H N P D I S C U S S I O N P A P E R Optimizing the Allocation of Resources among HIV Prevention Interventions in Honduras Girindre Beeharry, Nicole Schwab, Dariush Akhavan, Rosalinda Hernandez and Carla Paredes June 2002 OPTIMIZING THE ALLOCATION OF RESOURCES AMONG HIV PREVENTION INTERVENTIONS IN HONDURAS Girindre Beeharry, Nicole Schwab, Dariush Akhavan, Rosalinda Hernández and Carla Paredes June 2002 Health, Nutrition and Population (HNP) Discussion Paper This series is produced by the Health, Nutrition, and Population Family (HNP) of the World Bank's Human Development Network (HNP Discussion Paper). The papers in this series aim to provide a vehicle for publishing preliminary and unpolished results on HNP topics to encourage discussion and debate. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations or to members of its Board of Executive Directors or the countries they represent. Citation and the use of material presented in this series should take into account this provisional character. For free copies of papers in this series please contact the individual authors whose name appears on the paper. Enquiries about the series and submissions should be made directly to the Editor in Chief. Submissions should have been previously reviewed and cleared by the sponsoring department which will bear the cost of publication. No additional reviews will be undertaken after submission. The sponsoring department and authors bear full responsibility for the quality of the technical contents and presentation of material in the series. Since the material will be published as presented, authors should submit an electronic copy in a predefined format as well as three camera-ready hard copies (copied front to back exactly as the author would like the final publication to appear). Rough drafts that do not meet minimum presentational standards may be returned to authors for more work before being accepted. The Editor in Chief of the series is Alexander S. Preker (apreker@worldbank.org); For information regarding this and other World Bank publications, please contact the HNP Advisory Services (healthpop@worldbank.org) at: Tel (202) 473-2256; and Fax (202) 522-3234. ISBN 1-932126-67-8 © 2002 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 All rights reserved. ii Health, Nutrition and Population (HNP) Discussion Paper Optimizing the Allocation of Resources among HIV Prevention Interventions in Honduras Girindre Beeharrya ,Nicole Schwabb ,Dariush Akhavanc,Rosalinda Hernándezd and Carla Paredese aTask Team Leader, LCSHD, The World Bank, Washington D.C., USA. bJunior Professional Associate, LCSHH, The World Bank, Washington D.C., USA cPublic Health Specialist, Consultant, The World Bank, Washington D.C., USA dEpidemiologist, Consultant, The World Bank, Washington D.C., USA eConsultant, LCSHH, The World Bank, Washington D.C., USA Paper based on the workshop "Optimizing the allocation of resources among HIV prevention interventions in Honduras" Tegucigalpa, Honduras, May 13-16, 2002 Abstract: This paper presents a model that policymakers can use to determine the resource allocation that will prevent the maximum number of new HIV infections at any given budget level. The optimal allocation exercise was conducted in Honduras, where the epidemic is still concentrated in high-risk groups but has begun spreading into the general population. Most transmissions occur through heterosexual sex, followed by sex between men, and mother-to-child transmission. Adult prevalence is estimated at 1.4.%. The optimization exercise involves several steps:(a) choosing population subgroups targeted for intervention; (b) estimating the proportion of each subgroup that can be reached; (c) estimating the total number of new infections expected in each subpopulation; (d) defining the set of HIV prevention interventions to be considered; (e) estimating the unit cost of each intervention; and (f) estimating the expected effectiveness of each intervention. Most of the data required to run the model has to be guesstimated or derived from the literature. To address this challenge, a group of some forty local and international experts in HIV/AIDS met in Tegucigalpa in May 2002 and arrived at the consensus estimates used in this exercise. They based their estimates on data submitted by two local epidemiologists who had conducted an extensive literature search prior to the workshop. The results from this collective exercise show that for limited HIV prevention budgets (below $500,000), condom social marketing and condom distribution prevent the maximum number of HIV infections. If the HIV prevention budget is between $750,000 and $2.5 million, then Information Education and Communication (IEC) targeted at high risk groups, HIV counseling and access to rapid testing, and IEC for the Garifunas should also be part of the country's prevention strategy. The exercise shows that some prevention interventions are unattractive even when the HIV prevention budget increases to $10 million.Summary in 300 words maximum. Keywords: worlds1, worlds2, words3, words4, words5 (five sets of words max) Disclaimer: The findings, interpretations and conclusions expressed in the paper are entirely those of the authors, and do not represent the views of the World Bank, its Executive Directors, or the countries they represent. iii Correspondence Details: Girindre Beeharry; The World Bank, 1818 H Street NW, Washington D.C., 20433, USA; Tel: (202) 473-6595 Nicole Schwab, The World Bank, 1818 H Street, N.W. Mail Stop I7-700 ; Tel: (202) 458- 9892; Fax: (202) 552-0050; Email: nschwab@worldbank.org; Web: www.worldbank.org/lac Daruish Akhavan Carla Paredes Rosalinda Hernandez Author; Address; Tel: (***) ***-****; Fax: (***) ***-***-****; Email: ***@**** Web: www.***** iv Table of Contents ABBREVIATIONS AND ACRONYMS............................................................................. vi ACKNOWLEDGEMENTS............................................................................................... vii FOREWORD...................................................................................................................... ix EXECUTIVE SUMMARY................................................................................................. xi CHAPTER 1: BACKGROUND AND STUDY OBJECTIVES............................................ 1 INTRODUCTION..............................................................................................................1 OBJECTIVES OF THE STUDY.........................................................................................2 THE HIV/AIDS EPIDEMIC IN HONDURAS.....................................................................2 THE COST OF HIV/AIDS PREVENTION, TREATMENT AND CARE..............................6 CHAPTER 2: METHODOLOGY...................................................................................... 8 SUMMARY DESCRIPTION OF THE MODEL..................................................................8 THE TEN STEPS INVOLVED IN THE OPTIMIZATION EXERCISE.................................8 CAVEATS.......................................................................................................................10 CHAPTER 3: ESTIMATING INPUTS FOR THE MODEL............................................. 13 STEPS 1 AND 2...............................................................................................................13 STEP 3............................................................................................................................14 STEP 4............................................................................................................................15 STEP 5............................................................................................................................16 STEP 6............................................................................................................................17 STEP 7............................................................................................................................18 CHAPTER 4: RESULTS AND DISCUSSION................................................................... 20 RESULTS........................................................................................................................20 SENSITIVITY ANALYSIS..............................................................................................24 DISCUSSION..................................................................................................................25 APPENDIX I ­ MATHEMATICAL DESCRIPTION OF THE MODEL.......................... 29 THE CORE MODEL .......................................................................................................29 ADDITIONS TO THE CORE MODEL.............................................................................30 Reserved Spending....................................................................................................... 30 Correction for Effects of Concurrent Interventions......................................................... 30 Secondary Infections.................................................................................................... 31 APPENDIX III: DERIVATION OF INCIDENCE FROM PREVALENCE DATA: EXAMPLE OF MSM ........................................................................................................ 35 APPENDIX IV: BACKGROUND DATA FOR STEPS 1,2,3 AND 6.................................. 37 STEPS 1 AND 2: DATA SOURCES.................................................................................37 STEP 3: DETAILS OF SOURCES AND ESTIMATION PROCEDURES...........................39 STEP 6: DETAILS OF UNIT COST CALCULATION.......................................................43 APPENDIX V. SENSITIVITY ANALYSIS....................................................................... 49 TableV.2 Low and High Effectiveness Figures ..................................................................50 APPENDIX VI: LIST OF WORKSHOP PARTICIPANTS.............................................. 55 BIBLIOGRAPHY.............................................................................................................. 57 v ABBREVIATIONS AND ACRONYMS AIDS Acquired Immune Deficiency Syndrome ARV Antiretrovirals ASONAPVSIDAH Asociación Nacional de Personas Viviendo con SIDA en Honduras CDC Centers for Disease Control COMVIDA/SPS Comunidad y Vida de San Pedro Sula CNS Cuentas Nacionales en VIH/SIDA CSW Commercial Sex Worker DALYs Disability-Adjusted Life Years FSCS Fraternidad Sampedrana de Lucha Contra el SIDA GTZ Deutsche Gesellschaft für Technische Zusammenarbeit GmbH HIV Human Immunodeficiency Virus IDU Intravenous Drug User IEC Information, Education, Communication INE Instituto Nacional de Estadísticos IHSS Instituto Hondureño de Seguridad Social IAEN International AIDS Economics Network IDB Inter American Development Bank MDM Médicos del Mundo MSF Médicos Sin Fronteras MSM Men who have Sex with Men NGO Non-Governmental Organization PAHO Pan American Health Organization PASCA Proyecto Acción SIDA de Centroamérica PASMO Pan American Social Marketing Organization PLWHA Persons Living with HIV/AIDS. STD / STI Sexually Transmitted Disease / Infection UNAIDS Joint United Nations HIV/AIDS Program UNICEF United Nations Children's Fund USAID United States Agency for International Development WB World Bank YLL Years of Life Lost vi ACKNOWLEDGEMENTS This study was prepared by Girindre Beeharry (Task Team Leader, LCSHD), Nicole Schwab (LCSHD), Dariush Akhavan, Rosalinda Hernández, and Carla Paredes (consultants). Comments on the study concept and final draft were provided by Enrique Zelaya (UNAIDS), Paloma Cuchi (UNAIDS/PAHO), Stan Terrell, Logan Brenzel (USAID), Andre Medici, Ernest Massiah (IDB), Edward Kaplan (Yale School of Management), Isabel Noguer (Spanish National HIV/AIDS Program), Carol Scotton (CDC), Debrework Zewdie (HDNVP), Ruth Levine, Anabela Abreu, Helen Saxenian (LCSHD), Son Nam Nguyen (EASHD), Juan Rovira (HDNHE), and Mead Over (DECRG). Charlie Griffin (SASHD) and Helena Ribe (LCC2C) strongly supported the realization of this study. This study was based on a workshop which was made possible thanks to the support of Dr. Manuel Sandoval, Vice-Minister of Health of Honduras, Dr. Humberto Cosenza, Secretary for External Cooperation in the Ministry of Health, and Dr. Enrique Zelaya (UNAIDS). The staff of the World Bank Resident Mission, particularly Joseph Owen, Rebeca Santos, Iris Medina and Karla Lopez, greatly facilitated the organization of the workshop. Ramiro Nuñez, Natalia Moncada, and Mary Dowling (LCSHD) took excellent care of the logistical aspects of the workshop. The team wishes to extend its gratitude to all the participants who made time in their busy schedules to attend the workshop in Tegucigalpa. The authors are also grateful to the World Bank for having published the Report as an HNP Discussion Paper. vii viii FOREWORD (Alex will provide one) Alexander S. Preker Chief Economist Health, Nutrition, and Population (HNP) Editor of the HNP Publication Series ix x EXECUTIVE SUMMARY Faced with a growing HIV/AIDS epidemic and limited resources for prevention, policymakers need to decide how to distribute funds among different HIV prevention interventions to achieve the maximum impact on the epidemic. This paper presents a model that policymakers can use to determine the resource allocation that will prevent the maximum number of new HIV infections at any given budget level. A large group of stakeholders in Honduras generated the data needed to run the model through consultations and consensus. The underlying model can be downloaded from www.worldbank.org/lachealth. The optimal allocation exercise was conducted in Honduras, the country most affected by the HIV/AIDS epidemic in Central America, with an adult prevalence rate of 1.4% (Chapter 1). In Honduras, the epidemic is still concentrated in high-risk groups but has begun spreading into the general population. Most transmissions occur through heterosexual sex, followed by sex between men, and mother-to-child transmission. The optimization exercise involves several steps (Chapter 2): (a) choosing population subgroups targeted for intervention; (b) estimating the proportion of each subgroup that can be reached; (c) estimating the total number of new infections expected in each subpopulation; (d) defining the set of HIV prevention interventions to be considered; (e) estimating the unit cost of each intervention; and (f) estimating the expected effectiveness of each intervention. Most of the data required to run the model has to be guesstimated or derived from the literature. To address this challenge, a group of some forty local and international experts in HIV/AIDS met in Tegucigalpa in May 2002 and arrived at the consensus estimates used in this exercise. They based their estimates on data submitted by two local epidemiologists who had conducted an extensive literature search prior to the workshop (Chapter 3). The results from this collective exercise (Chapter 4) show that for limited HIV prevention budgets (below $500,000), condom social marketing and condom distribution prevent the maximum number of HIV infections. If the HIV prevention budget is between $750,000 and $2.5 million, then Information, Education and Communication (IEC) targeted at high risk groups, HIV counseling and access to rapid testing, and IEC for the Garifunasa should also be part of the country's prevention strategy. The exercise shows that some prevention interventions are unattractive even when the HIV prevention budget increases to $10 million (see Table 1). aThe Garifunas are an African-Caribbean population living along the Atlantic Coast of Honduras. xi Table 1. Optimal Allocation at Budgets of $500,000, $3 Million, and $10 Million & ons Prison. STIs Soc. Rights Pregnant Garifuna MSM, of Safety Intervention Counseling Testing IEC Women IEC IEC Adolescents IEC CSW, Workplace Interventi Prevention MTCT Syndromic Mgt. Condom Distribution Condom Marketing Blood Human Optimal $0 $0 $0 $0 $0 $0 $0 $0 $250k $250k $0 $0 Allocation at $500,000 Optimal $1.2 $0 $500k $0 $500k $0 $0 $0 $500k $250k $0 $0 Allocation at $3M 5M Optimal $2M $0 $750k $0 $500k $0 $1.25M $3M $750k $500k $500k $750k Allocation at $10M Drawing on local expertise and consensus, the exercise shows that a substantial dent in the epidemic can be achieved even with limited budgets, provided that these are appropriately channeled. If $1 million is invested in condom distribution, condom social marketing, and IEC for high risk groups (commercial sex workers (CSW), men who have sex with men (MSM), and prisoners) a total of about 5,100 HIV infections can be prevented. By contrast, $9 million distributed equally among the other 9 interventions, averts about 3,200 HIV infections. Guaranteeing funding for a minimal package of cost-effective interventions thus ensures a great impact on the course of the epidemic. A sensitivity analysis indicates that the need to prioritize condom distribution, condom social marketing, and IEC for high-risk groups is robust to changes in the underlying parameters (Appendix V). A second conclusion from the exercise is that seeking to reach a greater proportion of the high risk groups pays off in terms of the total number of infections that can be averted. For example, under the current assumption that only 50% of commercial sex workers, 30% of men who have sex with men and 55% of the Garifunas can be reached, even with $10 million allocated optimally, only about 12,000 new HIV infections can be averted. But if 100% of these three population groups could be reached, almost twice as many new HIV infections could be averted with the same budget. Graph 1 shows the impact of being able to reach more people in these three groups for budget levels of $1, $5 and $10 million. Graph 1. Impact of Reaching More People in High Risk Groups 20.000 with 15.000 Infections Allocation 10.000 of 5.000 No Prevented Optimal - $1M $5M $10M HIV Prevention Budget 50%CSW, 30%MSM, 55%Garifunas 100% all three groups xii Beyond a certain level of funding for each intervention (the exact levels are given by the model), the government should therefore develop strategies (outreach programs, stigma reduction, decriminalization of certain groups, expansion of service provision, etc.) to reach a greater percentage of high-risk groups, rather than continue delivering the interventions as designed originally. Because policymakers cannot allocate prevention budgets purely on the basis of cost- effectiveness, the model includes additional features. First, it allows them to see how the number of infections prevented changes with alternative resource allocation strategies. The study shows that resource allocation strategies that differ from the optimal scenario lead to a significantly lower number of infections prevented. Second, the model allows policymakers to reserve some funds for specific prevention interventions and submit the remainder of the prevention budget to the optimization process. Each country is faced with a different series of political constraints and social values that make departures from the "optimal" allocation of resources necessary. The model allows policymakers to measure the effect of these constraints in terms of foregone prevention opportunities. The exercise also generates a framework for program monitoring and evaluation. It sets out clearly the interventions and their key components, the population groups that will be targeted, the expected effectiveness and unit cost of each intervention, the number of people that they aim to reach, and the expected impact. These provide ready indicators that can be monitored and updated as the prevention strategy is implemented and new data are collected, giving rise to a dynamic and transparent framework for guiding HIV prevention. In addition, several elements of the