Po] icy RESEARCH WORKING PAPER 2321 Algorithnms for Purchasiing "Demand" for AIDS vaccines varies by level of risk and by AIDS Vaccines national wealth. At-risk individuals in poor countries suffer on both counts. David Bishai Providing funds to develop Maria K. Lin and distribute AIDS vaccines C. W. B. Kiyong,a should be a global concern. The World Bank Development Research Group Poverty and Hunmani Resources and Human Development Network Hea]th, Nutrition, and Population Team April 2000 POLICY RESEARCH WORKING PAPER 2321 Summary findings Bishai, Lin, and Kiyonga delineate two different Under an "equity" model - allocating vaccines to algorithms for the purchase of AIDS vaccines, to show everyone in the world at high risk as if they had the how differences in policy objectives can greatly affect financial resources of Western Europeans - vaccine projections of the number of courses of vaccine that will would be offered to 4.7 billion people. For a Western be needed. European man, reducing the risk of HIV/AIDS would be They consider a hypothetical vaccine costing S10 to a $789 concern; in Africa, the cornparable risk would be produce, and offering 60 percent, 75 percent, and 90 a $48,577 crisis. percent reductions in the risk of HIV for 10 years. For The authors conclude that financing AIDS vaccines each of the world's 10 major geographic divisions, they solely on the fixed budget of a ministry of health rmeans use published estimates of the risk of AIDS, the value of large vulnerable populations wouldn't receive the medical costs averted, and the value of potential vaccine. Allocating the vaccine based on society's ability productivity losses. to pay would still exclude many poor infants who would Under the "health sector" algorithm - in which probably be immunized if they were born in more purchases are made to minimize the impact of AIDS/HIV developed regions. on government health spending - 766 million courses Policymakers concerned about equity in health care of vaccine would be purchased. Under the "societal" must redouble efforts to making the financing of algorithm - in which purchases are made to minimize development and distribution of AIDS vaccines a global, the impact of AIDS!HIV on health spending and GDP - not a regional, concern. more than 3.7 billion courses of vaccine w-ould be purchased. This paper was commissioned by the World Bank AIDS Vaccine Task Force, co-chaired by Poverty and Human Resources, Development Research Group and the Hlealth, Nutrition, and Population Team, Hunman Development Network. Copies of this paper are available free from the World Bank, 1818 H Street NW., Washington, DC 20433. Please contact Patricia Sader, room MC3-556, telephone 202-473-3902, fax 202-522-1153, email address psader o,worldbank.org. P'olicy Research Working Papers are also posted on the Web at www.worldbank.org/research//workingpapers. David Bishai may be contacted at dbishai@jhsph.edu. April 2000. (32 pages) The Policy Research WVorking Paper Series disseminiates the findings of work in progress to encourage the exchange of' ideas ahout develop-nent issues. An objective of the series is to get the findings otet quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the counltries they represent. Produced by the Policy Research Dissemination Center Algorithms for Purchasing AIDS Vaccines David Bishai M.D., M.P.H., Ph.D. Maria K. Lin RPh., MPH. 2 CWB Kiyonga M.D. 3 'Author for Correspondence and Reprint Requests: Johns Hopkins University School of Hygiene and Public Health Departnent of Population and Family Health Sciences 615 N. Wolfe St. Baltimore, M:aryland 21205 Fax: 410 955-0792 Phone 410 955-7807 Email dbishailipedu Grant Support: World Bank and the Hopkins Population Center, NIH Grant 5P30HD06268-25 2Johns Hopkins University School of Hygiene and Public Health Department of International Health 615 N. Wolfe St., Box 231 Baltimore, Maryland 21205 Grant Support: World Bank 3 Office of the Minister of Health Kampala, Uganda Grant Support: The Bill & Melinda Gates Institute for Population and Reproductive Health KEY WORDS: AIDS, HIV, VACCINE, VACCINE STRATEGY, VACCINE POLICY, COSTS, DEVELOPING COUNTRIES, HIGH-RISK POPULAT[ONS Acknowledgments: The authots gtatefully acknowledge the suggestions of seminar participants at the World Bank, The Welch Center, and The International Health Economics Association. Special thanks are due to Martha Ainsworth, Amie Batson, Don Burke, Jose Esparza, Mark Kane, Bob Lawrence, Richard Mahoney, Philip Musgrove, Ken Nelson, Mead Over, and Tomas Philipson. All errors are our own. The authors wish to express their gratitude the World Bank for financial support and to the Bill and Melinda Gates Institute for Population and Reproductive Health which supports Dr. Kiyonga's attendance at the Johns Hopkins School of Public Health. Views expressed in this document do not reflect the opinions of either the World Bank or the Bill and Melinda Gates Foundation. 2 introduction In this paper we attempt to answer the question, "'If an AIDS vaccine arrived in the world on 1 January 2000, what are the ramifications of two different decision algorithms, in terms of who would get the vaccine and how many would be needed?" Anticipating the answer to this question will aid policy makers as they prepare for the efficient and equ.table use of an AIDS vaccine. An effective AIDS vaccine does not yet exist. Foxecasting what will happen when a vaccine is developed is difficdlt. Unforeseen changes in the fundamnental determinants of demand for a vaccine will occur betiween now and the time when a vaccine is released. These determinants not only include the epidemiology of AIDS/HIV, the world's, demographic makeup, and the economic resources available for health, but also the properties of th. vaccines themselves. Developments in vaccine production capacity and technology could change the determinants of supply. Because all of these unknown factors are interrelated, an attempt to simultaneously offer 10-year forecasts would lack credibili.ty. Thus, rather than project what demand will be in the year 2010, we profile who the most likely vaccine tecipients would be in the year 2000 ac:cording to 2 different algorithms. We enumerate likely vaccine recipients under each algorithm as a function of possible vaccine price regimes. METHODS Health Sector versus Societal Perspective To select algorithms for AIDS vaccine allocation that are relevant to what might actually occur, we employ what is known from empirical studies of national decisions to purchase other vaccines. It: has been shown from studies of the uptake cf Hepatitis B and Hemophilus Influenza B vaccines that a courntry's adoption of a vaccine depends on the cost of the vaccine, the incidence of the disease, the GDP/capita, and the cost of the consequences of the disease (Brooks et al. 1999; Miller and Flanders 1999). There are two possible perspectives on the costs of an illness: 1) The health sectorperspective in which the cost of an illness is the let present value of the sum of the costs of 3 the medical care required by the victims; and 2) The socetalperirpective in which the cost of an illness is not only the medical costs but the lost productivity of workers and families affected by the disease (Gold et al. 1996). Neither of these perspectives ordinarily