a whole. The aim of this study was to measure the influence of dyspepsia on work productivity of people within the Brazilian workforce. Methods: Adult patients were enrolled if they met the Roma III criteria for uninvestigated dyspepsia. All patients answered a demographic questionnaire. Productivity impairment was measured by the Work Productivity and Activity Impairment questionnaire. Subjects underwent upper gastrointestinal endoscopy and were classified as having functional or organic dyspepsia. The study protocol was approved by the Ethics Committee of Hospital de Clínicas de Porto Alegre, Brazil. Results: Eight hundred fifty patients with dyspepsia were evaluated: 628 were women (73.9%); mean age was 46.4 ⫾ 12.9 years; 387 (45.5%) were active workers. Introduction Symptoms of dyspepsia affect up to 40% of the adult population in the Western world [1–3]. According to the Rome III Consensus, diagnosis of dyspepsia is established if patients report at least one episode per week of epigastric pain or discomfort (postprandial fullness or early satiety) during the past three months. Symptoms must be present for more than six months. Most affected individuals do not have structural or biochemical abnormalities that can explain their symptoms and thus are classified as having functional or nonulcer dyspepsia [4,5]. The most common organic causes are gastroesophageal reflux disease, peptic ulcer disease, and gastric cancer. Another important category is uninvestigated dyspepsia, which consists of dyspeptic symptoms previous to diagnostic investigation. Dyspepsia accounts for 8.3% of visits to primary care physicians [6], although more than half of these patients do not seek Among active workers, 32.2% mentioned that dyspepsia had caused absenteeism from work during the preceding week and 78% reported a reduction of the work productivity (presenteeism). The lost work productivity score was 35.7% among all employed patients. The affect on work productivity was similar between patients with functional or organic dyspepsia. Conclusions: Our study showed an important influence of dyspepsia on work productivity. We did not find any statistically significant difference on the influence on work between patients with organic dyspepsia and functional dyspepsia. The social impact of these findings is underscored by taking into account the prevalence (up to 40%) of this condition in Brazil. Keywords: dyspepsia, esophagitis, peptic ulcer, productivity. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. medical assistance [7]. Drugs prescribed predominantly for dyspepsia made up 7% of the primary care prescribing budget in the United Kingdom in 2004. The indirect costs are even more impressive. The reduction in quality of life is similar to that seen in patients with mild heart failure [8]. Dyspepsia can be severe enough to cause sick leave from work (absenteeism), as well as jeopardize productivity while working (presenteeism). In Sweden, a study [7] estimated the influence of dyspepsia, gastroesophageal reflux disease, and peptic ulcer disease on the budget of that country. Using a top-down approach, the study found that 1.9% of all costs from sick leave from work in Sweden was attributed to dyspepsia, corresponding to US$144 million in 1997. The coding system was not able to differentiate among dyspepsia, peptic ulcer disease, and gastroesophageal reflux disease as the cause of the absenteeism [7]. Another Swedish study [9] found that the average functional dyspepsia patient was responsible for 26 more days of lost production than the average employee without functional dyspepsia. In Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Guilherme Becker Sander, Av Bage, 919/202, Porto Alegre - Rio Grande do Sul, 90460-080 Brazil. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.021 VALUE IN HEALTH 14 (2011) S126 –S129 the Leeds HELP study [8], conducted in the United Kingdom, 2% of 8473 randomly selected primary-care patients reported taking time off work because of dyspepsia, costing society annually £17.91 per person. Brook et al. [10] objectively measured the productivity of blue-collar workers and found that employees with functional dyspepsia produced 12% fewer units per hour than controls. A prospective study of dyspepsia from Southern General Hospital in Glasgow, Scotland [11], showed that each employed patient with functional dyspepsia loses on average 18.3 weeks of work per year. Although dyspepsia is very prevalent, the effect of dyspepsia on work absenteeism and presenteeism has not been assessed with validated tools. The influence of gastroesophageal reflux disease on work has been extensively studied, showing significant effects in patients with moderate to severe disease [12–14], but there is a lack of similar information for patients with functional dyspepsia. The aim of this study was to evaluate the influence of dyspepsia on work absenteeism, presenteeism, and productivity in daily life and to explore effect modifiers for the relationship between dyspepsia and productivity costs such as demographic data, severity of symptoms, and etiologic categories of dyspepsia. Study design This was a secondary study nested in the screening phase of the Helicobacter Eradication Relief of Dyspeptic Symptoms (HEROES) trial. HEROES trial was a randomized double-blind placebo-controlled clinical trial (ClinicalTrials.gov No. NCT00404534) carried out in a single academic hospital, the Hospital de Clínicas de Porto Alegre in Rio Grande do Sul, Brazil. The protocol was approved by the local institutional review board. Written informed consent was obtained from all patients before enrollment. Patients Primary care patients were recruited through newspaper, radio, and television advertising as well as through primary care clinic referrals. Patients of either sex were enrolled in the study if they were aged 18 years or older and had a diagnosis of dyspepsia according to the Roma III criteria. The symptoms must have been present for more than six months, with at least one episode per week of epigastric pain or burning or discomfort (postprandial fullness or early satiety) during the past three months. Exclusion criteria included predominant symptoms of heartburn or irritable bowel syndrome; alarm symptoms (weight loss ⬎ 10% of the previous weight, anemia, bleeding, or positive physical examination findings suggesting organic disease); history of peptic ulcer, upper gastrointestinal surgery, or biliary colic; relevant comorbidities; and alcohol or drug abuse. We also excluded patients unable to answer the study questionnaires. Outcomes measures Data on health-related reduction in work absenteeism and productivity while at work and when performing daily activities were obtained using the generic version of the Work Productivity and Activity Impairment (WPAI) Questionnaire. The WPAI has six questions regarding if the person is currently employed, questions regarding hours absent from work for health reasons, hours absent from work for other reasons, and number of hours worked and questions evaluating the influence of health problems on productivity at work and at daily activities. It has a validated Brazilian Portuguese version shown to be a reliable and valid tool to measure the influence of health problems on productivity [15]. The dyspeptic symptoms score was evaluated by trained investigators using the Porto Alegre Dyspeptic Symptoms Questionnaire (PADYQ) [16,17]. This 11-question instrument assesses the frequency, duration, and intensity of dyspepsia symptoms during the S127 preceding 30 days. Score ranges from 0 (no symptoms) to 44 (severe symptoms) and were validated in Brazilian Portuguese, showing high levels of internal consistency, reproducibility, responsiveness, face validity, discriminant validity, and concurrent validity. Study procedures Patients self-completed the WPAI questionnaire during the screening visit of the HEROES trial. After that, patients consulted with a trained investigator and provided demographic and clinical data, including information for the PADYQ questionnaire. Upper gastrointestinal endoscopies were performed at screening by two endoscopists. According to endoscopic findings, patients were classified into two etiologic subgroups: organic dyspepsia (if reflux esophagitis, peptic ulcer, celiac disease, malignancy, or achalasia was diagnosed) or functional dyspepsia. Statistical methods WPAI answers were computed to produce four scores: percentage of work missed for dyspepsia symptoms (absenteeism); percentage impairment at work for dyspepsia symptoms (presenteeism); the lost work productivity score (summarizing absenteeism and presenteeism) and impairment of daily activities due to dyspepsia. Absenteeism was calculated as the number of hours absent from work divided by the total work hours (total work hours ⫽ hours absent from work ⫹ hours actually worked) multiplied by 100, whereas presenteeism was calculated as the number of hours worked multiplied by the percentage reduction in productivity. The lost work productivity score was calculated as the sum of hours absent from work and reduced productivity at work (presenteeism) divided by the total work hours multiplied by 100. The percentage reduction in daily activities was used to measure impairment of daily activities due health related problems. The lost work productivity score was transformed in monetary values multiplying the score by the median income per capita, published by the Brazilian Institute of Geography and Statistics. The values are shown in American dollars, using the exchange rate of July 1, 2010 ($1 ⫽ 1.8 BRL). Continuous variables were expressed as means and standard deviations and were analyzed using the t test for independent samples. Qualitative variables were expressed as percentages and were compared using Fisher’s exact test. All two-tailed P values less than 0.05 were considered statistically significant. All variables that had P values that were less than 0.20 on univariate analysis were included in a forward stepwise logistic regression to assess their affect on the absenteeism, presenteeism, and daily activities. Correlation of dyspepsia and influence on work productivity was calculated with Spearman’s rank correlation coefficient. All analyses were performed using PASW Statistics (version 18.0. 2009, SPSS Inc., Chicago, IL). Results Patients Overall, 1151 screening visits were performed, and 850 patients fulfilled Rome III criteria for uninvestigated dyspepsia and were included in the study. Sixteen patients refused to undergo endoscopy and could not have their underlying etiology defined. The mean age of the included patients was 46.4 ⫾ 12.9 years and 73.9% were women. Overall, 45.5% (387 out of 850) of the study population was employed. Among patients with functional dyspepsia and organic dyspepsia 48.2% and 44.4% were employed, respectively (P ⫽ 0.37). Other demographic data are shown in Table 1. S128 VALUE IN HEALTH 14 (2011) S126 –S129 Table 1 – Demographic and clinical characteristics of Brazilian patients with uninvestigated dyspepsia. Variable Uninvestigated dyspepsia (N ⫽ 850) Age (mean y ⫾ SD) Range ⬎40 y Woman White race Education ⬎ 10 y Smoking status Never smoked Current/former Alcohol consumption No consumption Current/former Coffee drinkers Duration of dyspepsia ⬎ 5 y Etiology of dyspepsia* Organic Functional Family income ⬍ $500/month $500–$1000/month ⬎$1000/month 46.4 ⫾ 12.9 18–82 596 (70.1) 628 (73.9) 663 (78.1) 386 (45.4) 476 (56) 374 (43) 700 (82.3) 150 (17.7) 567 (66.7) 411 (48.4) 197 (23.2) 637 (74.9) 291 (34.1) 313 (36.8) 246 (29) Note. Data are No. (%) unless otherwise indicated. * Available data for 834 out of 850 subjects (98.1%). Impact of dyspepsia on productivity Among employed patients, 32% (124 out of 387) reported lost hours of work in preceding week because of dyspepsia (absenteeism). The average number of work hours lost due to dyspepsia-related absenteeism was 2.63 ⫾ 6.95 hours per week among all employed patients. The percentage of work missed because of dyspepsia (absenteeism) was 8.12%. Among employed patients, 78% (302 out of 387) report reduced productivity while working, reporting 34.7% less productivity in average, whereas mean presenteeism was 12.1 ⫾ 13.8 hours per week. The lost work productivity score was 35.7% among all employed patients. A summary of the results is shown in Table 2. Effects of functional dyspepsia versus organic dyspepsia on productivity Among employed patients, 30.9% of patients with functional dyspepsia reported absenteeism versus 34% of organic dyspepsia patients (P ⫽ 0.609). The mean number of lost hours was 2.34 ⫾ 6.32 per week for patients with functional dyspepsia versus 3.05 ⫾ 7.13 hours per week for patients with organic dyspepsia (P ⫽ 0.36). In the same sample, 77% of patients with functional dyspepsia reported presenteeism versus 79.8% of patients with organic dyspepsia (P ⫽ 0.77). There were no differences of lost work productivity score between patients with functional and organic dyspepsia (P ⫽ 0.41, data not shown). Correlation of dyspepsia intensity and productivity loss There was no difference in the proportion of those reporting absenteeism between those with lower and higher PADYQ scores (29.7% vs. 32.7%; P ⫽ 0.74), but higher rates of presenteeism was reported by those with higher scores (62.5% vs. 83.2%; P ⫽ 0.003). The mean loss of productivity while working was 23.6% versus 37.3%, respectively (P ⫽ 0.001). A weak positive linear correlation was noted between dyspepsia score and lost work productivity score, with an r value of 0.21 (P ⫽ 0.006). Monetary value Considering that the gross domestic product per capita in Brazil crossed the 10,000 dollars/year barrier recently, is estimated that the total productivity impairment (absenteeism ⫹ presenteeism) is costing US$3570 annually, or US$297.50 monthly per employed dyspeptic patient. The cost of absenteeism was US$20.97 monthly and the cost of presenteeism US$276.52 monthly (Table 2). Sensitivity analysis Sensitivity analysis of absenteeism was performed considering the data about absence compensating mechanism reported by Severens et al. [18]. According to this study, 75% of the absenteeism caused by dyspepsia is compensated (mainly by colleagues during normal working hours). The absenteeism cost considering this compensation was calculated to be US$5.24 per month. Because dyspeptia symptoms are intermittent, we performed a conservative sensitivity analysis of presenteeism taking in account that 15% of the time a patient is experiencing dyspepsia. Fifteen percent was chosen because it is the minimum frequency need to fulfill the Rome III criteria for dyspepsia (1 day/week). In this sensitivity analysis, the presenteeism cost is reduced to US$41.48 per month. In this conservative analysis, the total productivity loss due to dyspepsia falls to US$46.72 monthly or 4.95% of total productivity of employees with functional dyspepsia. Influence of dyspepsia on daily life Reduced productivity while carrying out daily life activities was also considerable. Overall, 83.8% (729 out of 850) of all patients reported reduced productivity during daily life (Table 2). The mean reduction was 40.5% ⫾ 29.03%. The influence of dyspepsia on daily life was higher among unemployed (43.0% ⫾ 29.5%) compared to employed patients (37.6% ⫾ 26.2%; P ⫽ 0.007). Multivariate analysis To examine the influence that several potentially important prognostic factors have on absenteeism, presenteeism, or daily life activities, the following factors were first examined individually by univariate analysis: sex, age, race, etiology of dyspepsia (organic or functional), years of education, familiar income, basal PADYQ score (ⱕ 20 vs. ⬎ 20), type of dyspepsia (ulcer-like or dismotility), coffee and alcohol consumption, smoking, and duration Table 2 – Summary of results. Variable Effect on work productivity Active workers, n (%) Patients that reported absenteeism, n (%) Number of work hours lost, (mean ⫾ SD) Patients that reported presenteeism, n (%) Lost productivity in average, (%) Lost work productivity score, (%) Monetary value Brazilian GDP per capita Absenteeism monthly cost per patient* Presenteeism monthly cost per patient† Impact of dyspepsia on daily life Patients who reported impact on daily life, n (%) Results 387 (45.5) 124 (32) 2.63 ⫾ 6.95 302 (78) 34.7 35.7 US$10.000 US$5.24–20.97 US$41.48–297.50 729 (83.8) GDP, Gross domestic product. * Sensitivity analysis of absenteeism was performed considering 75% of the absenteeism in dyspepsia is compensated [18]. † Sensitivity analysis of presenteeism was performed considering dyspepsia is intermittent. VALUE IN HEALTH 14 (2011) S126 –S129 of dyspepsia. Sex, age, years of education, duration of dyspepsia, basal PADYQ score, and coffee consumption were included in the multivariate analysis. Being a woman (P ⫽ 0.044) and having a higher dyspepsia score (P ⫽ 0.013) were retained in the final stepwise logistic regression analysis model as associated with daily life activity. Age younger than 40 years (P ⫽ 0.047) and higher dyspepsia score (P ⫽ 0.002) were associated with presenteeism. No variable was associated with absenteeism in multivariate analysis. S129 used the bottom limit of the dyspepsia definition for frequency to perform this calculation, but we deliberately did so to show the financial effects of dyspepsia in an unquestionable manner. Conclusions Our study shows that either organic or functional dyspepsia has a large influence on the productivity of Brazilian workers. Cost-effective measures of prevention, investigation, and treatment of dyspepsia should be prioritized in the planning of health care. Discussion This study shows that dyspepsia has a notable affect on work productivity and daily life activities. Our study analysis suggests that each employed person with dyspepsia has lost 35.7% of his or her potential productivity, considering absenteeism and presenteeism. Even the most conservative sensitivity analysis shows a reduction of productivity of 4.95%. In Brazil, the prevalence of dyspepsia symptoms is 40%, accordingly a population-based transversal study [19]. This means that in the best case scenario 1.98% of the national workforce is compromised due to dyspepsia symptoms. The best scenario estimated an effect on the Brazilian gross domestic product of US$30 billion. We also observed a substantial reduction of productivity while carrying out daily life activities. Functional dyspepsia has been neglected, in part, by the absence of associated mortality and absence of financial interests. Our data show that there is no difference between the effects on work productivity of patients with functional or organic dyspepsia. Because functional dyspepsia is three times more prevalent than organic dyspepsia, the lack of attention that functional dyspepsia has received from health authorities is conceivable. In Brazil, there are no clinical protocols of care aimed at the treatment of functional dyspepsia. Furthermore, endoscopic procedures are difficult to access and are low-paying procedures. Brook et al. [10] showed that employees with functional dyspepsia have an additional 0.83 absence days per year. This value is well below what we found. It is worth considering that the methodology of Brook et al.’s study [10] had sensitivity only to identify full-day absence from work. We believe that patients with symptoms of dyspepsia more often have to leave work early, but hardly need to miss the entire day. Brook et al. [10] also showed a reduced daily productivity of blue-collar workers of 12%. Our study analysis shows a presenteeism of 34%. These two results are comparable. Lerner et al. [20] showed a 2:1 relationship between illnessrelated self-reported productivity loss while at work and objectively measured productivity loss. Furthermore, we believe that the effects of dyspepsia in white-collar workers may be even greater than shown in blue-collar workers because the loss in both ability to focus and creativity are the main complaints reported. Surprisingly, the loss of productivity in daily activities was higher among unemployed persons. This may mean greater attention to professional activities by those who are employed, or that, for some, the reason for being unemployed may be related to more severe dyspepsia. Our study has some limitations. Half of patients with dyspepsia had never consulted a doctor, and their symptoms may have less impact on productivity at work than the population studied by us. The WPAI has a visual analogue scale to measure presenteeism and this kind of scale is susceptible to measurement biases such as spacing-out bias and end-aversion bias. Furthermore, Lerner et al. [20] found that for each 10% increase in self-reported work limitation, there is 4% to 5% reduction in objectively measured productivity. The sensitivity analysis we conducted was very conservative and directed precisely to minimize this potential bias. Probably our results underestimate presenteeism because we Acknowledgment A gastroscope was obtained through an unrestricted donation from Aché Laboratórios Farmacêuticos. REFERENCES [1] Jones RH, Lydeard SE, Hobbs FD, et al. Dyspepsia in England and Scotland. Gut 1990;31:401–5. [2] Talley NJ, Zinsmeister AR, Schleck CD, Melton LJ, 3rd. Dyspepsia and dyspepsia subgroups: a population-based study. Gastroenterology 1992;102:1259 – 68. [3] Olmos JA, Pogorelsky V, Tobal F, et al. Uninvestigated dyspepsia in Latin America: a population-based study. Dig Dis Sci 2006;51:1922–9. [4] Tack J, Talley NJ, Camilleri M, et al. Functional gastroduodenal disorders. Gastroenterology 2006;130:1466 –79. [5] Talley NJ, Stanghellini V, Heading RC, et al. Functional gastroduodenal disorders. Gut 1999;45 (Suppl. 2):II37– 42. [6] Maconi G, Tosetti C, Stanghellini V, et al. Dyspeptic symptoms in primary care. An observational study in general practice. Eur J Gastroenterol Hepatol 2002;14:985–90. [7] Agreus L, Borgquist L. The cost of gastro-oesophageal reflux disease, dyspepsia and peptic ulcer disease in Sweden. Pharmacoeconomics 2002;20:347–55. [8] Moayyedi P, Mason J. Clinical and economic consequences of dyspepsia in the community. Gut 2002;50(Suppl. 