model, such as the cost structure and efficacy of the interventions, can be changed; program monitoring and supervision should therefore focus on improving them. The exercise provides a flexible tool that can be adapted to different countries. It relies on local expertise and consensus, which generates ownership. This tool focuses on cost-effectiveness, arguably the most important, but not the only criterion policymakers may want to factor into their resource allocation decisions. Its results are meant to inform policy discussions, but should not be interpreted too rigidly. Users should not extrapolate the optimal allocation results from Honduras to other contexts, but conduct the exercise to build consensus and facilitate policy discussions among local experts. To summarize: a big dent in the epidemic can be made with a small number of interventions (condom social marketing, condom distribution, and IEC for high risk groups) at relatively low cost. A bigger dent still can be achieved if efforts are made to reach a greater proportion of high- risk groups. During the consensus workshop in Honduras, participants called for a `minimum' package of interventions that would include the three or four most cost-effective interventions for priority funding. Honduras recently applied for the first round of funding from the Global Fund to Fight AIDS, TB, and Malaria. The proposal was for $41 million, with $27 million for HIV/AIDS prevention, care, and treatment. Against stiff competition (385 applications were submitted) the Honduran proposal was recommended for funding following revisions. It is hoped that, as a result of the workshop and further policy dialogue, the key interventions that were not part of the Global Fund application will be included in the country's HIV/AIDS prevention plans, and will be funded by the Global Fund and other sources of external assistance, as well as domestic resources. Honduran and other policymakers now possess a model that can help them to: simulate the effect of alternative resource allocations; generate consensus around the HIV prevention interventions that will have the greatest impact on the epidemic; and monitor and evaluate the outcome of the chosen strategy. xiii xiv CHAPTER 1: BACKGROUND AND STUDY OBJECTIVES INTRODUCTION The emergence of the HIV/AIDS epidemic has exerted considerable stress on the national capacity to control infectious diseases. Traditional control techniques (vector control, environmental control, curative care) are all but irrelevant for an epidemic whose speed is predominantly determined by behavior (itself a function of sociological, cultural and economic determinants) rather than by environmental and biological factors. Health authorities have been stretched in many ways, as they have found themselves having to: develop expertise in relatively new areas such as behavior change; seek access to the populations (marginalized and criminalized) that most urgently needed help; and explicitly speak to two of the most powerful taboos in most societies: sex and death. For these reasons, public health authorities have had to build alliances with non-traditional partners (NGOs, religious bodies, private sector organizations, community leaders, other sectoral ministries, etc.) who could help them reach marginalized populations effectively and foster behavior change. Enlisting networks of commercial sex workers and soccer players to promote condom use speaks to the ability of public health institutions around the world to adapt to the particularities of the epidemic and to adopt non-traditional control techniques. The failure to prevent the spread of the epidemic ­ in other words, to effect behavior change such as delayed sexual debut, partner reduction, or consistent condom use ­ has had well-documented catastrophic consequences in many countries. Health authorities in developing countries have been stretched in yet another way: finding the financial resources to prevent the further spread of the epidemic in a context of scarce resources. Ministries of Health have routinely found themselves faced with two equally difficult tasks: reallocating resources away from other health programs and into HIV/AIDS programs, and convincing Ministries of Finance of the need for additional funds for HIV prevention, care, and treatment. Treatment with antiretrovirals, which constitutes the standard of care in rich countries, continues to be extremely expensive for poor countriesb. In a context where resources are scarce and will remain soc, it is critically important that whatever funds are available are spent wisely. In the case of prevention, funds are spent well when they contribute to preventing the maximum number of new HIV infections. The purpose of this study is to help Honduras define what mix of prevention strategies, at given budget levels, would avert the maximum number of new infections. The study does so by defining the universe for prevention in the country (i.e., the number of new infections expected by population groups), examining the relative cost and effectiveness of different strategies to prevent new infections, and allocating resources to interventions on the basis of the number of infections they can prevent at a given budget level. bDespite the recent sharp decrease in prices . cRelative to the cost of a comprehensive program of prevention, care, and treatment: see Marseille et al (2002)48. 1 OBJECTIVES OF THE STUDY The model is intended to help policymakers allocate limited resources for HIV among prevention programs. This study is a follow-up to a recent World Bank study on HIV/AIDS in Latin America104 which provides information on, among other things, the epidemic status, prevalence in high-risk groups,the distribution of reported AIDS cases by mode of transmission, resources and technical assistance available, and the main challenges and areas for improvement in prevention, care, epidemiological surveillance and blood safety. The model draws on the techniques elaborated in No Time to Lose37 and has the following characteristics: Aims to prevent the largest number of new HIV infections.d For the purpose of this study, the optimal resource allocation is defined as that which "will have the important property that no alternative division of the prevention budget could better achieve the overriding HIV prevention goal"37, where the overriding HIV prevention goal is to avert the largest number of new HIV infections. Focuses on prevention programs only. The model does not consider tradeoffs between funding care and treatment instead of preventione. It supposes that the policymaker has already earmarked funds for prevention. Analyzes alternative allocations differentiated by strategy and target group. Target population subgroups are identified by characteristics such as risk behavior, sexual orientation, sex, language and location. As described below, in Honduras, target groups include, among others, commercial sex workers, men who have sex with men, and the Garifunas. These subgroups have significantly higher rates of HIV/AIDS prevalence than the overall population. Enables other allocation criteria such as equity or political concerns to be valued explicitly in comparison to the optimal allocation. For example, if a certain amount of funding were to be reserved for a particular group or program, the resulting impact on the number of infections averted can be measured. This paper is organized as follows. In the remainder of this chapter, the characteristics of the epidemic in Honduras and the cost of prevention and care are briefly discussed. In Chapter 2, the overall methodology is presented, the steps involved in estimating the data for the model are discussed in some detail, and the limitations of the optimization exercise are examined. In Chapter 3, the data estimation process in Honduras is presented and discussed. In the concluding Chapter 4, the results from applying the optimization model to the data from Honduras are presented and discussed. THE HIV/AIDS EPIDEMIC IN HONDURAS Of the 40 million people living worldwide with HIV/AIDS at the end of the year 2001, 1.8 million lived in Latin America and the Caribbean96. In Latin America alone, in 2001, 0.5% of the population (1.4 million people) lived with HIV/AIDS, 130,000 adults and children were newly infected with HIV and 80,000 died104. The country most affected in Central America is Honduras, with an estimated prevalence rate of 1.4%. Honduras has just 17% of the population of Central America, but about 43%f of all dIn "No Time to Lose", Kaplan defines the optimal allocation of available public resources for HIV prevention, as one that prevents the maximum number of new HIV infections. eFor a discussion of the trade-offs between prevention and treatment, see Marseille et al (2002)48. 2 reported HIV/AIDS cases in the region. Whereas the reported AIDS incidence is 62 per million inhabitants for Latin America, the incidence in Honduras is almost twice as high, with 102 reported AIDS cases per million inhabitants in 2001. Similarly, the country's HIV incidence rose from 50 cases per million at the end of the 1980s to 404 per million in 1998 . 77 In most of Latin America, the epidemic is concentrated in high-risk groups, but in Honduras the epidemic has begun spreading out beyond those groups104. The main mode of transmission is heterosexual sex, whereas in South America other modes of transmission predominate. This makes the epidemic in Honduras similar to the pattern in the Caribbean region, which after Sub- Saharan Africa is the most affected region in the world. In Honduras, AIDS is the second leading cause of hospitalization and death after injuries due to violence, and has been the leading cause of death in women of childbearing age since 1997. It was estimated that at the end of 1999, 63,000 adults between the ages of 15 and 49 were living with HIV/AIDS and 4,200 had died of AIDS, leaving 13,599 orphans. This implies a mortality rate from HIV/AIDS of 665 per million inhabitants104. Available data on HIV/AIDS prevalence and incidence suffer from underreporting, which is evidenced by the fact that people who access services for the first time are usually at an advanced stage of the infection. Although Honduras was the first country in the region to implement a country-wide epidemiological surveillance system with free HIV testing available to the population, fluctuations in reported data over the years reflect failures in the system84. It has been estimated that between 30 and 50% of cases are still not reported2. HIV was likely introduced to the Honduran population through multiple channels. Among the first 100 reported AIDS cases, more than two thirds were acquired through heterosexual contact, close to 20 were men who have sex with men (MSM), and other modes of transmission accounted for much smaller proportions (2 intravenous drug users, 2 cases of mother-child transmission, 1 blood transfusion recipient)21,84. The following chart presents the distribution by mode of transmission of all AIDS cases reported in Honduras as of December 2001:83.6% heterosexual transmission, 7.4% sex between men and 6.4% mother to child transmission. Blood transfusions accounted for 0.6% of cases, drug users for 0.1%, and the mode of transmission in 1.9% of cases was unknown. fAs reported by National HIV/AIDS Program managers from Central America, confirmed during a PAHO/PASCA- sponsored meeting in San Salvador, May 20-24, 2002. 3 Chart 1.1 Modes of Transmission in Reported AIDS Cases, Honduras (1985-2001) Blood Transfusion Drug Users Mother to Child 0,6% 0,1% 6,4% Unknown Sex between 1,9% Men 7,4% Heterosexual 83,6% Source: Reproduced from the World Bank, 2002 The pattern seems to be moving towards more heterosexual transmission: of the cases reported in 2001, 88% (576) were heterosexual transmission, 5% (29 cases) homosexual transmission, and 7% (47 cases) vertical transmissiong. As a result of the predominantly heterosexual mode of transmission, almost as many women as men are affected by the disease in Honduras. The male to female ratio of HIV/AIDS infection dropped from 1.7 in 1994 to 1.2 in 2001h, the lowest ratio in Latin America. Within the general population, the group most affected are adults between 20 and 39 years (Graph 1.1). The epidemic is mainly concentrated in the cities of San Pedro Sula, Tegucigalpa, and along the central corridor between the capital and the northern and southern coasts (Graph 1.2). Although no data are currently available for rural areas, all regions have reported HIV/AIDS cases and the epidemic is very likely spreading to the rural areas. Graph 1.1 AIDS Cases by Age Groups Honduras, 1985-2001 3000 2694 2496 2500 1926 2000 1792 Cases 1500 1120 1000 668 670 AIDS 449 500 398 168 263 256 60 65 0 00-04 05-09 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-Mas Ignor. Age Groups gHowever, underreporting of homosexual transmission is likely in Honduras. In-depth interviews of 100 cases formerly classified as "heterosexual" transmission in Honduras uncovered that 53 of them had sex with another man. Behavioral surveys of MSM consistently reveal that many of them also have relations with women. 4 Graph 1.2 AIDS Cases by City - Honduras 1985- 2001 4000 3747 3500 3000 2645 Cases 2500 2000 AIDS 1500 No 1000 491444 312 326 297 500 271 261305160 132 0 Source: Secretaría de Salud de Honduras, STI/HIV/AIDS/TB Department One factor that increases the risk of HIV transmission is the high incidence of STDs in the country. In Central America, the incidence of curable STDs for the 15-49 age group is 130 per million 104 . In Honduras, the HIV prevalence among STD patients in 1989-92 was 14% in the department of Cortes40. In addition, condom use has been found to be variable in risky sexual encounters: in the year 2000, 11% of a sample of 400 CSW reported having had unprotected contacts with a new client in the last 12 months, and 16% had unprotected contacts with known clients31. By comparison, in a sample of 300 MSM, the average percentage of reported unprotected contacts was 53% (See Appendix IV for more figures on condom use). In conclusion, the epidemic in Honduras is worse than in other parts of Latin America. There are clearly identifiable risk groups and regions where prevalence is high and the trend suggests that the epidemic is still growing, albeit at a reduced rate2 (see Graph 1.3). Graph 1.3 Total AIDS Cases - Honduras 1985-2001 900 14000 year/ 800 12000 700 AIDS 600 10000 Cases 500 8000 400 6000 Cases AIDS 300 cumulated 200 4000 of of 100 2000 No. 0 0 No. 1985 1987 1989 1991 1993 1995 1997 1999 2001 Women Men Cumulated Source: Secretaría de Salud de Honduras, STI/HIV/AIDS/TB Department hCurrent statistics from the STI/AIDS National Program. 5 THE COST OF HIV/AIDS PREVENTION, TREATMENT AND CARE The HIV/AIDS epidemic has rapidly become a problem that transcends the realm of personal tragedy to become a major public health issue and in some cases even a concern for a country's overall economic development. The two principal reasons for the economic impact of the epidemic are: (i) it affects adults in the most productive years of their livesi, and (ii) treatment for associated opportunistic infections and antiretroviral therapy is very expensive and needs to be sustained for the rest of the patient's life . 26 Most countries have limited resources to use to prevent new infections and mitigate the effect of HIV/AIDS on households and systems, let alone to subsidize treatment costs. Furthermore, the additional resource demands on the public sector associated with the HIV/AIDS epidemic have usually not been translated into an increased budget for health104. This suggests that spending on HIV/AIDS may be crowding out other activities in the health care sector . The amount and composition of national HIV/AIDS expenditure in Honduras is shown in Table 1.1 below. It indicates that so far, the greatest burden has fallen on households, who spend a lot on treatment and antiretroviral drugs. This spending pattern has severe equity implications in a country where two-thirds of households were estimated to live below the poverty line in 199962. Table 1.1 Spending on HIV/AIDS in Honduras in 1999 Million US$ As % of total Per capita Total health expenditure in Honduras $344 $ 56 Public expenditure on health 57 $127 37% $21 Total expenditure on HIV/AIDSj $22.1 $3.5 Public expenditure on HIV/AIDS $3.4 15.5% $0.5 Household (hh) expenditure on HIV/AIDS $12.2 55% $1.9 % hh spending on treatment 92% $11.2 % hh spending on antiretroviral therapy 50% $6.1 Total health expenditure as % of GDP 7.2% HIV/AIDS expenditure as % of total health expenditure 6.1% Source: Secretaría de Salud. Cuentas Nacionales en VIH/SIDA: Honduras 1999 To better grasp the financial implications of the AIDS epidemic, one can compare public expenditure on HIV/AIDS in Honduras to the cost of providing treatment to the population and implementing an effective prevention package. Assuming antiretroviral therapy costs US$1,000 per patient per year, the country would need US$63 million per year,k equivalent to 20 times the iA study of 3 Central American countries (see González) suggests that the productive life of a worker is diminished by 60-70% if he has AIDS and does not receive ARV therapy. jTo calculate the total budget for HIVAIDS as a percentage of public health expenditure, low and high estimates from Cuentas Nacionales en VIH/SIDA Honduras(CNS)1999 were averaged. The average between low and high CNS estimates will be used in the remainder of the document when referring to CNS figures. kThis excludes all the infrastructure, testing, and training costs to support the widespread introduction of ARV. 6 current public expenditure for HIV/AIDS and about half of the current total public budget for healthl. The cost of prevention efforts, on the other hand, is much lower. Broomberg15 estimates that to implement a comprehensive HIV/AIDS prevention packagem in the Latin America and Caribbean region would cost $0.75 per capita. At this price, prevention efforts to reach the whole Honduran population would cost US$ 4.7 million, which is 1.4 times the current public expenditure on HIV/AIDS or 3.5% of the total public health budgetn. In addition to costing less, prevention is the principal way to influence the future trend of the epidemic, and mitigate the future cost to society resulting from new infections. These numbers clearly show that the cost of a fully comprehensive HIV/AIDS prevention, treatment and care program is beyond the financial capacity of Honduras, even with greatly increased external financial assistance from, for example, the Global Fund to Fight AIDS, Tuberculosis and Malaria. A less than comprehensive program is more affordable, but this means that policymakers must prioritize among competing requests for program funding. For lack of clear decision criteria, policymakers often find themselves investing in HIV/AIDS programs in response to particular interests and pressure groups (donors, NGOs, religious organizations, certain constituencies, etc.) rather than to maximize the impact on the epidemic. Alternatively, an "egalitarian" policymaker, reacting to the interest of special groups, might distribute resources available for HIV prevention proportionally to the number of AIDS cases in different sub-groups. Paradoxically, this "egalitarian" resource allocation could make everyone (even lower-risk groups) worse off, i.e. at greater risk of developing an infection than if all resources had been targeted to a single high-risk group56. In summary, in a world with competing interests and multiple possible interventions, the model provides policymakers with decision guidelines for achieving the greatest impact from domestic and external funding for HIV prevention. lCalculated as 63,000 current AIDS cases times an average cost of $1,000 per case per year compared to a public HIV/AIDS budget of $3.4 million and a total public budget for health of $134 million. m The package includes the following strategies identified as major cost effective interventions in the literature reviewed by the authors: 1. Promotion of safer sex behaviors through mass media programs; 2. Education of prostitutes and their clients, and provision of condoms; 3. Provision of combined sex/HIV education in secondary schools; 4. Ensuring safety of blood transfusions through screening of donated blood for HIV; 5. Provision of STD treatment services; 6. Provision of condoms through social marketing programs; 7. Prevention of unsafe drug use behaviors through needle exchange/bleach provision programs. nFor these estimates: $0.75 was multiplied by Honduras' population estimate for 1999 of 6.3 million = $4.7 million and compared to $3.4 million (public expenditure on HIV/AIDS), and $134 million (total public health expenditure), respectively. 