includes the immense intangible costs of pain, suffering, and bereavement that are borne by individuals. The decision that countries will eventually have to make regarding the purchase of the AIDS vaccine could be motivated by either of these two perspectives on the cost of disease. The model that currently appears to be most prevalent for vaccine purchase decisions by health ministries is the health sector perspective. The government health ministry is allocated a fixed budget and asked to allocate purchases for all items (including vaccines) to maximize a population's health under that budget. Health ministries may consider other objectives besides maximizing population health. Phoolcharoen et al (1999) consider a case where the health rministry saw its primary role as allocating vaccine to minimize the contagion externality regardless of the impact of AIDS on its own operating budget. Although such policies may be adopted in the future and may prove to be valid disease control strategies, there are few historical occurrences of vaccine policies selectively targeting those most likely to be contagious among a vast pool of susceptibles. Our algorithms whose principal objective is lirniting the budgetary impact of AIDS will account for the budgetary ramifications of special populations with a high propensity to spread disease. This does lead to different priorities from policies which make the minimization of secondary spread the principal objective. AIDS Vaccine Characteristics We base our assumptions about the characteristics of an AIDS vaccine on known properties of current vaccines. Vaccination efficacy will be sequentially modeled at 60%, 75%, and 90% efficacy with a duration of exactly 10 years. For example a recipient of a vaccine with 75% efficacy experiences an immediate 75% reduction in the risk of contracting AIDS over the next decade. Lacking any basis to predict how risk behavior may change as a result of vaccination we assume it will be constant. With vaccines of low efficacy, vaccine induced increases in risky behavior could 4 have serious repercussions for policy (Anderson and Garnet: 1996). Were we to apply our algorithms to vaccines of lower efficacy they would certainly require adaptation to account for behavioral effects of vaccine. There is an urgent need for erapirical data on the behavioral responses of participants in vaccine trials. Eventual responders cannot be distinguished from non-responders ex ante. We assume adverse effects of the vaccine are temporary and self-limited. We assume transportation and administration costs are no different for the AIDS vaccine than for current vaccines. We assume that the vaccine's protection will last at least 10 years. We do not consider here any applications of the vaccine to reduce viral shedding or viral load in those already infected. We assume that administering the vaccine to a susceptible pregnant woman protects the woman and is safe for the fetus. Infants born to pregnant vaccine recipients are susceptible until they themselves are vaccinated. We profile potential vaccine recipients by first divicing the world's population geographically into the major geographical regions established by the UN. We arbitrarily classify as "less developed countries" (LDCs): North Africa/Middle East, sub-Saharan Africa, South/South-East Asia, Eastern Europe/Central Asia, East Asia/Pacific, Caribbean, and Ladin America. We arbitrarily classify as "more developed countries" (MIDCs): Western Europe, North America, and Australia/New Zealand. There are obiious grounds for many specific exceptions. The one exception we did allow was to considerJapan, Korea, and China separately from the East Asian Region to narrow the otherwise wide variation in HIV economics and risk in this region. Within each region we disaggregate the general population into 4 groups on the basis of age and sex. In 'able 1 we propose a possible further breakdown of the groups most likely to be considered individually for vaccine receipt due to their high risk. Table 1 indicates there is little comprehensive data upon which to base population estimaies for high risk groups such as commercial sex workers (CSWs), injection drug users (IDU s), and men who have sex with men (MSM)¢ in each region of the world. The likelihood that these populations would be targeted by 5 public health campaigns could influence the allocation of an AIDS vaccine. Here we analyze models that incorporate estimates of the numbers of comrnercial sex workers, IDUs, and MSM, but we note that the small size of these groups relative to the general population make estimates of the total need for vaccine relatively robust to their inclusion or exclusion in the model. In this paper we present the details relevant to the general populations and selected high-risk populations. Decision Rule Algorithm We developed the following simple algorithm to profile the regional subgroups as vaccine recipients or non-recipients based on the "marginal benefit" (MB) of allocating AIDS vaccines to that group. Group j is a "vaccine recipient" if MB,>P; otherwise it is a non-vaccine recipient. [Rule 1] Where P is total price of a complete course of vaccine (including administration costs) and MBi is rnarginal benefit of extending vaccination to include group j. MBj is given as follows: MB, (Health Sector Perspective) =E X Ilo x (1+Nj ) x [PV(HC,)] -VC1 [Equation 1] Where: E = vaccine efficacy from 0-100%, assumed=75% for exactly 10 years in baseline analysis. Ijlo decadal incidence of HIV expressed as the discounted likelihood that the average person in group j will seroconvert in the next 10 years. Nj = the number of secondary cases that infected individuals in this group cause over 10 years PV = denotes present value computed at a 3% discount rate HCj = health care costs between the onset of HIV/AIDS and death VCj = vaccine administration cost We call Equation 1 the "health sector perspective" because it might reflect the perspective of a minister of health searching for investments which miniimize the drain that a given disease poses to an arbitrarily fixed health sector budget. We also compute MB1j from a societal perspective alternative that includes lost productivity. MBj (Societal Perspective) =E x ljlo x (1+Ni ) x [PV(HCi + Wj))] -VCi [Equation 2) Where: Wj = lost lifetime wage after the onset of total disability from AIDS. 6 Comrrputing MB1 according to equation 2 for each subgroup in each region produces what we call the "societal perspective", because the presence of lost productivity in the estimates makes MB1 reflect the perspective of a typical minister of finance searching for investments that would maximize the country's GDP. 