4):iv10 –2. [9] Nyren O, Adami HO, Gustavsson S, et al. Social and economic effects of non-ulcer dyspepsia. Scand J Gastroenterol 1985;109 (Suppl.):41–7. [10] Brook RA, Kleinman NL, Choung RS, et al. Functional dyspepsia impacts absenteeism and direct and indirect costs. Clin Gastroenterol Hepatol 2010;8:498 –503. [11] Crean GP, Holden RJ, Knill-Jones RP, et al. A database on dyspepsia. Gut 1994;35:191–202. [12] Wahlqvist P, Carlsson J, Stalhammar NO, Wiklund I. Validity of a Work Productivity and Activity Impairment questionnaire for patients with symptoms of gastro-esophageal reflux disease (WPAI-GERD)—results from a cross-sectional study. Value Health 2002;5:106 –13. [13] Gisbert JP, Cooper A, Karagiannis D, et al. Impact of gastroesophageal reflux disease on work absenteeism, presenteeism and productivity in daily life: a European observational study. Health Qual Life Outcomes 2009;7:90. [14] El-Dika S, Guyatt GH, Armstrong D, et al. The impact of illness in patients with moderate to severe gastro-esophageal reflux disease. BMC Gastroenterol 2005;5:23. [15] Ciconelli RM, Soarez PC, Kowalski CC, Ferraz MB. The Brazilian Portuguese version of the Work Productivity and Activity Impairment: General Health (WPAI-GH) Questionnaire. Sao Paulo Med J 2006;124: 325–32. [16] Sander GB, Mazzoleni LE, Francesconi CF, et al. Development and validation of a cross-cultural questionnaire to evaluate nonulcer dyspepsia: the Porto Alegre Dyspeptic Symptoms Questionnaire (PADYQ). Dig Dis Sci 2004;49:1822–9. [17] Sander GB, Mazzoleni LE, Ott EA, Francesconi CF. Symptom-based outcome measure for nonulcer dyspepsia. Am J Gastroenterol 2005; 100:2365– 6. [18] Severens JL, Laheij RJ, Jansen JB, et al. Estimating the cost of lost productivity in dyspepsia. Aliment Pharmacol Ther 1998;12:919 –23. [19] Sander G, Francesconi C, Mazzoleni L, Lopes M. An unexpected high prevalence of non-investigated dyspepsia in Brazil: a populationbased study. Gut 2007;56 (Suppl III): A195. [20] Lerner D, Amick BC 3rd, Lee JC, et al. Relationship of employee-reported work limitations to work productivity. Med Care 2003;41:649 –59. VALUE IN HEALTH 14 (2011) S130 –S132 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Evaluación de la Calidad de Vida en Pacientes con Linfoma no Hodgkin y Cáncer Colo-Rectal en Diferentes Etapas Clínicas Atendidos en el Instituto Mexicano del Seguro Social Luz-Ma-Adriana Balderas-Peña, MD, PhD1,*, Daniel Sat-Muñoz, MD2, Iris Contreras-Hernández, MD, MSc3, Pedro Solano-Murillo, MD4, Guillermo-Allan Hernández-Chávez, MD, MSc4, Ignacio Mariscal-Ramírez, MD4, Martha Lomelí-García, Chemist, Pharm4, Margarita-Arimatea Díaz-Cortés, Eng5, Joaquín-Federico Mould-Quevedo, PhD6, Ulises Palomares-Chacón7, César-Adrián Balderas-Peña8, Oscar-Miguel Garcés-Ruiz, MD4, Gilberto Morgan-Villela, MD, MSc4 1 Unidad de Investigación Médica en Epidemiología Clínica, UMAE Hospital de Especialidades del Centro Médico Nacional de Occidente, Instituto Mexicano del Seguro Social, Departamento de Farmacobiología, División de Ciencias Básicas, Centro Universitario de Ciencias Exactas e Ingenierías, Universidad de Guadalajara; 2Oncología Quirúrgica, Hospital General Regional No. 46, Instituto Mexicano del Seguro Social, Departamento de Morfología, División de Ciencias Básicas, Centro Universitario de Ciencias de la Salud, Universidad de Guadalajara; 3Unidad de Investigación en Economía de la Salud del Centro Médico Nacional Siglo XXI, Instituto Mexicano del Seguro Social, México, D.F.; 4División de Oncología, UMAE Hospital de Especialidades del Centro Médico Nacional de Occidente, Instituto Mexicano del Seguro Social, Guadalajara, México; 5Unidad de Investigación y Estudios de Posgrado, División de Electrónica y Computación, Centro Universitario de Ciencias Exactas e Ingenierías, Universidad de Guadalajara; 6Instituto Tecnológico de Estudios Superiores de Monterrey, Campus Ciudad de México, Monterrey, México; 7Coloproctología, UMAE Hospital de Especialidades del Centro Médico Nacional de Occidente, Instituto Mexicano del Seguro Social, Guadalajara, México; 8Coloproctología, Hospital General Regional No. 110, Instituto Mexicano del Seguro Social, Guadalajara, México A B S T R A C T Introduction. In Mexico during 2008, were reported 127,604 new cancer cases, 6,347 of them were colorectal cancer cases and 4,276 NonHodgkin Lymphoma (NHL) cases. Objective. To Evaluate Health Related Quality of Life in Non-Hodgkin Lymphoma and colorectal cancer cases in different clinical stages, attended in a High Specialty Medical facility at the Instituto Mexicano del Seguro Social, during a 13 month period. Results. 162 patients were included, 56.8% (n⫽92) with NHL and 43.2% (n⫽70) with colorectal cancer. The scores obtained in the NHL group were: Global health status/QoL: 67.75 (⫾27.55), physical functioning 69.64 (⫾29.98), role functioning 71.38 (⫾33.73), emotional functioning 69.7 (⫾26.57), cognitive functioning 75.36 (⫾28.01), social functioning 79.35 (⫾29.38), fatigue 35.27 (⫾28.27), nausea and vomiting 13.41 (⫾21.85), pain 28.08 (⫾30.25), dyspnea 19.20 (⫾32.11), insomnia 30.80 (⫾38.03), appetite lost 26.45 (⫾36.16), constipation 19.20 (⫾32.11), diarrhea 12.32 (⫾26.48), financial difficulties 26.09 (⫾35.57). In colorec- Introducción Durante el año 2008 en México se reportaron 127,604 casos incidentes de cáncer (tasa de incidencia 117.54/100,000 habitantes), de los cuales 6,347 correspondieron a cáncer colo-rectal y 4,276 a linfoma no Hodgkin (LNH) [1]. En Jalisco se reportaron 9,278 casos nuevos de cáncer, entre ellos 395 casos de cáncer colo-rectal y 169 casos de LNH; del total de neoplasias malignas el Instituto Mexicano del Seguro Social (IMSS) reportó el diagnóstico inicial en el 28.58% (2,652 casos) [2]. tal cancer patients the scores were: Global health status/QoL: 68.21 (⫾24.46), physical functioning 67.38 (⫾30.45), role functioning 65.48 (⫾35.70), emotional functioning 66.43 (⫾26.84), cognitive functioning 78.57 (⫾26.49), social functioning 75.24 (⫾31.05), fatigue 37.78 (⫾31.62), nausea and vomiting 20.00 (⫾28.32), pain 37.38 (⫾34.45), dyspnea 11.90 (⫾26.64), insomnia 28.09 (⫾35.73), appetite lost 23.81 (⫾36.40), constipation 19.05 (⫾32.88), diarrhea 20.95 (⫾31.17), financial difficulties 34.76 (⫾38.67). Conclusions. With these basal results is important a follow-up with special attention to the treatment and attendance processes, in patients with this neoplasms and their impact on the quality of life. Palabras Claves: colo-rectal cancer, lymphoma, non-Hodgkin QLQ-C30, process of attendance, quality of life. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. En los sistemas de salud pública como el Mexicano el retardo en el diagnóstico de cáncer y el inicio de los tratamientos, por años ha sido considerado una responsabilidad de las instituciones de salud [3]; sin embargo, apenas se ha empezado a tomar en cuenta la influencia de las normas culturales, la respuesta de los pacientes ante la presencia de los síntomas y la educación en salud sobre los resultados del tratamiento en los pacientes con cáncer como son: respuesta a clínica y/o patológica, supervivencia global, supervivencia libre de enfermedad [3–7] y los PRO (patient reported outcomes) como la calidad de vida [4 –7]. Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Título corto: Quality of life in NHL and colorectal cancer at the IMSS. * Autor de correspondencia: Balderas-Peña Luz-Ma-Adriana, 1000 Belisario Domínguez, Colonia Independencia, Guadalajara, Jalisco, México 44340; Tel: 52 33 3618 2661; Fax: 52 33 3663 1834. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.031 VALUE IN HEALTH 14 (2011) S130 –S132 En mayo del año 2003 la INCTR (acrónimo en inglés de: International Network of Cancer Treatment and Research) se dio a la tarea de reunir la información disponible en los pacientes con LNH atendidos en países con economías emergentes en cuanto a calidad de vida y otros PRO, sin embargo, solo se incluyeron países asiáticos (India, Arabia Saudita, Turquía, Egipto y Kuwait) [8]. El cáncer colo-rectal también es una entidad donde es fundamental brindar atención que mejore la calidad de vida [9]. Los síntomas son comunes e importantes en cáncer de colon tardío, cuando el pronóstico es pobre, pero menos comunes y menos obvios en etapas tempranas de la enfermedad; pues el cuadro depende de la localización y el tamaño tumoral y la presencia o no, de metástasis [9 –11]. En los últimos años se han observado variaciones en la supervivencia y la mortalidad en pacientes con cáncer, lo que condicionó que esta situación se abordara desde la perspectiva concepto sociológico de la contención, la cual contextualiza la relación existente entre las sensaciones corporales y los síntomas, con lo que el sujeto integra una respuesta que considera adecuada a estos, y favorece el establecimiento de una relación dinámica de factores relacionados con las situaciones sociales específicas de cada sujeto, la historia de vida, sus expectativas a futuro y su concordancia con valores y explicaciones social y culturalmente aceptables en su contexto [12]. En México al igual que lo descrito en otros países en desarrollo existen limitantes de recursos que en su momento pudieran determinar los resultados en la evaluación de la calidad de vida en pacientes con cáncer, pero hasta el momento este tipo de estudios no han sido publicados en la población de pacientes mexicanos con LNH ó cáncer colo-rectal. El objetivo de este estudio fue evaluar la calidad de vida a través del cuestionario EORTC QLQ-C30 (versión 3) [13], en una población de pacientes con linfoma no Hodgkin y cáncer colo-rectal en diferentes etapas clínicas atendidas una Unidad Médica de Alta Especialidad (UMAE) del Instituto Mexicano del Seguro Social (IMSS). Métodos Diseño Estudio transversal en 162 pacientes: 92 con LNH y 70 con cáncer colo-rectal atendidos en la División de Oncología de una UMAE del IMSS en Guadalajara, Jalisco, México en el lapso comprendido entre 1° de enero del 2008 y el 31 de enero del 2009. El proyecto fue aprobado por la Comisión Nacional de Investigación Científica del Instituto Mexicano del Seguro Social. Recolección de la información Se analizaron: género, edad, etapa clínica (EC) de la enfermedad al diagnóstico, escolaridad, estado marital y co-mórbidos. Se registro también, la realización o no de cirugía, y/o la administración o no de quimioterapia y/o radioterapia, el número de ciclos de quimioterapia, el número de sesiones de radioterapia y si los pacientes recibieron o no apoyo psicológico, grupos de autoayuda, orientación nutricional y terapia física-rehabilitación; se aplicó la encuesta EORTC (European Organisation for Research and Treatment of Cancer) QLQ-C30 (versión 3) [13], previamente validada por la propia EORTC para población mexicana. Análisis estadístico El cuestionario está compuestos por escalas multi-ítem y de un ítem [14,15]. EORTC QLQ-C30 incluye las siguientes escalas multiítem: cinco funcionales (física, de rol, cognitiva, emocional y social), tres de síntomas (fatiga, dolor y náusea-vómito) y una que evalúa el estado de salud global/calidad de vida; y también seis escalas de un ítem. Cada escala multi-ítem incluye un conjunto S131 diferente de ítems y ninguno de ellos se incluye en más de una escala. Todas las escalas (multi-ítem y un ítem) están construidas de manera similar. El promedio bruto (raw score) de cada ítem individual es sumado y en las escala multi-ítem dividido entre el número de ítems que integran la escala; esos puntajes de las escalas son transformados linealmente para obtener un puntaje de 0 a 100 acorde a las fórmulas e instrucciones proporcionadas en el EORTC QLQ-C30 Scoring Manual [14,15]. Un puntaje más alto implica un mayor nivel de la respuesta. Una puntuación alta para una escala funcional representa un nivel más alto de función (mayor nivel de salud), una puntuación más alta en el estado de salud global/calidad de vida, implica una mejor calidad de vida. Una calificación más alta en las escalas de síntomas representa una mayor sintomatología y/o la presencia de más problemas de salud. Los tiempos de atención se cuantificaron en días; se calcularon medianas y percentiles 25 y 75, Los puntajes del cuestionario EORTC QLQ-C30, se analizaron a través de promedios y desviaciones estándar; para identificar las diferencias entre las EC y la escolaridad se realizó la prueba de normalidad de Shapiro-Wilk y posteriormente el estadístico ANOVA, con Bonferroni como análisis post Hoc. Las diferencias por estado marital (unido/no unido), antecedentes personales de cáncer (si/no) y co-mórbidos [sí (diabetes mellitus 2, hipertensión arterial, enfermedades cardiovasculares, enfermedades reumáticas y otros)/no] se analizaron con T de student. Se consideró significativo un valor p menor o igual a 0.05. Los datos fueron analizados en Excel 2007 y en SPSS versión 16.0 (SPSS. Chicago, IL, USA). Resultados De los 162 pacientes estudiados el 56.8% (n⫽92) correspondieron a LNH y el 43.2% (n⫽70) a cáncer colo-rectal. La población de pacientes con LNH presentó las siguientes características: edad promedio de 53.1 (⫾18.09) años, 52.2% (n⫽48) del género femenino y 47.8% (n⫽44) masculino, unidos (casados-unión libre) en 74.7% (n⫽68), con educación básica en 59.8% (n⫽55) y media superior en 23.9% (n⫽22). El15.22% (n⫽14) se encontraron en EC-I, 18.5% (n⫽17) en EC-II, 23.91% (n⫽22) en EC-III, 28.3% (n⫽26) en EC-IV, y 14.1% (n⫽13) con enfermedad no clasificable (cuadros 1 y 2 a Material Complementario en: doi:10.1016/j. jval.2011.05.031). Los puntajes de las escalas los pacientes con LNH fueron los siguientes: Estado global de salud: 67.75 (⫾27.55), función física 69.64 (⫾29.98), función de rol 71.38 (⫾33.73), función emocional 69.7 (⫾26.57), función cognitiva 75.36 (⫾28.01), función social 79.35 (⫾29.38), fatiga 35.27 (⫾28.27), náusea-vómito 13.41 (⫾21.85), dolor 28.08 (⫾30.25), disnea 19.20 (⫾32.11), insomnio 30.80 (⫾38.03), pérdida del apetito 26.45 (⫾36.16), constipación 19.20 (⫾32.11), diarrea 12.32 (⫾26.48) y dificultades financieras 26.09 (⫾35.57). Los pacientes con cáncer colo-rectal tuvieron las siguientes características: edad promedio de 56.03 (⫾12.97) años, 58.6% (n⫽41) del género femenino y 41.4% (n⫽29) masculino, unidos (casados-unión libre) en 76.8% (n⫽53), con educación básica en 58.8% (n⫽40) y superior en 25% (n⫽17). El 2.9% (n⫽2) se encontraron en EC I, 22.9% (n⫽16) en EC II, 27.14% (n⫽19) en EC III, 24.3% (n⫽17) en EC IV, y 22.9% (n⫽16) con enfermedad no clasificable. Los puntajes de las escalas los pacientes con cáncer colo-rectal fueron: Estado global de salud: 68.21 (⫾24.46), función física 67.38 (⫾30.45), función de rol 65.48 (⫾35.70), función emocional 66.43 (⫾26.84), función cognitiva 78.57 (⫾26.49), función social 75.24 (⫾31.05), fatiga 37.78 (⫾31.62), náusea-vómito 20.00 (⫾28.32), dolor 37.38 (⫾34.45), disnea 11.90 (⫾26.64), insomnio 28.09 (⫾35.73), pérdida del apetito 23.81 (⫾36.40), constipación 19.05 (⫾32.88), diarrea 20.95 (⫾31.17) y dificultades financieras 34.76 (⫾38.67). En ninguna de las escalas se encontraron diferencias al agrupar a los pacientes por patología o por etapa clínica (p⬎0.050) S132 VALUE IN HEALTH 14 (2011) S130 –S132 (Cuadro 3, 4 y 5 Material Complementario en: doi:10.1016/j. jval.2011.05.031). 10.1016/j.jval.2011.05.031 o si es un artículo impreso, estará en www.valueinhealthjournal.com/issues (seleccione el volumen, número y artículo). Conclusiones Este estudio presenta la descripción de 92 pacientes con LNH y son el 54.44% (92/169 casos) de los casos incidentes en el 2008 en el Estado de Jalisco [2], así como de 70 sujetos con cáncer colo-rectal que corresponden al 17.72% (70/395 casos), lo que hace de estos resultados extrapolables a la población atendida con estas patologías dentro del IMSS en el estado de Jalisco y representativos de estas neoplasias en el Instituto el cual tiene adscritos a 43’583,112 de derechohabientes en México [16] que en su mayor parte son trabajadores asalariados (desde obreros, hasta profesionistas y funcionarios) y sus familias como beneficiarios (padres, cónyuges e hijos). En las puntuaciones observadas en la escala que evalúa el Estado General de Salud/Calidad de vida no se encontraron diferencias entre los sujetos con LNH (n⫽92) y cáncer colo-rectal (p 0.912), con puntajes superiores a 65; tampoco se observó diferencia en las escalas funcionales las cuales fueron en todos los casos mayores a 65 (p NS), ni en las escalas de síntomas las cuales tuvieron puntajes menores a 30 en todas las escalas (p NS), a excepción de aquella correspondiente a dificultades financieras que fue la que mostró puntuaciones más altas en ambas patologías y alcanzó 26.09 en los pacientes con LNH y 34.76 en los pacientes con cáncer colo-rectal. Estos resultados son importantes en el contexto de que ambas patologías son atendidas en las diferentes Unidades Médicas de Alta Especialidad del IMSS que atienden población adulta y ocupan, LNH el primer lugar como neoplasia linfática en México y el cáncer colo-rectal la tercera causa, solo superado por cáncer de mama y cáncer de próstata. Esta situación hace que sea de vital importancia el estudio de la calidad de vida como un objetivo primario al evaluar el impacto de los tratamientos y las maniobras de intervención en estos pacientes, aunados a los resultados en salud clásicos como respuesta al tratamiento (patológica y clínica), supervivencia global y supervivencia libre de enfermedad. Ante los resultados basales obtenidos en esta revisión es importante dar un seguimiento implementando mejoras en los tratamientos de los pacientes portadores de estas neoplasias y evaluar si esto incrementa su calidad de vida, la cual por otra parte se encuentra a la par de las referidas en otros estudios de calidad de vida en el mundo como es el caso del estudio de Avery et al., en el cual la calidad de vida y el estado general de salud se reportan con un puntaje de 56 y las escalas funcionales van de 32 a 79.5 y las de síntomas de 24.2 a 53.8 [17]. Agradecimientos Los autores agradecemos la participación del grupo de encuestadores (Rosa-Emilia Ramírez-Conchas, Martha-Cristina BalderasPeña, Miguel-Ángel Martínez-López, Adolfo-Leonardo GómezBalderas, Alfredo Prieto-Moreno y Febe-Eréndira Balderas-Peña) en la realización de este proyecto, pues sin su capacidad laboral no hubiese sido posible obtener los resultados aquí presentados. Fuentes de financiamiento: El presente trabajo fue financiado por Fundación IMSS A.C. Materiales Complementarios Material complementario que acompaña este artículo se puede encontrar en la versión en línea como un hipervínculo en doi: REFERENCIAS [1] Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Can 2010;127:2893–917. [2] Gutiérrez-Carranza A, Carranco-Ortiz BG, Sandoval-Urban E, Hernández-Sánchez JE, Ramírez-Anguiano VM, González-Álvarez JA, Pérez-Gómez HR, Gálvez-Gálvez B, Camacho-Cortés JL, González-Díaz O, Carmona-Quintero JC, García-Mejía D, Álvarez-Iñiguez PY, VélezGómez E, Delgado-Lamas JL, Maldonado-Hernández HR, LópezMariscal AR, Ulloa-Robles E. Registro Estatal de Cáncer. Jalisco 2008. Secretaría de Salud Jalisco. Dirección General de Salud Pública. Disponible en: http://www.jalisco.gob.mx/wps/wcm/connect/ f347130040963031b746b79c8da0b43f/Cancer2008.pdf?MOD⫽ AJPERES&CACHEID⫽f347130040963031b746b79c8da0b43f. [Último acceso 13 de diciembre de 2010]. [3] Pérez G, Porta M, Borrell C, et al. Interval from diagnosis to treatment onset for six major cancers in Catalonia, Spain. Cancer Detect Prev 2008;32:267–75. [4] Lis CG, Rodeghier M, Grutsch JF, Gupta D. Distribution and determinants of patient satisfaction in oncology with a focus on health related quality of life. BMS Health Serv Res 2009;9:190. [5] Kleeberg UR, Feyer P, Günther W, Behrens M. Patient satisfaction in outpatient cancer care: a prospective survey using The PASQOC questionnaire. Support Care Cancer 2008;16:947–54. [6] Gotay CC, Kawamoto CT, Bottomley A, Efficace F. The prognostic significance of patient-reported outcomes in cancer clinical trials. J Clin Oncol 2008;26:1355-63; Lipscomb J, Gotay CC, Snyder CF. Patientreported outcomes in cancer: a review of recent research and policy initiatives. CA Cancer J Clin 2007;57:278 –300. [7] Lipscomb J, Gotay CC, Snyder CF. Patient-reported outcomes in cancer: a review of recent research and policy initiatives. CA Cancer J Clin 2007;57:278 –300. [8] Naresh KN, Advani S, Adde M, et al. Report of an International Network of Cancer Treatment and Research workshop on non-Hodgkin’s lymphoma in developing countries. Blood Cells Mol Dis 2004;33:330–7. [9] Calva-Arcos M, Acevedo-Tirado MT. Revisión y actualización general en cáncer colorrectal. Anales de Radiología México. 