7 CHAPTER 2: METHODOLOGY SUMMARY DESCRIPTION OF THE MODEL The model's methodology is based on No Time to Lose37 ­ referred to as the core model ­ with additions described below. The methodology poses a classic maximization problem: how does one maximize the number of new infections prevented given a production function for HIV prevention and a resource constraint? The mathematical formulation and solution to this problem are presented in Appendix Io. The model calculates the impact of allocating resources among different interventions and, for each budget level, chooses the allocation which yields the greatest number of prevented HIV infections. The model uses information about resource constraints (i.e. the available budget) and the production functions for HIV prevention. The production functions transform "dollars spent" into "infections averted" to tell us the number of infections that can be prevented in a particular target group through a specific intervention at different budget levels. The production functions are built using estimates of the epidemiology of HIV and the cost and effectiveness of specific prevention programs for each target subgroup. If these estimates can be obtained for each of the interventions, intervention-specific production functions can be generated. These are added to generate an aggregate production function which is used to solve the constrained maximization problem. The core model was extended to accommodate two important features. First, a step was added to map subpopulations onto prevention interventions. This step allows us to present the results in terms of allocation of resources among prevention interventions rather than among population subgroups. This addition is because policymakers are more likely to manipulate funding levels by interventions than by population subgroups. A second addition to the core model is the inclusion of secondary infections. Prevention in some groups has a large multiplier effectp. For example, consistent use of condoms by commercial sex workers prevents infections among CSWs and among their clients and clients' partners. Interventions targeted to groups with large multipliers will have bigger total preventive effects, which makes it important to include secondary infections. THE TEN STEPS INVOLVED IN THE OPTIMIZATION EXERCISE 1. Define target population groups. The target groups and their respective sizes are identified and the proportion of each group that can realistically be reached by a given intervention is estimated. The definition of groups is based on the identification criteria and intervention strategies currently used in Honduras, which are related to risk behaviour and susceptibility to interventions. 2. Estimate HIV incidence rates for each group. The number of new HIV infections (also called primary infections) that would occur in each population group each year is estimated. Several methods have been developed worldwide to estimate incidence rates in high-risk populations. These include cohort studies, cross-sectional samples, back calculation from reported AIDS oThe model can be downloaded from: www.worldbank/lachealth pFor details, see Box 2.6, in World Bank 1997: Confronting AIDS. 8 incidence to historical HIV infection rates, study of immunological markers for HIV infection, and snapshot analysis 10,14, 42, 45 . However, for the purpose of this study, these methods were not available. It was thus necessary to rely on the available information from previous studies in the country, and on the judgment of experts working in this area. For some target groups, the HIV incidence rate was estimated from available prevalence data using the AIDSProj Spreadsheet Model . This model generates epidemiological projections 92 based on available data and derives incidence figures from prevalence data. Appendix III presents an example of this methodology. 3. Calculate the number of secondary infections that arise from primary infections. Since an HIV infected person can transmit the infection to others, when estimating the benefits of preventing an HIV infection, it is necessary to take into account both the benefits for the immediately affected individual and also people who s/he might otherwise have infected, incorporating potentially significant `multiplier effects'103. Studies have shown that some low-risk groups might be better protected by interventions that target linked high-risk groups than by direct interventions56. Secondary infections were calculated from data on the infectivity of each contact, the infectivity duration, the frequency of contacts, and the presence of STIs. These parameters differ significantly from one group to the other (see Appendix I). 4. Define the prevention interventions to be included in the analysis. In this step, local experts identified a list of prevention interventions on the basis of national experience (existing program design) and international best practice. These include collective interventions that reach across several population groups simultaneously, and interventions aimed at specific groups of individuals. For example, a mass media campaign targets the whole population, while an IEC strategy for Garifunas exclusively targets this group. 5. Estimate the composition of the population reached by each intervention. For the total population reached by each intervention, experts estimated what percentage of each target population group was reached. 6. Estimate the number of people reached by each intervention at different funding levels. n this step, realistic budget increments were defined and the corresponding number of people that can be reached by each intervention was estimated. This was done using information on the average cost of reaching a person through a particular intervention and subjectively estimating the saturation point, beyond which it is infinitely costly to reach more of the population. This can be represented graphically as follows: Saturation of a Target Population Individuals of Reached No. Budget Allocated to Intervention 9 7. Estimate the effectiveness of each HIV prevention strategy included in the analysis. In this step, the percentage of new HIV infections that can be prevented within the population reached by a given intervention with a defined budget level was estimated, drawing on available program effectiveness data in Honduras and other countries. 8. Calculate the number of new (primary and secondary) infections averted for each budget increment and for each intervention. This step is calculated automatically by the model using inputs 1 through 7. A correction factor is then applied to account for the fact that the effectiveness of an intervention in isolation is not the same as its effectiveness in parallel with other interventions. To illustrate this, suppose a country has an incidence of 1,000 cases of HIV infection per year. Applying a hypothetical Intervention A which prevents 30% of these infections, one would be left with an annual incidence of 700. A second intervention (Intervention B) with an effectiveness of 25% would now prevent 25% of 700 infections, and not 25% of the original 1,000 infections. Therefore, the collective effectiveness of multiple interventions is less than the sum of the effectiveness of each intervention. To account for this, a correction factor for concurrent effectiveness is applied, as detailed in Appendix I. 9. Derive the optimal resource allocation scenario. From the results of step 8, the model derives the optimal allocation across prevention programs for each budget level. 10. Incorporate additional constraints such as reserved spending. The model is designed to find the resource allocation among different target groups that will maximize the number of new infections prevented. However, it is difficult for policymakers to base their resource allocation decisions solely on cost-effectiveness criteria. Other criteria (such as equity) and constraints (social or political) come into play, which may require policymakers to reserve certain amounts of funding for particular groups or programs even if this means departing from the optimal allocation. So we included scenarios with reserved spending for certain target interventions and compared the resulting number of new infections averted with the optimal resource allocation scenario. This additional step allows policymakers to value their decision to depart from the optimal allocation of resources explicitly in terms of foregone averted HIV infections. CAVEATS 1. Due to the lack of available data, many of the model inputs were estimated subjectively by the experts. The exercise supposes that estimation by consensus is an acceptable second-best in the absence of data. Policymakers in Honduras have little information on the size of some of the key target populations (men who have sex with men, Garifunas, etc), the number of new infections expected to occur in each of these subgroups, and the effectiveness of different prevention interventions, making it difficult to estimate program effectiveness and baseline infection rates by subgroup. Some of these estimates were therefore generated subjectively in two steps. First, all the relevant data that could be obtained from the country's surveillance system, national statistics, and from local and international studies were collected. On the basis of a literature review, two local epidemiologists produced an initial estimate of all the variables that feed into the model. Second, these initial estimates were reviewed by a large group of Honduran and international experts during a 3-day workshop that agreed on consensus estimates for all input values. 2. The model is of greater use to policymakers because it bases allocation decisions on interventions rather than on population subgroups, but this generates technical difficulties. 10 To ensure a correct optimization process, corrections are included in the model that make sure that once a group has been saturated (i.e. that all its members have been reached by the intervention) additional funds allocated to the same intervention are redirected to the members of another group that has not been saturated yet. This issue is discussed in more detail in Appendix II. 3. The model does not take into account the positive effects of one intervention on the effectiveness of another. The effectiveness of voluntary counseling and testing can be increased, for instance, if it is preceded by an information campaign. Factoring in such synergies could be a follow-up improvement of the model. 4. Another limitation of the model is that it does not correct for double counting of people who have more than one specific characteristic that puts them at risk. Each characteristic (e.g. being a CSW, pregnant, a Garifuna, in prison, etc.) is targeted by different interventions using specific strategies that address the defined risk. To illustrate this, consider the example of a Garifuna adolescent MSM. He will be exposed to interventions targeting adolescents, but also to those targeting the Garifuna and MSM groups. We chose not to make the population groups exclusive because each of these three characteristics is important in its own right when determining how to target the individual through a given intervention. The consequence of this non-exclusiveness is that the sum of new infections expected in the next year in the high-risk groups does not correspond to the number of new infections that would be given by an epidemiological model. However, this does not have a significant impact on the allocation of resources as the size of the population groups does not affect the cost- effectiveness of the interventions. The implications for resource allocation arise once we reach the saturation point for each population group, which would be marginally different if the populations were exclusive. 5. The model weighs all infections prevented equally, irrespective of the mode of transmission or the characteristics of the person who has been saved from infection. In other words, the infection of a baby through maternal-child transmission and a new infection acquired through a commercial sex act are weighed equally. The results of the model ­ whose purpose is to find the resource allocation strategy that maximizes the number of new infections averted ­ must be discussed widely to allow for corrections in the weights assigned to an infection in each population groupq. To help in this discussion, the model was run using averted Years of Life Lost (YLL) as its objective function. This procedure gives more weight to the prevention of an infection among babies and adolescents than the prevention of an adult infection. 6. A related issue is that linking costs to the impact of prevention activities over a specific timeframe is very complex.36 If two interventions have the same cost and the same effectiveness in changing behavior, the one that has more immediate benefits will be prioritized by the model. For example, let us suppose an IEC strategy changes the behavior of adolescents as effectively as it changes that of the general population. Since the benefits accruing to the adolescents will materialize over a longer time frame (i.e. at the time when they are more sexually active and thus more at risk to become infected and transmit the infection), the model will consider IEC in the general population to be more effective. qSee Hammer (1993) for a discussion of the gap between people's preferences and the valuations of life implicit in cost-effectiveness ratios. 11 7. The exercise was conducted using average cost-effectiveness rather than marginal cost effectiveness: in other words, it is assumed that it is as easy to reach the first 100 persons in any population group as the next 100 persons in that group. In addition, fixed costs were not included in the calculation of the average cost. These two simplifications had to be made because of data limitations. Instead, the experts were asked to determine what proportion of each subgroup could be reached with the prevention programs, i.e., define saturation points for each subgroup. A unique average cost was used for reaching the population until this saturation point and the cost of reaching the population beyond the saturation point was considered to be infinite. A useful follow-up exercise to this study could be to relax this constraint. 8. More fundamentally, this model belongs to a controversial class of models that seeks to orient public spending on the basis of relative cost-effectiveness. The use of cost-effectiveness as a valid criterion for guiding public spending decisions is still controversialr. Cost-effectiveness is not used here to prove that public financing is needed but to prioritize among interventions once it has been determined that they merit public financing. Despite the limitation arising from the nature of the data and the estimates used, the model's broad recommendations are reliable and offer a significant improvement in allocation over current practices. It is thus correct to prioritize the interventions that the model shows to be several orders of magnitude more cost-effective than others. However, small differences in the cost-effectiveness of interventions, while suggestive, are more difficult to uphold. With these limitations in mind, we turn in the following chapter to the estimation of model inputs for Honduras, and in a subsequent chapter to a description and analysis of the results. rSee especially Hammer (1997), Musgrove (1999), Jack (2000), and Musgrove (2000). 12 CHAPTER 3: ESTIMATING INPUTS FOR THE MODEL This chapter describes the procedure used to estimate the data that were fed into the optimization model. Over a period of one month, two local experts first prepared an initial estimate of the parameters and functions needed to run the model on the basis of an extensive literature search and expert consultations. Their estimates (and sources) were then submitted to a group of about 40 national and international experts working in the field of HIV/AIDS during a three-day workshop from May 13 to 15, 2002 in Tegucigalpa (see Appendix VI for the list of participants). The experts spent the first two days of the workshop analyzing each estimate, debating them, and then modifying them according to their own experience and knowledge of data sources. For each of the ten steps of the model, workshop participant broke into groups of 5 to 8 to arrive at a first consensus estimate of the model inputs. Each working group then submitted its results to the rest of the workshop participants. The whole assembly then proceeded to approve the final estimates for each step according to one of the following previously agreed-upon procedures: (i) consensus among all experts; (ii) taking the average from the figures estimated by each group; (iii) taking the median from the figures estimated by each group. Workshop participants belonged broadly to three categories : (a) epidemiologists and prevention program specialists from the Ministry of Health (central and regional levels) and the IHSS; (b) persons representing local and international NGOs working on HIV/AIDS prevention and care; and (c) economists and epidemiologists belonging to international aid organizations. The participants covered a wide knowledge and motivation spectrum. While the consensus exercise and the fact that the participants were asked to work from the data obtained from the literature search contributed to reduce bias, it is impossible to completely eliminate such bias. STEPS 1 AND 2 Step 1 involves: (a) defining the population groups to be targeted; (b) estimating the size of each population group; and (c) estimating the proportion of each group that can be reached by a given intervention. In step 2, the number of primary infections that are expected to occur in each population group over the next 12 months (incidence) is estimated. The following population groups were included: commercial sex workers (CSW), men who have sex with men (MSM), prisoners, the Garifuna population, adolescents, pregnant women, babies born of HIV+ mothers, blood transfusion recipients, and the rest of the population. Table 3.1 below summarizes the consensus estimates by workshop participants for steps 1 and 2 (see Appendix IV for data sources). 