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
[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
[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
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.
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
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
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
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-
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.
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
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
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
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
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
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.
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
[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.
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.
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).
doi:10.1016/j.jval.2011.05.014
S14
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-
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.
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.
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
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.
doi:10.1016/j.jval.2011.05.034
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
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
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].
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.
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.
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].
doi:10.1016/j.jval.2011.05.018
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
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.
the online version as a hyperlink at doi:10.1016/j.jval.2011.05.018,
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
[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.
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.
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].
* Address correspondence to: Alessandra Maciel Almeida, Prof. Antônio Aleixo Street, 760/602 – Bairro de Lourdes, Belo Horizonte, MG,
Brazil. CEP 30180-150.
doi:10.1016/j.jval.2011.05.011
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
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.
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,
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).
the online version as a hyperlink at doi:10.1016/j.jval.2011.05.011
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
[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.
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.
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.
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.
doi:10.1016/j.jval.2011.05.024
S30
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.
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).
(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.
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 $19,799.
The analysis was robust to different deterministic and probabilistic
sensitivity analyses, except for the cost of ICD and for battery life.
Conclusions: The results varied considerably depending on the cohort
and discretely according to the health system. ICD could be cost-effective in Argentina, mainly in the MADIT I patients.
Palabras Claves: cardioverter-defibrillator, cost-effectiveness, cost-utility, health care utilization, sudden death.
entre U$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.
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]
doi:10.1016/j.jval.2011.05.030
S34
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
Tabla 2 – Costos utilizados en el modelo.
Público
(AR$)
Costo de la colocación del CDI
Costo del recambio de
generador
Costo anual del seguimiento
sin CDI
con CDI prevención
secundaria
con CDI prevención
primaria
Costo promedio de las
complicaciones
S. Social
(AR$)
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
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 cardíaca (2,83-8,93%)
de CDI (AR$20.000-100.000)
MADIT II
Mortalidad cardíaca (3,46-6,30)
de CDI (AR$20.000-100.000)
SISTEMA PÚBLICO
SEGURIDAD SOCIAL
SISTEMA PRIVADO
ICER (AR$/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.
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
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
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].
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.
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.
doi:10.1016/j.jval.2011.05.032
S40
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
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.
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
[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.
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.
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].
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.
doi:10.1016/j.jval.2011.05.033
S44
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
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 *
Atención médica
Consulta asma controlada †
Consulta asma no contralada †
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
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.
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.
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.
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.
doi:10.1016/j.jval.2011.05.023
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.
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.
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.
sclerosis in Switzerland. Eur J Health Econ 2006;7(Suppl. 2):S86 –95.
sclerosis in Spain. Eur J Health Econ 2006;7(Suppl. 2):S65–74.
sclerosis in the United Kingdom. Eur J Health Econ 2006;7(Suppl. 2):
S96 –104.
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.
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.
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].
* 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].
doi:10.1016/j.jval.2011.05.010
S52
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-
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
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
(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
smokers
– Population COPD incidence (age and sex
specific)
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
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
(b)
250
500
450
400
350
300
250
200
150
100
50
0
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
90
100
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
0
10
20
30
40
50
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
(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
[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].
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.
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.
* 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.
doi:10.1016/j.jval.2011.05.007
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
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)
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)
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.
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.
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.
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-
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.
doi:10.1016/j.jval.2011.05.025
S66
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
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$1) [36].
Una selección de los parámeros empleados en el análisis del
caso base se presenta en la Tabla 1.
Se probó el impacto en las estimaciones del modelo por las siguientes variantes en las características de las vacunas: una menor
inmunogenicidad de la PCV-10 (de 0,98 a 0,95), así como la presencia de efectos indirectos en la PCV-10 y la ausencia de efectos
indirectos en la PCV-7 y la PCV-13.
Resultados
Bajo las consideraciones realizadas en el caso base, la estrategia
basada en la PCV-7 presenta un costo incremental negativo (ahorros al sistema de salud) de US$ 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
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
%
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$/evento
Otitis meda [33]
14.946,9
5.651,6
16.541,1
1.560,9
831,6
Costos de vacunación
US$/dosis
PCV-7 [34]
PCV-10 [33]
PCV-13 (supuesto)
0,0
30,0
3,0
3,0
%
Otitis meda [25, 26]
Efectos directos de la PCV-7 en ENI [27]
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$1.307/niño vacunado, siendo mayor en 20 y 80%
al estimado para las PCV-7 y PCV-10, respectivamente.
El efecto económico y en salud de un menor factor de inmunogenicidad de la PCV-10 (0,95), así como la consideración de efectos indirectos en la PCV-10 y no consideración de efectos indirectos en las
PCV-7 y PCV-13 se presenta en la Tabla 3. A pesar de la disminución de
inmunogenicidad de la PCV-10 esta alternativa presenta un costo incremental negativo (US$ -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
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
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.
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.
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.
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$1,92 bilhão [5].
Apesar dos inúmeros esforços para organização e gerenciamento do
PMAC, não se conhece o número de usuários cadastrados e, principalmente, de usuários atendidos por patologia [6]. Isto se deve ao
processo de autorização de fornecimento dos medicamentos, realizado por meio do formulário de Autorização de Procedimento de Alto
Custo (APAC), que serve como cadastramento do usuário no banco de
dados nacional para fins gerenciais e de cobrança. A APAC apresenta
caráter contábil e é focada no medicamento e não no indivíduo.
Desta forma, indivíduos em uso contínuo de medicamentos podem
ter inúmeras APAC’s, dificultando a análise de suas carac-
Título resumido: Gastos do Ministério da Saúde do Brasil.
* Autor de Correspondência: Cristina Mariano Ruas Brandão, Grupo de Pesquisas em Farmacoepidemiologia, Av. Antônio Carlos, 6627,
Campus Pampulha, Belo Horizonte/MG, Brazil; Tel (31)3409-6861; Fax: (31)33574868.