2009;1:99 –115. [10] Marianne Korsgaard, Lars Pedersen, Søren Laurberg. Delay of diagnosis and treatment of colorectal cancer-A population-based Danish Study. Cancer Detect Prev 2008;32:45–51. [11] Marianne Korsgaard, Lars Pedersen, Henrik Toft Sørensen, Søren Laurberg. Delay of treatment is associated with advanced stage of rectal cancer but not of colon cancer. Cancer Detect Prev 2006;30:341–6. [12] Sand-Andersen R, Paarup B, Vedsted P, Bro F, Soendergaard J. ‘Containment’ as an analytical framework for understanding patient delay: A qualitative study of cancer patients’ symptom interpretation processes. Soc Sci Med 2010;71:378 – 85. [13] EORTC QLQ-C30 (versión 3). Disponible en: http://www.eortc.be/home/ qol/downloads/f/C30/QLQ-C30%20SpanishMexican.pdf. [Ultimo acceso 24 de diciembre de 2010[. [14] Fayers PM, Aaronson NK, Bjordal K, Groenvold M, Curran D, Bottomley A, on behalf of the EORTC Quality of Life Group. The EORTC QLQ-C30 Scoring Manual (3rd Edition). Published by: European Organisation for Research and Treatment of Cancer, Brussels 2001.EORTC QLQ-C30 Scoring Manual. Disponible en: http://www. eortc.be/home/qol/downloads/f/RV/RV_complete.pdf. [Último acceso 24 de diciembre de 2010]. [15] Aaronson NK, Ahmedzai S, Bergman B, et al. The European Organisation for Research and Treatment of Cancer QLQ-C30: A quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst 1993;85:365–76. [16] Instituto Mexicano del Seguro Social. Seguridad y solidaridad social. Promedio de Servicios otorgados en un día típico. Total nacional. Enero a septiembre de 2010. Disponible en: http://www.imss.gob.mx/ dpm/dis/Tabla.aspx?Srv⫽M00-1&ID⫽SCES006_001_001&OPC⫽opc03. [Último acceso 30 de diciembre de 2010]. [17] Avery KNL, Metcalfe C, Nicklin J, et al. Satisfaction with care: an independent outcome measure in surgical oncology. Ann Surg Oncol 2006;13:817–22. VALUE IN HEALTH 14 (2011) S133–S136 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Calidad de Vida en Mujeres Mexicanas con Cáncer de Mama en Diferentes Etapas Clínicas y su Asociación con Características SocioDemográficas, Estados Co-Mórbidos y Características del Proceso de Atención en el Instituto Mexicano del Seguro Social Daniel Sat-Muñoz, MD1, Iris Contreras-Hernández, MD, MSc2, Luz-Ma-Adriana Balderas-Peña, MD, PhD3,*, Guillermo-Allan Hernández-Chávez, MD, MSc4, Pedro Solano-Murillo, MD4, Ignacio Mariscal-Ramírez, MD4, Martha Lomelí-García, Chemist, Pharm4, Margarita-Arimatea Díaz-Cortés, Eng5, Joaquín-Federico Mould-Quevedo, PhD6, Alma-Rosa López-Mariscal, MD7, Sergio-Emilio Prieto-Miranda, MD8, Gilberto Morgan-Villela, MD, MSc8 1 Oncología Quirúrgica, Hospital General Regional No. 46, Instituto Mexicano del Seguro Social, Departamento de Morfología, División de Ciencias Básicas, Centro Universitario de Ciencias de la Salud, Universidad de Guadalajara, Guadalajara, Mexico; 2Unidad de Investigación en Economía de la Salud del Centro Médico Nacional Siglo XXI, Instituto Mexicano del Seguro Social, Mexico, D.F., México; 3Unidad de Investigación Médica en Epidemiología Clínica, UMAE Hospital de Especialidades del Centro Médico Nacional de Occidente, Instituto Mexicano del Seguro Social, Guadalajara, Mexico; Departamento de Farmacobiología, División de Ciencias Básicas, Centro Universitario de Ciencias Exactas e Ingenierías, Universidad de Guadalajara; 4División de Oncología, UMAE Hospital de Especialidades del Centro Médico Nacional de Occidente, Instituto Mexicano del Seguro Social, Guadalajara, Mexico; 5Unidad de Investigación y Estudios de Posgrado, División de Eléctronica y Computación, Centro Universitario de Ciencias Exactas e Ingenierías, Guadalajara, Mexico; 6Instituto Tecnológico de Estudios Superiores de Monterrey, Campus Ciudad de México, Monterrey, Mexico; 7Oncología Médica, Hospital General Regional No. 110, Instituto Mexicano del Seguro Social, Guadalajara, Mexico; 8Educación e Investigación en Salud, Hospital General Regional No. 46, Instituto Mexicano del Seguro Social, Guadalajara, Mexico A B S T R A C T Introduction: Quality of Life is the most studied PRO (patient reported outcome) in cancer patients. With early diagnosis and better treatments in breast cancer, this entity has been transformed in a chronic disease with longer survival. The joint effects of diseases and treatment on quality of life are each day more important to consider in survival patients. Objective: To evaluate Quality of Life, Socioeconomic factors, co-morbidities, and the attendance process impact on quality of life in breast cancer women with different clinical stages attending at the Instituto Mexicano del Seguro Social using the EORCT QLQ-C30 Results: The scores of EORTC QLQ-C30 (v3) were: Global health status / QoL: 73.47 (⫾20.81), physical functioning 76.98 (⫾20.85), role functioning 76.60 (⫾27.57), emotional functioning 64.53 (⫾26.81), cognitive functioning 74.47 (⫾26.02), social functioning 84.96 (⫾23.20), fatigue 31.94 (⫾25.45), nausea and vomiting 19.49 (⫾26.93), pain 28.95 (⫾27.27), dyspnea 15.29 (⫾24.62), insomnia 35.13 (⫾32.10), appetite lost 18.04 (⫾28.75), 18.04 (⫾28.75), constipation 19.20 (⫾32.11), diarrhea 12.9 Introducción La tercera causa de mortalidad general en México es el cáncer [1]. En este país, con población de 112’322,757 habitantes; 57’464,459 (51.16%) son mujeres y su esperanza de vida al nacer en 2009 fue de 77.6 años [2]. De acuerdo a las estadísticas de la IARC (International Agency for Research on Cancer) para 2008 en México, al igual que en Europa y en países desarrollados, cáncer (⫾24.25), financial difficulties 40.57 (⫾37.26). The scores with EORTC QLQ-BR23 were: body image 74.84 (⫾31.69), sexual functioning13.73 (⫾22.55), sexual enjoyment 32.86 (⫾36.17), future perspectives 51.69 (⫾38.00), systemic therapy side effects 30.82 (⫾20.71), breast symptoms22.85 (⫾23.49), arm symptoms 27.53 (⫾24.75), upsert by hair loss 43.80 (⫾44.01). Conclusions: Clinical stage in breast cancer is associated with differences in the scores from fatigue, nausea and vomiting and financial difficulties according to the evolution of the disease and the physical detriment associated. Socio-demographic features were related role functioning, fatigue and pain in single women with higher scores. Palabras Claves: breast cancer, process of care, QLQ-C30, QLQ-BR23, quality of life. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. de mama ocupó el segundo lugar en prevalencia y primer lugar de incidencia en mujeres (13,939 casos) [3–5]. Este padecimiento se comporta como una patología crónica [6], en el que es fundamental el mantenimiento de una buena calidad de vida como uno de los objetivos del proceso de atención de mujeres que viven con cáncer de mama. La calidad de vida es el PRO (patient reported outcomes) más estudiado en cáncer [7–9]. Al mejorar el diagnóstico y la detección Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Título corto: Quality of life in breast cancer women at the IMSS. * Autor de correspondencia: Luz-Ma-Adriana Balderas-Peña, 1000 Belisario Domínguez, Colonia Independencia, Guadalajara, Jalisco, México 44340; Tel: 52 33 3618 2661; Fax: 52 33 3663 1834. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.027 S134 VALUE IN HEALTH 14 (2011) S133–S136 temprana del cáncer de mama, la supervivencia se incrementa, y los efectos conjuntos de la enfermedad y el tratamiento sobre la calidad de vida se vuelven cada vez más importantes para las pacientes [10,11]. Actualmente ha disminuido la duración del tratamiento; el tipo y cantidad del mismo se ha modificado y la mujer se reintegra de forma temprana a sus actividades; pero su calidad de vida se puede ver afectada por los síntomas asociados al manejo, por la percepción y/o aceptación de la enfermedad y las modificaciones que el padecimiento impone al estilo de vida [12]. Los factores asociados con mayor frecuencia a una disminución en la calidad de vida de las mujeres con cáncer de mama, son: fatiga, menopausia y/o los síntomas de ésta, ambos relacionados con los tratamientos adyuvantes; así como alteraciones en la función cognitiva [13–17]. Otros factores que a largo plazo impactan en la calidad de vida de las supervivientes de cáncer de mama son la morbilidad psicológica y la no adaptación social, así como los problemas financieros [11]. En estas mujeres la enfermedad modifica sus expectativas de vida a futuro, (posibilidad de la maternidad, sexualidad, deterioro de la imagen corporal, sensación de feminidad y aceptación por parte de su pareja), debido al tratamiento quirúrgico, médico y/o radiante y la actividad de la enfermedad [10]. Estas expectativas son consideradas un factor predictivo de adaptación psicosocial a largo plazo. [10,14,15]. El tipo de cirugía y su asociación con la calidad de vida muestran resultados inconsistentes, pues algunos señalan que las variantes de la técnica no tienen impacto significativo en la calidad de vida, pero si en la imagen corporal, la función sexual y en la funcionalidad del brazo del lado afectado [8,12,18]. Dichas inconsistencias se deben al empleo de diferentes instrumentos de medición de la calidad de vida. [7,19 –21]. Conforme aumentan las mujeres supervivientes al cáncer de mama, se ejerce una profunda influencia en la estructura y el funcionamiento de los sistemas de salud de los diferentes países. La investigación de secuelas físicas y psicosociales del cáncer, su tratamiento, el impacto de aspectos socio-demográficos, co-mórbidos y características del proceso de atención en la calidad de vida de las pacientes es una prioridad para los sistemas de salud [22], por lo que su estudio debe ser un objetivo primario al evaluar tratamientos para cáncer mamario [3,8]. El objetivo de este estudio fue evaluar la calidad de vida en mujeres con cáncer de mama en diferentes etapas clínicas (EC) atendidas en el Instituto Mexicano del Seguro Social (IMSS) y el impacto de variables socio-demográficas, co-mórbidos y características del proceso de atención en ésta. Métodos Diseño Estudio transversal con 314 casos incidentes de cáncer de mama en diferentes EC, atendidos en oncología de dos hospitales generales regionales (HGR) y una Unidad Médica de Alta Especialidad (UMAE) del IMSS, en Guadalajara, Jalisco, México, en el período del 1° de enero del 2008 y al 31 de enero del 2009. Se aplicaron los cuestionarios EORTC QLQ-C30 y EORTC QLQ-BR23 [23,24], validados para población mexicana e hispanoparlante, respectivamente y un cuestionario de variables socio-demográficas y del proceso de atención. La EC, tiempos de atención y tratamientos se documentaron con el expediente clínico. El proyecto fue aprobado por la Comisión Nacional de Investigación Científica del IMSS. Recolección de la información Las pacientes fueron reclutadas en unidades de quimioterapia ambulatoria y consulta externa de oncología de los hospitales par- ticipantes. Se analizaron: edad, EC al diagnóstico, escolaridad, estado marital, co-mórbidos (diabetes mellitus 2, hipertensión arterial, enfermedades cardiovasculares, reumáticas y otros), características del proceso de atención: tiempo de atención en Unidad de Medicina Familiar (UMF), HGR y UMAE, tiempo entre diagnóstico y realización de una cirugía , entre indicación y administración de quimioterapia (QT) y/o radioterapia (RT). Se registro: realización o no de cirugía, y/o administración o no de QT y/o RT, número de ciclos de quimioterapia, número de sesiones de radioterapia; apoyo psicológico individual o en grupos de autoayuda, orientación nutricional, terapia física-rehabilitación (si/no); simultáneamente se aplicaron las encuestas citadas. Análisis estadístico Los cuestionarios están compuestos por escalas multi-ítem y de un ítem [23,24]. El cuestionario EORTC QLQ-C30 incluye escalas multi-ítem: cinco funcionales (física, de rol, cognitiva, emocional y social), tres de síntomas (fatiga, dolor y náusea-vómito) y una que evalúa el estado de salud global/calidad de vida; y también seis escalas de un ítem. Cada escala multi-ítem incluye un conjunto diferente de ítems y ninguno de ellos se incluye en más de una escala. Todas las escalas (multi-ítem y un ítem) están construidas de manera similar. El promedio bruto (raw score) de cada ítem individual es sumado y en las escala multiítem dividido entre el número de ítems que integran la escala; esos puntajes de las escalas son transformados linealmente para obtener un puntaje de 0 a 100 acorde a las fórmulas e instrucciones proporcionadas en el EORTC QLQ-C30 Scoring Manual [25,26]. Un puntaje más alto implica un mayor nivel de la respuesta. Una puntuación alta para una escala funcional representa un nivel más alto de función (mayor nivel de salud), una puntuación más alta en el estado de salud global/calidad de vida, implica una mejor calidad de vida. Una calificación más alta en las escalas de síntomas representa una mayor sintomatología y/o la presencia de más problemas de salud. El modulo EORTC QLQ-BR23 para cáncer de mama, está diseñado para pacientes en diferentes EC y bajo diferentes modalidades de tratamiento (cirugía, quimioterapia, radioterapia y hormonoterapia); comprende 23 preguntas que evalúan síntomas de la enfermedad, efectos secundarios del tratamiento, imagen corporal, función sexual y perspectivas de futuro. Incorpora cinco escalas multi-ítem: efectos adversos del tratamiento sistémico, síntomas braquiales, síntomas mamarios, imagen corporal y función sexual. Las escalas de un ítem evalúan: placer sexual, caída de cabello y perspectivas de futuro. El método de puntuación es el mismo que el empleado para las escalas funcionales, de síntomas y de un ítem en el cuestionario QLQ-C30 [25]. Los tiempos de atención se cuantificaron en días; se calcularon medianas y percentiles 25 y 75, Los puntajes de los cuestionarios EORTC QLQ-C30 y QLQ-BR23, se analizaron a través de promedios y desviaciones estándar; para identificar las diferencias entre las EC y la escolaridad se realizó la prueba de normalidad de ShapiroWilk y posteriormente el estadístico ANOVA, con Bonferroni como análisis post Hoc. Las diferencias por estado marital (unido/no unido), antecedentes personales de cáncer (si/no) y co-mórbidos (sí/no) se analizaron con T de student. Los tiempos de atención se expresaron en días, la realización y/o aplicación de un tratamiento u obtención de apoyo, se categorizaron de forma dicotómica (si/ no), se cuantificaron los ciclos de quimioterapia y/o sesiones de radioterapia. La asociación entre variables fue analizada con el coeficiente de correlación de Pearson (rP), Aquellas variables con significancia ⬍0.2 se incluyeron en un modelo multivariado de regresión lineal. Se consideró significativo un valor p menor o igual a 0.05. Los datos fueron analizados en Excel 2007 y en SPSS versión 16.0 (SPSS. Chicago, IL, USA). VALUE IN HEALTH 14 (2011) S133–S136 Resultados La edad promedio 52.23 (⫾10.55) años; pacientes 17 (5.4%) se encontraban en EC I, 94 (29.9%) en EC II, 101 (32.2%) en EC III, 47 (15%) en EC IV y 55 (17.5%) con enfermedad no clasificable. El estado marital de las pacientes fue: no unidas (solteras, divorciadas y viudas) en 30.9% (n⫽97) y unidas (casadas y en unión libre) 67.83% (n⫽213), el 1.27% (n⫽4) de las pacientes no especificó su situación marital. La escolaridad fue: analfabetas 2.9% (n⫽9), educación básica y media básica 57.3% (n⫽180), educación media superior 18.5% (n⫽58), educación superior 19.7% (n⫽62), cinco (1.59%) pacientes no especificaron su escolaridad. Los antecedentes personales de cáncer estuvieron presentes en 9 (2.9%) pacientes y se identificó la presencia de co-mórbidos en 210 (66.9%) mujeres. El proceso de atención mostró lo siguiente: tiempo entre primera consulta y confirmación de cáncer Md: 30 (1 a 124.5) días; tiempo de atención en UMF, Md: 1 (1 a 30) días; tiempo de atención en HGR, Md: 120 (30 a 392.5) días; tiempo de atención UMAE, Md: 91 (30 a 270) días; tiempo entre confirmación diagnóstica y cirugía, Md: 21 (14 a 34.5) días; tiempo entre prescripción y administración de QT, Md: 6 (3 a 14) días; tiempo entre la prescripción y aplicación de RT, Md: 10 (2 a 29.5) días. La mediana de ciclos de QT fue: 6 (6 a 8), y de sesiones de RT 25 (25 a 25). Las pacientes recibieron apoyo psicológico en 5.1% (n⫽16), en grupos de auto-ayuda 0.6% (n⫽2), orientación nutricional 5.4% (n⫽17), y fisioterapia 0.3% (n⫽1). El alfa de Cronbach fue ⬎0.65 en las escalas multi-ítem de ambos cuestionarios. Los resultados de la EORTC QLQ-C30 fueron: Estado General de Salud/Calidad de vida: 73.47 (⫾20.81), Funciones: física: 76.98 (⫾20.85), de rol: 76.60 (⫾27.57), emocional: 64.53 (⫾26.81), cognitiva: 74.47 (⫾26.02) y social: 84.96 (⫾23.20), fatiga: 31.94 (⫾25.45), náusea-vómito 19.49 (⫾26.93), dolor 28.95 (⫾27.27), disnea 15.29 (⫾24.62), insomnio 35.13 (⫾32.10), pérdida de apetito 18.04 (⫾28.75), constipación 22.02 (⫾30.62), diarrea 12.9 (⫾24.25) y dificultades económicas 40.57 (⫾37.26). Para EORTC QLQ-BR23: imagen corporal 74.84 (⫾31.69), función sexual 13.73 (⫾22.55), placer sexual 32.86 (⫾36.17), perspectivas de futuro 51.69 (⫾38.00), eventos adversos del tratamiento sistémico 30.82 (⫾20.71), síntomas mamarios 22.85 (⫾23.49), síntomas braquiales 27.53 (⫾24.75) y disgusto por la pérdida del cabello 43.80 (⫾44.01). Al analizar los resultados de los cuestionarios agrupando a las pacientes por EC se observó diferencia en las escalas: Fatiga con diferencias entre la EC IV y las EC II y III [43.56 (⫾29.68), 29.78 (⫾24.59) y 30.51 (⫾23.40) respectivamente; p 0.019]; náusea-vómito con diferencias entre las EC III y IV [15.74 (⫾20.82) y 29.44 (⫾37.88) respectivamente; p 0.035]; dificultades financieras con diferencias entre el grupo de pacientes con enfermedad no clasificable y en EC III [27.73 (⫾37.36) y 47.88 (⫾33.74) respectivamente; p 0.012] (cuadro 1 a Material complementario en: doi:10.1016/j.jval. 2011.05.027). No se observaron diferencias en las escalas del cuestionario QLQ-BR23 (cuadro 2 a Material complementario en: doi: 10.1016/j.jval.2011.05.027). Al analizar los puntajes agrupando a las pacientes por estado marital, se encontraron diferencias en la función de rol entre las pacientes unidas y no unidas: [80.17 (⫾25.09) y 69.00 (⫾31.23) respectivamente; p⫽0.002]; en fatiga [29.94 (⫾25.01) y 36.27 (⫾26.36) respectivamente; p⫽0.043] y en dolor [26.70 (⫾25.95) y 34.56 (⫾29.57) respectivamente; p⫽0.019]. La escala de pérdida de apetito mostró diferencias entre las pacientes con educación básica y media básica [20.42 (⫾30.72)] respecto a las que tuvieron educación superior [9.27 (⫾19.54); p 0.018]. No se observaron diferencias entre tener o no antecedentes personales de cáncer y/o la presencia de co-mórbidos (cuadro 3 a Material complementario en: doi:10.1016/j.jval.2011.05.027). En el cuestionario QLQ-BR23, se observó diferencia en eventos adversos del tratamiento sistémico, cuando las pacientes tenían el antecedente personal de cáncer previo, respecto a las que no (46.03 S135 (⫾31.68) y 31.05 (⫾20.92) respectivamente; p 0.034; cuadro 4 a Material complementario en: doi:10.1016/j.jval.2011.05.027). Al asociar el puntaje del estado global de salud (EGS)/calidad de vida y el proceso de atención, se encontró asociación con el tiempo de atención en UMF (rP:-0.194; p 0.005) y tiempo de atención en UMAE (rP:-0.130; p 0.42), lo que representó una disminución de -0.2 (p 0.008) en el puntaje EGS/Calidad de vida lo que desde el punto de vista clínico no impactó en el puntaje. Conclusiones En Jalisco durante el 2008 se reportaron 1251 casos incidentes de cáncer de mamá [25]; la población estudiada representó el 25.1% de éstos en la Entidad, por lo que los datos son extrapolables al IMSS y al sistema de salud a nivel nacional. La etapa clínica de la enfermedad condicionó diferencias en los puntajes de: fatiga, nausea-vómito y dificultades financieras, situación que está en relación directa con la historia natural del padecimiento y el deterioro físico asociado a éste. Las variables socio-demográficas mostraron impacto en la función de rol, fatiga y dolor donde las mujeres no unidas, que tuvieron mayores puntuaciones en las escalas. Este resultado puede ser reflejo del impacto del estado emocional y las condiciones psicológicas en la calidad de vida en una mujer sin pareja y sin una red social de apoyo; lo cual es evidenciado por los bajos porcentajes de asistencia psicológica individual, grupos de autoayuda, consejería nutricional y fisioterapia, en el contexto del manejo institucional. La escolaridad impactó en la escala de pérdida de apetito, con mayor puntaje en mujeres con educación básica, comparadas con las universitarias. Esta situación puede asociarse a diferencias en el estilo de vida y el nivel de ingresos de estos grupos. Los eventos adversos del tratamiento sistémico, fueron mayores en mujeres con una neoplasia maligna previa a la actual. No se observó asociación entre el proceso de atención y el estado de salud global. Agradecimientos Los autores agradecemos la participación del grupo de encuestadores (Rosa-Emilia Ramírez-Conchas, Martha-Cristina BalderasPeña, Miguel-Ángel Martínez-López, Adolfo-Leonardo GómezBalderas, Alfredo Prieto-Moreno y Febe-Eréndira Balderas-Peña) en la realización de este proyecto, pues sin su capacidad laboral no hubiese sido posible obtener los resultados aquí presentados. Fuentes de financiamiento: El presente trabajo fue financiado por Fundación IMSS A.C. Materiales Complementarios Material complementario que acompaña este artículo se puede encontrar en la versión en línea como un hipervínculo en doi: 10.1016/j.jval.2011.05.027 o si es un artículo impreso, estará en www.valueinhealthjournal.com/issues (seleccione el volumen, número y artículo). REFERENCIAS [1] Instituto Nacional de Estadística y Geografía. Anuario estadístico de los Estados Unidos Mexicanos 2009. http://www.inegi.org.mx/prod_serv/ contenidos/espanol/bvinegi/productos/integracion/pais/aeeum/2009/ Aeeum091.pdf. [Último acceso 23 de diciembre de 2010]. S136 VALUE IN HEALTH 14 (2011) S133–S136 [2] Datos preliminares. Censo de Población y vivienda 2010. Instituto Nacional de Estadística y Geografía. http://www.inegi.org.mx/. [Último acceso 23 de diciembre de 2010]. [3] Den Oudsten BL, Van Heck GL, Van der Steeg AFW, et al. The WHOQOL-100 has good psycometric properties in breast cancer patients. Journal Cli Epidemiology 2009;62:195–205. [4] Globocan 2008. Fast Stats. Mexico. http://globocan.iarc.fr/factsheets/ populations/factsheet.asp?uno⫽484. [Último acceso 23 de diciembre de 2010]. [5] Albert US, Koller M, Lorenz W, et al. Quality Circle. Quality of life profile: from measurement to clinical application. Breast 2002;11: 324 –34. [6] Maza-Fernandez ME, Vecchi-Martini E. History, overview and challenges of the breast cancer movement in Mexico. Salud Publica Mex 2009;51(Suppl. 