13 Table 3.1 Workshop Consensus Estimates for Steps 1 and 2 No. of new infections Proportion that can be Population Groups Population per year reached Commercial Sex Workers 14,462 347 50% Men who have sex with Men 57,116 971 30% Garifunas 142,760 1,713 55% Adolescents 1,507,687 1,508 60% Prisoners 12,518 125 80% Pregnant Women 188,179 188 66% Babies Born of HIV+ Mothers 2,823 1,129 66% Workers in their Workplace 106,530 106 75% Blood Transfusion Recipients 18,246 119 100% Rest of the population 4,434,241 4,434 40% Total 6,484,562 10,641 These subgroups by no means constitute an exhaustive list. If it is believed that, say, street children are particularly at risk and/or require specific interventions, they can also be included in the model if the data required for their inclusion can be estimated. Along these lines, the following target groups were considered for inclusion but rejected: Intravenous Drug Users (IDUs): While in many countries this is a high risk group, in Honduras, passive surveillance systems have only detected 9 cases of HIV infection through this route since 1985. In the last 4 years, there were no reported cases of infection through intravenous drug use. Consequently, IDUs were not included as a target group77. Mobile Populations: The only study performed in this group55 found no evidence linking migration to AIDS. In Honduras, a 1998 study85 on truck drivers found a prevalence of 1.1%, and a 2001 study30 among sailors showed a prevalence of 0.5%. Because prevalence among these mobile groups does not exceed that among the general population, they were not included as separate risk groups in the model. STEP 3 In this step, the number of secondary infections that arise from the primary infections estimated in step 2 is calculated. To estimate the number of secondary infections generated by primary infections in each population group, the following parameters must be estimated: the average number of sexual partners, the mean HIV prevalence of the sexual partners of each subgroup, the transmission probability per unprotected contact, protection (condom) effectiveness, the average number of unprotected sex acts per partner, and the average number of protected sex acts per partner (See Appendix I for mathematical details). During the workshop, the experts reviewed the data found in the literature on several of the required figures and estimated the missing values from their own expertise. Table 3.2 presents the consensus estimates. The sources for the parameters are given in Appendix IV. 14 Table 3.2 Infectivity, Prevalence and Behavioral Data MSM MSM Groups CSW Pregnant General (contacts (contacts w/ Garifunas Adol. Prisoners Women Workers Pop. with men) women) Transmission probability per unprotected contact (without 0.10% 1.0% 0.20% 0.15% 0.15% 0.15% 0.15% 0.15% 0.15% STDs) Adjusted transmission probability per unprotected 0.29% 2.60% 0.52% 0.42% 0.17% 0.23% 0.17% 0.17% 0.17% contact (with STDs) % always using a condom 45% 47% 47% 10% 17% 7% 17% 17% 17% % unprotected contacts 55% 53% 53% 90% 83% 93% 83% 83% 83% Condom effectiveness 95% 95% 95% 95% 95% 95% 95% 95% 95% Number of sexual partners 493 11 18 3.0 0.5 2.0 1 1.5 1.0 per year Number of contacts per sexual partner per year 1.3 20 7 52 12 52 52 104 104 Tot No of contacts / year 637 211 125 156 6 104 52 156 104 Expected years of life 6 7 7 6 8 6 8 7 7 No of secondary infections arising from one primary 6.4 16.5 2.5 3.0 0.1 1.2 0.6 1.4 1.0 infection in the subgroup Note that there are two ways of modeling secondary infectionst that produce significantly different numbers of secondary infections. The Bernoulli model was used in this exercise, but the optimization model was also run using the linear formulation: the resulting optimal allocation was the same in both cases. This study, however, should not be relied upon to provide policymakers with robust estimates of the absolute number of secondary infectionsu. STEP 4 In this step, the workshop participants were asked to define preventive interventions to be included in the model. Experts agreed to include the following interventions, in part because they are the ones included in the funding proposal the Government of Honduras submitted to the Global Fund for AIDS, Tuberculosis and Malaria5,94,67,68,69,70, 73,79. 1. HIV Counseling with Gender Focus and Access to Rapid Testing. sBlood transfusion recipients follow the same pattern as the general population except that their expected life after infection is 2 years, leading to 0.3 secondary infections per primary infection in this group. tThe first is a linear model that can be found in Over and Piot 1993 and the second is the Bernoulli model detailed in Pinkerton 1998. uIn addition, no discounting of secondary infections is carried out here, signifying that an infection prevented in a few years time has as much value to the policymaker as one prevented today. 15 2. Information, Education and Communication (IEC) for Pregnant Women. 3. IEC for the Garifuna Population. 4. IEC for Adolescents. 5. IEC for High-Risk Groups (CSW, MSM and Prisoners). 6. Interventions in the Workplace. 7. Strengthening the Vertical Transmission Prevention Program. 8. Syndromic Management of STIs. 9. Condom Distribution in High-Risk Groups. 10. Condom Social Marketing. 11. Blood Safety. 12. Supporting the Promotion and Defense of Human Rights. The strategies underlying each intervention are presented in step 6 below. Other interventions were discussed, including the treatment and care of people living with HIV/AIDS. Since the focus of this intervention is not primarily preventive, the group decided not to include it in the model. Note that not all the potentially cost-effective interventions are included in this exercise: only those indicated by the workshop participants are included. The model is flexible enough to include additional interventions provided the data required for their inclusion can be estimated. STEP 5 Out of the total population reached by a given intervention, workshop participants were asked to estimate the percentage in each population subgroup. For example, if an intervention reaches 100,000 people, they estimated how many of these would be CSW, how many MSM, etc. Table 3.4 presents the median of the estimates from each group, which was accepted as final data to feed into the model for this step. 16 Table 3.4 Percentage of the population reached by an intervention corresponding to each group of & CSW, MTCT Interventions Mgt. Soc. tion Safety Rights Pregnant Garifuna Adoles. MSM, Counseling Testing IEC Women IEC IEC IEC Prison. Workplace Interventions Preven Syndrom. STIs Condom Distribution Condom Marketing Blood Human Commercial Sex Workers 24% 35% 26% 29% 27% 23% Men who have sex with Men 23% 35% 25% 27% 24% 29% Garifunas 10% 100% 16% 18% 17% 13% Adolescents 5% 100% 5% 10% 10% 6% Prisoners 19% 30% 13% 16% 8% 16% Pregnant Women 5% 100% 100% 5% 10% Workers in their Workplace 6% 100% 5% 5% 1% Transfusion recipients 100% Rest of the population 9% 7% 10% 2% Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% STEP 6 In this step, the workshop participants were asked to define the specific set of activities in each of the prevention interventions defined in Step 4. The costs and unit costs of conducting these activities were then calculated on the basis of consultations with staff from the National HIV/AIDS Program and from the unit cost calculations made by the Instituto de Salud Pública of Mexico in January 200223,61,65,74,78,93. The costs are expressed in US dollars; an exchange rate of 1US$ = 16.40 Lempiras was used for currency conversion. Table 3.4 below summarizes the unit costs for each intervention. Appendix IV presents the details on how these costs were obtained. Note that the actual contents of the activities (e.g., what specific messages the IEC campaigns should contain) are not defined here. It is assumed that they correspond to international best practice. Recommendations to finance any particular intervention therefore have to be taken to mean that the activities that make up the intervention follow international norms. If the interventions depart from standard practice, then the effectiveness parameters have to be revised to reflect this. 17 Table 3.5 Unit Costs per Target Person of each Intervention Intervention Unit Cost 1. Counseling and Access to Rapid Testing $18.30 2. IEC for Pregnant Women $9.15 3. IEC for Garifunas $8.70 4. IEC for Adolescents $10.45 5. IEC targeted at High-Risk Groups (CSW, MSM and Prisoners) $11.10 6. Interventions in the Workplace $20.90 7. Strengthening the Vertical Transmission Prevention Program $636.20 8. Syndromic Management of STIs $49.00 9. Condom Distribution in High-Risk Groups. $3.60 10. Condom Social Marketing. $1.70 11. Blood Safety $24.10 12. Supporting the Promotion and Defense of Human Rights. $10.30 STEP 7 In this step, workshop participants were asked to review data on effectiveness of interventions by target group (in terms of the percentage reduction in new HIV infections), and modify the effectiveness parameters when necessary, according to their expertise. Initial figures submitted for their revision came from a review of the Literature Database for Evaluating HIV/AIDS Interventions12. Prior to the workshop, studies from this database were analyzed and the percentage reduction in risk reported in these studies was used as a proxy for a percentage reduction in new HIV infections and applied to the corresponding risk groups and interventions of the model. Table 3.6 below presents the average of the estimates on effectiveness from each group, which was accepted as final data to feed into the model for this step. 18 Table 3.6 Effectiveness of Interventions of & CSW, MTCT Mgt. Soc. ention Safety Pregnant Garifuna Adolescents MSM, Interv Counseling Testing IEC Women IEC IEC IEC Prison. Workplace Interventions Prevention Syndrom. STIs Condom Distribution Condom Marketing Blood Rights Commercial 18% 30% 24% 31% 24% 5% SexWorkers Men who have sex with Men 25% 28% 23% 29% 24% 5% Garifunas 16% 17% 22% 18% 11% 5% Adolescents 9% 38% 5% 12% 9% 2% Prisoners 30% 23% 23% 18% 16% 2% Pregnant 8% 25% 52% 11% 2% Women Workers in their Workplace 12% 23% 17% 17% 2% Transfusion Recipients 95% Rest of the 6% 15% 6% 2% population Source: Average from group estimates based on participants expertise and information from the Literature Database for Evaluating HIV/AIDS Interventions. The Futures Group International 2002. Note that the effectiveness of a given intervention may vary across different population groups depending on their receptiveness to the intervention and their readiness to change their behavior as a result of it. Step 7 was the final step of data estimation and validation. All the data presented in this chapter were subsequently fed into the optimization model. The results from the model are presented and discussed in the next chapter. 19 CHAPTER 4: RESULTS AND DISCUSSION RESULTS The following cost-effectiveness curves were generated using the estimates described in the previous chapter. Figure 4.1 Cost-Effectiveness Curves for HIV Prevention Effectiveness of Interventions 3.000 1. HIV Counseling and Access to Rapid Testing 2.500 2- IEC for Pregnant Women 3.- IEC for the Garifunas 2.000 4. IEC for Adolescents Infections 5.IEC for High-Risk Groups (CSW, MSM and A Prisoners) 1.500 6. Workplace Intervention Averted of 7 Strengthening of the Vertical Transmission Program 1.000 8.Syndromic Management of STIs Number 9. Condom Distribution in High-Risk Groups 500 10. Condom Social Marketing 11. Blood Safety 0 12. Promotion and Defense of Human Rights $0 $250.000$500.000$750.000 $1.000.000$1.250.000$1.500.000$1.750.000$2.000.000$2.250.000$2.500.000$2.750.000$3.000.000$3.250.000$3.500.000$3.750.000$4.000.000$4.250.000$4.500.000$4.750.000$5.000.000 Resources Assigned to Each Intervention From Figure 4.1, it can be seen that by spending, say, $1,000,000, on Condom Social Marketing, about 2,100 (primary + secondary) HIV infections could be averted; if the same amount was spent on workplace interventions, only about 30 infections would be averted. These differences exist because each intervention is characterized by a different cost, different effectiveness, and a different maximum amount of primary infections that it can hope to avert. The model utilizes these cost-effectiveness curves to generate the optimum allocation of resources at different levels of budget for HIV prevention. Table 4.1 below summarizes the optimal budget allocations among prevention interventions for HIV prevention budgets of up to $10 million. 20 Table 4.1 is read as follows: suppose the additional budget available for HIV prevention is $5 million, then the allocation of resources that will prevent the maximum number of new infections is that in which: $1.25 million is spent on counseling and testing, $750,000 on IEC for the Garifuna population, $500,000 on IEC targeted at MSM, CSWs and prisoners, $1 million on the syndromic management of STIs, $500,000 on condom distribution among groups with risky behavior, $500,000 on condom social marketing, and the remaining $500,000 on supporting the promotion and defense of human rights. At this budget level, according to the optimization model, no additional resources should be allocated to IEC for pregnant women, IEC for adolescents, workplace interventions, strengthening the prevention of vertical transmission, and blood safety. With this optimal allocation of $5 million, a total of about 8,400 primary and secondary infections can be prevented. Table 4.1 Optimal Resource Allocation for Each Budget Level in and and the Rights Testing of STIs (US$) Garifunas MSM Intervention and of Social Rapid Pregnant the Distribution Adolescents High-Risk Groups Human Safety Counseling to for (CSW, Transmission of Budget for for for HIV IEC IEC IEC Workplace Condom Blood Promotion Strengthening Condom Total 1. Access 2- Women 3.- 4. 5.IEC Groups Prisoners) 6. 7 Vertical Program 8.Syndromic Management 9. High-Risk 10. Marketing 11. 12. Defense $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $250.000 $0 $0 $0 $0 $0 $0 $0 $0 $250.000 $0 $0 $0 $500.000 $0 $0 $0 $0 $0 $0 $0 $0 $250.000 $250.000 $0 $0 $750.000 $0 $0 $0 $0 $250.000 $0 $0 $0 $250.000 $250.000 $0 $0 $1.000.000 $0 $0 $0 $0 $500.000 $0 $0 $0 $250.000 $250.000 $0 $0 $1.250.000 $250.000 $0 $0 $0 $500.000 $0 $0 $0 $250.000 $250.000 $0 $0 $1.500.000 $250.000 $0 $0 $0 $500.000 $0 $0 $0 $500.000 $250.000 $0 $0 $1.750.000 $500.000 $0 $0 $0 $500.000 $0 $0 $0 $500.000 $250.000 $0 $0 $2.000.000 $750.000 $0 $0 $0 $500.000 $0 $0 $0 $500.000 $250.000 $0 $0 $2.250.000 $1.000.000 $0 $0 $0 $500.000 $0 $0 $0 $500.000 $250.000 $0 $0 $2.500.000 $1.000.000 $0 $250.000 $0 $500.000 $0 $0 $0 $500.000 $250.000 $0 $0 $2.750.000 $1.000.000 $0 $500.000 $0 $500.000 $0 $0 $0 $500.000 $250.000 $0 $0 $3.000.000 $1.250.000 $0 $500.000 $0 $500.000 $0 $0 $0 $500.000 $250.000 $0 $0 $3.250.000 $1.250.000 $0 $500.000 $0 $500.000 $0 $0 $0 $500.000 $500.000 $0 $0 $3.500.000 $1.250.000 $0 $500.000 $0 $500.000 $0 $0 $0 $500.000 $500.000 $0 $250.000 $3.750.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $0 $500.000 $500.000 $0 $250.000 $4.000.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $0 $500.000 $500.000 $0 $500.000 $4.250.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $250.000 $500.000 $500.000 $0 $500.000 $4.500.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $500.000 $500.000 $500.000 $0 $500.000 $4.750.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $750.000 $500.000 $500.000 $0 $500.000 $5.000.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $1.000.000 $500.000 $500.000 $0 $500.000 $5.250.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $1.250.000 $500.000 $500.000 $0 $500.000 $5.500.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $1.500.000 $500.000 $500.000 $0 $500.000 $5.750.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $1.750.000 $500.000 $500.000 $0 $500.000 $6.000.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $2.000.000 $500.000 $500.000 $0 $500.000 $6.250.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $2.250.000 $500.000 $500.000 $0 $500.000 $6.500.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $2.500.000 $500.000 $500.000 $0 $500.000 $6.750.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $2.750.000 $500.000 $500.000 $0 $500.000 $7.000.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $2.750.000 $750.000 $500.000 $0 $500.000 $7.250.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $0 $3.000.000 $750.000 $500.000 $0 $500.000 $7.500.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $250.000 $3.000.000 $750.000 $500.000 $0 $500.000 $7.750.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $500.000 $3.000.000 $750.000 $500.000 $0 $500.000 $8.000.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $750.000 $3.000.000 $750.000 $500.000 $0 $500.000 $8.250.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $1.000.000 $3.000.000 $750.000 $500.000 $0 $500.000 $8.500.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $1.000.000 $3.000.000 $750.000 $500.000 $250.000 $500.000 $8.750.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $1.000.000 $3.000.000 $750.000 $500.000 $500.000 $500.000 $9.000.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $1.250.000 $3.000.000 $750.000 $500.000 $500.000 $500.000 $9.250.000 $1.250.000 $0 $750.000 $0 $500.000 $0 $1.250.000 $3.000.000 $750.000 $500.000 $500.000 $750.000 $9.500.000 $1.500.000 $0 $750.000 $0 $500.000 $0 $1.250.000 $3.000.000 $750.000 $500.000 $500.000 $750.000 $9.750.000 $1.750.000 $0 $750.000 $0 $500.000 $0 $1.250.000 $3.000.000 $750.000 $500.000 $500.000 $750.000 $10.000.000 $2.000.000 $0 $750.000 $0 $500.000 $0 $1.250.000 $3.000.000 $750.000 $500.000 $500.000 $750.000 The results indicate that the first additional $500,000 available in Honduras for HIV prevention should be spent on condom social marketing and the distribution of condoms to groups with risk behavior (CSWs, MSM, prisoners, Garifunas, and adolescents). If $1 million dollars were available, it should be split 21 among the following three interventions: $500,000 for IEC targeted to high-risk groups, and $250,000 each for condom social marketing, and distribution of condoms to groups with high-risk behavior. The optimal allocation of resources reveals different funding strategies for different interventions. The amount allocated to some interventions, such as counseling and access to rapid testing, keeps increasing as the available HIV prevention budget increases. By contrast, the optimal amount allocated to other interventions, such as IEC for high-risk groups, increases sharply at first but then remains constant with further increases in the budget. For a third category of interventions, such as workplace interventions, the model recommends no investment, even when the prevention budget reaches $10 million: i.e., even at these levels, other interventions prevent more infections at a lower cost. The results of the optimal allocation exercise are presented in another format in Figure 4.2 below. The number of infections prevented with each budget increment is shown along with the cumulative number of new infections prevented. Because the resources are allocated by prioritizing cost-effective interventions, it follows that the first budget increments prevent more infections. As the budget increases, less cost- effective interventions, which are able to prevent fewer infections, start to be funded. Since the most cost- effective interventions are funded first, additional increments prevent relatively fewer new infections and the cumulative number of infections prevented starts tapering off. Figure 4.2 Total number of Infections Prevented HIV Infections Prevented with Optimal Allocation 2.500 14.000 12.000 2.000 10.000 Increment Prevented 1.500 Budget 8.000 by Infections of 6.000 1.000 No. Prevented HIV Infections 4.000 Prevented with Optimal Allocation Infections 500 Cumulated by Budget 2.000 Increment 0 0 Cumulated No. of $0 Infections Prevented with $2.000.000$4.000.000$6.000.000$8.000.000 Optimal Allocation $10.000.000$12.000.000$14.000.000$16.000.000$18.000.000$20.000.000$22.000.000$24.000.000$26.000.000$28.000.000$30.000.000$32.000.000$34.000.000$36.000.000$38.000.000$40.000.000 Total HIV Prevention Budget The model contains two additional features of interest to policymakers. The first allows the policymaker to compare alternative resource allocation decisions with each other and with the optimal allocation, in terms of infections averted. The second allows the policymaker to reserve part of the budget for certain interventions and submit the rest of the prevention budget to optimization. Here is an illustration of the use of these features. 