'Equations 1 and 2 conform loosely tc perspectives that might be taken by hypothetical social planners deciding to allocate AIDS vaccine to subgroups of a region. Since a planner pursuing either of these simplified objectives would end up explicitly rationing vaccine to individuals based on their economic status, neither algorithm is ethically appropriate if an equitable distribution is of value. To indicate the degree of inequity in vaccine distribution inherent in these two simple 2lgorithrrms we recalculate MBj throughout the world based on each group's AIDS/HIV incidence, but with Western European values for medical spending and lost productivity. We call this exercise the "equit perrpective'. To implement the algorithm we gathered second&ry data for each regional subgroup on HIV incidence(Bernard et: al. 1998; UNAIDS and WHO 1998a), medical care costs (Mann, Tarantola and Netter 1992), and GDP/Capita (World Bank 1998). Sensitivity analysis of models incorporating high- risk groups demonstrated that estimates of the number of secondary cases spread by CSWs and IDUs was unnecessary-according to the algorithm their own high risk rendered them vaccine recipients whether cr not they infected any secondary casqs. We assume that for the general population that the number of secondary cases caused is negligible and make the approximation that Nj =0 After using the algorithms to profile each subgroup as a "recipient" or "non-recipient", as a function of the vaccine price, we tabulated the cumulative population of recipients across all subgroups, based on demographic data for cohorts define d by age and sex. (United Nations 1997) Medical Care Costs We assume that in the developed world, a patient would begin to incur medical costs on average about 2 years after seroconverting and incur them for 10 years before succumbing to 7 disease(Curran et al. 1988). In the developed countries significant medical costs are incurred through the use of highly active antiretroviral therapy (HAARI) for asymptomatic seropositive patients. The two-year duration reflects delays in seroconversion and delays in diagnosis of seroconverions. Prior to the development of HAART and improvements in opportunistic disease prophylaxis, the time between seroconversion and death was thought to be 10 years (Bartlett 1998). Responders to HAART may have dramatically prolonged survival, but it is too soon to tell. Indeed the very best answer to the question, "How long between seroconversion and death in the developed world in 1999?" is "Nobody knows." It is likely that HAART will increase the total medical costs per case of AIDS, so we treat this possibility in our sensitivity analysis. We assume that in developing countries, an adult patient would begin to incur medical costs 5 years after seroconverting and incur them for an average of 2 years (Morgan et al. 1997). The natural history of HIV/AIDS in children regarding mean time to AIDS and mean time to death does not appear to deviate markedly from adults(Barnhart et al. 1996). Our model neglects possible regional differences in care seeking behavior and opportunistic infections that might lead adult groups to spend more on treatments than children. Little is known about these differentials. Thus medical costs are assumed to be the same for all ages and sexes. We discount the medical costs and lost wage costs by 3%. Estimates of the actual medical care costs of AIDS in each region are derived as shown in the appendix. Productivity Costs We assume that adults who seroconvert would withdraw from the labor force after six years in developed countries, and after four years in developing countries. In withdrawing from the labor force these patients would cease to add valuable goods and services to the economy. We assume that in every country for adults age 15-49 that the average age for the onset of AIDS is 30 and that retirement age is 65. AIDS thus yields 35 years of productivity loss. We compute the present value of this loss using a discount rate of 3%. As a measure of the value of these lost goods and services, 8 we use the regional GDP/Capita and apply it to both men and women. GDP/capita is a coarse measure of economic well being; however, there are few alternative measures. The global perspective of this paper makes it infeasible to account for the complex adaptation any individual economy may make to the loss of many productive workers (Over et al. 192). Furthermore, the morbidity and mortality of AIDS impose an intangible burden of pain and suffering that the GDP cannot possibly reflect. Vaccine Administration Costs A full consideration of the delivery costs would include the costs of marketing the vaccine to individual target groups (Cutts, Orenstein and Bernier 1992; Kim-Farley and EPI Team 1992). However, we sus pec: that in many parts of the world, the word about the vaccine will spread quickly- -along with inisinformation about its safety and efficacy. Mduch of the marketing costs will depend on what sorts of local rumors are spread about the vaccine and how deeply they are entrenched (Nichter 1995). Lacking a basis to estimate marketing costs we focus on administration costs. In developed cDuntries, we take as an estimate the physician fee ($4.21) for therapeutic injection in the U.S. (CPT 90782) (Flealth Care Financing Administration 1998). In less developed countries, the administration cost of adding an additional vaccine to the EPI program has been estimated at $0.50 (Hall et al. 1993). Results Table 2 presents present value estimates of the lifetime cost of a single case of AIDS in each region of the world for children, teens, adult men, and addlt women. The costs are presented for both the societal and the health sector perspective. These estimated medical costs range from a low of $38 per case of E[IV/AIDS in sub-Saharan Africa to a high of nearly $300,000 per case in North America. I'he large variation in costs is due to both variation in the availability of costly treatments and variation in survival. 9 The societal costs are lower for infants than for children because the eventual lost lifetime productivity is discounted by more years. The societal costs are higher for children than adults because more years of productivity are lost on average for a child or teenager who is infected. The Benefit of AIDS Vaccines Given our estimates of HIV seroconversion risk, vaccine efficacy, and the economnic loss, we can compute MB for each target group using Equation 1 for the health sector perspective, and Equation 2 for the societal perspective. These results are displayed in Tables 3 and 4. Reviewing the health sector perspective, the groups with negative benefits (Table 3) are generally at a very low immediate risk of HIV, and primarily are infants of low risk mothers or those people residing in countries where the lack of treatments for HIV/AIDS makes the medical cost of the disease artificially low. Comparing the actual estimates of vaccination benefits to the benefits developing countries would gain if their medical spending were that of Western Europe indicates that as more medical treatments begin to be provided to HIV/AIDS patients the value of an AIDS vaccine in developing countries will only increase. Table 4 shows the net present benefits of vaccination from the societal perspective. The table shows that if the goal is to maximriize the GDP, there are few population groups anywhere in the world with an HIV risk so low that vaccine administration costs exceed the benefits. A person in a high-risk region like sub-Saharan Africa faces a greater than 10% chance of annihilation from HIV/AIDS. Thus, it is not surprising that the benefit of removing this threat is a large fraction of the remaining lifetime incomes shown in Table 2. The values in Tables 3 and 4 should be interpreted as the returns per individual from an investment in AIDS vaccine, not as a prediction of what individuals would actually spend to obtain these returns. Poor