2):s329 –34. [7] Arraras JL, Illarramendi JJ, Manterola A, et al. Evaluación de la calidad de vida a largo plazo en pacientes con cáncer de mama en estadios iniciales mediantes los cuestionarios de la EORTC. Rev Clin Esp 2003; 203:577– 81. [8] Montazeri A. Health-related quality of life in breast cancer patients: A bibliographic review of the literature from 1974 to 2007. J Experimental & Clin Cancer Res 2008;27:32. [9] Montazeri A. Quality of life data as prognostic indicators of survival in cancer patients: an overview of the literature from 1982 to 2008. Health Qual Life Outcomes 2009,7:102. [10] Avis NE, Carwford S, Manuel J. Quality of life among younger women with breast cancer J Clin Oncol 2005;23:3322–30. [11] Hoopman R, Muller MJ, TerweeCB, et al. Translation and validation of the EORTC QLQ.C30 for use among Turkish and Moroccan ethnic minority cancer patients in the Netherlands. Eur J Cancer 2006;42:1839 – 47. [12] Arndt V, Merx H, Stegmaier Ch, et al. Persistence of restrictions in quality of life from the first to the third year after Diagnosis in women with breast cancer. J. Clin Oncol 2005;23:4945–53. [13] Bower JE, Ganz PA, Desmond KA, et al. Fatigue in breast cancer survivors: Ocurrence, correlates, and impact on quality of life. J Clin Oncol 2000;18:743–53. [14] Tchen N, Juffs HG, Downie FP, Yi DQ, et al. Cognitive function, fatigue, and menopausal symptoms in women receiving adjuvant chemotherapy for breast cancer. J Clin Oncol 2003;21:4175– 83. [15] Mar Fan HG, Houéde-Tchen N, Yi DQ, et al. Fatigue, menopausal symptoms, and cognitive function in woman after adjuvant chemotherapy for breast cancer: 1- and 2- year follow-up of a prospective controlled study. J Clin Oncol 2005;23:8025–32. [16] Ahmed RL, Prizment A, Lazovich D, et al. Lymphedema and quality of life in breast cancer survivors: The Iowa Women=s Health Study. J Clin Oncol 2008;26:5689 –96. [17] Kim SH, Son BH, Hwang SY, et al. Fatigue and depression in diseasefree greast cancer survivors: Prevalence, correlates, and association with quality of life. J Pain & Symptom Management 2008;35:644 –55. [18] Groenvold M, Petersen MA, Idler E, et al. Psychological distress and fatigue predicted recurrence and survival in primary breast cancer patients. Breast Cancer Res Treat 2007;105:201–19. [19] Hormes JM, Lytle LA, Gross CR, et al. The Body Image and Relationships Scale: Development and Validation of a measure of body image in female breast cancer survivors. J Cli Oncol 2008;26:1269 –74. [20] Gordon LG, Battistutta D, Scuffham P, et al. The impact of rehabilitation support services on health-related quality of life for women with breast cancer. Breast Cancer Res Treat 2005;93:217–26. [21] EORTC QLQ-C30 (versión 3). http://www.eortc.be/home/qol/ downloads/f/C30/QLQ-C30%20SpanishMexican.pdf. [Último acceso 24 de diciembre de 2010]. [22] EORTC QLQ - BR23 http://www.eortc.be/home/qol/downloads/f/BR23/ BR23%20Spanish.pdf. [Último acceso 24 de diciembre de 2010]. [23] Fayers PM, Aaronson NK, Bjordal K, et al. On behalf of the EORTC Quality of Life Group. The EORTC QLQ-C30 Scoring Manual (3rd Edition). Published by: European Organisation for Research and Treatment of Cancer, Brussels 2001.EORTC QLQ-C30 Scoring Manual. http://www.eortc.be/home/qol/downloads/f/RV/RV_complete.pdf. [Último acceso 24 de diciembre de 2010]. [24] Aaronson NK, Ahmedzai S, Bergman B, et al. The European Organisation for Research and Treatment of Cancer QLQ-C30: A quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst 1993;85:365–76. [25] Gutiérrez-Carranza A, Carranco-Ortiz BG, Sandoval-Urban E, et al. Registro Estatal de Cáncer. Jalisco 2008. Secretaría de Salud Jalisco. Dirección General de Salud Pública. Disponible en: http://www.jalisco.gob.mx/wps/wcm/ connect/f347130040963031b746b79c8da0b43f/Cancer2008.pdf?MOD⫽ AJPERES&CACHEID⫽f347130040963031b746b79c8da0b43f. [Último acceso 13 de diciembre de 2010]. VALUE IN HEALTH 14 (2011) S137–S140 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval The Costs of Type 2 Diabetes Mellitus Outpatient Care in the Brazilian Public Health System Luciana R. Bahia, MD, PhD1,*, Denizar Vianna Araujo, MD, PhD1, Beatriz D. Schaan, MD, PhD2, Sérgio A. Dib, MD, PhD3, Carlos Antônio Negrato, MD, PhD4, Marluce P.S. Leão, MD5, Alberto José S. Ramos, MD6, Adriana C. Forti, MD, PhD7, Marília B. Gomes, MD, PhD1, Maria Cristina Foss, MD, PhD8, Rosane A. Monteiro8, Daniela Sartorelli, PhD8, Laércio J. Franco, MD, PhD8 1 Universidade do Estado do Rio de Janeiro, Rio de Janeiro, Brazil; 2Universidade Federal do Rio Grande do Sul/HCPA, Porto Alegre, Brazil; 3Universidade Federal de São Paulo, Sao Paulo, Brazil; 4Associação de Diabetes de Bauru, Bauru, Brazil; 5Universidade Estadual de Santa Cruz, Itabuna, Brazil; 6Universidade Federal de Campina Grande, Campina Grande, Brazil; 7Universidade Estadual do Ceará, Fortaleza, Brazil; 8Universidade de São Paulo, Ribeirão Preto, Brazil A B S T R A C T Objective: The prevalence of type 2 diabetes has shown a significant increase in parallel with health care costs. The objective of the Brazilian Study on Diabetes Costs (ESCUDI study) was to estimate direct and indirect costs of type 2 diabetes outpatient care in the Brazilian Public Health Care System. Methods: Data were collected from different levels of health care in eight Brazilian cities in 2007. A total of 1000 outpatients were interviewed and had their medical records data analyzed. Direct medical costs included expenses with medications, diagnostic tests, procedures, blood glucose test strips, and office visits. Nonmedical direct costs included expenses with diet products, transportation, and caregivers. Absenteeism, sick leave, and early retirement were classified as indirect costs. Results: Total annual cost for outpatient care was US$2108 per patient, out of which US$1335 per patient of direct costs (63.3%) and US$773 per patient of indirect costs (36.7%). Costs escalated as duration of Introduction The prevalence of diabetes in all age groups worldwide had been estimated to reach 2.8% in 2000 and 4.4% by 2030. By then, the total number of people with diabetes is projected to rise from 171 million to 366 million [1]. Nearly two-thirds of individuals with diabetes live in developing countries such as Brazil, India, and China, where this number is expected to increase during the next two decades [2]. A single nationwide study was carried out in Brazil in the late 1980s and results showed 7.6% prevalence of diabetes in people aged 30 to 69 years [3]. It is remarkable that nearly 50% of those diagnosed with diabetes were not aware of their disease and only 25% reported receiving care. More recent regional studies showed an increase in prevalence rates: 12.1% in the city of Ribeirão Preto [4], 12.4% in Porto Alegre [5], and 13.5% in São Carlos [6]. The Brazilian Public Health System (SUS) consists of a network of district or basic health care units, hospitals, laboratories, and blood banks essentially institutionalized to provide all Brazilian citizens with comprehensive care. diabetes and level of health care increased. Patients with both microvascular and macrovascular complications had higher costs (US$3199 per patient) compared to those with either microvascular (US$2062 per patient) or macrovascular (US$2517 per patient) complications only. The greatest portion of direct costs was attributed to medication (48.2%). Conclusions: Diabetes treatment leads to elevated costs both to Brazilian Public Health Care System and society. Costs increased along with duration of disease, level of care and presence of chronic complications, which suggested a need to reallocate health resources focusing on primary prevention of diabetes and its complications. Keywords: Brazil, costs, outpatient care, type 2 diabetes. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. Because direct and indirect costs of diabetes in Brazil are unknown, public health policymakers and managers cannot properly assess the real needs of patients with diabetes. Therefore, different management approaches are inadequately evaluated and resources are not correctly and sufficiently allocated. Moreover, the difficulty in accessing health services makes it even harder to measure the actual burden of diabetes in Brazil. In 2007, the Brazilian Society of Diabetes created a working group with the purpose of estimating diabetes care costs in Brazil. Initially, the group conducted the Brazilian Study on Diabetes Costs (ESCUDI study) to estimate direct and indirect costs of outpatient care of a selected sample of type 2 diabetes patients in the SUS. Subjects and methods A retrospective study based on data collected from different levels of care (primary, secondary, and tertiary care units) in eight Brazilian cities was carried out during the year 2007. Those cities were selected based on geographical criteria (northwest, south, and southeast), where health services were fairly organized, medical Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: L. R. Bahia, Rua Visconde de Pirajá 547-501, Ipanema, Rio de Janeiro, ZIP 22410-003, Brazil. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.009 S138 VALUE IN HEALTH 14 (2011) S137–S140 records were available at different levels of care, and research on diabetes was being conducted by local experienced investigators. The study design included 50 patients per level of care, a total of 100 patients from medium-sized cities (Bauru, Itabuna, and Campina Grande) and 150 patients from large cities (Fortaleza, Rio de Janeiro, Porto Alegre, Ribeirão Preto, and São Paulo). In the city of São Paulo, the only patients included were the ones from the secondary and tertiary levels; altogether the final sample comprised 1000 patients. The inclusion criteria were being older than age 30 years and having been followed-up at the health center for at least 1 year. Exclusion criteria were being pregnant, older than age 75 years, or having other diseases (human immunodeficiency virus, cancer, or hepatic failure) that could significantly affect the analysis of resource use. Patients scheduled to attend the centers were randomly selected for a face-to-face interview, which consisted of a questionnaire developed by the main research group. The basis of such research instrument came from a Pan American Health Organization questionnaire, which had been previously used in the Caribbean. The questionnaire was tested with 15 individuals and amendments were made. It was administered in two phases: patient data were collected through personal interviews, and then data were collected by systematically reviewing medical records. The subjects were classified into groups: those who had microvascular (retinopathy, neuropathy, or nephropathy) and/or macrovascular (cerebrovascular, coronary, or peripheral artery disease) complications. Such subdivision could only be possible once they had reported signs or symptoms of any of those diseases or had information in their medical records. To participate in the study, all subjects read and signed an informed consent form. The study was previously approved by the Research Ethics Committee of the University of São Paulo/Ribeirão Preto campus. Investigators conducted the research in accordance with the principles of the Declaration of Helsinki. Estimated costs The economic effects of outpatient diabetes care were assessed from the SUS and society perspectives over the course of 1 year. Direct and indirect costs were taken into account. There were costs in Brazilian currency (R$) attributed to all resources utilized for the year 2007, which were then converted into US dollars during the analysis using a purchasing power parity basis for 2005: US$1 ⫽ R$ 1.4 [7]. Direct costs were estimated using a bottom-up approach for primary data collection [8] and divided into medical and nonmedical costs. The direct medical costs assessed included medications, diagnostic tests, procedures, medical supplies (such as blood glucose test strips), visits with health professionals (physicians, nurses, nutritionists, physical therapists, dentists, and psychologists), and hospital costs for emergency room visits (including provider fees only). Medications used were categorized into four groups: diabetes and obesity, cardiovascular and dyslipidemia, psychiatric, and others (all other classes of medications). The direct nonmedical costs assessed included expenses with artificial sweeteners and diet products, patient transportation to attend clinic visits and laboratory testing, and expenses for hiring a temporary caregiver during a patient recovery period. Due to the wide variety of urban means of transport available (bus, train, subway, car), a minimum amount corresponding to two regular bus fares (US$3.10) was set to assess transportation expenses. Total direct outpatient costs of diabetes care were defined as the sum of all direct medical and nonmedical costs. Primary data collection on hospitalization costs was not performed. Medical procedures, health professionals’ visits, and supplies were assessed through reimbursements given to health centers based on SUS management of Procedures, Medications, and Special Supplies and Equipment System [9]. As for the assessment of medications covered by SUS, the weighted mean of the last three medication purchases made by public health units in 2007 was based on the Brazilian Ministry of Health price database [10]. Medications purchased at discount pharmacies cost nearly as much as those covered by SUS. Medications purchased at private pharmacies included the price of generic medications, if available; if not, the corresponding brand-name drug was obtained online through a publication on medication prices with an added tax of 18% [11]. The indirect costs assessed were absenteeism and resulting loss of productivity for the patients and their caregivers, sick leave, and early retirement. The human capital approach was adopted to estimate indirect costs. The mean monthly income and the value of working hours were used to calculate workdays lost [8]. Statistical analysis Data collected through questionnaires were electronically compiled by Epi-info 2000 (Centers for Disease Control and Prevention, Atlanta, GA). The statistical analysis was performed using SAS/ STAT version 9 (SAS Institute, Inc, Cary, NC). Variance analysis models were constructed to compare quantitative variables from the different levels of care, diabetes complications, and different disease durations. Fisher’s exact test was carried out for comparison. A 5% significance level was set. Results Sample characteristics There were a greater proportion of women receiving medical care (66.5% women vs. 33.5% men). Mean age was 59.0 ⫾ 9.1 years and mean duration of diabetes diagnosis was 11.2 ⫾ 7.8 years. The self-reported prevalence of coronary artery disease was 70.4%, with a higher prevalence at the tertiary level compared to the secondary and primary levels (P ⬍ 0.001). The prevalence of selfreported arterial hypertension (79.6%), cerebrovascular disease (7.6%), and hypercholesterolemia (70.4%) were similar at all different levels of care. Home blood glucose monitoring was reported by 34.3% of the sample (n ⫽ 343). Information on chronic diabetes complications was not available in 38% of the sample (n ⫽ 380) due to missing information in the medical records. Out of the remaining patients (n ⫽ 620), 28.9% had at least one microvascular complication (micro group; n ⫽ 289), 17% had at least one macrovascular complication (macro group; n ⫽ 171) and 16% had at least one microvascular and one macrovascular complication associated (micro/macro group; n ⫽ 160). The mean monthly income in the sample was US$543.35 and the income of each day of work was US$18.11. Direct costs Total direct outpatient cost of diabetes care was US$1335 per patient/year, out of which US$1014 per patient/year expended on direct medical costs and US$332 per patient/year on non-medical costs. (Table 1 in Supplemental Materials at: doi:10.1016/j. jval.2011.05.009). Total annual cost of medications for 1000 patients was US$747,356; US$563,506 paid by the public health system (SUS) (75.4%) and US$183,849 paid by patients in private pharmacies (24.6%). The expenses per patient/year with different medication groups were: US$249 with diabetes/obesity (n ⫽ 959); US$397 with cardiovascular/dyslipidemia (n ⫽ 905); US$321 with psychiatric conditions (n ⫽ 170), and US$285 with other medications (n ⫽ 335). (See Table 2 in Supplemental Materials at: doi:10.1016/j.jval.2011.05.009). The total cost of exams and procedures summed US$1216 per patient/year and the cost for health professionals’ consultations was US$794 per patient/year. VALUE IN HEALTH 14 (2011) S137–S140 The cost of home blood glucose monitoring was US$102,748 (US$299 per patient/year; n ⫽ 343). Annual expenses with artificial sweeteners and diet products amounted to US$258,617 (US$286 per patient). Expenses related to blood pressure monitoring were reported by 62 patients (US$59 per patient) and only 15 patients reported expenses with hiring a temporary caregiver (US$599 per patient). Patient transportation costs for attending consultations and/or undergoing diagnostic tests totaled US$50,964 (US$52 per patient). There was an increment in direct costs with diabetes progression and presence of diabetes complications (data not shown). Indirect costs Total indirect costs were US$773,212 (US$773 per patient/year), which corresponds to 36.7% of total diabetes costs. (See Table 1 in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.009). The lost of workdays corresponded to a loss of productivity of US$437 per patient/year from those who reported having an income (n ⫽ 829). A total of 32 (3.2%) and 74 (7.4%) subjects reported diabetesrelated sick leave and early retirement, respectively, associated with US$103,680 and US$410,702 expenses paid by the government. Job loss due to diabetes was reported by 127 subjects (12.7%). There was an increment in indirect costs with diabetes progression and presence of diabetes complications (data not shown). Total costs In the sample studied, the sum of direct and indirect costs amounted to US$2,108,287 for 1,000 patients per year (US$2,108 per patient). Total costs of diabetes care was US$ 1144 per patient at the primary care level, US$2445 at the secondary level and US$2810 at the tertiary level, with significant differences among total costs at the primary level and those at the secondary and tertiary levels (P ⬍ 0.01). (See Fig. 1A in Supplemental Materials found at: doi: 10.1016/j.jval.2011.05.009). In regard to diabetes duration, there was an increment in total costs as diabetes progressed: US$1971 per patient with diabetes duration 10 years or fewer; US$2173 per patient in those with diabetes duration between 10 and 19 years and US$2544 per patient in those with 20 years or more of diabetes duration (P ⬍ 0.01 among all the groups). (See Fig. 1B in Supplemental Materials found at: doi:10.1016/j.jval.2011.05.009). Total costs per patient from diabetes complications were as follows: US$2062 per patient in the micro group (n ⫽ 289); US$2517 per patient in the macro group (n ⫽ 171) and US$3199 per patient in the micro/macro group (n ⫽ 160) (P ⬍ 0.01 among all the groups). (See Fig. 1C in supplemental Materials found at: doi:10.1016/j.jval.2011. 05.009). Discussion ESCUDI is the first study conducted in Brazil showing a breakdown of costs of type 2 diabetes outpatient care in the SUS. Data were collected in eight cities with the purpose of describing health care in different Brazilian regions (southern, southeastern, and northeastern). Total cost of diabetes outpatient care during a 1-year period was US$2108 per patient/year, out of which 63.3% was direct costs and 36.7% was indirect costs. As expected, costs increased as diabetes progressed (23% cost increment in those with 10 or fewer years of diabetes duration compared to those with diabetes duration of 20 years or more) and with the presence of chronic complications (25% cost increment in those with both complications compared to those with only one microvascular or macrovascular complication). Similar results were described in a previous cost analysis of chronic diabetes complications [12]. S139 There was no difference in direct costs between secondary and tertiary levels of care, but in general, costs were higher than the primary care level, where majority of patients are being treated. A great and significant portion of costs (36.7%) were due to indirect costs of diabetes, which also include hidden expenditures paid by society that are usually little known and not included in government budget analyses. When comparing the annual costs made by the individuals (nonmedical direct costs, purchased medications, and loss of productivity) with their mean annual income (US$6520), we could demonstrate that 14.4% of their income was used for treatment. The greatest portion of direct costs was attributed to medication (48.2%). Although the Brazilian health system has the responsibility to provide drug treatment for any chronic disease, 24.6% of the patients bought medicines from private pharmacies. Previous studies on costs of diabetes clearly show that the greatest proportion of costs is attributed to hospital admissions. The CODE-2 study [13] demonstrated the medical costs of diabetes care in eight European countries based on primary data collection and database analysis. The estimated cost per patient per year was €2834 and the largest portion of these costs was attributed to hospital admissions (55%), mostly due to chronic diabetes complications, whereas only a small portion accounted for medication (4%). Those European results are hardly comparable to our Brazilian ones because these two studies had different approaches and the former included costs of hospital admissions, which are very high. The most recent American study on diabetes costs [14] shows an estimated cost of US$174 billion in 2007, out of which US$116 billion was spent on health resources and US$58 billion was attributed to loss of productivity. Similar to the CODE-2 study [13], the largest portion of costs were attributed to hospital admissions (50%), followed by medication and medical supplies (12%) [14]. Again, these results cannot be appropriately compared to the Brazilian ones because different methods were used (top-down approach vs. bottom-up approach). The ratio of direct cost to indirect cost was similar in the Brazilian and American studies, reinforcing the heavy burden of diabetes to individuals and society. Rosa et al. [15] investigated the costs of public hospital admissions in Brazil during 2 years (1999 –2001) using diabetes as the main diagnosis and the method of attributable risk associated to all admissions. The estimated number of hospital admissions was 836,300 per year (49.3/104 inhabitants), with a cost of US$243,900 per year (14.4 million/104 inhabitants). Public expenditure related to hospital admissions of individuals with diabetes was expressive (2.2% of the Brazilian Ministry of Health budget) [15]. Our study has some limitations, such as lack of knowledge on the actual situation of patients with diabetes regarding their regional distribution, access to care, and disease severity, as well as the fact that cities were randomly selected for data collection and are not representative of the whole country. Because the patient selection process was not ideal, some selection bias could be observed, such as that the majority of included patients were women or workers. Moreover, considering the lack of information on chronic complications in medical records, we analyzed the results separately. By doing so, we conclude that, regarding level of care, age, social level, or other identifiable factor that can be considered a systematic selection bias, there were no differences between those individuals with available data which allowed the classification of the presence of chronic complications (n ⫽ 620) and those without (n ⫽ 380). For the above reasons, results obtained in this sample may not accurately reflect average costs of diabetes care in the whole country. As a cost-of-illness study, the results were only descriptive, without comparisons to other populations. The possibility to compare the costs in a similar group without diabetes would in fact be much more informative to managers of the SUS. However, there are no published cost studies with primary data collection of the general Brazilian population that could be used to compare to our results. Because our S140 VALUE IN HEALTH 14 (2011) S137–S140 study did not collect data from patients without diabetes, we cannot ensure that the costs were attributed exclusively to diabetes. The findings of our study indicate a serious economic threat posed by diabetes that public authorities and all social sectors have to face. The noninclusion of hospital costs in the analysis resulted in an underestimate of the actual total cost of diabetes care in Brazil. The high percentage of indirect costs reveals hidden losses posed by diabetes that need to be taken into account. However, even with the inclusion of nonmedical and indirect costs in the study, not all the harm caused by diabetes were considered. There are intangible costs such as pain, suffering, and loss of quality of life that were not shown. The increasing incidence and prevalence of diabetes evidenced by epidemiologic studies worldwide represents a growing burden that most health systems are unable to deal with. There is an imperative need to develop and improve interventions toward prevention of diabetes and its complications and to reorganize resources to improve the effectiveness of health care. If this is not done, the public financing of diabetes treatment will be unviable in a near future, with deleterious consequences for the health of the Brazilian population. Acknowledgments The authors thank Dr Marcus Tambascia, Dr Walter Minicucci, and Dr Leão Zagury for the encouragement received for this project; Ms Anna Maria e Ms Kariane from the Brazilian Diabetes Society for her efforts; and Daiane Guaina and Henrique Ceretta for providing statistical analysis. The authors also thank the ESCUDI investigators: Alberto José C. Ramos, Daniel G. M. S. Leão Júnior, Debora F. B. Leite, Fábio C. V. Nascimento, Gewdy D. Lima, Giovanny R. C. Vasconcelos, Graciele Sbruzzi, Graziela H. Pinto, Gustavo T. Barros, Jaquelini M. Sauer, Lorena L. Silva, Maria Cristina G. Matos, Maria Júlia C. L. N. Araújo, Maria Roseneide S. Torres, Natallie M. Monteiro, Neila S. Amaral, Pâmela S. Almeida, Paula Beatriz O. Soares, Paula S. Souza, Roberta A. Cobas, Rennah G. dos Santos, Tânia D. Soares, Thaíse S. Andrade, and Thiago M. Brito. Sources of financial support: Main support for this research was provided by the Brazilian Diabetes Society, with additional support from Sanofi Aventis and Eli Lilly pharmaceuticals. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi:10.1016/j.jval.2011.05.009, or if hard copy of article, at www.valueinhealthjournal.com/issues (select volume, issue, and article). REFERENCES [1] Wild S, Roglic G, Green A, et al. Global prevalence of diabetes: estimates for the year 2000 and projections for 2030. Diabetes Care 2004;27:1047–53. [2] Yach D, Stucker S, Brownell KD. Epidemiological and economic consequences of the global epidemics of obesity and diabetes. Nature Med 2006;12:62– 6. [3] Malerbi DA, Franco LJ. Multicenter study of the prevalence of diabetes mellitus and impaired glucose tolerance in the urban Brazilian population aged 30 – 69 yr. The Brazilian Cooperative Group on the Study of Diabetes Prevalence. Diabetes Care 1992;15:1509 –16. [4] Costa MT, Torquato G, Montenegro R, et al. Prevalence of diabetes mellitus and impaired glucose tolerance in the urban population aged 30-69 years in Ribeirão Preto (São Paulo), Brazil. Sao Paulo Med J 2003; 121:224 –30. [5] Schaan BD, Harzheim E, Gus I. Cardiac risk profile in diabetes mellitus and impaired fasting glucose. Rev Saude Publica 2004;38:529 –36. [6] Bosi PL, Carvalho AM, Contrera D, et al. Prevalence of diabetes and impaired glucose tolerance in the urban population of 30 to 79 years of the city of São Carlos, São Paulo. Arq Bras Endocrinol Metab 2009;53: 726 –32. [7] World Bank: 2005 international comparison program.Available from:http://siteresources.worldbank.org/ICPINT/Resources/summarytables.pdf. [Accessed May 10, 2009]. [8] Rice DP. Estimating the cost of illness. Washington, DC: US Government Printing Office, 1966. Health Economic Series No 6. PHS Publication No. 947– 6. [9] Ministério da Saúde. DATASUS. Sistema de Gerenciamento da Tabela de Procedimentos, Medicamentos e OPM do SUS (SIGTAP). Competência: dezembro/2007. Disponível em. Available from: http:// sigtap.datasus.gov.br/Table-unificada/app/sec/inicio.jsp. [Accessed February 10, 2009]. [10] Brazilian Ministry of Health. http://bvsms.saude.gov.br/php/index. php. [Accessed February 10, 2009]. [11] Revista de ciencia y tecnologia para la farmacia del siglo XX1. Available from:http://www.revistakairos.com/revista/bra/default_ bra.asp [Accessed February 10, 2009]. [12] Williams R, Van Gaal L, Lucioni C. Assessing the impact of complications on the costs of Type II diabetes. Diabetologia 2002;45(Suppl.):S13–7. [13] Jönsson B. Revealing the cost of Type II diabetes in Europe. Diabetologia 2002;45(Suppl.):S5–12. [14] American Diabetes Association. Economic Costs of Diabetes in the US 2007. Diabetes Care 2008;31:1–20. [15] Rosa RS, Schmidt MI, Duncan BB, et al. Internações por Diabetes Mellitus como diagnóstico principal na Rede Pública do Brasil, 1999 – 2001. Rev Bras Epidemiol 2007;10:465–78. VALUE IN HEALTH 14 (2011) S141–S146 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval The Use of a Decision Board to Elicit Brazilian Patients’ and Physicians’ Preferences for Treatment: The Case of Lupus Nephritis Mirhelen Mendes de Abreu, MD, PhD1,*, Amiran Gafni, PhD2, Marcos Bosi Ferraz, MD, PhD3 1 Federal University of São Carlos, São Carlos, SP, Brazil; 2McMaster University, Hamilton, ON, Canada; 3Federal University of São Paulo, São Paulo, SP, Brazil A B S T R A C T Objectives: To find preferences for treatment expressed by lupus patients and physicians (who were asked to assume they have lupus) and to explore if certain variables explain these preferences. Methods: One hundred seventy-two patients and 202 physicians were interviewed using a lupus nephritis decision board that describes the treatment options and their potential benefits and risks. Clinical and sociodemographic variables were collected. Participants were asked to indicate their preferred treatment and provide justification for their choice. Descriptive statistics, t tests, and Pearson’s chi-square tests were used to determine the significance of differences in the decisions made by the two groups. A logistic regression model determined which factors contributed to treatment decisions. Results: The average age of study participants was 34 ⫾ 8 years for patients and 31 ⫾ 7 years for physicians. Sixty-eight percent of patients and 96% of physicians (P ⬍ 0.001) se- Introduction The growing involvement of patients in their health care decisions has given them a more active role in the patient–physician relationship [1,2]. Many studies have looked at this theme and show that different factors will influence patient preferences for their health care [3– 6]. Some studies have looked at the relevance of parameters such as pain, clinical history, socioeconomic context, and the willingness to accept the risk that clinical decisions might have adverse outcomes [6 –9]. Many authors have also pointed to discrepancies between patient and physician preferences in terms of the therapeutic handling of different diseases, including the techniques used to approach this point [10,11]. These issues are parameters that must be looked at from the point of view of preference-based medicine [5–11]. Systemic lupus erythematosus (SLE) is a chronic inflammatory autoimmune disease of varying progress and prognosis [12]. Lupus nephritis (LN) is the renal complication of SLE. It occurs in 40% to 60% of patients at some stage of the disease [12,13]. Survival has increased in recent decades due to a better understanding of the mechanisms of the disease and the therapeutic arsenal available [12,14]. The target of treatment is to promote remission of renal disease and impede its progression to endstage renal disease [12,13]. All of the prognostic studies have identified LN as a predictor of a negative outcome [12– lected the oral option. Patients and physicians justified their choice of treatment using different arguments (P ⬍ 0.001 in each case). Logistic regression showed that risk potential (P ⬍ 0.001) and a history of joint involvement (P ⫽ 0.011) were the arguments used most often to explain a patient’s decision and the risk of side effects was most relevant among physicians (P ⬍ 0.001). Conclusions: Using a decision board, patients and physicians were found to have different preferences for treatment when faced with the same treatment options. Further, the variables that influence their preferences are different. Keywords: decision aids, patient preference, shared decision making, systemic lupus erythematosus. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. 14]. Therapeutic handling of LN remains controversial due to the different risks and benefits resulting from the use of the immunosuppressive drugs available to treat LN [12–14]. Several studies have been exploring patient preferences on the context of lupus disease. However they usually do not use a decision tool to help the process of decision making [15–21]. A decision support tool, such as a decision board (DB), displays a number of clinical issues and the consequences of each option to enable solving the issue using a process that is both standardized and free of bias [22]. The DB is a simple tool that displays more than one therapeutic option to be valued and decided on [22–25]. It is low cost and can be easily updated. The development and validating of a Brazilian DB for LN for patients in Brazil was the first step in our study and has been described in a previous article [24]. Although several studies have attempted to assess the existence of discrepancies between physician and patient preferences regarding the same health care issue, most have used physicians who specialize in the disease in question [15–21]. In our study, we deliberately did not use rheumatologists for the reasons explained in more detail below. The objectives of this study were: 1) to find out the preferences for treatment expressed by SLE patients and by physicians who were asked to assume that they themselves had SLE, when faced with two options to treat LN using the Brazilian DB for LN, and 2) to Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Mirhelen Mendes de Abreu, Federal University of São Carlos, Medicine Department, Washington Luis Highway, Km 235, SP 310, São Carlos, Brazil 13565-905. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.015 S142 VALUE IN HEALTH 14 (2011) S141–S146 Fig. 1 – Schematic representation of the decision board. Initially, only Chart A if fully exposed. Charts B and C are shown with only the subtitles visible. The content is filled out during the course of the interview. (A) Summary presentation of disease information, with an emphasis on lupus nephritis. (B) A description of the treatment options: method of administration, likelihood of remission and a list of the eight potential side effects that the literature reports as having the highest prevalence. The interviewee then selects, from the list of eight side effects, those which she finds most bothersome. (C) Data on the probability of experiencing each of the three side effects selected by the individual interviewee as the most undesirable. explore if certain variables explain these preferences and if these variables differ between the patients and the physicians. Materials and Methods This was a cross-sectional study [26]. Patients were selected consecutively on the date they were seen at a university tertiary hospital clinic, which is an authority in the country on the care of complex diseases like SLE. Over the period of one year, patients and physicians were invited to take part in the study. Nonrheumatologist physicians were selected from the personnel files of the same institution. An Email message was sent to each of these physicians describing the nature of the study and inviting them to come in for an interview. Those who agreed to participate and met the criteria for selection were included in the study. Interviews with consenting physicians were scheduled during normal working hours at the hospital. The following criteria were used to include patients: women between ages 18 and 50 years with a 12-month or longer diagnosis of SLE according to the modified classification criteria published by the American College of Rheumatology and any clinical manifestation of the disease. In addition, patients had to be able to read and write, according to their own statement, agree to participate in the study, and sign a statement of free and informed consent. Patients presenting active lupus psychosis or any form of cognitive disability that would make it difficult for them to understand the questionnaires were excluded, as were patients whose records were unavailable at the time of the interview. The following criteria were used to include physicians: female academic physicians, practitioners, and physicians in training. Rheumatologists were excluded from the study to avoid a bias based on specific knowledge of the disease or the practices in the services they work for. We only recruited female physicians because the disease primarily affects women. The LN DB was used during individual interviews [25]. The content was presented and added to during the course of the interview (Fig. 1). At the beginning of the interview participants were given a summary presentation of the disease. Following this, they were given information on the treatment options (Option 1- oral treatment, or Option 2 - intravenous treatment), including methods of administration, the chances of remission, and a list of the eight most prevalent side effects described in the literature. We did not include the name of the drugs to avoid bias. Option 1 (oral) referred to therapy with mycophenolate mofetil, and Option 2 to therapy with intravenous cyclophosphamide. Study participants were asked to select the three side effects that most bothered them out of the list of eight. In the last phase of the DB, participants were told the probability of each of the selected side effects for each treatment option. Following the standard presentation, the content was reviewed and interviewees were asked to select one of the treatment options. After this they were asked to justify their decision based on a list of preselected alternatives. Justifica- S143 VALUE IN HEALTH 14 (2011) S141–S146 tions were classified under one of four options: risk, effectiveness, risk/benefit trade-off, and practicality. Risk justifications were those based on the probability of each of the three side effects for each treatment option. Effectiveness justifications were those where participants justified their treatment choice based on the probability of disease remission. Risk/benefit trade-off justifications were those where patients, when choosing their preferred option, tried to consider all of the characteristics of both treatment options (i.e., their potential risks and effectiveness). Practicality justifications were those where the most important factor in choosing the treatment was how the drug was administered and how it would affect the interviewee’s day-to-day activities [15–25]. The methodology used to categorize the justifications was presented in an article on developing and validating the Brazilian LN DB [25]. Following the DB presentation, patients were asked to fill out the clinical and health related quality of life questionnaire. The clinical questionnaires used were Systemic Lupus Erythematosus Disease Activity Index [SLEDAI] [27], and Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SLICC–DI) [28]. Health-related quality of life was assessed using the Medical Outcomes Study 36-Item ShortForm Health Survey (SF-36). Previous studies had already translated and tested the validity of the Portuguese language version of this tool [29,30]. Socioeconomic variables were assessed using the Brazilian Economic Classification Criterion (BECC) questionnaire, a validated tool used to classify people into socioeconomic categories (ranging from A to E, with A being the highest) [31]. In our study we wanted to test the hypothesis that patients and physicians will choose different treatments for LN when presented with the same clinical scenario using the DB. This hypothesis was formulating based on the following premises: 1) the socioeconomic and cultural contexts are different for the physician and patient groups included in the study; and 2) physicians tend to have some technical information about the disease and a different frame of mind in term of understanding of diseases in general. By excluding rheumatologists we believe we came as close as possible to contrasting the medical approach with the lay person approach to assessing the value of a treatment. Initially we assessed the preferences of each group for treatment (ie, choice of preferred treatment) following the DB presentation. We then tested if a statistical difference between treatment preferences exists, after which we tried to find out which variables might explain the treatment preferences of each group. The following variables were used to test the hypothesis: 1) decision variables (i.e., side effects selected, treatment decision, and justification); 2) socioeconomic variables (i.e., income, the BECC, and years of schooling); 3) clinical variables (i.e., severity). In this study, we defined severity as two or more of the following: prior and/or current use of cyclophosphamide; prior and/or current use of other immunosuppressive agents; increased disease activity defined as an SLEDAI score of eight or more; and increased chronicity, defined as an SLICC-DI score of five or more. Severity was adopted as a dichotomous variable. The project was approved by the Committee for Ethics in Research of our institution. Statistical analysis This study belongs to a larger one. We developed and tested a DB and a willingness-to-pay tool to assess strengh of preference. The sample was calculated based on prior similar studies and an equal distribution of the values to be offered in the willingness-to-pay survey. The estimated sample was 150 interviewees for each group [28,32]. The descriptive analysis summarizes the qualitative and quantitative variables as appropriate. Patient and physician results were compared using Pearson’s chi-square and Student t tests [32]. For the purposes of this analysis, side effects were Table 1 – Demographic and socioeconomic characteristics of the study population*. Patients 1. Age – average (SD) 2. Race – N (%) Whtie Non-white 3. Marital status (N%) Single Married Divorced/separated Windowed Other† 4. No. of children 0–1 2–3 ⱖ4 5.BECC (income, in US$ equivalent) A1 (3000,00) A2 (2100,00) B1 (980,00) B2 (550,00) C (327,00) D (212,00) E (103.5,00) Physicians 34.3 (8) 31.0 (7) 68 (39.5) 104 (60.5) 136 (67.0) 68 (33.0) 58 (33.5) 95 (55.2) 16 (9.3) 2 (1.1) 1 (0.5) 147 (73.0) 50 (25.0) 0 (0.0) 2 (1.0) 3 (1.0) 103 (59.1) 62 (35.9) 7 (4.0) 185 (91.6) 16 (7.9) 1 (0.5) 0 (0.0) 2 (1.2) 7 (4.1) 31 (18.0) 87 (50.5) 44 (25.5) 1 (0.6) 41 (20.3) 67 (33.2) 50 (24.8) 32 (15.8) 12 (5.9) 0 (0.0) 0 (0.0) P ⬍ 0.001 for all comparisons. BECC, Brazilian Economic Classification Criterion. * Values are the number (%) unless otherwise indicated. † Other: Assigned (government donation, squatter’s UNIFESP resident housing). rights, grouped into risk of death that was also presented as a dichotomous variable (yes/no). Univariate analyses were completed to identify associations between variables and LN treatment decisions. Logistic regression analyses assessed which factors influence the treatment decision in each group. We assumed that Option 1 and Option 2 are the dependent variables. As explanatory variables we included side effects, justification for the option selected, socioeconomic variables, years of schooling, clinical severity variables (yes/no), and quality of life (SF-36 domains) [32,33]. Only associations found to be significant with univariate analyses were included in the logistic regression [30,31]. For all statistical tests we considered a level of significance of 5% or lower [32,33]. All of the statistical analyses were conducted using SPSS version 13.0 (SPSS Inc., Chicago, IL). Results We interviewed 172 patients and 202 physicians during 1 year. The average age was 34 ⫾ 8 years for patients and 31 ⫾ 7 years for physicians. All of the patients approached agreed to participate in the study and 8% of the physicians approached refused, all of them in training and claiming not to have time to participate. Everyone who participated in the survey provided data. Most of the patient group was made up of mulatto (i.e., Brazilian mixed race [35,36]) persons (45%) and most of the physician group (67%) was white. Demographic, quality of life, and clinical characteristics are described in Tables 1, 2, and 3, respectively. Regarding the BECC distribution, 76% of patients were classified as C and D and 18% as B2. In the physician group, the BECC distribution was class A2 (33.2%), followed by B1 (24.8%), and A1 (20.3%) (Table 1). All of the SF-36 domain scores were lower for patients than physicians (P ⬍ 0.001 for each comparison). The average time spent on the DB was 20 ⫾ S144 VALUE IN HEALTH 14 (2011) S141–S146 Table 2 – Health-related quality of life characteristic of the study population. Functional capacity Physical issues Pain Overall health Vitality Social aspects Emotional aspects Mental Health Patients (n ⫽ 172) Median (standard error) Physicians (n ⫽ 202) Median (standard error) 67,99 (1,91) 67,08 (2,07) 58,68 (1,92) 54,87 (1,69) 53,16 (1,77) 64,45 (2,19) 68,2 (2,16) 55 (1,82) 91,98 (0,63) 88,59 (1,33) 80,07 (1,37) 82,08 (1,01) 62,03 (1,23) 79,19 (1,46) 85,33 (1,33) 71,81 (1,13) Note: For all comparisons, P ⬍ 0.001. SF-36, The Medical Outcomes Study 36-Item Short-Form Health Survey. 