22 For an HIV prevention budget of $2 million, the optimal allocation among interventions is given in Table 4.2 below. A maximum of about 6,400 primary and secondary infections can be averted if the resources are allocated among interventions as given by the model, according to cost-effectiveness criteria. With this optimum allocation, it costs about $314 on average to prevent an infection. Table 4.2 Optimal Allocation with $2 million & STIs Soc. Prison. Pregnant Garifuna Adoles. MSM, of Safety Intervention Counseling Testing IEC Women IEC IEC IEC CSW, Workplace Interventions Prevention MTCT Syndrom. Mgt. Condom Distribution Condom Marketing Blood Human Rights Total Optimal $750k $0 0 $0 $500k $0 $0 $0 $500k $250k $0 $0 $2m Allocation Infections 942 0 0 0 1,641 0 0 0 2,189 1,596 0 0 6,368 Prevented Suppose the policymaker is not satisfied with this allocation of resources and finds it politically and socially more agreeable to allocate the same budget as per Table 4.3 below. The number of infections prevented with this allocation is about 1,800, at a cost of about $1,125 per infection averted. With this allocation, the policymaker is only able to prevent about 30% of the number of infections that could be prevented with the same budget (if it was allocated optimally as shown in Table 4.2 above). The model thus makes the opportunity cost of alternative resource allocation decisions explicit. Table 4.3 Alternative Allocation 1 & STIs Soc. Prison. Pregnant Garifuna MSM, of Safety Intervention Counseling Testing IEC Women IEC IEC Adolescents IEC CSW, Workplace Interventions Prevention MTCT Syndrom. Mgt. Condom Distribution Condom Marketing Blood Rights Total Allocation $250k $0 $250k $250k $250k $250k $250k $0 $0 $0 $250k $250k $2m Infections 319 0 192 8 970 6 70 0 0 0 69 143 1,777 Prevented Now suppose the policymaker decides that the above trade-off between efficiency and sensitivity to socio- political constraints is unacceptable because the government is not prepared to accept that 4,600 (=6,400­ 1,800) persons get infected when this could have been prevented. Suppose instead, that the government decides to adopt an "intermediate" strategy that puts some weight on both cost-effectiveness and other criteria. Say the policymaker wants to reserve a minimum budget for certain interventions for valid reasons (liability, rule of rescue, social values, and political constraints) while submitting the remainder of the HIV prevention budget to the optimization process. In the example given in table 4.4 below, the policymaker reserves a budget of $1.25 million for allocation according to criteria other than cost- effectiveness, and submits the remaining $750,000 to the optimization process. 23 Table 4.4 Alternative Allocation 2: With Reserved Spending & Prison. STIs Soc. Safety ting Pregnant Garifuna MSM, of Intervention Counseling Tes IEC Women IEC IEC Adolescents IEC CSW, Workplace Interventions Prevention MTCT Syndrom. Mgt. Condom Distribution Condom Marketing Blood Rights Total Reserved $0 $250k $0 $250k $0 $0 $250k $0 $0 $0 $250k $250k $1.25m Spending Optimal $0 $0 $0 $0 $250k $0 $0 $0 $250k $250k $0 $0 $0.75m Allocation of Non-reserved Spending Resulting $0 $250k $0 $250k $250k $0 $250k $0 $250k $250k $250k $250k $2m Total Resource Allocation Infections 0 9 0 8 970 0 70 0 1,870 1,596 69 143 4,735 Prevented This time almost 4,750 infections are prevented, at an average cost of about $420 per infection prevented. This is a great improvement upon the previous resource allocation since it allows 2.7 times more infections to be prevented. This allocation prevents about 75% as many infections as the optimal allocation. It is a matter for policymakers to decide whether this ­ or a similar "intermediate" ­ allocation represents the right trade-off between pure efficiency and other criteria they judge important for resource allocation decisions. SENSITIVITY ANALYSIS The exercise is based on a large number of "guesstimates" that can all be challenged legitimately. The results from this model must be interpreted with caution, and updated when better data become available. However, when the key parameters (unit costs, effectiveness and secondary infections) are varied within a reasonable range, the optimal allocation changes only marginally (see the sensitivity analysis in Appendix V). For example, when the unit cost of each intervention is doubled, the same set of interventions are prioritized by the model and the other groups of interventions appear even later (i.e., when the total budget is even larger). Similarly, the optimal allocation remains stable when the effectiveness of each intervention is varied according to alternative figures estimated during the workshop. Finally, when secondary infections are omitted from the model, the prevention of mother-to-child transmission and blood safety become more attractive interventions at lower budget levels. In addition to changes in these parameters, the results are robust to the use of `averted Years of Life Lost' instead of `averted HIV infections' as the objective function of the model. However, it should be stressed that the results of the exercise depend on the way the interventions have been defined. In this respect, if Syndromic Management of STIs is split into two interventions, one focusing on high-risk and the other on low-risk groups, then the syndromic management of STIs for high- risk groups would be included in the optimal allocation at much lower budget levels, but still without a change in the three priority interventions. The sensitivity analysis indicates that the core results of the optimal allocation exercise are robust. Indeed, many of the estimates do not have to be correct in absolute terms: as long as their relative values do not change much, the results from the optimization exercise are quite stable. Naturally, changing a parameter 24 (unit cost for example) of a single intervention, while keeping the others constant, will modify the optimal allocation if the intervention now becomes relatively more or less cost-effective than another. It is not possible to conduct a sensitivity analysis on each parameter of the model individually, but readers are encouraged to download the model and experiment with varying the values of parameters they entertain doubts about and see the effect of these changes on the optimal allocation results. DISCUSSION The model works well in the context of a workshop where all the stakeholders in HIV prevention are invited to test out and discuss the implications of alternative resource allocation proposals. During the workshop in Honduras, it was suggested that the results of the model could be used to establish a minimum package of key interventions that must be carried out because of their potential impact on the epidemic: this is equivalent to reserving spending for the most cost-effective measures and using the amount remaining from the HIV prevention budget to address socio-political considerations and to fund strategies with medium to long term impact. In the case of Honduras, a minimum package would include the following three interventions: condom distribution, condom social marketing, and IEC for high risk populations (CSW, MSM and prisoners). This result is robust to the sensitivity analyses mentioned in the previous section. To underline the importance of funding the most cost-effective interventions, one can compare the effect of allocating $1million exclusively among the three most cost-effective to allocating $9 million split equally among the other 9 interventions. In the first case, with a budget of $1 million (assigned to the three priority interventions according to the optimal allocation scenario) a total of about 5,100 HIV infections can be averted. In the second case, with a total budget of $9 million, less than two thirds as many, or 3,200 HIV infections are averted. This shows that guaranteeing the funding of a minimum package of priority interventions ensures a great impact on the future course of the epidemic. The Hondurans recently submitted a proposal for the first round of funding from the Global Fund to Fight AIDS, TB, and Malaria. The proposal was for $41 million, with $27 million for HIV/AIDS prevention, care, and treatment. Against stiff competition (385 applications were submitted) the Honduran proposal was recommended for funding following revisions. It is hoped that, as a result of the workshop and further policy dialogue, the key interventions that were not previously in the Global Fund application will be included in the country's HIV/AIDS prevention plans, and will be funded by the Global Fund and other sources of external assistance, as well as domestic resources. In the base case scenario, even with a total budget of $40 million, only 11,600 infections can be prevented (see Figure 4.2), which corresponds to only 30% of expected infections. This is because workshop participants estimated that only between 30% to 55% of the key population groups can be reached by the interventions. However, if this constraint is relaxed and we assume that 100% of the three key risk groups (CSW, MSM and Garifunas) can be reached, with a budget of $40 million, about 21,000 or almost twice as many HIV infections can now be prevented. This means, for instance, that if the government were to spend more than $500,000 on IEC targeted at high risk groups (which is the budget level at which the number of infections averted by this intervention reaches a plateau ­ see point A in Figure 4.1), it should spend it on developing strategies to reach a greater percentage of these groups, rather than continue delivering the intervention as designed originally. The following points are worth noting when analyzing the results presented above: § The results apply marginally, i.e., for incremental budget allocations, not for the overall budget allocations. For example, when the analysis suggests that no money should be allocated to a particular intervention, it does not mean that all the budget that is currently allocated to the intervention in question should be withdrawn, but that no additional resources should be allocated to the intervention. 25 § As suggested earlier, cost-effectiveness should not be the sole criterion for making resource allocation decisions. Although the analysis suggests that no investments should be made for improving blood safety until the prevention budget reaches $8.5 million, a policymaker may want to override this recommendation for several reasons. First, transmission by blood transfusion has a near-100% infectivity rate and an incubation period of only about 2 years: policymakers may therefore invoke the rule of rescuev to invest in blood safety. Second, the Ministry of Health is directly liable for HIV transmissions that occur through the blood transfusion system; by contrast, the Ministry of Health cannot be held directly liable for transmission of the virus through the sexual mode, since the latter depends upon individuals behaviors. Because of the rule of rescue and liability issues, the Ministry of Health may want to invest in ensuring that all blood units are HIV-free whether or not this constitutes a cost-effective measure. Similarly, policymakers may want to overrule the recommendation of not investing further in preventing mother-to-child transmissions for ethical reasons even though it may not be cost-effective. § Enabling factors may be key for a given intervention to achieve expected results. Political will and leadership, for instance, is needed for the adoption of certain HIV prevention strategies (especially those involving marginalized groups). In other words, a prevention strategy may need more than the interventions in this exercise. Similarly, certain interventions with low cost- effectiveness may provide an important synergistic effect to other interventions prioritized by the model. A mass media campaign, for example, may prevent few new infections directly, but might make people more receptive to subsequent more targeted interventions. These synergies constitute another reason why the model recommendations not to finance certain interventions at all should not be taken literally. § The recommendationsarevalidtotheextentthattheinterventionsconsistofthespecificstrategies detailed in Appendix IV. Should the package of strategies that constitute a particular intervention be altered for technical or socio-political reasons, then the cost of the intervention and its effectiveness should also be updated and the model re-run with the new parameters. In other words, IEC for Adolescents, for instance, should be taken to mean the sum of TV and radio spots, peer education, capacity building for teachers in sex education, and the distribution of sex education materials. Should, for a variety of reasons, these strategies be revised, then the cost and effectiveness parameters should also be altered to reflect the change in the content of the intervention. If the composition of the target population for a particular intervention changes (step 5), the effectiveness and cost data similarly must be revised. § Given that the difference in impact between a prevention program that allocates resources optimally and one that doesn't is potentially very large, it pays off to develop systems that generate the information critical to run the optimization model. In the absence of data, we were forced to rely on a problematic second-best solution: expert consensus. The periodic estimation of incidence rates in high-risk populations (through such techniques as cohort studies, cross-sectional samples, back calculation from reported AIDS incidence to historical HIV infection rates, study of immunological markers for HIV infection, and/or snapshot analysis 10,14,42,45) and the periodic evaluation of intervention effectiveness in various subgroups would particularly improve the information base and enable better decision making. Despite its limitations, the exercise provides a solid base for discussions about the appropriate allocation of resources for HIV prevention in Honduras. The model is a flexible tool that allows policymakers to engage constructively with the various stakeholders involved in HIV prevention. Policymakers can revise the parameters associated with each prevention intervention and include new interventions in the model. vThe rule of rescue suggests that interventions should be prioritized when their absence leads to almost certain death, as is the case for blood safety. See Musgrove (1999) for a discussion of the rule of rescue criterion for public spending on health care. 26 More importantly, it allows explicit comparisons (in terms of number of infections averted) to be made between alternative combinations of prevention strategies. The model thus allows policymakers to calculate the effect of departures from the optimal allocation (for social, ethical, and political constraints) on the epidemic. The exercise also generates a framework for program monitoring and evaluation. It sets out clearly the interventions and their key components, the population groups that will be targeted, the expected effectiveness and unit cost of each intervention, the number of people that they aim to reach, and the expected impact. These provide ready indicators that can be monitored and updated as the prevention strategy is implemented and new data are collected, giving rise to a dynamic and transparent framework for guiding HIV prevention. In addition, several elements of the model, such as the cost structure and efficacy of the interventions, can be changed; program monitoring and supervision should therefore focus on improving them. The exercise provides a flexible tool that can be adapted to different countries. It relies on local expertise and consensus, which generates ownership. This tool hinges on cost-effectiveness, arguably the most important, but not the only criterion policymakers may want to factor into their resource allocation decisions. Its results are meant to inform policy discussions, but should not be interpreted too rigidly. Users should not extrapolate the optimal allocation results from Honduras to other contexts, but conduct the exercise to build consensus and facilitate policy discussions among local experts. Honduran and other policymakers now possess a model that can help them: simulate the effect of alternative resource allocations; generate consensus around the HIV prevention interventions that have the greatest impact on the epidemic; and monitor and evaluate the outcome of the chosen strategy. 27 28 APPENDIX I ­ MATHEMATICAL DESCRIPTION OF THE MODEL THE CORE MODEL w The number of new infections averted by each intervention i for a budget level x is calculated with the following formula: q Ii (x) = i(x) * Ia(o) * min [(Ni(x) * Pia), (Popa *fa)] / Popa ] a=1 where: Ii (x) is the number of new infections averted by each intervention i for budget level x i(x) is the effectiveness of a prevention program i as a function of the budget allocation x. It represents the percentage of new HIV infections that can be prevented within the total population reached by a given intervention i with a defined budget levelx assigned to that intervention. i(x) is estimated in Step 7 of the model (see Chapter 3). q is the total number of population groups. q is defined in Step 1a. Ia(o) is the baseline number of infections in group a. Ia(o) is the sum of primary and secondary infections, as estimated in Steps 2 and 3 respectively. Ni(x) is the totalnumber of people reached by intervention i at budget level x. Ni(x) = (total available budget) / (unit cost of the intervention) The unit cost of the intervention is an average cost, which is estimated in Step 6. This simplification was used due to data limitations that prevented an estimation of the marginal cost of each intervention (see Chapter 2, Caveats). Pia is the percentage of the population reached by intervention i that corresponds to group a. Pia is estimated in Step 5. Popa is the total population of group a. Popa is estimated in Step 1b. fa is the maximum proportion of group a that can be reached. fa is estimated in Step 1c. and (Ni(x) * Pia) = (Popa *fa) We seek to maximize the number of infections averted across all interventions under a budget constraint. This problem can be represented as follows: n q Max (i(x) * Ia(o) * [ (Ni(x) * Pia) / Popa ] ) i =1 a=1 w Adapted from Kaplan 2000. 29 s.t. xi = B where: n is the total number of interventions. These are defined in Step 4. xi is the budget allocated to each intervention B is the total Budget available for HIV/AIDS prevention programs ADDITIONS TO THE CORE MODEL Reserved Spending To account for reserved spending for a particular intervention, in the constrained maximization problem, we fix: xi = ri for i = 1,2,3,..., n. where ri is the guaranteed allocation for each group. Correction for Effects of Concurrent Interventions Effects of positive or negative synergy aside, the effectiveness of an intervention in isolation is always different than its effectiveness in coexistence with other interventions. For example, six interventions, which in isolation would each prevent 20% of all new infections, will not prevent 120% of all new infections when acting together. The model corrects for this using the following formulas: (i) The corrected number of infections averted by all interventions acting on group a is: n n Corrected Iia = Iia * {1-[(1-Eff.1a)* (1-Eff.2a)*........* (1-Eff.na)]} i=1 i=1 where: Iia is the number of new infections averted by each intervention i in group a, based on the assumption of isolated effectiveness n is the total number of interventions Eff.ia is the % effectiveness of intervention i in averting overall infections in group a (ii) Similarly, the corrected total number of infections prevented by intervention i is: q q Corrected Iia = Iia * {1-[(1-Eff.1a)* (1-Eff.2a)*........