populations have difficulty financing such large investments despite their manifest importance. The parenthesized values in Tables 3 and 4 indicate the benefit that a Western European would derive from the AIDS vaccine if they faced the same risk as corresponding groups in othet 10 areas of the world. This exercise offers a lens with which tc view the economics of the AIDS epidemic from a standardized perspective. The relatively low risk of a Western European man makes reducing that HIN/AIDS risk a $789 nuisance, but if that same man were to confront the HIV/AIDS risk of an African then this threat would be a $48,577 crisis. Vaccine Demand Curves Each different algorithms is designed to label the various subpopulations of the world as "vaccine recipients" or "non-recipients" as the price of vaccine varies from $200 to $1.00 per course. The graphs enumerate the number of vaccine recio,ients as a function of the price of vaccine. Figure 1 does this according to the health sector decision algorithm. Figure 2 shows the "demand curve" that would be generated from the societal perspective algorithm. We use the term "demand curve" to suggest isomorphism to the typical price vs. quaintity graphs from economics. We do not mean to suggest that the quantity demanded on the horizontal axis will result from the aggregate decisions of individual households in a private market. However, country level decision-makers are very likely to make purchase decisions as a function of price. Country level heterogeneity in HIV risk and national resources will lead countries of low financial resources and/or low risk to defer vaccine adoption until the ptice becomes quite low. These curves ignore the constraints that poor populations will face in obtaining sufficient financing to purchase a quantity of vaccine that achieves an optimal allocation according to the algorithm. Estimates of these financing shortfalls are discussed below. The callout boxes in Figures 1 and 2 identify some of the main sub-populations that make the transition from non-recipients to recipients as the price goes down. The callout boxes are far from comprehensive as there are over 91 separate subpopulations being tracked (13 regions x 7 risk groups). The general pattern depicted by the callout boxes is that at high prices the vaccine recipients have very high risk or very high medical and productivity costs of AIDS. As the price drops below $25, (Health Sector Algorithm) and below $50 (Societal Algorithm) very large 11 populations from less developed countries convert to vaccine recipients and account for highly elastic price response at the lower price ranges. Sensitivity Tests To test the robustness of our model to our estimates of incidence we re-estimated the demand curves with the assumption that incidence was 25% lower and 25% higher for each population. These alternative demand curves are displayed as dotted lines in the figures. Testing sensitivity to the cost assumptions by altering assumed medical and productivity costs by ± 25% generates identical results to those plotted as dotted lines in Figures 1 and 2, due to the structure of the equations. Assumptions about vaccine efficacy also affect the number of vaccines distributed by the algorithm. The dotted lines in Figures 3 and 4 plot out how much more or less vaccine is distributed if the vaccine lowers risk by 90% or 60%. Discussion Limitations On a population basis, it is reasonable to conduct policy analysis using the notion of population risk. Caution is warranted however, however, because the risk of HIV/AIDS is not the same for every individual in a group. The possibility of safer lifestyles makes a vaccine an optional but not unique way to prevent HIV/AIDS. If one were to survey individuals on how much they would pay for an AIDS vaccine, responses would likely diverge from our values based on differences between an individual's own perceived HIV risk and the population risk measures our model would impute to him. Some individuals who are able to choose an absolutely risk free lifestyle may choose to never obtain an AIDS vaccine. For now, our model makes the arbitrary assumption that all individuals above age 49 are categorically able to eliminate their HIV risk through safe lifestyles. This achieves simplicity at the expense of disregarding the evidence that risk free lifestyles exist for 12 individuals in younger cohorts and that HIV continues to be transmittecd between individuals well beyond age 49. Policy Analysis for a $10 Vaccine Until an AIDS vaccine is discovered, the marginal cost of producing it is unknown. The claim by industry (DIupuy and Freidel 1990) that there are increasing returns to scale in the production of vaccine suggests that in the first few years o:f vaccine production, the cost could be significantly higher than for subsequent generations of vaccine users. Unless firms are given incentives to rapidly extend capacity, it could be an additional decade er thefirst vzacne before one could expect an AIDS vaccine whose marginal production cost for a course is less than $1.00. Perhaps after this decade, additional factories will have been built throughout the world to reap returns to scale. Suppose the marginal cost just to produce and deliver a single course of the first generation of vaccine is $10.00. The health sector allocation strategy suggests that 766 million people would obtain benefits in excess of $10.00 and hence would wish to buy the vaccine. The model predicts that of these roughly 235 miDlion are estimated to be in LDCs. In contrast, the societal allocation strategy sug,ests thalt 3.7 billion persons (3.3 bilion in LDCs) would obtain benefits in excess of $10.00. The equity model predicts that if everybody in the world had the resources of Western Europe, virtualy everybody in the world under the age of 49 (4.7 bilion out of 4.8 billion) would bear sufficient HIV risk to justify a $10.00 investment in AIDS vaccine. Which Vaccine Allocation Curve? Health ministries in regions like Africa would stil be reluctant to purchase vaccine at a $10 price because $10.00 per person would exhaust al of the average annual health ministry budget. Our model predicts that it would require a $9.00 purchase subsidy to make the most optimal vaccine purchase affordable for health ministries in sub-Saharan Africa. With a $9.00 subsidy, a typical sub- 13 Saharan health ministry could expect to recoup its own $1.00 outlay in the form of reduced medical expenditure for HIV/AIDS. In contrast to health ministries, our model predicts that treasury departments in every region of the world ought to be willing to spend $10.00 per citizen on AIDS vaccination for large portions of the teenage and adult populations. Such investment decisions would pay their way through enhanced survival of working populations. Indeed, there is a growing recognition that confronting AIDS is not just a health issue, but a development issue (World Bank 1997). An irnportant caveat is that both the health sector perspective and the societal perspective would allocate vaccine based on ability to pay just like any other commodity. Rationing by ability toD pay, based on the societal perspective model would still deny AIDS vaccine to roughly 700 million LDC infants who would have obtained earlier protection had they been born in wealthier