11 minutes for patients and 9 ⫾ 4 minutes for physicians. The frequency of the three worst side effects selected by patients was: cancer caused by the drug (44.2%), hair loss (21.6%), and severe infection (19.1%). Among physicians the order of selected side effects was cancer caused by the drug (45.5%), severe infection (33.1%), and sterility (12.5%). The difference in side effects selected by the two groups was statistically significant (P ⬍ 0.001), as shown in Table 4. Option 1 was the preferred treatment choice by 68% of patients and 98% of physicians (P ⬍ 0.001). Patients and physicians justified their decision based on risk (47.7% and 68.9%), effectiveness (12.2% and 2.0%), risk/benefit trade-offs (2.3% and 22.3%), and practicality (37.8% and 5.9%). Analyses show these differences are Table 3 – Clinical characteristics of lupus erythematosus patients included in the study (N = 172). Characteristic 1. Duration of disease (y) ⬍1 1–6 7–13 ⬎14 2. Lupus nephritis Yes 3. Cyclophosphamide use prior to or during the study Yes 4. Immunosuppressant drug use prior to or during the study Yes 5. ACR classification criteria 1. Malar rash 2. Discoid rash 3. Photosensitivity 4. Oral ulcer 5. Arthritis 6. Serositis 7. Renal disorder 8. Neurological disorder 9. Hematological disorder 10. Immunological disorder 11. Anti-nuclear antibody ACR, American College of Rheumatology. n % 9 67 52 37 5.2 38.9 30.2 21.5 129 75.0 94 55.0 129 75.0 21 11 27 6 144 45 77 16 36 105 165 12.2 6.4 15.6 3.5 83.7 26.0 44.8 9.3 21.0 61.0 96.0 Table 4 – Distribution of patient and physician selections regarding the scenarios presented by the decision board*. Characteristics Patients Physicians (n ⫽ 172) (n ⫽ 202) 1. Side-effects selected – N (5) Latrogenical cancer Sterility Severe infection Mild infection Nausea Vomiting Diarrhea Hair loss 2. Treatment decision Option 1 (Yes) 3. Treatment decision justification Risk Effectiveness Risk-benefit trade-off Practicality P 137 (79.7) 42 (24.4) 116 (67.4) 14 (8.1) 39 (22.7) 31 (18) 38 (22.1) 96 (55.8) 196 (97.0) 131 (64.9) 168 (83.2) 5 (2.5) 6 (3) 22 (10.9) 24 (11.9) 53 (26.2) ⬍0.001 ⬍0.001 ⬍0.001 0.013 ⬍0.001 0.049 0.008 ⬍0.001 117 (68%) 196 (98%) ⬍0.001 82 (47.6) 21 (12.0) 4 (2.0) 66 (38.) 141 (69.8) 4 (1.9) 45 (22.4) 12 (5.9) ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 * Values are number (%) unless otherwise indicated. statistically significant in each case (P ⬍ 0.001) (Table 4). Based on a univariate analysis, the variables selected for the multivariate patient model were: cancer, severe infection, decision justification, SLEDAI, skin involvement, joint involvement, maternity (yes/ no), number of inhabitants, socioeconomic class, income, and SF36. The final model was adjusted excluding variables step by step to arrive at a reduced model. Multivariate analysis shows that patient decisions were guided primarily by the potential risk posed by the drug, represented by a risk based justification, compared to an effectiveness-based justification (P ⬍ 0.001; odds ratio 31.8; 95% confidence interval 8.2–122.9) or a practicality based justification (P ⬍ 0.001); OR 6.0; 95% CI 2.5–14.2). Patients with prior joint involvement were less likely to select Option 1, compared with patients with no joint involvement (P ⫽ 0.011; OR 5.3; 95% CI 1.4 –19.5) (Table 5). In the physician group, the same explanatory variables were used to build the univariate model, with the exception of the clinical variables. This model showed that physicians who justified their decision based on risk tended to select Option 1, compared to Table 5 – Logistic regression model to assess the factors that influence lupus nephritis treatment preferences among real patients and physicians exposed to the decision board scenarios. Variable* Patients Justification Risk/effectiveness Risk/practicality Joint involvement Socioeconomic category C/AB D/AB Physicians Risk/practicality P Odds ratio 95% Confidence interval ⬍0.001 0.001 0.011 31.8 6.0 5.3 8.2 2.5 1.4 0.073 0.024 0.3 0.2 0.11 0.07 0.007 * Dependent variable: Option 1. 64 3.12 122.9 14.2 19.5 1.1 0.8 1332 VALUE IN HEALTH 14 (2011) S141–S146 those who made their decision based on practicality (P ⫽ 0.007), OR 64; 95% CI 3.1–1332). (See Table 5.) Age can also influence the decision: the older the study participant the more likely she was to select Option 1 (P ⫽ 0.08; OR 2.28; 95% CI 0.88 –5.8). Discussion This study addresses the preferences for treatment of two different groups facing the same scenario. Both groups consisted of individuals of the same sex and of a similar age group. The other demographic characteristics are different for the two groups, among them socioeconomic level and BECC scores (Table 1) [31]. The health-related quality of life profile also yielded different results for both groups. The patient group had the lowest SF-36 scores in all domains, probably as a result of the disease itself (Table 2). Because the institution was a tertiary center, the severity of the disease reflected this, as shown in Table 3. The two groups selected different side effect as being more important (Table 4). Regarding the treatment options, both groups tended to prefer Option 1. Results showed that the decisions were based on different justifications (P ⬍ 0.001), which were a reflection of what is important for each interviewee when making treatment decisions (Table 5). We asked physician to imagine they were patients (i.e., to imagine that they have LN). This strategy has strengths and limitations. The strengths lie in the fact that, by considering themselves patients, potentially the physicians will make their decisions from the point of view of a patient. Different from other studies that have assessed the preference of physicians and patients, we tried to assess the physician choice, encouraging them to place themselves on the other side and make their decisions accordingly [15–21]. By excluding rheumatologists, we avoid decisions that are influenced by clinical practice and well known evidence. In this way, we can elicit physician preferences and understand how they make their own health care decisions. This information may help us understand how medical knowledge and training can influence choices. This in turn can contribute to the design of education strategies that can help both patients and physicians communicate and deliberate while attempting to choose the best treatment for the patient [15–25]. This study was structured to simulate the practical clinical reality of the institution where it was developed. A previous study done in Brazil explored patients’ and physicians’ perceptions about rheumatoid arthritis care. A study by Ferraz et al. [34] observed that patient and rheumatologists had different opinion about the health care provided. This study had a limitation that the answers provided by the rheumatologists who participated in the study were based on the guidelines used during that period of time rather than on what really happened [18 –21,34]. A limitation of our study is that decisions made based on a hypothetical scenario may not necessarily reflect the decisions that would be made in a real situation. We believe, however, that the strengths outweigh the potential limitation. In addition, patients included in the study faced a hypothetical situation because knew they were making a decision as part of a scenario, and not an actual decision regarding their own treatment. This being the case, both groups faced hypothetical situations [15–21]. Regarding the side effects chosen, although the side effect most often selected as being the worst was cancer by both groups (44% for patients and 45% for physicians), the rest were quite different (P ⬍ 0.001). We still do not understand the factors that could influence these decisions [25]. Other studies that assess patient preferences suggest similar results, although none made use of a support tool. There is a need for studies to explore this area [25]. As a measure of agreement between patients and physicians, Kappa test was used to compare both groups’ choices regarding the three side effects chosen and the final decision. For the three S145 side effects chosen, Kappa results were 0.334, 0.202, and 0.220 (P ⬍ 0.001), respectively. In respect to the final decision, Kappa coefficient was 0.757 (P ⬍ 0.001). Both groups tended to prefer Option 1. This was, however, stronger among physicians and it was statistically significant (98% vs. 68% for patients; P ⬍0.001). There are also significant differences in their justifications (P ⬍ 0.001). Also, this study showed that both groups have in common the fact that the variable “decision justification” was the one that best explained the decision regarding treatment options. It can allow us to reflect that, even if the justifications used by the two groups were different, their influence on the decisions made reflect their concerns regarding the potential consequences of these decisions. These findings also show that different factors influence patient and physician decisions, as shown in the decision justifications [26,35– 44], and may be one of the factors explaining the miscommunication observed in patient–physician encounters [35,36,43– 45]. Regarding the other factors that influence preferences, study results also show that joint involvement and possibly socioeconomic class influence patient treatment decisions. These variables may reflect the impact of the chronic nature and how this influences patient decisions. This interpretation is speculative. The variables that explain what influences patient preferences are far from being completely understood and further research is called for to explore this issue. Among physicians, household income and BECC scores did not influence treatment decision. These data may reflect the homogeneous socioeconomic and cultural character of this group compared to the patient group, which is socioeconomically more diverse and made up predominantly of persons in socioeconomic categories C, D, and E [31]. One issue that may have influenced this group’s decision was the likelihood of the selected side effects. This strategy enabled making up a scenario of a severity proportional to the events mentioned [37– 41]. Our results show that physician and patient decisions are influenced by different factors, leading to potential discrepancies when facing the same issue to be assessed, namely the treatment of LN. Considering the chronic nature of SLE and its clinical peculiarities, we must question if such differences compromise communication between physician and patient in clinical practice, leading to poor treatment compliance [37– 45]. Conclusions Physician and patient decisions are influenced by different factors, leading to potential discrepancies when facing the same issue to be assessed. Treatment choice has to be discussed with patients, because individual preference seems to be determined by personal attitudes toward safety and convenience, by past experience, and by the perception of current disease status. Acknowledgments The authors thank Dr. Ruy Geraldo Bevilacqua and Angela Paes for help with statistics. Source of financial support: São Paulo Research Foundation (Fundação de Amparo à Pesquisa do Estado de São Paulo). REFERENCES [1] Elwyn G, Edwards A, Eccles M, Rovner D. Decision analysis in patient care. Lancet 2001;358:571– 4. [2] Aymé S, Kole A, Groft S. Empowerment of patients: lessons from the rare diseases community. Lancet 2008;371:2048 –51. [3] Eddy DM. Rationing by patient choice. JAMA 1991;265:105– 8. S146 VALUE IN HEALTH 14 (2011) S141–S146 [4] Hjermstad MJ, Fossa SD, Oldervoll L, et al. Fatigue in long-term Hodgkin’s disease survivors: a follow-up study. J Clin Oncol 2005;23: 6587–95. [5] Turk DC, Monarch ES, Williams AD. Cancer patients in pain: considerations for assessing the whole person. Hematol Oncol Clin North Am 2002;16:511–25. [6] Kendler D, Kung AW, Fuleihan GH, et al. Patients with osteoporosis prefer once weekly to once daily dosing with alendronate. Maturitas 2004;48:243–51. [7] Ryan M, Scott DA, Reeves C, et al. Eliciting public preferences for healthcare: a systematic review of techniques. Health Technol Assess 2001;5:1–186. [8] Slevin M, Plant H, Lynch D, et al. Who should measure quality of life, the doctor or the patients? Br J Cancer 1988;57:109 –12. [9] Rothwell PM, McDowell Z, Wong CK, Dorman PJ. Doctors and patients don’t agree: cross-sectional study of patients’ and doctors’ perceptions and assessments of disability in multiple sclerosis. BMJ 1997;314:1580 –3. [10] Suarez-Amazor ME, Conner-Spady B, Kendall CJ, et al. Lack of congruence in the ratings of patient’s health status by patients and their physicians. Med Decis Making 2001;21:113–21. [11] Befort CA, Greiner KA, Hall S, et al. Weight-related perceptions among patients and physicians: how well do physicians judge patients’ motivation to lose weight? J Gen Intern Med 2006;21:1086 –90. [12] D’Cruz DP, Khamashta MA, Hughes GR. Systemic lupus erythematosus. Lancet 2007;369:587–96. [13] Houssiau F. Thirty years of cyclophosphamide: assessing the evidence. Lupus 2007;16:212– 6. [14] Contreras G, Pardo V, Leclercq B, et al. Sequential therapies for proliferative lupus nephritis. N Engl J Med 2004;350:971– 80. [15] Fraenkel L, Bogardus ST, Wittink DR. Risk-attitude and patient treatment preference. Lupus 2003;12:370 –76. [16] Kwoh CK, Ibrahim SA. Rheumatology patient and physician concordance with respect to important health and symptom status outcomes. Arthritis Care Res 2001;45:372–7. [17] Neville C, Clarke AE, Joseph L, et al. Learning from discordance in patient and physician global assessments of systemic lupus erythematosus disease activity. J Rheumatol 2000; 27:675–79. [18] Alarcon GS, McGwin G Jr, Brooks K, et al. Lupus in Minority populations: nature versus nurture. Systemic lupus erythematosus in three ethnic groups. XI. Sources of discrepancy in perception of disease activity: a comparison of physician and patient visual analog scale scores. Arthritis Rheum 2002;47:408 –13. [19] Alarcon GS, Rodriguez JL, Benavides G Jr, et al. Systemic lupus erythematosus in three ethnic groups. V. Acculturation, health-related attitudes and behaviors, and disease activity in Hispanic patients from the LUMINA cohort. LUMINA Study Group. Lupus in minority populations, nature versus nurture. Arthritis Care Res 1999;12:267–76. [20] Yen JC, Neville C, Fortin PR. Discordance between patients and their physicians in the assessment of lupus disease activity: relevance for clinical trials. Lupus 1999;8:660 –70. [21] Aberer E. Epidemiologic, socioeconomic and psychosocial aspects in lupus erythematosus. Lupus 2010;19:1118 –24. [22] Charles C, Gafni A, et al. Treatment decision aids: conceptual issues and future directions. Health Expectations 2005;8:114 –25. [23] Whelan T, Gafni A, Charles C, Levine M. Lessons learned from the decision board: a unique and evolving decision aid. Health Expectation 2000;3:69 –76. [24] Curtis JR, Chen L, Harrold LR, et al. Physician preference motivates the use of anti-tumor necrosis factor therapy independent of clinical disease activity. Arthritis Care Res 2010;62:101–7. [25] Abreu MM, Gafni A, Ferraz MB. Development and testing of a decision board (DB) to help clinicians present treatment options to lupus nephritis patients in Brazil. Arthritis Care and Res 2009;61:37– 45. [26] Von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. PLoS Med 2007;4:1623–27. [27] Bombardier C, Gladman DD, Urowitz MB, et al. Derivation of the SLEDAI. A Disease Activity Index for lupus patients. Arthritis Rheum 1992;35:630 – 40. [28] Hochberg MC. Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus [letter]. Arthritis Rheum 1997;40:1725. [29] Gladman DD, Urowitz MB. The SLICC/ACR damage index: progress report and experience in the field. Lupus 1999;8:632–7. [30] Ciconelli RM, Ferraz MB, Santos W, et al. Brazilian-Portuguese version of the SF-36. A reliable and valid quality of life outcome measure. Brazilian J Rheumatol 1999;39:143–50. [31] Brazilian economic classification criterion. Available from: http://www.abep.org/codigosguias/ABEP_CCEB.pdf. [Accessed January 20, 2009]. [32] Walter SD, Eliasziw M, Donner A. Sample size and optimal designs for studies. Stat Med 1998;17:101–10. [33] Armitage P, Berry, G. Statistical methods in medical research. (3rd ed.). Oxford: Blackwell Science, 1994. [34] Ferraz MB, Ciconelli RM, Vilar MJ. Patient’s preference regarding the option of performing unselective liver biopsy following methotrexate treatment in rheumatoid arthritis. Clin Exp Rheumatol 1994;12:621–5. [35] Maio MC, Monteiro S, Chor D, Faerstein E, Lopes CS. [Ethnicity/race in the Pró-Saúde Study: comparative results of two methods of selfclassification in Rio de Janeiro, Brazil], in Portugese. Cad Saude Publ 2005;21:171– 80. [36] Mangels L, Neves L. Racial classification in Brazil: discrepancies between observed and self-identified race. Available from: http:// www.allacademic.com//meta/p_mla_apa_research_citation/1/8/2/2/8/ pages182287/p182287-1.php. [Accessed December 21, 2010]. [37] Berrios-Rivera JP, Street RL Jr, Garcia Popa-Lisseanu MG, et al. Trust in physicians and elements of medical interaction in patients with rheumatoid arthritis and systemic lupus erythematosus. Arthritis Rheum 2006;55:385–93. [38] Ross LF, Zenios S, Thistlethwaite JR Jr. Shared decision making in deceased-donor transplantation. Lancet 2006;368:333–7. [39] Verburg RJ, Mahabali SD, Stiggelbout AM, et al. High dose chemotherapy and hematopoietic stem cell transplantation: a study of treatment preference in patients with rheumatoid arthritis and rheumatologists. J Rheumatol 2002;29:1653– 8. [40] Garfield S, Smith F, Francis SA, Chalmers C. Can patients’ preferences for involvement in decision-making regarding the use of medicines be predicted? Patient Educ Couns 2007;66:361–7. [41] Rathore SS, Krumholz HM. Differences, disparities, and biases: clarifying racial variations in health care use. Ann Intern Med 2004; 141:635– 8. [42] Kilbourne AM, Switzer G, Hyman K, et al. Advancing health disparities research within the health care system: a conceptual framework. Am J Public Health 2006;96:2113–21. [43] Charles C, Gafni A, Whelan T, O’Brien MA. Cultural influences on the physician-patient encounter: the case of shared treatment decisionmaking. Patient Educ Couns 2006;63:262–7. [44] Yazdany J, Tonner C, Trupin L, et al. Provision of preventive health care in systemic lupus erythematosus: data from a large observational cohort study. Arthritis Res Ther 2010;12: R84. [45] Marko NF, Weil RJ. The role of observational investigations in comparative effectiveness research. Value Health 2010:13:989 –97. VALUE IN HEALTH 14 (2011) S147–S150 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jval Patrones de Tratamiento y Costo de Atención del Cáncer de Mama Avanzado Con Falla a Antraciclinas y Taxanos en 3 Hospitales Públicos de México Juan Alejandro Silva, MD1, Juan Francisco Gonzalez, MD2, Juan Enrique Bargalló, MD3, Gabriela Hernández-Rivera, MD4, Xóchitl Gómez-Roel, MD4, Sigfrido Rangel, MD4, Juan Jesús Vargas-Valencia5, Jonathan Martínez-Fonseca5, Bonnie Meyer Korenblat Donato, PhD6, Ariadna Juárez-García, PhD4,* 1 Hospital de Oncología Siglo XXI del IMSS, México D.F., México; 2Hospital Universitario “Dr José Eleuterio González” Monterrey, N.L., México; 3Instituto Nacional de Cancerología, México D.F., México; 4Bristol-Myers Squibb de México, México D.F., México; 5Econopharma Consulting S. A. de C. V, Mexico, Mexico D.F., Mexico; 6Bristol-Myers Squibb, Wallingford, CT, USA A B S T R A C T Objectives: In Mexico, breast cancer is the second leading cause of cancer mortality among females. For patients with advanced breast cancer (ABC) resistant to anthracyclines and taxanes (AT), there are limited treatment options. There is a scarcity of data regarding clinical management of this population and treatment costs at this stage of the disease. The objective of this study was to describe the treatment patterns of care for metastatic breast cancer after AT and the associated cost from the point-of-view of the Mexican Public Health Care Sector. Methods: Between January 1, 2004 and December 31, 2007, a retrospective cohort of adult female ABC patients resistant to AT was developed by reviewing and extracting key data from medical charts. We conducted a retrospective, transversal and descriptive analysis of the patient data. Target population data files were obtained from 414 patients from 3 public hospitals in México. Results: Capecitabine, vinorelbine and cyclophosfamide were the most commonly prescribed agents, however clinical drug therapy management of the disease was different Introducción En México, las neoplasias malignas representan un importante problema de salud. En el año 2006, los tumores malignos representaron la tercera causa de muerte con 63,888 defunciones. De ellos el cáncer de mama representa el 15% de las defunciones, FUCAM [1] e INEGI [2]. En el año 2002, Brandan y Villaseñor [3] reportaron 11,064 casos diagnosticados. Aunque en México esta neoplasia se reporta como la segunda con mayor incidencia, después del cáncer cervicouterino, varios estados de la República, lo reportan en primer lugar en mujeres de entre 40 a 54 años de edad, Secretaría de Salud [4], Cárdenas y Sandoval [5]. Se estima que únicamente el 10% de los casos son detectados en las etapas iniciales de la enfermedad, lo que conduce a un within and among the three hospitals included in the study. This difference translated into a disparity of prescription costs, ranging from an average of $122.22 pesos/patient/month (cyclophosfamide, IC 95% $94.43$150.01) to $37,835.53 pesos/patient/month (capecitabine⫹trastuzumab IC 95% $34,953.18-$40,717.88) for the first treatment after AT. Conclusions: The results highlight a lack of standardized care for patients and suggest that differences in treatment patterns are not only a reflection of scarcity of scientific data and diversity of prescription preferences among physicians but also of economic restrictions. Ultimately, there is a clear unmet medical need to be addressed through evidence-based medicine alternatives that support efficacy and cost effectiveness treatments. Palabras Claves: breast cancer, costs, health care utilization, treatment patterns. Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. diagnóstico tardío (etapas clínicas III y IV) y a un pobre pronóstico (progresión de la enfermedad de un 65%), más aun, 40% de las mujeres diagnosticadas con cáncer de mama en estapas clínicas II/III desarrollarán enfermedad metastásica, FUCAM [1]. En México, según el Plan Nacional de Seguro Médico dado a conocer por el presidente de la república, el costo del tratamiento de cáncer de mama es mayor a los 20 mil dólares por mujer, Puente [6], por lo que la necesidad de utilizar tratamientos efectivos basados en la mejor evidencia disponible es clara. La selección del tratamiento adecuado depende de diferentes factores, como: edad, estado funcional y enfermedades concomitantes, tipo de tratamiento adyuvante previo, intervalo libre de enfermedad, agresividad de la enfermedad, sitio, número y volumen de las metástasis, tratamiento previo y respuesta al mismo, receptores hormonales y sobre-expresión del Her 2-neu entre otros. Sin embargo, una vez que se han recibido Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. Título corto: Treatment patterns and costs in breast cancer. * Autor de correspondencia: Ariadna Juárez García, Gerente de Farmacoeconomía, Bristol-Myers Squibb México, Av Revolución No.1267, Col.Tlacopac, Del.Alvaro Obregón, México D.F. CP 01049; Tel: 52 (55) 53372842; Fax: 52 (55) 55931956. E-mail: [email protected]. 1098-3015/$36.00 – see front matter Copyright © 2011, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi:10.1016/j.jval.2011.05.029 S148 VALUE IN HEALTH 14 (2011) S147–S150 antraciclinas y taxanos, las opciones terapéuticas son más limitadas. Durante algunos años, Capecitabina fue el primer medicamento aprobado tanto en EUA como en México en pacientes con falla a antraciclinas y taxanos basado principalmente en la información obtenida en los estudios fase II, los cuales mostraron un beneficio en estos pacientes con una tasa de respuesta que varía de estudio a estudio de un 15 a un 50%, Supervivencia libre de progresión de 3.2 a 5.9 meses y una supervivencia global de 9.5 a 17 meses, Blum JL, et al. [7], Blum JL, et al. [8], Reichardt P, et al. [9] y Fumoleau P, et al. [10]. Desde el año 2007 en EUA y 2009 en México fue aprobado en este contexto de pacientes que ya fallaron a antraciclinas y taxanos la combinación de capecitabina con Ixabepilona basado en la eficacia mostrada en el estudio fase III en el cual se comparó ixabepilona ⫹ capecitabina versus capecitabina sola encontrando beneficio en cuanto a supervivencia libre de progresión (5.8 meses versus 4.2 meses p⫽0.0003) y tasa de respuesta objetiva (35% versus 14% p⬍0.0001) Thomas ES, et al. [11]. Vinorelbina tanto oral como inyectable está aprobada en México para su uso en pacientes con cáncer de mama, no así en EUA. La evidencia de la eficacia de este medicamento en pacientes con falla a antraciclinas y taxanos está basada principalmente en estudios fase II solo o en combinación con capecitabina con tasas de respuesta de un 20 a un 25% en monoterapia hasta un 58% en combinación, con una supervivencia libre de progresión de 3.4 a 9 meses y una supervivencia global de alrededor de entre 6 meses como monoterapia y hasta 27.2 meses en combinación. Martin M, et al. [12], Degardin M, et al. [13], Langkjer ST, et al. [14], Toi M, et al. [15] Estevez LG, [16], Ahn JH, et al. [17] y Winer EP, et al. [18], la mayor parte de la información está en relación a la vinorelbina inyectable, sin embargo la presentación oral a una dosis de 80mg/m2 se considera equivalente a la dosis intravenosa de 30mg/m2 Bourgeois H, et al. [19]. En cuanto a ciclofosfamida, la única información disponible en pacientes con falla a antraciclinas y taxanos es en combinación con otros medicamentos, especialmente combinado con metotrexate en un esquema metronómico el cual se refiere a la administración frecuente incluso diaria de agentes quimioterapéuticos a dosis significativamente menores a la máxima dosis tolerada, con pocos intervalos libres de medicamento. Esta forma de administración ha mostrado eficacia moderada con tasas de respuesta de alrededor de un 19% Colleoni M, et al. [20]. Sin embargo, podemos hipotetizar que en el contexto mexicano los pacientes con cáncer de mama avanzado posterior al manejo con antraciclinas y taxanos no reciben un tratamiento estandarizado basado en la mejor evidencia disponible, reflejo de esto, es el hecho de que las primeras guías de manejo del CMA en México fueron publicadas en el 2009. Con lo que surge la pregunta de investigación: ¿Cuáles son los esquemas de tratamiento citotóxicos empleados en el manejo del cáncer de mama avanzado posterior al manejo con antraciclinas y taxanos y el costo asociado desde la perspectiva de las instituciones de salud públicas? El responder a esta cuestión, proporcionará información que permita documentar los patrones de tratamiento en pacientes con cáncer de mama y potenciales beneficios clínicos asociados, el impacto económico del tratamiento del cáncer de mama en México, así como subrayar las posibles necesidades clínicas insatisfechas. Diseño del estudio Material y Métodos Definición de las variables Objetivo Conocer los esquemas de tratamiento citotóxicos empleados en el manejo de pacientes con cáncer de mama avanzado con falla a antraciclinas y taxanos, así como los costos asociados a nivel público en México. La identificación y cuantificación de los insumos y procedimientos para el costeo de los procesos realizados para el tratamiento del cáncer de mama avanzado con falla a antraciclinas y taxanos se realizó a través de un análisis transversal, retrospectivo y descriptivo, utilizando los expedientes clínicos de pacientes tratados de forma institucional. Los datos del examen médico, las características clínicas, tratamientos y usos de recursos fueron sistemáticamente recogidos a partir de la fecha de diagnóstico del cáncer de mama hasta la muerte o pérdida de contacto. La falla a antraciclinas y a taxanos se define como la progresión de la enfermedad con el inicio de un nuevo régimen de tratamiento a menos de 3 meses de la última dosis del régimen. Criterio de inclusión Los pacientes debieron tener un diagnóstico de cáncer de mama avanzado y tratamiento con antraciclinas y taxanos entre el 1 de enero del 2004 y el 31 de diciembre del 2007 y falla a los mismos. La elegibilidad en el estudio fue confirmada por la revisión de los expedientes clínicos. Criterios de exclusión Expedientes mal clasificados o incompletos. Lugar del estudio La población blanco se obtuvo a partir de 3 grandes centros hospitalarios de tercer nivel: Hospital de Oncología “Siglo XXI” del IMSS, Instituto Nacional de Cancerología y Centro Universitario contra el Cáncer – CUCC del Hospital Universitario “Dr. José Eleuterio González”. Por tratarse de una revisión retrospectiva de expedientes y no implicar ningún tipo de contacto con los pacientes, ninguno de los comités de éticas de las instituciones participantes solicitan consentimiento informado por parte de los pacientes. Tamaño de la muestra Se incluyeron 414 expedientes clínicos de pacientes con diagnóstico de cáncer de mama avanzado y tratamiento con antraciclinas y taxanos, del 1° de enero del 2004 al 31 de diciembre de 2007 y que cumplieron con los criterios de inclusión. El tamaño de muestra se definió como una proporción a partir del total de 15,109,348 casos estimados de cáncer de mama en México, esta cifra se estima considerando una prevalencia de 13,939 casos por cien mil habitantes según datos publicados por la Agencia Internacional para la Investigación del Cáncer (IARC, por sus siglas en inglés) publicados en GLOBOCAN [21] y una población estimada para México durante el año 2010 de 108,396,211 habitantes CONAPO [22]; la selección de los casos se realizó de forma aleatoria numerando el total de expedientes clínicos que cumplieron los criterios de inclusión en cada institución participante y se generaron la misma cantidad de números aleatorios con el software Excel 2007, la selección se realizó de acuerdo al orden de aparición en el listado generado. 1 n⫽ 1 ␦2 ⫹ 2 Zn⁄2 Sy2 n Donde: ␦ ⫽ Error absoluto ⫽ 0.05 N ⫽ 15,109,348 número total de caso en México n ⫽ 384 Las variables incluidas en la cédula de captura son: consultas médicas, medicamentos, exámenes de laboratorio y de gabinete, visitas del personal médico, radioterapia, consulta de urgencias y número de días de hospitalización. El instrumento diseñado recabó información para definir las características epidemiológicas de la muestra (genero, edad y lugar de origen, sitio de residencia). VALUE IN HEALTH 14 (2011) S147–S150 En cuanto a la enfermedad, se precisan los antecedentes asociados al cáncer de mama, la existencia de comorbilidades y los costos médicos directos, generados por concepto de atención a en la institución (hospitalización, estudios de laboratorio y gabinete). Análisis de los resultados clínicos Se reporta la estadística descriptiva con respecto a los patrones de tratamiento y la secuencia de la terapia, junto con los resultados asociados. Se describe la secuencia de las modalidades de tratamiento, incluyendo todos los fármacos administrados y utilización de recursos derivados de eventos adversos. Especialmente se identificó la frecuencia de uso de fármacos citotóxicos. Se describió como frecuencias y tasas, con posterior construcción del algoritmo de tratamiento. Análisis de costos de atención El costeo se realizó desde el punto de vista institucional, recolectando información de los costos médicos directos. La utilización de los recursos se determino tabulando todos los insumos y procedimientos registrados en el expediente clínico del paciente y, a continuación, se calcularon los costos médicos para cada paciente durante el tratamiento del cáncer de mama. El costo de los medicamentos se tomó de los costos publicados en el portal del Sistema Electrónico de Contrataciones Gubernamentales (Compranet: http://compranet.gob.mx). El costo de los procedimientos e insumos fueron tomados de los Costos Unitarios correspondientes al tercer Nivel de Atención Médica para el Instituto Mexicano del Seguro Social para el año 2009, publicados en el Diario Oficial de la Nación el viernes 6 de marzo de 2009. Todos los costos son reportados en pesos mexicanos. Aspectos éticos Se trata de un estudio basado en el análisis de información secundaria y se ajusta a los principios establecidos en la declaración “Helsinki V”. La información que se obtuvo de la revisión del expediente se protegió con un folio especial para mantener el nombre del paciente confidencialmente y además, cualquier resultado de éste que se brinde será siempre de forma general y nunca de un paciente en especial. Resultados Características basales S149 El tratamiento previo con antraciclinas y taxanos está reportado en la tabla 1 a Materiales Complementarios en: doi:10.1016/ j.jval.2011.05.029. Las antraciclinas se utilizaron principalmente en el contexto neoadyuvante y adyuvante, y los taxanos principalmente en contexto adyuvante y metastásico. La duración promedio del régimen de antraciclinas de los 3 centros fue de 4.92 ciclos y el de taxanos correspondió a 6.76 ciclos. La antraciclina más utilizada en los tres hospitales fue la epirrubicina (60.39%, IC 95% 55.0%- 65.8%) mientras que la doxorrubicina reportó una tasa más baja de uso (39.61% IC 95% 34.2%-45.0%). El taxano de mayor uso fue docetaxel (59.90% IC 95% 54.5%-65.3%) mientras que paclitaxel representó el 40.58% (IC 95% 35.2%-46%). Los tipos de recurrencia que se presentaron después de la terapia con AT fueron en un 93% de los casos metástasis a distancia y en un 25% recurrencia locoregional. Algunos pacientes presentaron ambos tipos de recurrencia. Tratamiento posterior a la falla con AT El tratamiento citotóxico, en las tablas 2 y 3 a Materiales Complementarios en: doi:10.1016/j.jval.2011.05.029 se observa la distribución de las alternativas utilizadas como primera y segunda línea después de haber fallado a AT, también se consideró dentro de las opciones incluidas en las tablas la terapia biológica en caso de haber sido utilizada. Como tratamiento de primera línea después de falla a AT, la opción más frecuente es capecitabina con un 54.92% (IC 95% 49.4%-60.4%) de los casos, con un costo mensual de $10,033.73 (IC 95% $9,055.00-$11,012.46) seguido de vinorelbina oral que representa el 11.75% (IC 95% 8.2%-15.3%) y un costo de $24,329.09 (IC 95% $23,297.38-$25,360.81); y en tercer lugar con 8.89% (IC 95% 5.7%-12.0%), vinorelbina inyectable, con un costo de $4,040.40 (IC 95% $3,539.08-$4541.72). El estudio demostró que la alternativa más costosa, trastuzumab ⫹ capecitabina, en esta etapa de la enfermedad asciende hasta $37,835.53 ( IC 95% $ 34,953.18- $40,717.88). Como alternativa de segunda línea el fármaco citotóxico más usado fue ciclofosfamida con el 29.77% (IC 95% 21.9%-37.6%) de los casos y un costo mensual de $114.11 (IC 95% $105.52- $122.69) seguido por vinorelbina oral con un 21.37% (IC 95% 14.4%-28.4%) de los casos y un costo de $25,074.64 (IC 95% $23,160.31-$26,988.96) y capecitabina con un 16.03% (IC 95% 9.7%22.3%) y un costo de $9,288.02 (IC 95% $8,782.12-$9,793.92). La alternativa más costosa de segunda línea fue la combinación de vinorelbina oral ⫹ trastuzumab inyectable, con un costo promedio de $51,960.86 ( IC 95% $47,401.19- $56,520.52). Los costos anteriores se estimaron tomando en cuenta las dosis utilizadas en cada paciente. Población de estudio y periodo observacional La población estudiada en las tres instituciones, (Hospital de Oncología “Siglo XXI” del IMSS, Instituto Nacional de Cancerología y Centro Universitario contra el Cáncer – CUCC del Hospital Universitario “Dr. José Eleuterio González”) en el periodo comprendido entre el 1 de enero del 2004 y el 31 de diciembre del 2007, incluyó a 414 pacientes con cáncer de mama avanzado previamente tratados con Antraciclinas y Taxanos y con falla a ellos. El mayor porcentaje de las pacientes dentro del estudio pertenecían al Hospital SXXI (66.90%) seguido por el Hospital universitario “Dr. José Eleuterio González” (18.11%) y el INCAN (14.97%). Los rangos de edad para la población estudiada se distribuye de la siguiente manera: de 20-29 años de edad es del 4%, de 30-39 del 12%, de 40-49 del 32%, de 50-59 del 25%, de 60-69 del 17%, de 70 años o más de 11%. El estadio clínico de presentación al primer diagnóstico más frecuente es el IIIA con 31.6%, seguido por el IIIB con un 24.3% y el IIIC en tercer lugar con un 18.5%, la etapa clínica IV representa el 11.4%. Esto es, el 85.8% de los pacientes son diagnosticados en estados avanzados de la enfermedad. Mientras que el 0.03% es diagnosticado en estadio clínico 0, el 3% en estadio clínico I y el 10.9% restante en estadio II. Discusión y Conclusiones Este estudio mostró las opciones de tratamiento del paciente oncológico utilizadas en algunas de las instituciones más importantes del país, y el costo derivado de su uso; y la gran variabilidad de uso de citotóxicos de Institución a Institución e incluso dentro de la misma Institución, siendo en forma general capecitabina, vinorelbina y ciclofosfamida los tres fármacos más utilizados. En México, desde 1994 se han llevado a cabo reuniones periódicas de Oncólogos con experiencia en el manejo de cáncer de mama para diseñar un Consenso de tratamiento, lo que ha permitido desarrollar algunas guías clínicas .En la primera reunión sobre tratamiento del cáncer mamario realizada en el 2007, se estableció que aquellos pacientes con falla a antraciclinas y taxanos podían ser candidatos a recibir capecitabina, gemcitabina o vinorelbina como opciones de tratamiento Erazo y Cárdenas (23). En el año 2009, Erazo y Silva-Uribe [24], se llevó a cabo una reunión para desarrollar recomendaciones de tratamiento en cáncer de mama metastático en las cuales en este mismo grupo de pacientes se recomendaba el uso de capecitabina sola o en combinación con Ixabepilona, vinorelbina o S150 VALUE IN HEALTH 14 (2011) S147–S150 gemcitabina. Sin embargo como se puede observar en este estudio, algunos de los más utilizados como la monoterapia con ciclofosfamida no están mencionadas ni como recomendación de expertos por la falta de evidencia para su uso, Erazo y Silva-Uribe [24] y Clinical Practice Guidelines in Oncology [25]. Los resultados demuestran que mientras que los tratamientos tempranos para pacientes con cáncer de mama son estándares y homogéneos, los tratamientos citotóxicos recibidos después de la falla a AT tienen gran variabilidad en pacientes con cáncer de mama avanzado. Así mismo esta varianza se ve reflejada en el consumo de recursos asignado a pacientes en estas etapas de la enfermedad que pueden ir desde los $51,960.86 pesos mensuales por paciente (aquellos que reciben vinorelbina oral ⫹ trastuzumab en contexto de 2a línea, IC 95% $47,401.19- $56,520.52) a los $114.11 pesos mensuales (si reciben ciclofosfamida en ese mismo contexto, IC 95% $105.52- $122.69). Los resultados subrayan una falta de estandarización en el tratamiento de estos pacientes. No hay otros estudios en México o Latinoamérica que hablen acerca de los patrones de tratamiento y su costo en el contexto de pacientes con cáncer de mama avanzado resistentes a antraciclinas y taxanos, sin embargo, Knaul FM et al. [26] en el año 2009 publicaron una cohorte de 1904 pacientes con diagnóstico de cáncer mamario en el Instituto Mexicano del Seguro Social, en diferentes etapas de la enfermedad y se reportaron los costos de su atención médica en ese tiempo, encontrando un costo de tratamiento general paciente/año menor para etapas tempranas lo que enfatiza la importancia del diagnóstico temprano en la reducción de costos en el tratamiento. Este estudio menciona numerosos fármacos utilizados en el manejo general de las pacientes en etapa IV (esquemas basados en antraciclinas, taxanos, carboplatino, gemcitabina, capecitabina, vinorelbina y trastuzumab) lo que demuestra la variabilidad de tratamientos utilizados en esta fase, hallazgo congruente con nuestro estudio. Más investigación es necesaria para conocer a fondo la causa de esta variabilidad. Sin embargo discusiones iniciales con los expertos sugieren que las diferencias van más allá de preferencias de prescripción clínica o falta de datos científicos que den soporte a la práctica clínica sino también la falta de acceso a estos medicamentos por problemas financieros o de abasto. Estos resultados reflejan en último caso una clara necesidad clínica insatisfecha, que tiene que ser resuelta a través de una atención equitativa y eficaz basada en una práctica de medicina basada en evidencia y guías clínicas locales para los pacientes atendidos en estas instituciones. Fuentes de financiamiento: Bristol-Myers Squibb Mexico financially supported this study. Materiales Complementarios Material complementario que acompaña este artículo se puede encontrar en la versión en línea como un hipervínculo en doi: 10.1016/j.jval.2011.05.029 o si es un artículo impreso, estará en: www.valueinhealthjournal.com/issues (seleccione el volumen, número y artículo). REFERENCIAS [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [1] Instituto de Enfermedades de la Mama, FUCAM. Cáncer de Mama. [Publicación en línea]. Disponible desde Internet en: ⬍http://www .fucam.org.mx/CAncer%20de%20mama%20Dr%20Sergio%20RodrIguez .pdf. [Acceso el 05-08-2010]. [2] INEGI. Estadísticas a propósito del Día Mundial Contra el cáncer. Datos Nacionales. 2008. [Publicación en línea]. Disponible desde Internet en: [26] www.inegi.gob.mx/inegi/contenidos/espanol/prensa/Contenidos/ estadisticas/2008/cancer08.doc. [Acceso el 05-08-2010]. Brandan ME y Villaseñor Y. Detección del cáncer de mama: estado de la mamografía en México. Cancerología 2006;1:147–162. Registro histopatológico de neoplasias malignas. Dirección General de Epidemiología, SSA, México. 2003. [Publicación en línea]. Disponible desde internet en: <http://www.dgepi.salud.gob.mx/diveent/ RHNM.htm. [Acceso el 05-08-2010]. Cárdenas SJ, Sandoval GF. Tercera revisión del consenso nacional sobre diagnóstico y tratamiento del cáncer mamario. GAMO 2008;7(Suppl 6):1–35. Puente B. Cáncer de mama: hábitos del occidente bajo escrutinio. 2008, [Publicación en línea]. Disponible desde internet en: < http://www.cimacnoticias.com/site/08050707-Cancer-de-mama-hab .33071.0.html. [Acceso el 05-08-2010]. Blum JL, Jones SE, Buzdar AU, et al. Multicenter phase ii study of capecitabine in paclitaxel-refractory metastatic breast cancer. J Clin Oncol 1999;17:485–93. Blum JL, Dieras V, Lo Russo PM, et al. Multicenter phase ii study of capecitabine in taxane-pretreated metastatic breast carcinoma patients. Cancer 2001;92:1759 – 68. Reichardt P, Von Minckwitz G, Thuss-Patience PC, et al. Multicenter phase ii study of oral capecitabine in patients with metastatic breast cancer relapsing after treatment with a taxane-containing therapy. Ann Oncol 2003;14:1227–33. Fumoleau P, Largillier R, Clippe C et al. Multicentre, phase ii study evaluating capecitabine monotherapy in patients with anthracyclineand taxane-pretreated metastatic breast cancer. Eur J Cancer 2004;40:536 – 42. Thomas ES, Gomez HL, Li RK. et al. Ixabepilone plus capecitabine for metastatic breast cancer progressing after anthracycline and taxane treatment. J Clin Oncol 2007;25:5210 –17. Martin M, Ruiz A, Muñoz M, et al. Gemcitabine plus vinorelbine versus vinorelbine monotherapy in patients with metastatic breast cancer previously treated with anthracyclinse and taxanes: final results of the phase III Spanish Breast Cancer Research Group (GEICAM) trial” Lancet Oncol 2007;8:219 –25. Degardin M, Bonneterre J, Hecquet B, et al. Vinorelbine as a salvage treatment for advanced breast cancer” Ann Oncol 1994;5:423– 6. Langkjer ST, Ejlertsen B, Mouridsen H, et al. Vinorelbine as first-line or second-line therapy for advanced breast cancer: a Phase I-II trial by the Danish Breast Cancer Cooperative Group. Acta Oncol 2008;47:735–9. Toi M, Saeki T, Aogi K, et al. Late phase II clinical study of vinorelbine monotherapy in advanced or recurrent breast cancer previously treated with anthracyclines and taxanes. Jpn J Clin Oncol 2005;35:310 –15. Estevez LG, Batista N, Sánchez-Rovira P, et al. A phase II study of Capecitabine and Vinorelbine in patients with metastatic breast cancer pretreated with anthracyclines and taxanes. Clin Breast Cancer 2008;8:149 –54. Ahn JH, Kim SB, Kim TW, et al. Capecitabine and Vinorelbine in patients with metastatic breast cancer previously treated with anthracycline and taxane. J Korean Med Sci 2004;19:547–53. Winer EP, Chu L, Spicer DV. Oral Vinorelbine in the treatment of advanced breast cancer. Semin Oncol 1995;22(2 Suppl. 