* (1-Eff.na)]} a=1 a=1 30 (iii) The corrected total number of infections prevented by all interventions is given by: n q Iia * {1-[(1-Eff.1a)* (1-Eff.2a)*........* (1-Eff.na)]} i =1 a=1 Secondary Infections The expected number of secondary infections in the general population that arise from one primary infection in a given subgroup (S) was calculated using the following Bernoulli formulax: S =m*(1-P)*[1-(1-An)n (1-Ak)k] Where: S is the expected number of secondary infections that arise from one primary infection m is the average number of sexual partners for a person in the population group under analysis P is the mean HIV prevalence among the sexual partners of a person in the population group under analysis An is the transmission probability per unprotected contact Ak is the transmission probability per protected contact where Ak = (1-e)*An e is condom effectiveness n is the average number of unprotected sex acts per partner k is the average number of protected sex acts per partner In addition, the effect on the transmission probability per contact of a co-infection with an ulcerous or non-ulcerous STD was included in the model using the following formula to calculate the transmission probability per unprotected contact An : An = T * (PSTDu*Mu + PSTDnu*Mnu) + T * (1 - PSTDu - PSTDnu) Where: T is the transmission probability with no STD co-infection PSTDu is the prevalence of ulcerous STDs in the subgroup PSTDnuis the prevalence of non ulcerous STDs in the subgroup Mu is the infectivity modification factor for ulcerous STDs Mnu is the infectivity modification factor for non-ulcerous STDs All the parameters for the calculation of S are presented in Step 3 of the Model. xFrom Pinkerton and Adamson 1998, and Pinkerton 2000. 31 32 APPENDIX II: THE EFFECT OF SUBGROUP SATURATION ON COST-EFFECTIVENESS Basing resource allocation decisions on specific subgroups of the population rather than interventions greatly simplifies the calculations involved. This is, however, not very useful for the policymaker since assigning certain amounts of resources to a subgroup does not solve the allocation question unambiguously; resources allocated to any specific subgroup can be spent in many different ways with different results. The approach adopted here is to base allocation on interventions and to define the composition of the population reached by each intervention. Both the interventions and their mapping onto subgroups must be defined to prevent ambiguity in the interpretation of the results of the model. By using interventions (which reach more than one subgroup) rather than subgroups as the basis for resource allocation, certain corrections have to be made to ensure a correct optimization process. An intervention, say counseling and access to rapid testing (see step 4), has a given composition of subgroups as its target population. As more money is allocated to this intervention, the smaller subgroups begin to get saturated. This means that the intervention has already reached the accessible population in the smaller subgroups (e.g., CSWs, MSM, prisoners). Budget increases for this intervention do not permit reaching more people in those subgroups. In other words, given enough resources, each intervention can reach a theoretical point where it cannot attain a single additional member out of a number of subgroups. The saturation of subgroups as explained above implies that the composition of the population attained by an intervention can change as the budget increases. The table below illustrates this point for a fictitious intervention that benefits three subgroups, two of which are small (1,500 CSW and 4,500 MSM) and the general population. The unit cost for the intervention is supposed to be US$10 per person. With $100,000, therefore, the intervention reaches 10,000 beneficiaries, of which 1,500 are CSWs and 1,500 are MSM. This means that the CSW population is saturated at this budget level, and as more funds become available for this intervention, they do not benefit the CSW population any longer. The same happens for MSM when $300,000 are allocated to the intervention. The Changing Composition of Population Reached: The Effects of Subgroup Saturation Number of Individuals Reached Percentage of Population Targeted by Intervention Reached Budget Level CSWs MSM General CSWs MSM General Population Population $100,000 1,500 1,500 7,000 15% 15% 70% $200,000 1,500 3,000 15,500 8% 15% 78% $300,000 1,500 4,500 24,000 5% 15% 80% $400,000 1,500 4,500 34,000 4% 11% 85% $500,000 1,500 4,500 44,000 3% 9% 88% 33 Note: Shaded cells indicate a subgroup at or beyond saturation point. As the budget increases, the model accounts for this by calculating the proportion of target population "not filled" by saturated subgroups, and increasing the composition by other subgroups to attain the correct number of individuals for that budget level. The implication for the optimization exercise is that the cost-effectiveness of an intervention changes (usually decreases) as the budget allocated to the intervention increases. Indeed, the effectiveness of the intervention is subgroup specific (see step 7); since the composition of the population benefiting from an intervention changes as the budget increases, the cost-effectiveness of the intervention also changes. The cost-effectiveness of the intervention tends to decrease because small, high-risk subgroups form a decreasing part of the population reached and thus the relative difficulty and cost of preventing each infection tends to increase. 34 APPENDIX III: DERIVATION OF INCIDENCE FROM PREVALENCE DATA: EXAMPLE OF MSM To estimate the HIV incidence for MSM, Garifunas and Prisoners, we applied the AIDSProj Spreadsheet Model92. This model uses information from surveillance in antenatal clinics, target group studies and general demographic data to generate incidence estimates on the basis of prevalence figures. As an example, the following graph presents the 1980-2010 projections for HIV prevalence and incidence in the MSM population in Honduras. Estimated Seroprevalence and Incidence in MSM (%) 30,0% 25,0% 20,0% 15,0% 10,0% 5,0% 0,0% 1980 1985 1990 1995 2000 2005 2010 Projected Prevalence Projected Incidence Reported Prevalence from Sentinel Surveillance The main inputs to generate the above graph include demographic and surveillance data from ante-natal clinics and information from specific studies in the MSM target group, as presented in the following tables: Population Projections Annual Population Growth Rate (%/yr) Total Population Adult Population in 1980y in 1980z 1980 2010 Honduras 3.70% 3.50% Tegucigalpa 3.19% 2.99% San Pedro Sula 268,349 139,998 3.27% 3.07% MSM 24,362 24,362 3.27% 3.07% Rest of country 2,953,567 1,529,462 3.81% 3.61% yPopulation data from the National Census. INE 2002. zAdult population defined as the population above 15 years 35 Ante-Natal Clinics and Sentinel Surveillance Data Tegucigalpa San Pedro Sula MSM 77,80,86,90 Ante-natal clinics surv. data Ante-natal clinics surv. data Year Prevalence Sample Standard Prevalence Sample Standard Prevalence Sample Standard size Error size Error size Error 1987 20.3% 187 5.8% 1988 1989 1990 3.4% 26.2% 49 12.3% 1991 0.20% 1292 0.2% 3.6% 416 1.8% 1992 0.30% 2.8% 400 1.6% 1993 0.30% 1025 0.3% 2.0% 400 1.4% 1994 4.0% 400 1.9% 1995 4.1% 400 1.9% 1996 1.03% 700 0.7% 1997 0.70% 2000 0.4% 1998 0.65% 771 0.6% 2.0% 1127 0.8% 8.0% 425 2.6% 1999 2000 2001 12.1% 230 4.2% In addition, information on: (a) the total population by sex and age with projections for the period 1980-2010; (b) the ratio of men to women; (c) the total number of AIDS cases from 1985 to 2001 by sex and age; (d) the population of the target group estimated for the year 1980; and (e) the annual population growth rate, from 1980 to 2010; were entered into the AIDSProj spreadsheet to create a correction factor adjusting for the fertility-inhibiting effect of HIV, male infections, and the population over the age of 49. The 1980-2001 data on reported AIDS cases per year permitted a comparison between projected and reported cases. The procedure used for Garifunas and prisoners is similar and uses the same general information, varying only in the data inputs particular to these groups. 36 APPENDIX IV: BACKGROUND DATA FOR STEPS 1,2,3 AND 6 STEPS 1 AND 2: DATA SOURCES % that Comments / Data Sources Population Populati No. of new can be Groups on infections per year reached aa Population Estimates Incidence Estimates Commercial 14,462 347 50% From "Estimating HIV/AIDS idem. Study gives incidence of 2.4% Sex Workers prevalence in Countries with low for CSW. level and concentrated epidemic: the example of Honduras" (to be published)91 Men who 57,116 971 30% Workshop experts estimated that Used the AIDSProj SpreadSheet have sex with 3% of male population in Model92,bb. Introduced prevalence data Men Honduras is gay. This was applied from 4 studies77,80,86,90: 1988=20.3%, to the adult male pop from INE 1990=26.2%, 1998=8%, and National Census 1988-2050 2001=12.1%. Correction factor was projections for 200239 (1,903,868) applied for sex, age and reduction in fertility. Obtained an incidence of 1.7% Garifunas 142,760 1,713 55% Experts estimated the % of Used the AIDSProj SpreadSheet Garifuna pop. in the regions of Model. Introduced prevalence data Cortes, Atlantida, Colon, Gracias a from only existing study87: 1998=8.4% Dios, and Islas de la Bahia82 and + data from surveillance studies77,80. applied these to the 2002 pop. Correction factor was applied for sex, projections from the INE National age and reduction in fertility. Obtained Census by municipality38. incidence of 1.2% Adolescents 1,507,68 1,508 60% Age group between 10 and 19 "Estimaciones de Proyección del 7 (WHO definition of adolescents) VIH/SIDA en Honduras 1988-2010"1. from INE National Census Gives 2002 incidence of 0,1% for the projections39. general pop. To date there are no studies among adolescents that give evidence that the incidence is higher in this group than in the general pop. Prisoners 12,518 125 80% From Daily Census of Prison Used the AIDSProj SpreadSheet Model Population83. Secretaría de with prevalence data38,75,77: Seguridad. Dirección General de 1988=0.62%, 1989=6.3% 1998=6.8% + Servicios Especiales. surveillance studies data80. Obtained 1% incidence aaThe values for Step 1c ­ estimating the proportion of each population that can realistically be reached given their characteristics ­ was obtained by averaging the estimates of all workshop participants. bbTo estimate HIV incidence for MSM, Garifunas and prisoners, the AIDSProj SpreadSheet Model was used. For details refer to http://www.tfgi.com/AIDSproj.asp. and see Appendix III. 37 Data Sources for Steps 1 and 2 (continued) No. of Propor- Comments / Data Sources Population new tion Groups Population infections that per year can be Population Estimates Incidence Estimates reached Pregnant 188,179 188 66% Expected births from INE National "Estimaciones de Proyección Women Census Projections39 (188,179). del VIH/SIDA en Honduras 66% of pregnant women are 1988-2010"1,88. Gives 2002 covered by the service provision incidence of 0,1% for the network of the Ministry of general pop, which was Health71,72 applied to pregnant women. Babies Born 2,823 1,129 66% We apply the HIV prevalence Assuming a mother to child of HIV+ figure of 1.4%1,88 to the total transmission probability of Mothers population of pregnant women 40%101 in the absence of any 188,179 to obtain the total number intervention, the expected of babies born of HIV+ women in number of HIV+ babies is the year and add the expected 1,129. number of new HIV+ pregnant women to obtain the total number of babies born to HIV+ women Workers in 106,530 106 75% This group was considered a Same as for pregnant women: their distinct one only because they are incidence of 0.1%1 Workplace already the target of a prevention program led by the National HIV/AIDS Program in coordination with Instituto Nacional de Seguridad Social (IHSS) and the Ministry of Labor4,66. Although the group refers to "workers", initially only workers of "maquilas" will be considered. Population figure from the Secretaría del Trabajo, Census of Maquila Workers (2002)8,9. % of HIV +ve blood donors Blood 18,246 119 100% The population corresponds to the =121 / 182,480 = 0.66% Transfusion number of blood transfusions per Infectivity of HIV positive Recipients year19. blood= 100%19. Rest of the 4,434,241 4,434 40% Difference between the total pop. Same as for pregnant women: population (obtained from the population incidence of 0.1%1 projections conducted by the Instituto Nacional de Estadística de Honduras39) and the above groups. Total 6,484,562 10,641 2002 total pop projections from Sum of new infections of INE census each group 38 STEP 3: DETAILS OF SOURCES AND ESTIMATION PROCEDURES IV. 1 Background data for the calculation of adjusted infectivity MSM MSM Group CSW (contacts (contacts Garifunas Adole PrisonersPregnant Women Workers GenPop w/ men) w/ women) HIV prevalence in subgroup 10.0% 12.1% 12.1% 8.4% 1.4% 6.8% 1.4% 1.4% 1.4% HIV prevalence in the partners 1.4% 12.1% 1.4% 8.4% 1.4% 1.4% 1.4% 1.4% 1.4% of the subgroup Transmission probability per non protected contact (without 0.10% 1.0% 0.20% 0.15% 0.15% 0.15% 0.15% 0.15% 0.15% STDs) % who reported having had a STD in last year 14% 12% 12% 15% 1% 4% 1% 1% 1% Prevalence of ulcerous STDs in HIV+ve individuals from 1.4% 2.5% 2.5% 2.2% 0.2% 0.6% 0.2% 0.2% 0.2% subgroup Prevalence of non-ulcerous STDs in HIV+ve individuals 13% 9.8% 9.8% 12.8% 1.0% 3.7% 1.0% 1.0% 1.0% from subgroup Infectivity modification factor for coexistence of ulcerous 60 30 30 30 30 30 30 30 30 STD Infectivity modification factor for coexistence of non-ulcerous 10 10 10 10 10 10 10 10 10 STD Adjusted transmission probability (infectivity) per non protected contact (with 0.29% 2.60% 0.52% 0.42% 0.17% 0.23% 0.17% 0.17% 0.17% STDs) 39 IV.2 Sources of the data presented in main text table 3.2 and in table IV.1 above Source HIV prevalence in subgroup Different studies were used to obtain prevalence data for the different groups: CSW91; MSM90; Garifunas87; prisoners75 ; adolescents, pregnant women and workers are assumed to have the same prevalence as the general population1 HIV prevalence in the partners of Same source as above, and the partners are assumed to be the general population the subgroup (P) except for MSM contacts with men, where they are MSM and for Garifunas. Transmission probability per non Source: AVERT Model63. Other sources were also consulted as a comparison56,60. protected contact (without STDs) ­ Infectivity. % who reported having had a STD HIV Seroepidemiological studies from the Secretaría de Salud were used for in last year CSW76, MSM86, Garifunas87, Prisoners75 and Pregnant Women88 . For adolescents, factory workers and the general population, figures from sentinel surveillance of pregnant women were applied88. Prevalence of ulcerous and non- Same sources as above + an epidemiological database of the STD/HIV Program ulcerous STDs in HIV+ve from the Secretaría de Salud73. For other groups, an average between % ulcerous and individuals from subgroup non-ulcerous STDs of groups with information was applied to the % who reported having an STD in the past year. Infectivity modification factor for Source: AVERT Model63. Other sources were also consulted as a comparison56,60 coexistence of ulcerous and non- ulcerous STDs % always using a condom HIV Seroepidemiological studies from the Secretaría de Salud were used for CSW76, MSM86, Garifunas87, Prisoners75 and Pregnant Women88 . For adolescents, factory workers and the general population, figures from sentinel surveillance of pregnant women were applied88. % unprotected contacts = 1 ­ (% always using a condom) Condom effectiveness Pinkerton 200060 Number of sexual partners and Estimated during the workshop. See details below. number of contacts per sexual partner Expected years of life Expert estimates during workshop. Number of secondary infections See Appendix I for a detail of the formula used to calculate the number of secondary arising from one primary infection infections. in the subgroup 40 IV.3 Sources and estimation procedure for no. of partners and no. of contacts Population group Number of Number of Number of Source partners contacts per contacts per partner year CSW 41.1 / month76 637 / (41.1*12) 637 Group work =1.3 MSM See detail below Group work Garifunas 3 156 / 3 = 52 156 (3 contacts/ Group work week) Adolescents 0.5 12 12*0.5 = 6 Group work Prisoners 2 104 / 2 = 52 104 Group work Pregnant Women 1 52 / 1 = 52 52 Group work Workers 1.5 156 / 1.5 =104 156 (3 contacts/ Group work week) General 1 104 / 1 = 104 104 (2 contacts/ 1.9 contacts/ population week) week= workshop survey average IV.4 Number of male sexual partners and contacts for MSM MSM who are MSM who are not CSW CSW Total number of contacts with male partners / year 290 182 (average from group work) % in each category90 27% 73% All MSM Total No. of contacts w/ male partners (weighted avge from above) 211 No. of male casual + stable partners (from PASMO background data) 10.8 No of contacts per male partner per year 19,6 41 IV.5 Number of female sexual partners and contacts for MSM Average No per month90 2 Average No per year (avge from group work) 18 No of contacts w/ female partners / year (avge from group work) 125 No. of contacts per female partner (=125 / 18) 7 % of Bisexual MSM (MSM who have sex with men and women)90 38% IV.6 Condom use among CSW91 Condom use Proportion With new clients 79% 33% With known clients 28% 67% Weighted average 45% IV.7 Condom use among MSM90 % who always use condoms MSM MSM who are Both CSW Stable partner 45% 24% 23% Known partner 23% 52% 47% New partner 62% 62% 62% 42 STEP 6: DETAILS OF UNIT COST CALCULATION The following tables present the components of each intervention, as defined during the workshop, with their associated costs and data sources. Intervention 1: Counseling and Access to Rapid Testing79,73,5 Activities Unit Cost Coverage Total Cost Counseling $6.8093 24,000cc $163,200 Testing $0.75dd 24,000 $18,000 Confirmatory Testing $0.75 9,512ee $7,134 M&E - Central Level $121.9565 9ff $1,098 M&E - Regional Level $146.3465 42gg $6,146 Training in Counseling $5,792.00 65,93 42 $243,264 Total Intervention 24,000 $438,842 Cost of Intervention Per Person Targeted $18.29 Intervention 2: IEC for Pregnant Women4 Activities Unit Cost Coverage Total Cost Radio Spots $16023 540 $86,400 TV Spots $54223 54 $29,268 News (Channel 6) $44023 54 $23,760 Peer Education74,93 $8 121,465hh $971,720 Total Intervention 121,465 $1,111,148 Cost of Intervention Per Person Targeted $9.15 ccThis is the number of persons that the National Reference Laboratory Director, Lic. Rita Meza, estimates will request testing over the next twelve months (April 10, 2002). ddCosts given by Lic. Rita Meza, Nat ional Reference Laboratory Director (April 10, 2002) and confirmed by Lic. Maria Gilma Rosales, Administrator of the National STI/AIDS Program of Honduras (April 20, 2002) eeExpected number of people testing positive (not counting newborn babies). ffCorresponds to the 9 Regions. ggCorresponds to the 42 Health Districts. hhCorresponds to the 66% of pregnant women who seek or have access to ante-natal care (Secretaría de Salud, 2002). 