countries. Unless world leaders agree to policies supporting public and private cooperation to correct inequiry, the AIDS vaccine is likely to be allocated as unequally as any other scarce commodity. Is There Enough AIDS Vaccine Research & Development? The AIDS vaccine allocation that we project is compatible with substanial profitability depending on the marginal cost of the vaccine. Profits in the pharmaceutical industry are difficult to predict because regulators retain tight control over firms' abilities to use the monopoly privileges they have earned through research. Although potential demand is important in projecting profit, potential regulations are equally important. It is unlikely that any pharmaceutical company could afford to behave like an unfettered monopolist in setting profit margins for the AIDS vaccine, but unless the vaccine production costs are exorbitant, a markup of $1.00 per course of vaccine would generate between $1-4 billion in profit. One route to higher profits is the ability to charge wealthier countries more than poorer countries for pharmaceuticals. Regulatory changes discussed by the EEC (Danzon 1997), and statements by US Congressmen (Russell 1997) threaten to erode this option. The recent experience of South Africa indicates that when poorer countries can obtain life-saving AIDS drugs at 14 lower prices, the temaptation to re-import them proves irresistible to politicians. Credible signals that international regulatory institutions were prepared to tolerate and enforce higher profit margins for a pharmaceutical firm. with a patent on the AIDS vaccine would substantiaLLy inflate the estimates of future profitability and stimulate private investment in A]DS vaccine research. Conclusion Th.is paper has offered depictions of two models of AIDS vaccine allocation based on ability to pay and one alternative model of alLocation based on equity. The ability topay modelwas based on predictors of vaccine uptake such as incidence, medical spending, and GDP/capita that were proven empirically to have been associated with regional uptake of Hepatitis B vaccine and Hemophilus Influenza 13 vaccine (Miller and Flanders 1999). The equiy modelwas based on what demand would be if populations with higher the HIV/AIDS risk had the financial resources of Western Europeans. The model reveals that large disparities in the attention given to HIV/AIDS could be based on global resource inequalities. The relatively low risk of a Western European man makes reducing this HIV/AIDS risk a $789 concern, but if that same man were transplanted to sub-Saharan Africa HIV/AIDS would be a $48,577 crisis. Our f6iadinig that financing the AIDS vaccine sclely within the fixed budget of a ministry of health could exclude large and vulnerable populations fzom vaccine receipt offers a strong reminder that HIW/AIDS must be considered a development issue affecting an entire economy. Nevertheless we find that allocation of AIDS vaccine b:tsed on societal ability to pay would still exclude large numbers of poor infants who would be inmunized if they were born in more developed regions. Policymakers concerned with equityr in health need to redouble efforts to make financing the human confrontation with AIDS a global, rather than a regional, issue. 15 References Anderson, R. M., and CG. P. Garnett. 1996. "Low-efficacy HIV vaccines: potential for community-based intervention pIogrammes ". Lancet 348:1010-3. Barnhart, H. X.., M. B. Caldwell, P. Thomas, L. Mascola, I. C'rtiz, H. W. Hsu, J. Schulte, R. Parrott, Y. Maldonado, and R. Byers. 1996. "Natural history of human immunodeficiency virus disease in perinatally infected children: an analysis from the Pediatric Spectrum of Disease Project." Pediatrics 97:710-6. Bartlett, John. 1998. Medical Management of HI VInfection. Baltmore: Port City Press. Bernard, RP, U Zellweger, J Wang, and B Somaini. 1998. "WXorld Status & Prospects of Recent Dynamics of Adult HIV Spread." in 12th World AIDS Conference Geneva. Geneva. Biggar, R. J., and P. S. ]Rosenberg. 1993. "HIV infection/AIDS in the United States during the 1990s." Clin Infect Dis 17 Suppl 1:S219-23. Brooks, A, FT Cutts, J Justice, and G Walt. 1999. "Policy Study of Factors Influencing the Adoption of New and Underutilized Vaccines in Developing Countries.". San Francisco and London: University of Califcrnia at San Francisco London School of Hygiene. Brown, Tim, and Werasit Sittitrai. 1995. "The Impact of HIV in Children in Thailand." . Bangkok: Thai Red Cross Society. China Ministry of Health, and UN Theme Group on HIV/AIDS in China. 1997. "China Responds to AIDS: HIV/AIDS Situation & Needs Assessment Report." : China Ministry of Health. Crofts, N., G. Reid, and P. Deany. 1998. "Injecting drug use and HIV infection in Asia. The Asian Harm Reduction Network ." Aids 12:S69-78. Curran, J. W., H. NV. J affe, A. M. Hardy, W. M. Morgan, R. M. Selik, and T. J. Dondero. 1988. "Epidemiology of H]:V infection and AIDS in the United States." dcience 239:610-6. Cutts, Felicity T., Walter A. Orenstein, and Roger H. Bernier. 1992. "Causes of Low Preschool Immunization Coverage in the United States." Annual Review ofPublic Health 13:385-398. Danzon, P. 1997. Pharmaceutical Price Regulation: National Poli ies vs. Global Interests. Washington, DC: AEI Press. Datta, P., J. E. Embree, J. K. Kreiss, J. 0. Ndinya-Achola, Mq. Braddick, M. Temmerman, N. J. Nagelkerke, G. M:aitha., K K. Holmes, P. Piot, and et al. 1994. 'Mother-to-child transmission of human immunodeficiency virus type 1: report from the Nairobi Study ." J Infect Dis 170:1134-40. Dupuy, J.M., and L. Freidel. 1990. "Lag Between Discovery and Production of New Vaccines for the Developing World." Lancet 336:733-34. Gold, Marthe R, Joanna Siegel, Louise Russell, B, and Milton C. Weinstein. 1996. Cost-effectiveness in Health and Medisine. New York: Oxford University Press. Gray, R. H., IM. J. Wawer, D. Serwadda, N. Sewankambo, C. Li, F. Wabwire-Mangen, L. Paxton, N. Kiwanuka, G. Kigozi,J. Konde-Lule, T. C. Quinn, C. A. Gaydos, and D. McNairn. 1998. "Population-based study of fertility in women with HIV-1 infection in Uganda." Lancet 351:98-103. Guerena-Buwgueno, F., A. S. Benenson, andJ. Sepulveda-Amor. 1991. "HIV-1 prevalence in selected Tijuana sub-populations." Am J Public Health 81:623-5. 17 Hall, A. J., R. L. Roberston, P. E. Crivelli, Y. Lowe, H. Inskip, S. K Snow, and H. Whittle. 1993. "Cost- effectiveness of hepatitis B vaccine in The Gambia." Trans R Soc Trop Med Hyg 87:333-6. Hanenberg, R., and W. Rojanapithayakom. 1998. "Changes in prostitution and the AIDS epidemic in Thailand." AIDS Care 10:69-79. Health Care Financing Administration. 1998. "Public Use Data National Physician Fee Schedule Payment Amount File.". Isomura, S., and M. Mizogami. 1992. "The low rate of HIV infection in Japanese homosexual and bisexual men: an analysis of HIV seroprevalence and behavioural risk factors." Aids 6:501-3. Kim-Farley, R., and EPI Team. 1992. "Global Immunization." Annual Reviews ofPublic Health 13:223-37. MacDonald, M., A. D. Wodak, R. Al, N. Crofts, P. H. Cunningham, K. A. Dolan, M. Kelaher, W. M. Loxley, I. van Beek, and J. M. Kaldor. 1997. "HIV prevalence and risk behaviour in needle exchange attenders: a national study. The Collaboration of Australian Needle Exchanges." MedJAust 166:237- 40. MacKellar, D., L. Valleroy,J. Karon, G. Lemp, and R. Janssen. 