doi:10.1016/j.jval.2011.05.028
S72
Tabela 1 – Distribuição dos gastos individuais com medicamentos de Alto Custo, de acordo com o sexo e ano de
seguimento. Ministério da Saúde, 2000 –2004.
Seguimento
Feminino (n⫽388.471)
Masculino (n⫽222.948)
Gasto mensal per capita (R$)
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)
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).
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.
S74
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
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
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
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-
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.
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.
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.
doi:10.1016/j.jval.2011.05.019
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
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.
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.
[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.
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.
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-
* Address correspondence to: Luciana Clark, Rua Tranquillo Prosperi, 143, Bairro Santa Genebra II, Campinas – SP, Brazil 13084-778.
doi:10.1016/j.jval.2011.05.013
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
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%.
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.
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.
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
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.
doi:10.1016/j.jval.2011.05.022
S86
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-
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
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.
10.1016/j.jval.2011.05.022 si es un artículo impreso, estará en
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.
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.
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.
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].
doi:10.1016/j.jval.2011.05.017
S90
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.
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.
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
[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.
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.
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.
doi:10.1016/j.jval.2011.05.008
S94
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.
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.
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].
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.
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
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.
doi:10.1016/j.jval.2011.05.026
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
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)
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.
10.1016/j.jval.2011.05.026o si es un artículo impreso, estará en
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.
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.
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,
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.
doi:10.1016/j.jval.2011.05.006
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
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
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$)
Utilities
Heart failure patient utility (New York Heart Association
Class II-III)
Discount rate (%)
3752
1798
54
363
1538
4187
1,876–5,628
899–2,697
27–81
182–545
769–2,306
2,093–6,280
0.80
5
0.78–0.82
3–7
Source
15
15
Meta-analysis
Meta-analysis
Health care insurance
plan, primary data
Health care insurance
plan, Primary data
17
17
17
17
17
Estimated
16
Estimated
S104
Table 3 – Model predicted cost, effectiveness, and cost-effectiveness of competing strategies.
Total cost (Int$)
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.
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.
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
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].
[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.
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.
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
article.
* Address correspondence to: Luciane Nascimento Cruz, Rua Ramiro Barcelos, 2350/Building 21, Porto Alegre, RS, Brazil 90035-903.
doi:10.1016/j.jval.2011.05.012
S109
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
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.
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
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
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
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.
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.
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
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.
doi:10.1016/j.jval.2011.05.035
S116
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-
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.
10.1016/j.jval.2011.05.035o si es un artículo impreso, estará en
REFERENCIAS
[1] Cascade EF, Kalali AH, Thase ME. Use of antidepressants. expansion
beyond depression and anxiety. Psychiatry 2007;4:25– 8.
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.
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.
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.
doi:10.1016/j.jval.2011.05.016
S120
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
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
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.
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.
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].
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.
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.
doi:10.1016/j.jval.2011.05.020
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
(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.
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.
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.
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.
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
* Address correspondence to: Guilherme Becker Sander, Av Bage, 919/202, Porto Alegre - Rio Grande do Sul, 90460-080 Brazil.
doi:10.1016/j.jval.2011.05.021
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
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 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.
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.
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,
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.
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].
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.
doi:10.1016/j.jval.2011.05.031
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.
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.
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
(Cuadro 3, 4 y 5 Material Complementario en: doi:10.1016/j.
jval.2011.05.031).
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
Fuentes de financiamiento: El presente trabajo fue financiado
por Fundación IMSS A.C.
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
[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
[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.
[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.
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,
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.
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
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.
doi:10.1016/j.jval.2011.05.027
S134
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.
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].
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).
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
Fuentes de financiamiento: El presente trabajo fue financiado
por Fundación IMSS A.C.
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
[2] Datos preliminares. Censo de Población y vivienda 2010. Instituto
Nacional de Estadística y Geografía. http://www.inegi.org.mx/. [Último
[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.
[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].
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.
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
* Address correspondence to: L. R. Bahia, Rua Visconde de Pirajá 547-501, Ipanema, Rio de Janeiro, ZIP 22410-003, Brazil.
doi:10.1016/j.jval.2011.05.009
S138
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].
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.
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
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
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.
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.
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.
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
* 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.
doi:10.1016/j.jval.2011.05.015
S142
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
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.
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
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
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
[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.
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.
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.
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
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.
doi:10.1016/j.jval.2011.05.029
S148
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).
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
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.
10.1016/j.jval.2011.05.029 o si es un artículo impreso, estará en:
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
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.

in health

Transcripción

Documentos relacionados

Our Physicians/Nuestros Medicos Our Locations/ Nuestras Clinicas

Senhores - Bass fishing trips at Amazing Lake El Salto Mexico by

J.A. Del Río

The 40 Annual - UNIS-UN

Crece el número de niños con diabetes en Colorado