5):72– 8, discussion 78-9. Bourgeois H, Vermorken J, Dark G, et al. Evaluation of oral versus intravenous dose of Vinorelbine to achieve equivalent blood exposures in patients with solid tumours. Cancer Chemother Pharmacol 2007;60:407–13. Colleoni M, Rocca A, Sandri MT, et al. Low-dose oral methotrexate and cyclophosphamide in metastatic breast cancer: antitumor activity and correlation with vascular endothelial growth factor levels. Ann Oncl 2002;13:73– 80. Incidencia y mortalidad del cáncer a nivel mundial en 2008, publicado por GLOBOCAN, organismo perteneciente a la organización mundial de salud, en su sitio web: http://globocan.iarc.fr/factsheets/cancers/ breast.asp#INCIDENCE. [01-08-2010]. Datos publicado por CONAPO, República Mexicana: Indicadores demográficos 1990-2050, en su sitio web: http://www.conapo.gob.mx/. [01-08-2010]. Erazo AE, Cárdenas-Sánchez J et al. Primera Reunión sobre Tratamiento Médico del Cáncer Mamario. GAMO 2007; 6(Suppl 1):1–11. Erazo AE, Silva-Uribe M. Recomendaciones en cáncer de mama metastásico. GAMO 2009;8(Suppl. 2):1–16. Clinical Practice Guidelines in Oncology. Breast cancer v.2.2010. National Comprehensive Cancer Network, v.2.2010. [Publicación en línea]. Disponible desde internet en: <http://www.nccn.org/ professionals/physician_gls/f_guidelines.asp. [Acceso el 10-12-2010]. Knaul FM, Arreola-Ornelas H, Velázquez E, et al. El costo de la atención médica del cáncer mamario: el caso del Instituto Mexicano del Seguro Social. Salud Pública de México, 2009;51(Suppl. 2):S286-95. Reviewer Acknowledgement Value in Health would like to thank the following for reviewing manuscripts for this Special Issue: Jose Luis Aguilar, VITAMEDICA, Mexico D.F., Mexico Urko Aguirre, Hospital Galdakao-Usansolo, Galdakao, Vizcaya, Spain Felipe Aizpuru, Hospital de Txagorritxu, Vitoria-Gasteiz, Spain Rafael Alfonso, University of Washington, Seattle, Washington, USA Maria Cristina Amorim, University of São Paulo, São Paulo, Brazil Zuleika Aponte Torres, Novartis Pharmaceutica, SA, Barcelona, Spain Denizar Araujo, State University of Rio de Janeiro, Rio de Janeiro, Brazil Vicent Balanzá-Martínez, CIBERSAM University of Valencia, Valencia, Spain Luz-Maria Adriana Balderas-Peña, Instituto Mexicano del Seguro Social, Guadalajara, Jalisco, Mexico Javier Ballesteros, University of the Basque Country, UPV-EHU, Leioa, Spain Eva Baro Ramos, 3D Health Research, Barcelona, Spain Virginia Becerra Bachino, Boehringer Ingelheim, Sant Cugat Del Valles, Spain Clara Bermúdez-Tamayo, Escuela Andaluza de Salud Pública, Granada, Spain Juan Blackburn, Baxter Export Corporation, Fort Lauderdale, FL, USA Cristina M. Ruas Brandão, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil Alexandra Brentani, Faculdade de Medicina da Universidade de São Paulo, São Paulo, Brazil Bernardo Briones, Novartis, Mexico D.F., Mexico David Bruhn, Lilly, San Diego, CA, USA Jefferson Antonio Buendia Rodriguez, Universidad de Buenos Aires. Instituto de Efectividad Clínica y Sanitaria. Buenos Aires, Argentina Jennifer Bueno, Investigaciones para la salud y la equidad. Bogota, Colombia Heidy A. Cáceres, Pfizer Colombia, Bogota, Colombia Joaquín Enzo Caporale, IECS - CENEXA UNLP, Buenos Aires, Argentina J. Jaime Caro, United BioSource Corporation, Lexington, MA, USA Jean Carter, University of Montana, Missoula, MT, USA Alfonso Casado, Dynamic Solutions, Madrid, Spain Mariangela Cherchiglia, Universidade Federal De Minas Gerais, Belo Horizonte – MG, Brazil Otávio Clark, Medinsight-Evidencias Brazil, São Paulo, Brazil Laura Cortès, Instituto Mexicano del Seguro Social, Tlaquepaque, Jalisco, Mexico Luciane Cruz, Federal University of Rio Grande do Sul, Porto Alegre, Brazil Jesús Cuervo, BAP HEALTH, Oviedo, Asturias, Spain Gilberto De Lima Lopes, Jr., Johns Hopkins University, Singapore Grazielle Dias Silva, Secretaria de Estado de Saúde de Minas Gerais, Belo Horizonte, Brazil Margareth Eira, Instituto de Infectologia Emílio Ribas, São Paulo, Brazil Antonio Escobar Martinez, Hospital of Basurto, Bilbao, Spain Laia Febrer, Bayer Healthcare, Sant Joan Despí, Barcelona, Spain Lara Ferreira, University of the Algarve, Faro, Portugal Marcelo Fonseca, Axia.Bio, São Paulo, Brazil Ana Luisa G. Gomes Coelho, Universita Cattolica del Sacro Cuore, Roma, Curitiba, Brazil Rosa María Galindo Suárez, Pfizer Mexico, Mexico D.F., Mexico Oscar Gamboa, Instituto Nacional de Cancerología, Bogota, Colombia Ignacio García Tellez, Roche Mexico, Mexico D.F., Mexico Fernando Garcia-Contreras, Unidad de Investigación en Economía de la Salud, Mexico D.F., Mexico Graciela García-Mahias, Cenabast Chile, Santiago, Chile Jorge Gomez, GSK LatinA, Victoria, Argentina Fernando Gusmão-Filho, IMIP - Instituto Materno-Infantil Prof. Fernando Figueira, Recife, Brazil Iñaki Gutierrez-Ibarluzea, Basque Government, Vitoria-gasteiz, Spain Rosina Hinojosa, Universidad Nacional Mayor de San Marcos, Lima, Peru Andrey Kulikov, Moscow Medical Academy n.a. I.M.Setchenov, Moscow, Russia Hui-Chu Lang, National Yang Ming University, Taipei, Taiwan S152 VALUE IN HEALTH 14 (2011) S151–S152 Iñaki Lete, Hospital Santiago Apóstol, Álavaitoria, Álava, Spain Egidio Lima Dórea, Hypertension Outpatient Clinic, University Hospital, University of São Paulo, São Paulo, Brazil Luis Lizan, Outcomes’10, Castellon, Spain Márcio Machado Dias Ferreira, GSK Brazil, Rio de Janeiro, Brazil Pedro Magalhães, Porto Alegre – RS, Brazil Javier Mar, Hospital Alto Deba, Mondragon, Spain Claudio A. Méndez, Universidad Austral de Chile, Valdivia, Chile Joan Mendivil, IMIM-Hospital del Mar, Barcelona, Spain Gisela Morales, Correos de México, Mexico D.F., Mexico Claudia Morales Moreno, Mutual SER EPS-S, Cartagena, Colombia Joaquín Federico Mould-Quevedo, Pfizer Mexico, Mexico D.F., Mexico Marcelo Nita, Bristol Myers Squibb, São Paulo, Brazil Jose Luis Olvera-Gomez, Mexican Institute of Social Security, Mexico D.F., Mexico Luis Pereira, University of the Algarve, Faro, Portugal Carlos Polanco, Merck Serono, Madrid, Spain Jessica Presa, Wyeth Ind. Farmacêutica, São Paulo, Brazil Carina Ramos, Ministry of Health, Brasilia, Brazil Pablo Rebollo Alvarez, BAP Health Outcomes Research, Oviedo, Spain Fernanda d’ Athayde Rodrigues, UFRGS - Federal University of Rio Grande do Sul, Santa Maria, Rio Grande do Sul, Brazil Evelyn Rodriguez, Bayer HealthCare Pharmaceuticals, Montville, NJ, USA Diego Rosselli, Universidad Jorge Tadeo Lozano, Bogota, Colombia Carlos Rubio Terres, HEALTH VALUE, Madrid, Spain Mario Giorgio Saggia, Johnson & Johnson, São Paulo, Brazil Jorge Salazar, Novartis Farmacéutica S.A, Barcelona, Spain Jean Siebenaler, i3 Innovus, Milton, FL, USA Fabiano Souza, ICESP, São Paulo, Brazil Stephen Stefani, UNIMED, Porto Alegre, Brazil Maíra Takemoto, ANOVA, Rio de Janeiro, Brazil Gabriela Tannus, Axia.Bio, São Paulo, Brazil Vanessa Teich, MedInsight & Evidências, Rio de Janeiro, Brazil Diana R. Téllez Sánchez, Fundación ESENSA Research Center, Bogota, Colombia David Thompson, i3 Innovus, Medford, MA, USA Armando Vargas, R A C Salud Consultores, SA de CV, Mexico D.F., Mexico Luis Alberto Vera Benavides, Oficina de Salud Basada en la Evidencia / Facultad de Medicina. Universidad Austral de Chile. Valdivia, Chile Miguel Angel Villasis-Keever, Mexican Institute of Social Security, Mexico D.F., Mexico Vicente Zanon, HOSPITAL DR. PESET, Valencia, Spain Victor Zarate, Centre for Health Economics, University of York, York, UK GUIDE FOR AUTHORS Value in Health is a peer-reviewed publication of the International Society for Pharmacoeconomics and Outcomes Research. Its mission is to provide a forum for the advancement and dissemination of knowledge and research in pharmacoeconomics and the health-related outcomes of disease and treatment processes. The journal therefore solicits original contributions in applied and theoretical pharmacoeconomics, and in the theory, measurement, and analysis of the health-related outcomes relevant to forwarding scholarly and public dialogue about the assessment of value in health and health care. In keeping with its broad mission, Value in Health will also accept methodology papers and critical reviews of empirical and theoretical literature in pharmacoeconomics and outcomes research. Value in Health does not consider papers reporting data series or data sets that do not include appropriate statistical confidence intervals and/or other measures of statistical imprecision. Value in Health also does not consider papers reporting modeling results that do not include sensitivity analysis of key and influential model parameters. Authors for whom English is a second language may choose to have their manuscript professionally edited before submission or during the review process. Authors wishing to pursue a professional English-language editing service should make contact and arrange payment with the editing service of their choice. For more details regarding the recommended services, please refer to http://support.elsevier.com/. I. ETHICS IN PUBLISHING For information on Ethics in Publishing and Ethical guidelines for journal publication see http://www.elsevier.com/publishingethics and http:// www.elsevier.com/ethicalguidelines. II. CONFLICT OF INTEREST All authors must disclose any financial and personal relationships with other people or organisations that could inappropriately influence (bias) their work. Examples of potential conflicts of interest include employment, consultancies, stock ownership, honoraria, paid expert testimony, patent applications/registrations, and grants or other funding. See also http://www.elsevier.com/conflictsofinterest. III. SUBMISSION DECLARATION Submission of an article implies that the work described has not been published previously (except in the form of an abstract or as part of a published lecture or academic thesis), that it is not under consideration for publication elsewhere, that its publication is approved by all authors and tacitly or explicitly by the responsible authorities where the work was carried out, and that, if accepted, it will not be published elsewhere including electronically in the same form, in English or in any other language, without the written consent of the copyright-holder. IV. RETAINED AUTHOR RIGHTS As an author you (or your employer or institution) retain certain rights; for details you are referred to: http://www.elsevier.com/authorsrights. V. FUNDING BODY AGREEMENTS AND POLICIES Elsevier has established agreements and developed policies to allow authors whose articles appear in journals published by Elsevier, to comply with potential manuscript archiving requirements as specified as conditions of their grant awards. To learn more about existing agreements and policies please visit http://www.elsevier.com/fundingbodies. VI. MANUSCRIPT SUBMISSION AND SPECIFICATIONS To submit a manuscript to Value in Health, please go to: http://mc. manuscriptcentral.com/vih. For assistance, authors may contact the Value in Health editorial office at: [email protected]. If submissions are larger than 500 KB, they should be compressed using PKZIP or WINZIP. Authors will be required to assign copyright of their papers. Copyright assignment is a condition of publication and papers will not be passed to the publisher for production unless copyright has been assigned. An appropriate copyright assignment form can be found at the following address: http://www.ispor.org/publications/value/Value-In-Health-CopyrightTransfer-Form_2011.pdf. A faxed copy of this completed and signed form is acceptable; fax to 609-219-0774 or email to: [email protected]. If excerpts from other copyrighted works are included, the author(s) must obtain written permission from the copyright owners and credit the source(s) in the article. Elsevier has preprinted forms for use by authors in these cases: please consult http://www.elsevier.com/permissions. Each Submission should contain separate documents as follows: i. COVER LETTER. The cover letter should include: 1) title of the manuscript; 2) name of the document file(s) containing the manuscript and the software (and version) used; 3) name and all contact information for the corresponding author and a statement as to whether the data, models, or methodology used in the research are proprietary; 4) names of all sponsors of the research and a statement of all direct or indirect financial relationships the authors have with the sponsors; and 5) if applicable, a statement that the publication of study results was not contingent on the sponsor’s approval or censorship of the manuscript. ii. TITLE PAGE. The title page should contain the following: 1) title; 2) full names (first and surname) of all authors including academic degrees and affiliation(s); 3) name, mailing and email addresses, telephone and fax numbers of corresponding author (with whom all correspondence will take place unless other arrangements are made); 4) all sources of financial or other support for the manuscript (if no funding was received, this should be noted on the title page); 5) at least four key words for indexing purposes; and 6) a running title of not more than 45 characters including spaces. iii. MANUSCRIPTS. Manuscripts must be written in English, typed in either Microsoft Word (Version 5.0 or later) or WordPerfect (version 5.1 or later). Please note that Word 2007 is not yet compatible with journal production systems. Unfortunately, the journal cannot accept Microsoft Word 2007 documents until such a time as a stable production version is released. Please use Word’s ’Save As’ option therefore to save your document as an older (.doc) file type. Manuscripts should be in 8.5x11inch page format, double-spaced with 1-inch margins on all sides and size 10 font (Arial or Times New Roman fonts are preferred). Minimal formatting should be used, i.e., no justification, italics, bold, indenting, etc. There should be no hard returns at the end of lines. Double-spacing after each element is requested (e.g., headings, titles, paragraphs, legends). There is no limit on manuscript length, but length in terms of clarity and conciseness will be considered in the editorial process. The ‘Uniform Requirements for Manuscripts Submitted to Biomedical Journals’ should be consulted for specific style issues not addressed here (www.acponline.org, Ann Intern Med 1997;126:36-47). a. ABSTRACT. An abstract of 250 words or less is required, summarizing the work reported in the manuscript. Original research manuscripts should use a structured format for the abstract, i.e., Objectives, Methods, Results, and Conclusions. b. TEXT. The body of the manuscript should be divided into sections that facilitate reading and comprehension of the material. This should normally include sections with the major headings: Introduction, Methods, Results, Conclusions, Acknowledgments (if needed), and References. There should be no footnotes. Figures (inclusive of figure legends) and Tables must be submitted each as separate documents. c. REFERENCES. References should be listed in a separate section and numbered consecutively with Arabic numerals in the order in which they are cited in the text. Citing unpublished or non-peer-reviewed work such as abstracts and presented papers is discouraged. Personal communications may be indicated in the text as long as written acknowledgment from the authors of the communications accompanies the manuscript. Reference style should follow that of Index Medicus. Spell out single-word journals and abbreviate all others according to the style of Index Medicus. If there are more than four authors, use only the names of the first three, followed by et al. The three most common types of references are illustrated below for example. Journal article: Surname and initials of author(s), title of article, name of journal, year, volume number, first and last page. Arocho R, McMillan CA. Discriminant and criterion evaluation of the U.S.-Spanish version of the SF-36 Health Survey in a Cuban-American population with benign hyperplasia. Med Care 1998;36:766 –72. Book: Surname and initials of author(s)/editor(s), title and subtitle, volume, edition (other than first), city, publisher, year. Johnston J. Econometric Methods (3rd ed.). New York: McGraw-Hill, 1984. Chapter in Book: Surname and initials of author(s), title of chapter, author(s)/editor(s) of book, title of book, volume, edition (other than first), city, publisher, year. Luce BR, Manning WG, Siegel JE, et al. Estimating costs in costeffectiveness analysis. In: Gold MR, Siegel JE, Russell LB, et al., eds., Cost-effectiveness in Health and Medicine. New York: Oxford University Press, 1996. iv. TABLES. Tables should be clearly labeled, neatly typed, and easy to GUIDE FOR AUTHORS – continued understand without reference to the text. Each should be doublespaced and presented on a separate page. Statistical estimates should indicate parameter estimates and, as appropriate, t ratios or standard error, statistical significance, sample size, and other relevant information. All abbreviations must be explained below each table. Each table should be numbered and have a self-explanatory title. v. FIGURES. Figures should each be submitted as a separate image file, not embedded in the manuscript document or in a slide presentation. Cite figures consecutively, as they appear in the text, with Arabic numbers (Figure 1, Figure 2, Figure 3A, etc.). If, together with your accepted article, you submit usable color figures then the Journal will ensure, at no additional charge, that these figures will appear in color on the Web (e.g., ScienceDirect and other sites) regardless of whether or not these illustrations are reproduced in color in the printed version. There will be a charge for color reproduction in print; you will receive information regarding the costs from Elsevier after receipt of your accepted article. Please indicate your preference for color in print or on the Web only. Each figure must be assigned a brief title (as few words as possible, and reserving abbreviations for the legend) as well as a legend. The corresponding legend should be typed doublespaced on a separate page. All symbols, arrows, and abbreviations must be explained in the legend. Please submit files with a resolution of at least 300 DPI. Line artwork should contain a resolution of least 1000 DPI. Elsevier recommends submitting figures in the following formats: TIFF, JPG, EPS, and PDF. Please be sure to delete any identifying patient information such as name, social security number, etc. Photographs in which a person’s face is recognizable must be accompanied by a letter of release from that person explicitly granting permission for publication in the Journal. For any previously published material, written permission for both print and electronic reprint rights must be obtained from the copyright holder. For further explanation and examples of artwork preparation, see Elsevier’s Author Artwork Instructions at www.elsevier.com/artwork. vi. SUPPLEMENTARY MATERIAL. You may submit appendixes that describe either methods or results in more detail if these are needed for clarity of understanding by either peer reviewers or readers. If appropriate, materials suitable for Web publication but not print publication (eg, audio or video files, see below) can also be submitted. If you do so, indicate the particular reasons for the appendix and whether you are submitting it for possible Web publication or simply for peer review purposes. Value in Health accepts audio and video files as ancillaries to the main article. Audio files should be in .mp3 format; the recommended upper limit for the size of a single file is 10 Mb. Video files should be submitted in .mpg or .mp4 format; the recommended upper limit for the size of a single file is 10 Mb. Any alternative format supplied may be subject to conversion (if technically possible) prior to online publication. vii. SURVEY INSTRUMENT. For papers analyzing preferences, ViH requires the submission of a copy of the survey instrument (translated into English in case of different original language) used to generate the preference data. This is to help in the review process and the survey instrument need not appear in a final publication. If the authors wish the questionnaire to be published with the paper, it should be submitted through ScholarOne Manuscripts as part of the paper. If they do not wish it to be published, it should be submitted through ScholarOne Manuscripts as Supporting Information and then will be sent to the reviewers as a reviewer’s appendix. VII. DATA, MODELS, AND METHODOLOGY All authors must agree to make their data available at the Editor’s request for examination and re-analysis by referees or other persons designated by the Editor. All models and methodologies must be presented in sufficient detail to be fully comprehensible to readers. VIII. AUTHOR ANONYMITY From September 15, 2003, it is the policy of Value in Health that peer review of submitted manuscripts is double blinded, i.e., the reviewers do not know the names of the authors of manuscripts and the authors do not know the names of the reviewers. Blinded reviews are common practice at many important scientific and medical journals. IX. THE REVIEW PROCESS All manuscripts deemed appropriate for Value in Health after initial screening will be reviewed by at least two peer reviewers. The objective of the journal is to complete peer review and reach editorial decision within ten to twelve weeks of submission, at which time the corresponding author will receive written notification, including anonymous reviewer commentary. X. AUTHOR TRACKING SERVICES Authors may track accepted articles at http://www.elsevier.com/ trackarticle and set up e-mail alerts to inform them when an article’s status has changed. Contact details for questions arising after acceptance of an article, especially those relating to proofs, will be provided by the publisher. XI. PROOFS Proofs will be sent electronically to the Authors to be carefully checked for printer’s errors. Substantive changes or additions to the edited manuscript cannot be allowed at this stage. Corrected proofs should be returned to the publisher within 2 days of receipt. XII. OFFPRINTS The corresponding author, at no cost, will be provided with a PDF file of the article via e-mail. For an extra charge, paper offprints can be ordered via the offprint order form which is sent once the article is accepted for publication. The PDF file is a watermarked version of the published article and includes a cover sheet with the journal cover image and a disclaimer outlining the terms and conditions of use.