1996. "The Young Men's Survey: methods for estimating HIV seroprevalence and risk factors among young men who have sex with men." Public Health Rp 111:138-44. Mann, Jonathan, Daniel Tarantola, and Thomas W. Netter (Eds.). 1992. AIDS in the World. Cambridge, MA: Harvard University. Mastro, T. D., D. Kitayaporn, B. G. Weniger, S. Vanichseni, V. Laosunthorn, T. Uneklabh, C. Uneklabh, K. Choopanya, and K. Limpakarnjanarat. 1994. "Estimating the number of HIV-infected injection drug users in Bangkok: a capture--recapture method ." Am J Public Health 84:1094-9. McCarthy, M. C., F. S. Wignall, J. Sanchez, E. Gotuzzo, J. Alarcon, I. Phillips, D. M. Watts, and K. C Hyams. 1996. "The epidemiology of HIV-1 infection in Peru, 1986-1990." Aids 10:1141-5. Miller, Mark, and D Flanders. 1999. "A model to estimate the probability of Hepatitisb- and Haemopbilus influenzae type b- vaccine uptake into national vaccination programs." Vaccine Forthcoming. Morgan, D., G. H. Maude, S. S. Malamba, M. J. Okongo, H. U. Wagner, D. W. Mulder, and J. A. Whitworth. 1997. "HIV-1 disease progression and AIDS-defining disorders in rural Uganda." Lancet 350:245-50. Nichter, Mark. 1995. "Vaccinations in the Third World: A Consideration of Community Demand." Social Science and Medicine 41:617-632. O'Connor, C. C., G. Berry, R. Rohrsheim, and B. Donovan. 1996. "Sexual health and use of condoms among local and international sex workers in Sydney." Genitourin Med 72:47-51. Over, Mead, Randall P. Ellis,Joyce H. Huber, and Orville Solon. 192. "The Consequences of Adult Ill Health." Pp. 161-199 in The Health ofAdults in the Developing World, edited by Richard G. A. Feachem, Tord Kjellstrom, Christopher J.L. Murray, Mead Over, and Margaret Phillips. Washington, DC: World Bank. Phoolcharoen, Wiput, Viroj Tangcharoensathien, Siriwan Pitayarangsarit, Sukhontha Khongsin, Vijj Kasemsup, and Sripen Tantivess. 1999. "The Potential Demand for an AIDS Vaccine in Thailand." . Nonthaburi, Thailand: Health Systems Research Institute. Reddy, Marlita (Ed.). 1994. StatisticalAbstract of the World. Detroit, MI: Gale Research Inc. Russell, Philip K. 1997. "Economic Obstacles to the Optimal Utilization of an AIDS Vaccine." Journal of the InternationalAssociation of Physicians in AIDS Care September. Sy, F. S., C. L. Chng, S. T. Choi, and F. Y. Wong. 1998. "Epidemiology of HIV and AIDS among Asian and Pacific Islander Americans." AIDS Educ Prev 10:4-18. UNAIDS, and WHO. 1998a. Report on the global HIV/IAIDS epidemic - June 1998. 18 UNAIDS, and WHO. 1998b. UNATDS/ WIO HI V/AIDS epidemic update - December 1998. United Nations. 1 997. Demographic Yearbook 1995. New York: United Nations. World Bank. 1997. Confronting AIDS: Public Priorities in a Global'Epidemic. Washington, DC: Oxford University Press. World Bank. 1998. World Development Report, 1998. New York: Oxford University Press. 19 Table 1. Tar et Groups and Rationale. See Appendix for Population Size Estimates. Groups Analyzed Rationale Data Sources for Estimating Population 1. Infants/Toddlers Can add AIDS vaccine to standard (United Nations 1997) Aged 0-4 battery of Expanded Program on Immunization (EPI) vaccines. Can enfotce receipt with school entry requirements 2. Children/Teens Transition period from low risk to high (United Nations 1997) Aged 5-14 risk Many Accessible through schools 3. Women, High risk group (United Nations 1997) aged 15-49 Often accessible through antenatal care 4. Men, High risk group (United Nations 1997) aged 15-49 5. Not targeted Older Adults Generally lower risk IGenerally lower contagion to others Other Target Grou s Medical Personnel Highly accessible through medical system (Reddy 1994) Highly motivated to obtain vaccine Men Who Have Sex With High risk group (Biggar and Rosenberg 1993; China Men High rate of contagion Ministry of Health and China 1997; Accessible via NGOs in developed Isomura and Mizogami 1992; countries MacKellar et al. 1996; Sy et al. 1998) Military Staff Highly accessible through military (Reddy 1994) medical facilities Military & strategic interest in their immunity IV Drug Users High risk group (China Ministry of Health and China High rate of contagion 1997; Crofts, Reid and Deany 1998; Guerena-Burgueno, Benenson and Sepulveda-Amor 1991; MacDornald et al. 1997; Mastro et al. 1994; McCarthy et al. 1996) Commercial Sex Workers Highest risk group in many areas (Brown and Sittitrai 1995; China High rate of contagion Ministry of Health and China 1997; Hanenberg and Rojanapithayaakom 1998; O'Connor et al. 1996) 20 Table 2: Present value lost per new case of AIDS by age and by region. [1] Region Infants/Toddlers School Children Average for Adults (age 0-4) (age 5-14) (age 15-49) Western Europe, Medical $255,014 $255,014 $255,014 Western Erope. Societal III $573,552 $683,102 $583,263 North Africa & Middle East, Medical $1,335 $1,335 $1,335 I North Africa & Middle East, Societal $103,300 $1 38,368 1 $11 7,55 9 SbD-Saharan Ariica, Medic&-UL $38 $38 $38 Sub-Saharan Africa, Societal $16,140 $21,678 $18,386 South & South-East Asia, Medical $441 $441 $441 South & South-East Asia, Societal $53,829 $72,190 $61,277 Eastern Europe & Central Asia, Medical $5,035 $5,035 $5,035 Eastern Europe & Central Asia, Societal $31,274 $40,298 $34,934 China, Medical $1,896 $1,896 $1,896 China, Societal $10,972 $14,093 $12,238 Japan, Medical $150,591 $150,591 $150,591 Japan, Societal $801,166 $1,024,909 $821,001 Korea, Medical $67,031 $67,031 $67,031 Korea, Societal $218,860 $271,077 $223,489 Ramt Asin & Pacific- Medical $2,820 $2,820 $2,820 East Asia & Pacific, Societal $25,151 $32,831 $28,266 Australia & New Zealand, Medical $122,163 $122,163 $122,163 Australia & New Zealand, Societal $409,977 $508,961 $418,752 North America, Medical $299,894 $299,894 $299,894 North America, Societal $650,616 $771,235 $661,309 Caribbean, Medical $3,322 $3,322 $3,322 Caribbean, Societal $68,946 $91,515 $78,101 Latin America, Medical $1,942 $1,942 $1,942 Latin Amnerica, Societal $42,683 $56,694 $48,366 [11 Discounting at 3%/. Note that this places less weight on the productivity losses of children who acquire AIDS. [2] "Societal" refers to the sum of medical and productivity losses Table 3: Health Sector Perspective Estimates. Net Expected Benefit of Vaccination by Group and By Region. Numbers in parentheses are the values that would be obtained if the ability to pay in that region were equal to that of Western Europe Negative values indicate that the savings to the medical sector from vaccination do not exceed vaccine delivery costs. More Developed Countres Infants/Toddlers Childrenrreens Women Men Female CSW MSM IDUs (age 0-4) (age 5-14) (age 15-49) (age 15-49) Western Europe (WE) $31.61 $105.39 $87.13 $342.87 $4,945.84 $14,784.09 $12,625.79 ($31.61) ($105.39) ($87.13) ($342.87) ($4945.84) ($14784.09) ($12635.79) Australia & NZ (A & NZ) -$2.10 $16.02 $0.51 $72.00 $176.21 $8,682.82 $652.14 ($0.19) ($38.02) ($5.64) ($154.87) ($372.40) ($18129.78) ($1365.91) North America (N. Am) $87.94 $262.94 $209.51 $850.66 $61.61 $10,899.69 $12,471.42 ($74.15) ($222.96) ($177.52) ($722.73) ($51.76) ($9267.86) ($10604.38) Japan -$3.74 $95.23 $0.83 $77.66 $1,071.98 NA NA -($3.77) ($163.82) ($3.97) ($133.07) ($1817.88) NA NA S. Korea -$3.72 $44.46 $1.19 $31.05 $474.94 NA $15,898.62 -($3.14) ($179.16) ($15.54) ($129.13) ($1817.88) NA ($60495.45) Less Developed Countmes North Africa & Middle East (NA & MC) -$0.41 -$0.06 -$0.15 $0.89 $0.85 NA $12.44 ($17.04) ($82.63) ($66.01) ($265.53) ($257.17) NA ($4380.80) Sub-Saharan Africa (SSA) -$0.10 $1.07 $2.61 $2.67 $15.83 NA -$10.50 ($2,656.19) ($10,513.90) ($20,818.01) ($21238.58) ($109414.76) NA NA South & South East Asia (SSEA) -$0.30 $1.23 $1.32 $4.59 $98.17 NA $240.50 ($112.70) ($999.10) ($1,051.71) ($2945.69) ($57110.57) NA ($145285.01) East Europe & Central Asia (EE & CA) -$0.22 $9.74 $8.14 $31.80 $1,187.34 NA $524.60 $13.66 $518.01 $437.36 ($1635.70) ($60163.24) NA ($26595.52) East Asia & Pacific (EA & Pac) -$0.45 $0.11 $0.46 $6.12 $19.65 $382.46 $668.40 (excluding China, Japan, S. Korea) ($4.19) ($54.46) ($86.02) ($598.58) ($51919.56) NA ($52952.65) China -$0.46 $0.78 $0.17 $4.44 $13.04 NA $449.20 ($4.40 $171.90 $90.04 ($63.44) ($5638.66 ($1352.82) ($4194.79) Caribbean $5.12 $27.91 $36.63 $76.03 $675.81 NA $689.27 ($430.80) ($2,180.90) ($2,849.86) ($5874.73) ($51919.56) NA ($52952.65) Latin America (L Am) $0.05 $4.05 $2.86 $14.35 $61.92 $317.41 $375.68 ($71.09) ($597.18) ($440.87) ($1948.86) ($8196.25) ($41744.06) ($49395.82) Table 4: Societal Perspective Estimates. Net Expected Benefit of Vaccination by Group and By Region.Numbers in parentheses are the values that would be obtained if the ability to pay in that region were equal to that of Western Europe. Negative values indicate that the savings to the GDP from vaccination do not exceed vaccine delivery costs. More Developed Countries Infants/roddlers Childrenlreens Women Men Female CSW MSM IDUs (age 0-4) (age 5-14) (age 15-49) (age 15-49) lWestern Europe (WE) $76.34 $289.38 $204.69 $789.62 $11,330.35 $33,819.38 $28,895.81 ($76.34) ($289.38) ($204.69) ($789.62) ($11330.35) ($33819.38) ($28905.81) Australia & New Zealand (A & NZ) $2.87 $80.08 $11.97 $257.01 43 91 $2Q 77320 $2,245.641 ($5.69) ($108.92) ($18.33) ($359.63) ($857.18) ($41471.60) ($3129.52) North America (N Am) $195.71 $682.81 $467.07 $1,880.91 $140.92 $24,040.42 $27,506.31 ($172.03) ($604.30) . ($411.45) ($1658.43) ($123.80) ($21202.75) ($24259.60) Japan -$2.61 $671.32 $22.35 $441.18 $5,862.10 NA NA -($3.21) ($445.89) ($14.51) ($311.06) ($4163.25) NA NA S. Korea -$3.08 $0.00 $13.31 $112.85 $1,592.85 NA $53,016.81 -($1.80) ($488.67) ($40.97) ($300.76) ($4163.25) NA ($138369.76) Less Developed Countries North Africa & Middle East (NA & MC) $6.61 $44.61 $30.15 $122.10 $118.25 NA $2,008.66 ($38.95) ($222.19) ($151.61) ($607.96) ($588.84) NA ($10020.35) Sub-Saharan Africa (SSA) $167.64 $893.29 $1,500.50 $1,530.82 $7,888.27 NA -$10.50 ($5,974.66) ($28,164.31) ($47,615.30) ($48577.24) ($250252.55) NA NA South & South East Asia (SSEA) $23.40 $232.47 $252.33 R7n7 43 13 722SX NA $34.899.80 ($254.11) ($2,677.11) ($2406.10) ($6737.97) ($130623.14) NA ($332294.47) East Europe & Central Asia (EE & $1.24 $81.44 $59.48 $223.64 $8,241.34 NA $3,642.89 CA) ($31.34) ($1,388.44) ($1,000.96) ($3741.80) ($137605.15) NA ($60829.55) East Asia & Pacific (EA & Pac) -$0.04 $6.58 $9.09 $65.90 $201.46 $3,838.83 $6,705.43 (excluding China, Japan, S. Korea) ($10.04) ($146.73) ($197.38) ($1369.71) ($118750.33) NA ($121113.20) China -$0.29 $9.03 $3.84 $31.36 $86.94 NA $2,902.80 ($10.53) ($461.32) ($206.57) ($1518.05) ($12897.31) ($3094.79) ($9594.91) Caribbean $116.11 $782.33 $872.46 $1,798.86 $15,900.65 NA $16,217.05 ( $969.54) ($5842.79) ($6,518.80) ($13437.25) ($118750.33) NA ($121113.20) Latin America (L Am) $11.48 $132.38 $83.21 $369.22 $1,554.12 $7,916.88 $9,368.13 ($160.52) $1,600.51) ($1,008.99) $4458.06) ($18746.99) ($95477.07) $112978.07) Figure 1. Health Sector Perspective Algorithm for AIDS Vaccine Distribution. Callout boxes are not comprehensive but suggest the principle populations receiving vaccine as a function of price. Details and region abbreviations listed in Table 3 Dotted lines show sensitivity of demand estimates to variation in HIV incidence rates. The Global Distribution of AIDS Vaccines Health Sector Perspective Algorithm 150 ~~ -4 7 = ~N. Am. Adults, teens, CSWs, IDUs. WE: Men, CSWs; EE & CA: IDUs & CSWs l 150~~~~~~~~~~ LAm. CSWs, Japan: CSWs A&NZ: IDUs (hin: IDUs SSEA. IDUs| i 125- i O1 00- l ~10 US . WE: Women N.Am Children i; ;;751 I ) __ __I_ _... i , 50 | J _ WE: Children CEE&CA: Men 25 ...! LAmn Men | 0 } I t~~~~~~~~~~~~~~~~~---- \e.t Nn ...... ......-- - - ........... ................ ......... 0 500,000,000 1,000,000,000 1,500,000,000 2,000,000,000 2,500,000,000 3,000,OO),000 Courses of Vaccine Distributed (Global) I - Baseline -.-.-.- 25% Hiaher Incidence 25% Lower incidence 24 Figure 2. Societal Perspective Allocation Algorithm for AIDS Vaccine Distribution Callout boxes are not comprehensive but suggest the principle populations receiving vaccine as a function of price. Details and abbreviations listed in Table 4. Dotted lines show sensitivity of demand estimates to variation in HIV incidence rates. The Global Distribution of AIDS Vaccine Societal Perspective Algorithm I__ EE&CA: IDU, CSW, Men NA&MC: IDU China: IDU $200 ------ _ SSA., WE, N.Am, SSEA: Men, Women, Teens N Am: Children LAm:Men W $150 ~ _~4 o . 9 l SSA: Children $100 LA:Wmn EE&CA, A&N:Tes - WE: Children China: Men L $50 _ . E )EA: Chillen EA&Pac: Teens '*-* $0 -_ - __I 1,000,000,000 2,000,000,000 3,000,000,00C 4,000,000,000 5,000,000,000 Courses of Vaccine Distributed (Global) E--Baseline ------- 25% Higher Incidence -- 25% Lower Incidence 25 Figure 3. Sensitivity Tests for Vaccine Efficacy based on Health Sector Algorithm. Dotted lines show sensitivity of demand estimates to vanation in AIDS Vaccine Efficacy he Global Distribution of AIDS Vaccines Health Sector Perspective With Varying Vaccine Efficacy 150 __ 0 7 5 -_ 0- - o 0 0 500,000,000 1,000,000,000 1,500,000,000 2,000,000,000 2,500,000,000 3,000,000,000 Courses of Vaccine Demanded (Global) Baseline - 90% Vaccine Efficacy 60% Vaccine Efficac 26 Figure 4. Sensitivity Teests for Vaccine Efficacy based on Societal Algorithm. Dotted lines show sensitivity of demand estimates to variation in AIDS Vaccine Efficacy T he Global Distribution of AIDS Vaccines Societal Perspective With Varying Vaccine Efficacy $200 I L__ _ IT $150 - _ *1. 0 g $10 00 - 3,0 4 10000,0 ,000,000,000 3,000,000,000) 4,000,000,000 s,0000o,000oo Courses of Vaccine Dernanded (Global) [-_ Baseline - 90% Vaccine Efficacy 60% Vaccine Efficacy 27 Appendix A: Estimating AIDS Incidence over 10 Years by Population Subgroup We base our incidence rates on UNAIDS estimates for regional populations of the world (Bernard et al. 1998; UNAIDS and WHO 1998b). It is important to point out that the most common identifiable risk factor for HIV/AIDS in most parts of the world is being a heterosexual adult who has had sex with multiple partners or with a partner with multiple partners. In those few areas where the bulk of incident HIV cases have some other identifiable risk factor, attributing UNAIDS incidence estimates to the general population results in an overestimate of risk. Sensitivity analyses show that demand at higher vaccine prices is more sensitive than at prices under $10.00. As improved HIV incidence data are assembled our model will be updated. Using UNAIDS estimates of incidence rates for adults in the major regions of the world, (Bernard et al. 1998; UNAIDS and WHO 1998b) we make