43 Intervention 3: IEC for Garifuna Population74,93,23 Activities Unit Cost Coverage Total Cost Radio Spots $9.7523 540 $5,265 Peer Education $8 93,000ii,39 $744,000 Theatre 23 $25,525 $25,525 Training of Traditional Medicine Practicioners5 $32,000 -- $32,000 Total Intervention 93,000 $806,790 Cost of Intervention Per Person Targeted $8.68 Intervention 4: IEC for Adolescents Activities Unit Cost Coverage Total Cost Radio Spots $160 540 $86,400 TV Spots $542 54 $29,268 News $440 54 $23,760 Peer Education $8 1,507,687jj $12,061,496 Comics (Fotonovelas) $1.50 45,000kk $67,500 Teacher Training in Sex Education $6893 47,336ll,8 $3,218,848 Curriculum Revision79 $2,000 3 $6,000 Reproduction of Sex Ed. Material for Teachers $5.30mm 47,336 $250,881 Total Intervention 1,507,687 $15,744,153 Cost of Intervention Per Person Targeted $10.44 Intervention 5: IEC Targeted at CSW, MSM and Prisoners Activities Unit Cost Coverage Total Cost Peer Education $893 84,096nn $672,768 Training in Counseling $5,79265,93 9oo $52,128 Sex Education Workshops $5,79265,93 36pp $208,512 Total Intervention 84,096 $933,408 Cost of Intervention Per Person Targeted $11.10 iiCorresponds to the Garifuna population age 10-49. jjPopulation 10-19. kkCorresponds to 25% of adolescents who are in school (12% of 180,922-Banco Central de Honduras 2002) llNumber of teachers in Honduras mm Estimate from Lic. Maria Gilma Rosales, Administrator of National STI/AIDS Program (May 14, 2002) nnSum of populations of CSW, MSM and Prisoners. ooOne for each of the 9 regions. ppA total of 12 workshops for each of the three subgroups. 44 Intervention 6: Workplace Interventions Activities Unit Cost Coverage Total Cost Posters $0.1723 5,000 $850 Brochures $0.1523 200,000 $30,000 Peer Education79 $8.0093 106,5309 $852,240 Training for workers and managers $2,317.00qq 576rr $1,334,592 Video Master $2,439.0023 1 $2,439 Video Reproduction $12.2023 100 $1,220 Audio Messages $25.2023 100 $2,520 Total Intervention 106,530 $2,223,861 Cost of Intervention Per Person Targeted $20.88 Intervention 7: Prevention of Vertical Transmission Activities Unit Cost Coverage Total Cost M&E - Central Level $61.0065 18ss $1,098 M&E - Local Level $3,073.00tt,65 2tt $6,146 Elisa Diagnostic Tests $3.8078,93 124,198hh $471,953 Breast Milk Substitute $270.0079,93 1,863uu $503,059 AZT $81.00ww 1,863 $150,918 (Nevirapine) vv ($0.0061) ($0) Care for HIV+ Mothers $28.0066,74 1,863 $52,169 Total Intervention 1,863 $1,185,342 ($1,034,425) Cost of Intervention Per Person Targeted $636.19 ($555.19) qqCost estimate for a 2.5 day workshop for 25 persons, obtained from Lic. Maria Gilma Rosales, Administrator of National STI/AIDS Program (April 20, 2002). rrMonthly workshops in each of the 48 factories for one year, to develop the peer education capacity. ss2 M&E experts x 9 regions. ttCorresponds to 2 monitoring trips per year x 1.5 days each x 2 specialist x 42 health districts, at a cost of $3,073 per specialist per year. uu1,863 newborns from HIV+ mothers (2,823 HIV+ve women in 2002 and 66% coverage - see steps 1 and 2). vvFor a discussion on the efficacy of different treatment regimens see: Farley et al 2000. ww Personal communication with Dr. Enrique Zelaya, UNAIDS Advisor for Central America, June 12 2002 45 Intervention 8: STI Syndromic Management Activities Unit Cost Coverage Total Cost Training in Syndromic Management $5.792,0065,93 42xx $243.264 Training in Counseling $5.792,0065,93 42 $243.264 Detection of syphilis, gonorrhea, chlamydia and hepatitis $5,6067 26.197yy $146.703 Counseling $6,8093 26.197 $178.140 Treatment of STI $15,0074,93 26.197 $392.955 Condom Distribution $0,3393 26.197 $8.645 Information Campaign $70.024,0023 1 $70.024 Total Intervention 26.197 $1.282.995 Cost of Intervention Per Person Targeted $48,97 Intervention 9: Condom Distribution in Populations with Risk Behavior Activities Unit Cost Coverage Total Cost Condoms to CSW $0.1193,zz 9,212,294aaa $1.013.352 Condoms to MSM $0.11 14,793,044bbb $1.627.235 Condoms to Garifunas $0.11 22,270,560ccc $2.449.762 Condoms to Adolescents $0.11 9,046,122ddd $995.073 Condoms to Prisoners $0.11 1,301,872eee $143.206 Total Intervention 1,734,543fff $6,228,628 Cost of Intervention Per Person Targeted $3.59 xx One 5 day workshop for each of the 42 Health Districts yy Corresponds to the number of cases of STI covered by the Secretaría de Salud in 200173. zz Cost includes condom, transportation, storage and distribution. aaaCorresponds to 14,462 CSW x 637 contacts per year (see Step 3. above) bbbCorresponds to 57,116 MSM x 259 (=62% x 211 + 38% x 259) contacts per year (see Step 3. above) cccCorresponds to 142,760 Garifunas x 156 contacts per year (see Step 3. above) dddCorresponds to 1,507,687 adolescents x 6 contacts per year (see Step 3. above) eeeCorresponds to 12,518 prisoners x 104 contacts per year (see Step 3. above) fffCorresponds to the total number of CSW, MSM, Garifunas, adolescents and prisoners 46 Intervention 10: Condom Social Marketing Activities Unit Cost Coverage Total Cost Condom Social Marketing $1.70 100,000ggg $170,000ggg Total Intervention 100,000 $170,000 Cost of Intervention Per Person Targeted $1.70 Intervention 11: Blood Safety Activities Unit Cost Coverage Total Cost Blood Screening $22.00hhh 35,000hhh $770,000 M&E - Central Leveliii $61.00 18 $1,098 M&E - Local Leveljjj $73.17 32 $2,341 Mass Media Campaign $70,024.0023 1 $70,024 Total Intervention 35,000 $843,463 Cost of Intervention Per Person Targeted $24.10 Intervention 12: Support the Promotion & Defense of Human Rights79 Activities Unit Cost Coverage Total Cost Radio America Spots $3223 156kkk $5,008 Radio HRN Spots $3223 156kkk $4,930 Tevicentro $54223 54 $29,268 Channel 6 News $44023 54 $23,760 La Prensa $11723 30lll $3,510 La Tribuna $11723 30lll $3,510 Master Video $2,43923 1 $2,439 Video Reproduction $12.2023 25 $305 Citizen Networks $73,12579 -- $73,125 Strengthening of HIV+ Networksmmm $400,00079 -- $400,000 Total Intervention 52,980nnn $375,854 Cost of Intervention Per Person Targeted $10.30 ggg Cost and coverage information obtained from Lic. Luis Fernández (PASMO), April 12, 2002. hhh Cost information obtained from Lic. Liz Vinelli (Director, Red Cross Blood Bank, Honduras) May 14, 2002. iii2 persons x 1 visit x 2.5 days x 9 regions, at a cost per person of $61.0065. jjj2 persons x 2 visits x 1.5 days x 16 hospital laboratories, at a cost per person of $73.1765. kkk26 spots/ per month x 6 months. lll30 3x5 slots in B&W over 6 six months. mmm Includes IEC/Rights of HIV+ Peer Education, at a total cost of $200,00079 + $8 * 25,000 persons (peer education). nnn We assume the beneficiary population to be the partners of HIV+ individuals, using a 1:1 proportion. Applying the 2002 adult prevalence of 1.4% to the population >15 years, we obtain 52,980 beneficiaries. 47 48 APPENDIX V. SENSITIVITY ANALYSIS The sensitivity analysis examines the changes in the optimal allocation that result from; (i) increasing the unit costs of interventions; (ii) varying the effectiveness of the interventions; and (iii) excluding secondary infections from the model. In addition, we examine whether the results are robust to using `averted Years of Life Lost' instead of `averted HIV infections' as the objective. (i) Varying unit costs Table V.1 below shows the impact of more or less doubling the unit costs of all interventions Table V.1 Sensitivity of the Optimal Allocation to Varying Unit Costs & STIs Soc. Prison. Pregnant Garifuna MSM, of Safety Intervention Counseling Testing IEC Women IEC IEC Adolescents IEC CSW, Workplace Interventions Prevention MTCT Syndrom. Mgt. Condom Distribution Condom Marketing Blood Rights Base Unit $18.29 $9.15 $8.68 $10.44 $11.10 $20.88 $636.19 $48.97 $3.59 $1.70 $24.10 $10.30 Costs Optimal $0 $0 $0 $0 $500k $0 $0 $0 $250k $250k $0 $0 Allocation at $1m Opt.Allocation $1.25M $0 $750 $0 $500k $0 $0 $1M $500k $500k $0 $500k at $5m Opt.Allocation $2M $0 $750 $0 $500k $0 $1.25M $3M $750k $500k $500k $750k at $10m Unit Costs $36.00 $18.00 $17.00 $21.00 $22.00 $42.00 $1,200.00 $100.00 $8.00 $4.00 $48.00 $20.00 twice as High Optimal $0 $0 $0 $0 $250k $0 $0 $0 $500k $250k $0 $0 Allocation at $1m Opt Allocation $2.25M $0 $0 $0 $750k $0 $0 $0 $1.25M $750k $0 $0 at $5m Opt Allocation $2.25M $0 $1.25M $0 $750k $0 $0 $2.5M $1.25M $1M $0 $1M at $10m Two interesting results emerge from this exercise. First, for low HIV prevention budgets, the results are stable: the same three interventions (condom social marketing, distribution of condoms to groups displaying risky behavior, and IEC targeted at CSW, MSM and prisoners) are chosen, although the amounts allocated to each varies. Second, as the budget available for HIV prevention increases, doubling unit costs has the effect of demarcating the interventions into four groups even more clearly. A first group of four interventions is consistently prioritized (condom social marketing, condom distribution to risk groups, IEC targeted at CSW, MSM and prisoners, and counseling and access to rapid testing). When these interventions start being saturated, then another group of three interventions are included: IEC for the Garifuna population, the promotion and defense of human rights, and the syndromic management of STIs. Above a budget of $16 million, these interventions, in turn, start being saturated, and another group of two interventions become good buys: prevention of MTCT and blood safety. As in the base scenario, a fourth group of three 49 interventions doesn't appear even with very generous prevention budgets: IEC for pregnant women, IEC for adolescents, and workplace interventions. In broad terms, these results are quite robust to changes in unit costs: according to cost-effectiveness criteria, for limited prevention budgets, investing in interventions 1, 5, 9, and 10 prevents the maximum number of new infections; for medium prevention budgets, interventions 3, 8 and 12 should be included; and when the prevention budget is very high, the inclusion of interventions 7 and 11 is warranted. (ii) Varying the effectiveness of interventions During the workshop, different work groups came up with rather different estimates of effectiveness. It was decided as a consensus that the average of these estimates would be used as the base scenario. However, considering that these estimates are based primarily on the experience of the participants with little literature to draw on, we perform a sensitivity analysis using the estimates from two of the 5 workshop working groups, with figures on the low and high side, respectively, instead of the average across all groupsooo. Tables V.2 and V.3 below present the `low' and `high' effectiveness figures per intervention and population groupppp, followed by the impact of using these low and high figures on the optimal allocation at budget levels of US$1 and US$10 million. TableV.2 Low and High Effectiveness Figures of & CSW, MTCT Mgt. Soc. Safety Pregnant Garifuna Adolescents MSM, Intervention Counseling Testing IEC Women IEC IEC IEC Prison. Workplace Interventions Prevention Syndrom. STIs Condom Distribution Condom Marketing Blood Rights Commercial 12% 28% 15% 24% 24%qqq 5% SexWorkers 60% 40% 40% 60% 24% 5% Men who have 25% 25% 15% 24% 24% 5% sex with Men 60% 40% 40% 50% 24% 5% Garifunas 12% 20% 12% 12% 20% 5% 50% 20% 40% 40% 11% 5% Adolescents 10% 25% 5% 9% 10% 2% 5% 70% 5% 10% 9% 2% Prisoners 23% 20% 15% 9% 16% 2% 60% 35% 50% 55% 16% 2% Pregnant 8% 2% 60% 5% 2% Women 8% 90% 90% 13% 2% Workers in their 8% 21% 5% 18% 2% Workplace 20% 30% 20% 17% 2% Transfusion 95% Recipients 100% oooThe two groups that most divert from the average figures were chosen to perform the sensitivity analysis and their entire set of estimates used, to preserve their relative ranking of effectiveness. pppFor the base effectiveness figures, refer to Table 3.6 in Chapter 3. qqqGiven that the `low effectiveness' group didn't estimate effectiveness figures for Condom Social Marketing and Rights, workshop averages are used for these two interventions. The same applies to the `high effectiveness' groups estimates of Counseling and Testing and Syndromig Management of STIs in pregnant women. 50 Rest of the 8% 5% 6% 2% population 5% 10% 6% 2% 51 Table V.3 Sensitivity of the Optimal Allocation to Varying the Effectiveness of Interventions & STIs Soc. Prison. Pregnant Garifuna MSM, of Safety nterventionI Counseling Testing IEC Women IEC IEC Adolescents IEC CSW, Workplace Interventions Prevention MTCT Syndrom. Mgt. Condom Distribution Condom Marketing Blood Rights Optimal Allocation at $1m Base $0 $0 $0 $0 $500k $0 $0 $0 $250k $250k $0 $0 scenario Low effectiv. $0 $0 $0 $0 $500k $0 $0 $0 $250k $250k $0 $0 Estimates High $0 $0 $0 $0 $500k $0 $0 $0 $250k $250k $0 $0 effectiv. Estimates Optimal Allocation at $10m Base $2M $0 $750 $0 $500k $0 $1.25M $3M $750k $500k $500k $750k scenario Low effectiv. $1.25M $0 $750 $0 $500k $0 $1.25M $3M $750k $750k $500k $1.25M Estimates High $2.75M $0 $750 $0 $500k $0 $1.25M $3M $750k $500k $0 $500k effectiv. Estimates As can be seen in the Table V.3, with a budget of $1 million, using different effectiveness figures for each intervention has no impact on the optimal allocation of resources. The only difference relates to the number of infections averted which now is 4,850 with the low effectiveness figures and 6,600 with the high effectiveness figures as compared with the base scenario of 5,100 HIV infections averted. With a budget of $10 million, a variation in the estimated effectiveness of the interventions has a marginal impact on the optimal allocation: compared to the base scenario, in the low effectiveness scenario, $750,000 is removed from counseling and testing and allocated to the promotion of human rights; with the high effectiveness figures, the funds allocated to counseling and testing are now increased by $750.000 by reducing the amounts allocated to blood safety and the promotion of human rights. This shows that the interventions prioritized by the model are robust to changes in effectiveness, with marginal reallocations of money among other interventions at high budget levels. (iii) Omitting Secondary Infections As discussed in Chapter 2, by including secondary infections into the model, we take into account the multiplier effects that result from preventing an infection in an individual who will then no longer infect others. To calculate these secondary infections, we again relied on the expert judgment of workshop participants, especially with respect to the number of sexual contacts and partners for individuals in each group. The resulting figures suggest that an HIV+ MSM will infect almost 3 times as many other individuals in his remaining lifetime as a CSW (17.5 compared with 6.5 secondary infections respectively). Similarly, an HIV+ CSW will infect twice as many people as an HIV+ Garifuna, 6 times as many people as an HIV+ person from the rest of the population, and 60 times as many as an infected adolescent. As a result of incorporating secondary infections, the relative importance of CSW, MSM and Garifunas in preventing the further spread of the epidemic is emphasized. This is reflected in the optimal allocation scenario, where interventions targeting these groups are prioritized. If secondary 52 infections had not been included, the model would have missed out on the importance of these groups in the fight against the epidemic and prioritized interventions directed at other groups earlier. The effect of removing secondary infections from the model is shown in Figure V.1 below presenting the cost-effectiveness curves of interventions without secondary infections. Effectiveness of Interventions (Secondary Infections not Included in the Model) 450 400 1. HIV Counseling and Access to Rapid Testing 2- IEC for Pregnant Women 350 3.- IEC for the Garifunas 300 4. IEC for Adolescents Infections 250 5.IEC for High-Risk Groups (CSW, MSM and Prisoners) 6. Workplace Intervention Averted 200 of 7 Strengthening of the Vertical Transmission Program 150 8.Syndromic Management of STIs Number 100 9. Condom Distribution in High-Risk Groups 10. Condom Social Marketing 50 11. Blood Safety 0 12. Promotion and Defense of Human Rights $0 $250.000$500.000$750.000 $1.000.000$1.250.000$1.500.000$1.750.000$2.000.000$2.250.000$2.500.000$2.750.000$3.000.000$3.250.000$3.500.000$3.750.000$4.000.000$4.250.000$4.500.000$4.750.000$5.000.000 Resources Assigned to Each Intervention Figure V.1 With the exclusion of secondary infections, the prevention of MTCT and blood safety are shown to be relatively more cost-effective than in the base scenario. The optimal allocation is modified accordingly, with investments in MTCT becoming part of the optimum package of interventions at an HIV prevention budget of $1.25 million (compared with $7.5 million in the base case). Similarly investments in blood safety are included in the optimum package of interventions at an HIV prevention budget of $2.25 million (compared with $8.5 million in the base case). Note, however, that the three priority interventions identified in the base scenario, namely condom distribution to high risk groups, condom social marketing, and IEC targeting high risk groups, remain the most cost-effective ones for the first few budget increments when secondary infections are excluded. Also note that even without secondary infections, IEC for pregnant women, IEC for adolescents, and workplace interventions are not included in the optimal allocation for prevention budgets up to at least $10 million. These results are also illustrated by Table V.4 showing the optimal allocation with and without secondary infections at $1 and $5 million. 53 Table V.4 Optimal Allocation at $1 and $5 Million without Secondary Infections & STIs ion Soc. Prison. Pregnant Garifuna MSM, of Safety Intervention Counseling Testing IEC Women IEC IEC Adolescents IEC CSW, Workplace Interventions Prevention MTCT Syndrom. Mgt. Condom Distribut Condom Marketing Blood Rights Optimal Allocation at $1M Base $0 $0 $0 $0 $500k $0 $0 $0 $250k $250k $0 $0 scenario No $0 $0 $0 $0 $250k $0 $0 $0 $500k $250k $0 $0 Secondary Infections Optimal Allocation at $5M Base $1.25M $0 $750 $0 $500k $0 $0 $1M $500k $500k $0 $500k scenario No $750M $0 $750 $0 $500k $0 $1.25M $500k $750k $500k $500k $0 secondary Infections (iv) Years of Life Lost (YLL) The model was run both with `HIV infections averted' and `YLL averted' as the objective function. By using YLL instead of infections averted, we ascribe more weight to the prevention of an HIV infection among newborns and adolescents than to the prevention of an HIV infection in the rest of the population. The analysis shows that using YLL instead of infections as the objective function does not alter the resource allocation decisions at all, for prevention budgets up to about $30m, suggesting that ascribing more weight to infections among newborns and adolescents is not sufficient to overcome the handicap of low cost-effectiveness of interventions that target them. Because using YLL instead of infections does not modify the allocation results, we only present the results in terms of HIV infections averted. 54 APPENDIX VI: LIST OF WORKSHOP PARTICIPANTS The workshop was facilitated by Carla Paredes (Epidemiologist, Consultant), Rosalinda Hernández (Epidemiologist, Consultant), Dariush Akhavan (Public Health Specialist, Consultant), Nicole Schwab (Junior Professional Associate, LCSHD) and Girindre Beeharry (Task Team Leader, LCSHD). NAME INSTITUTION 1 Ada Josefina Rivera Staff, Instituto Hondureño de Seguridad Social (IHSS) 2 Alicia Almendárez Coordinator, Asociación Colectivo Violeta 3 Athos Barahona Representative, UNAIDS 4 Berta E. Álvarez Staff, Honduras National HIV.AIDS Program 5 Carlos Leiva Representative, International Development Cooperation, Catholic Institute for International Relations (CID/CIRRT) 6 Cesar Núñez Director, Proyecto Acción SIDA en Centro América (PASCA) 7 Dionel Guzmán Coordinator, Asociación Nacional de Personas Viviendo con VIH/SIDA (ASONAPCSIDA) 8 Elio Sierra Staff, IHSS 9 Fatima Valle Consultant, PAHO 10 Flor Matute Representative, Médicos Sin Fronteras 11 Glenda Ruiz Consultant, World Bank 12 Helen Saxenian Lead Economist, World Bank 13 Hilton Trochez Staff, IHSS 14 Humberto Cosenza Executive Secretary for External Cooperation, Ministry of Health 15 Irma Benavides Coordinator, Casa Alianza 16 Isabel Noguer Technical Advisor, Spanish National AIDS Plan Secretariat 18 Joany García Coordinator, Comunicación y Vida de San Pedro Sula (COMVIDA/SPS) 19 Jorge Fernández Consultant, GTZ/SIDA Y COMUNICACIÓN 20 Jorge Flores Coordinator, Comunidad Gay Sampedrana 21 Jorge Valle Staff, IHSS 22 José A. Izazola Director, Fundación SIDA para Latinoamérica y Centro América (SIDALAC) 23 Julia Elena Rico Valladares Staff, IHSS 24 Lerida Carías Representative, Organización Gaviota 25 Lesby Castro Coordinator HIV/AIDS Program Región Sanitaria Metropolitana/ Ministry of Health 55 NAME INSTITUTION Metropolitana/ Ministry of Health 26 Luis Medina General Coordinator of Health Prevention Programs, Ministry of Health 27 Luis Roberto Escoto Director, UNICEF Honduras 28 Luisa Maria Pineda Epidemiologist, Hospital Mario Catarino Rivas 29 Manuel Sandoval, Vice-Minister, Ministry of Health 30 Manuel Sierra Epidemiologist, Unidad de Investigación, Universidad Nacional Autónoma de Honduras 31 Marco Antonio López Coordinator, Organización Colectivo Violeta 32 Marco Urquía Head of STI Program, Ministry of Health 33 Mayte Paredes Staff, STI/HIV/AIDS National Program, Ministry of Health 34 Norma Galindo Representative, UNICEF 35 Paloma Cuchi Regional Advisor on Situation Analysis and Program Development, UNAIDS/PAHO 36 PJ Ponj Coordinator, Médicos Sin Fronteras 37 Ramon Jeremías Soto Consultant, PASCA 38 Rebeca Santos Consultant, World Bank 39 Rene López Coordinator, Organización Fraternidad Sampedrana de la Lucha contra el SIDA (FSCS) 40 Rodulio Perdomo Researcher, Centro de Estudios para el Desarrollo y la Participación Social 41 Stanley Terrell Technical Advisor in AIDS and Child Survival, USAID Guatemala 42 Victor Borjas Coordinator of HIV/AIDS Program, Region 3 43 Xiomara Sierra Representative, Caritas Honduras 44 Yanuario García Consultant, Programa de Reforma del Sector Salud/World Bank 45 José Enrique Zelaya Inter-Country Program Adviser for Central America, UNAIDS Gaviota, COMVIDA, Caritas are NGOs that work with HIV/AIDS-infected persons. Violeta is an organization dedicated to Gay rights. Casa Alianza works with street children. 56 BIBLIOGRAPHY 1. Aguilar S., Hernández R., Soto R.J., Nuñez C., Stover J. Projection of HIV/AIDS in Honduras 1980-2010. Tegucigalpa: Secretaría de Salud; 1999. 2. Aguilar S., Hernández R., Soto RJ., Nuñez C., Stover J., Proyecciones de la Epidemia de VIH/SIDA, Honduras 1980-2010. Secretaría de Salud / PASMO / The Futures Group. 2001. 3. Akhavan, D. Economic Analysis of HIV/AIDS prevention in Brazil: Final Report. The World Bank. (unpublished) March 1998. 4. Álvarez, B. Análisis de Estrategias de Comunicación Educativa para la Prevención de VIH/SIDA en Honduras. Tegucigalpa: Secretaría de Salud, Fundación Fomento en Salud; 2002 (en prensa). 5. Asociación el Buen Pastor/ONUSIDA/OPS/Secretaría de Salud. Proyecto Educación Preventiva y Fortalecimiento de la Respuesta Local en las Comunidades Garífunas. Aspectos Sociodemográficos, Conocimientos, Creencias, Actitudes y Prácticas Relacionadas con la Transmisión del ITS/VIH en las Comunidades Garífunas de: Travesía, Bajamar, Triunfo de la Cruz, Santa Fe y Limón. Tegucigalpa: Asociación el Buen Pastor, ONUSIDA, OPS, Secretaría de Salud; Noviembre del 2000. 6. Avila, C. Epidemiology and Public Health. SIDALAC. www.sidalac.org.mx/english/publications/vancouver/avila.htm 7. Baltussen, R. et al. Generalized Cost-Effectiveness Analysis: A Guide. World Health Organization. Global Program on Evidence for Health Policy. Geneva: WHO; 2002. 8. Banco Central de Honduras. Honduras en Cifras 1998-2000. Tegucigalpa: Banco Central de Honduras; 2002. 9. Banco Central de Honduras. Marco Legal de la Actividad Maquiladora 1998- 2000. Tegucigalpa: Banco Central de Honduras; 2000. Anexo Nº 6. 10. Bibus, D. et al. A Model for Distributing HIV Prevention Resources. AIDS Public Policy J. 1994; 9 (4): 197-206. 11. Boerma, T. and Bennett, J. Costs of district AIDS programs. 21: 353-370. In HIV prevention and AIDS care in Africa. Eds. Ng'weshemi J., Boerma T, Bennett J., and Schapink D. Amsterdam: Royal Tropical Institute. 12. Bollinger, L. Cooper-Arnold, K. and Stover, J. Literature Database for Evaluating HIV/AIDS Interventions. The Futures Group International. March 2002. Available at: www.tfgi.com 13. Brookmeyer, R. and Gail, MH. AIDS Epidemiology: A quantitative Approach. Oxford: Oxford Universty Press; 1994. 57 14. Brookmeyer, R. et al. Estimation of current human immunodeficiency virus infection from a cross-sectional survey using early diagnostic test. American Journal of Epidemiology. 1995; Vol. 141. 15. Broomberg, J. et al. Economic analysis at the global level : a resource requirement model for HIV prevention in developing countries. Health Policy. 1996; 38: 45-65. 16. Committee on HIV Prevention Strategies in the US. No time to lose: Getting more from HIV prevention. Washington, D.C.: National Academy Press; 2000. 17. Corby, H. and Schneider, M. HIV Prevention Interventions: What Works and What Doesn't? Available at: http://hivinsite.ucsf.edu/InSite 18. Creese, A. et al. Cost-effectiveness of HIV/AIDS interventions in Africa: a review of the evidence. The Lancet. 2002 (in press). 19. Cruz Roja Hondureña. Informe Banco de Sangre. Tegucigalpa: Cruz Roja Hondureña; Mayo 2002. 20. De Moya, EA. et al. Measuring the costs and benefits of HIV screening systems in Trinidad, Tobago, the Dominican Republic and the Philippines. Amsterdam: VIII Int'l AIDS Conference; July 1992 [abs POD5402]. 21. Departamento ETS/VIH/SIDA, Secretaría de Salud de Honduras. 22. Epidemiology of AIDS and HIV infection in the Caribbean, Bulletin of the Pan American Health Organization. 1989; 23 (1-2): 42-49. 23. Estudio: Índice de Multimedia para Medios de Comunicación de Honduras. Tegucigalpa: 2001. 24. Farley, T. et al. Efficacy of Antiretroviral Regimens for Prevention of Mother to Child Transmission of HIV and Some Programmatic Issues. Background paper prepared for Technical Consultation on New Data on the Prevention of Mother to Child Transmission of HIV and their Policy Implications. Geneva: WHO; October 2000. 25. Forsythe, S. et al. Measuring the costs and benefits of targeted condom distribution programs in Latin America and Caribbean Countries. Amsterdam: VIII Int'l AIDS Conference; July 1992 [abs POD5403]. 26. González, E. Implicaciones económicas de la epidemia del VIH/SIDA y racionalidad económica para la prevención. In: El SIDA en México: oportunidades de la sociedad civil para disminuir su impacto. 27. Hammer, J. Prices versus protocols in public health care. World Bank Economic Review. 1997; 11 (3). 28. Hammer, J. The economics of malaria control. The World Bank Research Observer. 1993; 8 (1):1-22. 58 29. Health in the Americas, 1998 Edition, Volume II. Honduras p. 331-342. 30. Hernández, R. Casisi, D. Soto, RJ. Sierra, M. Pinel, R. Borjas, V. Amador, N. Meza, R. Mendoza, S. Estudio seroepidemiologico de VIH, Sífilis y Hepatitis B en Marinos de Puerto Cortes. Honduras 2001. Tegucigalpa: Secretaría de Salud; 2001. 31. Hernández, R. Soto, R.J. Campa, A. Pinel, R. Borjas, V. Sierra, M. Amador, N. Meza, R. Baum, M. Divergence in Maturity Of the HIV Epidemic in Female Sex Workers in Two Major Areas of Honduras. Abstract for XIV Int'l AIDS Conference. Barcelona: Secretaria de Salud and Florida International University; July 2002. 32. Holtgrave, D. and Kelly, J. Preventing HIV/AIDS among High-Risk Urban Women: The Cost-Effectivenes of a Behavioral Group Intervention. Am J Public Health. 1996; 86 (10): 1442-1445. 33. Holtgrave, D. ed. Handbook of Economic Evaluation of HIV Prevention Programs. New York, NY: Plenum Press; 1998. 34. Holtgrave, D. et al. Effectiveness and Efficiency of HIV prevention services: an overview: Public Health Rep. 1995; 110: 134-146. 35. Holtgrave, D. et al. Quantitative Economic Evaluations of HIV-Related Prevention and Treatment Services: A Review. Risk. 1994; 5 (1):29-47. 36. IAEN. Methods and Models: Cost-Effectiveness Analysis for HIV Prevention. 2002. available at: http://www.iaen.org/costs 37. Institute of Medicine. No Time to Lose: Getting More from HIV Prevention. 2000. Available at: http://books.nap.edu/catalog/9964.html 38. Instituto Nacional de Estadísticas. Informe preliminar de los datos del censo nacional de población del 2001: utilizado en la elaboración de planes estratégicos del 2002 por la Secretaría de Salud de Honduras. Tegucigalpa: INE; 2002. 39. Instituto Nacional de Estadísticas. Proyecciones de Población de Honduras por Sexo y Edad. 1988-2050. Tegucigalpa: INE; 1988. 40. International Programs Center, Population Division, U.S. Census Bureau, HIV/AIDS Surveillance Data Base, June 2000. Available at: http://www.census.gov/ipc/www/ 41. Jack, W. Public spending on health care: how are different criteria related? A second opinion. Health Policy. 2000; 53: 61-67. 42. Janssen, R.S. et al. New testing strategy to detect early HIV-1 infection for use in incidence estimates and for clinical and prevention purposes. JAMA. 1998; 281 (20). 59 43. Jha, P. et al. The evidence base for interventions to prevent HIV infection in low and middle-income countries. CMH Working Paper Series, 2001; Paper No. WG5:2. 44. Kahn, J. The Cost-Effectiveness of HIV Prevention Targeting: How much Bang for the Buck? Am J Public Health. 1996; 86 (12): 1709-1712. 45. Kaplan, E. and Brookmeyer, R. Snapshot estimators of recent HIV incidence rates. Operations Research. 1999; 47 (1): 29-37. 46. Kaplan, E. Economic analysis of needle exchange. AIDS. 1995; 9 (10): 1113- 1119. 47. Kaplan, E. Economic Evaluation and HIV Prevention Community Planning: A Policy Analyst's Perspective. In: Holtgrave, ed. Handbook of Economic Evaluation of HIV Prevention Programs. New York, NY: Plenum Press; 1998:177-193. 48. Marseille, E. Hofmann, P.B. Kahn, J.G. HIV prevention before HAART in sub- Saharan Africa. The Lancet. 2002; 359: 1851-1856. 49. MEASURE. Transition Planning Review. MEASURE. Evaluation / Tulane University. USAID Cooperative Agreement no. HRN-A-00-97-00018-00. 50. Moses, S. et al. Controlling HIV in Africa: effectiveness and cost of an intervention in a high-frequency STD transmitter core group. AIDS. 1991; 5:407-4. 51. Musgrove, P. Cost-Effectiveness as a criterion for public spending on health: a reply to William Jack's `second opinion'. Health Policy. 2000; 54: 229-233. 52. Musgrove, P. Public spending on health care: how are different criteria related? Health Policy. 1999; 47: 207-223. 53. Nagelkerke, N. et al. Modeling the HIV/AIDS epidemics in India and Botswana : The effect of interventions. CMH Working Paper Series. 2001; Paper No. WG5:4. 54. Oladosu, M. and Alden, S. Conocimientos, Actitudes y Prácticas en VIH-SIDA y Uso de Condón en Trabajadoras Sexuales y Hombres que Tienen Sexo con Hombres en Honduras. Tegucigalpa: PASMO; 2001. 55. ONUSIDA/PASCA. Migración y SIDA en Centro América, México y USA. Diagnóstico, Intervención y Evaluación. Guatemala: ONUSIDA/PASCA; 2000. 56. Over, M. and Piot, P. HIV infection and sexually transmitted diseases. In: Jamison, DT et al. ed. Disease Control Priorities in Developing Countries. New York, NY: Oxford University Press; 1993:455-527. 60 57. Pan American Health Organization, Special Program for Health Analysis. Regional Core Health Data Initiative; Technical Health Information System. Washington DC: OPS; 2001. 58. Panel on the Evaluation of AIDS Interventions. Evaluating AIDS Prevention Programs. Washington, D.C.: National Academy Press; 1991. 59. Pinkerton, S. and Abramson, P. The Bernoulli-Process Model of HIV Transmission: Applications and Implications. In: Handbook of Economic Evaluation of HIV Prevention Programs. Ed. Holtgrave, New York, NY: Plenum Press; 1998. 60. Pinkerton, S. et al. Secondary HIV transmission rate in a mixed-gender sample. Intl J STD & AIDS. 2000; 11: 38-44. 61. Potential Industry/Multi-Country HIV/AIDS Programme Collaboration. February 27, 2002 (fact sheet) MMWR, 2002. Available at: http://www.worldbank.org/wbi/B-SPAN/sub_potential_AIDS.htm 62. Poverty Reduction Strategy Paper. Government of Honduras. 2001. 63. Rehle, T. et al. AVERT : a user friendly model to estimate the impact of HIV/sexually transmitted disease prevention interventions on HIV transmissions. AIDS. 1998; 12 (Supplement 2). 64. Salomon, J. and Murray, C. Modeling HIV/AIDS epidemics in sub-Saharan Africa using seroprevalence data from antenatal clinics. Bulletin of the World Health Organization. 2001; 79 (7) : 596-607. 65. Secretaría de Finanzas. Acuerdo Ejecutivo No. 01719 del 27 de Septiembre del 2002. Escala de Viáticos para Movilizaciones Internas de Personal del Estado. Secretaría de Finanzas. Tegucigalpa. Septiembre 2001. 66. Secretaría de Salud, Fundación Fomento en Salud, AID. Impacto Socioeconómico del VIH/SIDA en Honduras 2001. Tegucigalpa: Secretaría de Salud/FFS/AID; octubre 2001. 67. Secretaría de Salud, ONUSIDA. Aceleración del acceso a Atención y Apoyo para Hondureños Viviendo con VIH/SIDA. Tegucigalpa: Secretaría de Salud, ONUSIDA; Octubre del 2001. 68. Secretaría de Salud, OPS. Habilidades de Autoapoyo y Autocontrol Emocional Módulo III. Tegucigalpa: Secretaría de Salud, OPS; 2001. 69. Secretaría de Salud, OPS. Técnicas y Principios de Consejería en ITS/VIH/SIDA. Módulo II. Tegucigalpa: Secretaría de Salud, OPS; 2001. 70. Secretaría de Salud, Organización Panamericana de la Salud. Manual Temático-Metodológico de Formación de Consejeros y Consejeras en ITS/VIH/SIDA. Módulo I: Proceso Metodológico. Tegucigalpa: Secretaría de Salud, OPS; 2001. 61 71. Secretaría de Salud. Boletín de Información Estadística de Atención Ambulatoria en Salud. 2001. Tegucigalpa: Secretaría de Salud; 2002, p. 23. 72. Secretaría de Salud. Boletín de Información Estadística de Atención Hospitalaria en Salud. 2001. Tegucigalpa: Secretaría de Salud; 2002. 73. Secretaría de Salud. Cuadro Resumen de las Infecciones de Transmisión Sexual Atendidas de Enero a Diciembre del 2001. Tegucigalpa: Secretaría de Salud; 2002. 74. Secretaría de Salud. Cuentas Nacionales en VIH/SIDA: Honduras 1999. Tegucigalpa: Secretaría de Salud, SIDALAC; Enero 2001. 75. Secretaría de Salud. Estudio de Prevalencia de VIH, Hepatitis B y Sífilis en 2000 Privados de Libertad, 1988. Tegucigalpa: Secretaría de Salud; 2000. 76. Secretaría de Salud. Estudio Seroepidemiológico de Sífilis, Hepatitis B y VIH en Trabajadores Comerciales del Sexo Ambulantes de las Ciudades de Comayagua, La Ceiba, San Pedro Sula, Tegucigalpa y Puerto Cortés. Tegucigalpa, Honduras 2000. 77. Secretaría de Salud. Informe Anual de Incidencia de VIH/SIDA. Honduras 1985-2001. Tegucigalpa: Secretaría de Salud; 2002. 78. Secretaría de Salud. Laboratorio Central de Referencias. Tegucigalpa: Secretaría de Salud; Mayo 2002. 79. Secretaría de Salud. Propuesta: El Fondo Global para combatir el SIDA, la tuberculosis y la malaria. Tegucigalpa: Secretaría de Salud; Febrero 2002. 80. Secretaría de Salud. Resumen de Resultados Relevantes: Estudios Epidemiológicos en ETS/VIH/SIDA, 1986-1999. Tegucigalpa: Secretaría de Salud; 2002. 81. Secretaría de Salud/ASHONPLAFA/CDC/USAID. Encuesta Nacional de Salud Masculina 1996. Informe Final de un grupo científico. Tegucigalpa: Secretaría de Salud/ASHONPLAFA/CDC/USAID; 1998. 82. Secretaría de Salud/CONPAH/ASDI/OPS/OMS. Salud de los Pueblos Indígenas de Honduras, Interculturalidad y Procesos de Convergencia Nacional. Iniciativa de salud de los Pueblos Indígenas. Tegucigalpa: Secretaría de Salud/CONPAH/ASDI/OPS/OMS; 1999. 83. Secretaría de Seguridad. Dirección General de Servicios Especiales Preventivos. Reporte Diario de la Población Penal. Informe Final. Tegucigalpa: Secretaría de Seguridad; Abril 2002. 84. Sierra, M. Epidemiología de VIH/SIDA en Honduras: Situación actual y perspectivas. Tegucigalpa: Unidad de Vigilancia Epidemiológica, Departamento de ETS/SIDA, Secretaría de Salud; 1998. 62 85. Sierra, M. Paredes, CM. Pinel, R. Branson, B. Meza, R. Mendoza, S. et al. Estudio Seroepidemiológico de VIH, Hepatitis B y Sífilis en Camioneros 1998. Tegucigalpa: Secretaría de Salud; 2000. 86. Sierra, M. Paredes, CM. Pinel, R. Branson, B. Meza, R. Mendoza, S. et al. Estudio Seroepidemiológico de VIH, Hepatitis B y Sífilis en 425 HSH 1998. Tegucigalpa: Secretaría de Salud; 2000. 87. Sierra, M. Paredes, CM. Pinel, R. Branson, B. Meza, R. Mendoza, S. et al. Estudio Seroepidemiológico de VIH, Hepatitis B y Sífilis en Garífunas de Triunfo de la Cruz, Bajamar, Sambo Creek y Corozal 1998. Tegucigalpa: Secretaría de Salud; 2000. 88. Sierra, M. Paredes, CM. Pinel, R. Branson, B. Meza, R. Mendoza, S. et al. Estudio Seroepidemiológico de VIH, Hepatitis B y Sífilis en 2000 MECP 1998. Tegucigalpa: Secretaría de Salud; 2000. 89. Soderlund, N. et al. The costs of HIV prevention strategies in developing countries. Bulletin of the World Health Organization, 1993; 71 (5) : 595-604. 90. Soto, J. et al. Estudio Multicéntrico en TCS y HSH en Ciudades Puertos de Honduras. Tegucigalpa: Secretaría de Salud/PASCA/OPS/USAID/Universidad de Washington; 2002 (documento en prensa). 91. Soto, RJ. Alvarenga, M. Walker, N. Garcia, JM. Zacarías, F. Estimating HIV/AIDS Prevalence in Countries with Low Level and Concentrated Epidemic: The example of Honduras. Tegucigalpa: PASCA; 2001 (en prensa). 92. Stover, J. AIDSProj Spreadsheet Model for using ANC Surveillance Data to Estimate and Project HIV Prevalence and AIDS Cases. USA: The Futures Group International; 1998. 93. Stover, J. Bollinger, L. Resource Needs for HIV/AIDS: Model for Estimating Resource Allocation of Prevention Interventions and Orphan Care. Guatemala: The Futures Group International; 2002. 94. Stover, J. The Effects of Behavior Change on Trends in HIV Incidence in Uganda and Kenya 2000. The Future Groups International; 2002. 95. The Futures Group International. Goals Model for estimating the effects of resource allocation decisions on the achievement of the Goals of the HIV/AIDS Strategic Plan. Available at: www.tfgi.com. 96. UNAIDS. AIDS epidemic update 2001. Available at: www.unaids.org 97. UNAIDS. Cost-Effectiveness Analysis and HIV/AIDS. UNAIDS Technical Update, August 1998. 98. UNAIDS. Costing Guidelines for HIV Prevention Strategies. Geneva: UNAIDS; 2000. 63 99. UNAIDS. Epidemiological Fact Sheets on HIV/AIDS and sexually transmitted infections. Honduras: UNAIDS; 2000 update. 100. Van Vliet, K. et al. The effectiveness of HIV prevention strategies under alternative scenarios : Evaluation with the STDSIM model. In: Confronting AIDS, ch 11 : 207-237. 101. WHO. Briefing Notes for UNGASS on HIV/AIDS. Prevention of mother-to- child transmission of HIV infection: Review of Evidence. WHO. 2001. 102. Willbond et al. The evidence base for interventions in the care and management of AIDS in low and middle-income countries. CMH Working Paper Series, 2001; Paper No. WG5:29. 103. World Bank. Confronting AIDS. Public Priorities in a Global Epidemic. Washington, DC: The World Bank; 1997. 104. World Bank. HIV/AIDS in Latin America. Washington, DC: The World Bank; 2002 (forthcoming). 64 About this series... This series is produced by the Health, Nutrition, and Population Family (HNP) of the World Bank's Human Development Network. The papers in this series aim to provide a vehicle for publishing preliminary and unpolished results on HNP topics to encourage discussion and debate. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations or to members of its Board of Executive Directors or the countries they represent. Citation and the use of material presented in this series should take into account this provisional character. For free copies of papers in this series please contact the individual authors whose name appears on the paper. Enquiries about the series and submissions should be made directly to the Editor in Chief Alexander S. Preker (apreker@worldbank.org) or HNP Advisory Service (healthpop@worldbank.org, tel 202 473-2256, fax 202 522-3234). For more information, see also www.worldbank.org/hnppublications. THE WORLD BANK 1818 H Street, NW Washington, DC USA 20433 Telephone: 202 477 1234 Facsimile: 202 477 6391 Internet: www.worldbank.org E-mail: feedback@worldbank.org