the assumption that the gender ratio of prevalent cases is a good approximation of the gender ratio of incident cases. This. assumption is an oversimplification of the complex dynamics of the AIDS epidemic, but it permits us to leverage what we know about prevalence among target groups at the regional level to offer a rough suggestion of annual attack rates in the adult target groups. Sensitivity analyses discussed in the text reveal how important these assumptions are in determining the results of the model. We assume that annual attack rate in the cohort of age 5-14 is 50% of adult incidence because over the subsequent decade, half of the life years of this cohort will be spent in the age group 15-24. We assume that incidence for a cohort age 0-4 is exclusively due to mother-child transmission. We use as the probability that the child of a seropositive mother will experience HIV seroconversion prior to age 10 the estimate of 42.8% and the probability that the infection occurred post-natally of 440/o (Datta et al. 1994). When we estimate decadal incidence in this group as 42.8% x 44%X HIV Prevalence in Women Age 15-49, we are assuming that births are as frequent for HIV+ as HIV- women although this may not be the case (Gray et al. 1998). Our sensitivity tests cover incidence rates in this cohort that are as much as 25% lower to cover this possibility. To convert annual attack rates, Ait, to decadal incidence, Ijlo we use the formula: 10 Ijo = (I- I jt )A,(1 + r)' t=l 28 where r is the discount rate. We adopt the approximation that 1-]:jii =l. Discounting is necessary because seroconversion t years into the ten year period postpones the costs by t additional years. Appendix Table Al shows the source data on attack rates and our estimates of the discounted decadal incidence rates for each group. Appendix Table Al.: Decacal Incidence Rates for General lopulation by Regional Group Region' Adult Incidence InfantrToddlers Children/reens Women Men (Betnard et al. 1998; (age 11-4) (age 5-14) (age 15-49) (age 15-49) UNAIDS and WHO 1998b) HIV Conversions per HIV Conversions per 100,000 Person-DECADES 100,000 Person-yrs WadernEurope 14.56 18.73 57.31 47.75 181.47 NoflhAfrica&MiddleEast 10.35 9.17 43.47 34.77 139.09 Sub-SaharanAfrica 1277.76 . 1,389.04 5,497.43 10,884.91 11,104.81 South& South-EastAsia 127.08 59.19 522.64 550.15 1,540.41 Etern Europe&CentralAsla 48.41 7.40 271.10 228.93 855.48 China 2.56 90.14 47.34 313.23 Japan 0.23 87.85 4.28 347.14 Korea 0.56 20.66 10.33 72.30 EastAu"a8Pacific 21.85 2.45 28.74 45.23 69.71 Ausria & NowZealand 5.16 2.30 22.08 5.15 83.17 North Amdeca 27.87 40.97 118.77 95.02 380.08 Canbean 268.31 225.51 1,140.54 1,490.30 3,071.85 Laln Amerca 72.90 37.43 312.50 230.77 1,019.22 Estimates cf incidence in CSWs, IDUs and MSM are more problematic. Clearly the incidence rates in these populations have been higher than those of the general adult population as reflected in their higher prevalence rates. Assuming that incidence rates and case fatality rates in both the high risk groups and general 29 population have always both been subject to the exact same historic trends permits the following approximation: PR=(Prevalencei /Prevalence) -IR=(Incidencei/Incidence) . Where subscripts ii and j denote two different risk groups. Although it is doubtful that this assumption holds it does offer us a means of estimating incidence in CSWs, IDUs, and MSM which we can then subject to rigorous sensitivity analysis. Our procedure is to multiply the adult population incidence estimates listed in the table above by the ratio of prevalence in the high risk prevalence in adult population. These incidence figures are tabulated below. We did not compute incidence for the high risk groups for regions where there was no reason to believe that the regional governments would be able to effectively target the high risk group. In each of these cases the high risk groups are subsumed within the general population. Fortunately for our estimates the population sizes in these groups are small and do no L contribute heavily to the overall number of doses of vaccine that are allocated by the algorithm. Appendix Table A2: Decadal Incidence Rates for High Risk Population by Regional Group Region CSWs [ MSM IDUs Westem Europe 0.025933523 0.077320318 0.066087963 North Atrica & Middle East 0.001347224 NA 0.022907548 sub-Saharan Africa 0.572075489 NA NA South & South-East Asia 0.298604078 NA 0.759622341 Eastern Europe & Central 0.314564878 NA 0.139056759 Asia China 0.009526755 NA 0.316321247 Japan 0.009526755 NA NA Korea 0.009526755 NA 0.316321247 East Asia & Pacific 0.009526755 0.181102604 0.316321247 Australia & New Zealand 0.001969122 0.094813205 0.007163664 North America 0.029262008 0.048478849 0.055466791 Carribean 0.271462984 NA 0.276864461 Latn America 0.042856538 0.218260594 0.258267671 30 Appendix B: Estimating The Medical Costs of AIDS by Region Because exising estimates of worldwide health care costE for AIDS assume that each AIDS patient consumes tie full complement of medical treatments that are statidard in that region, they provide an admitted overestimate (Mann, Tarantola and Netter 1992). We adjust these estimates downward by replacing the assumption of fiffl complement utilization by a health care access index. For each region of the world, we compute the access index as the ratio of physicans per capita in region j to the physician per capita ratio in Western Europe. Appendix Table B shows our revised AIDS health care costs for each person-year of AIDS. We inflate the Mann et il. (1992) cost estimates to 1997 dollars using the United States Consumer Price Index for medical care. We are aware that new treatments available, primarily in developed countries, have substantially increased the medical costs in these regions since 1992 and plan to update Table B as estimates of these new treatment costs are produjced. Appendix Table B: Ad'justed Medical Costs per Person Year of AIDS by Region Region Health Cost/Person CostVPerson Year of Access Adjusted Access Year of AIDS, AIDS $1997 $US [3] Medical Index [11 1990 Cost/Person Year of AIDS in $ 1997 Western Europe 1.00 $22,391 $31,716 $31,716 North Africa & Middle East 0.23 $2,446 $3,465 $809 Sub-Saharari Africa 0.04 $393 $557 $23 South & South-East Asia 0.11 $1,700 $2,408 $267 Eastern Europe, Ctrl. Asia[4] 1.42 $1,520 $2,153 $3,050 East Asia & Pacific excl 0.71 $1,700 $2,408 $1,708 China, Japan, &';. Korea [2] China [2] 0.48 $1,700 $2,408 $1,148 Japan 0.57 $23,160 $32,805 $18,729 Republic of Korea 0.25 $23,160 $32,805 $8,337 Australia & New Zealand 0.77 $14,015 $19,852 $15,193 North America 0.82 $31,995 $45,320 $37,298 Caribbean 0.66 $2,157 $3,055 $2,013 Latin America 0.42 $1,992 $2,822 $1,177 [1] Health Access Index for each region is Physicians per Capita relative to P ,ysicians per Capita in Westem Europe [2] Applying Mann et al (1992) figure for S. SE Asia [3] Adjusted using US Medical Care 1990-98 lnflator=1.42 (4] Central Asia has a surplus of physicians by Western European standards. Whether this translates into increased medical access is unclear. 31 Appendix C. Number of People in Each Regional Group In Appendix Table C. Displayed are the numbers of people in each region of the world according to the major age and sex groups considered in the model. Data are from UNAIDS, 1998 (UNAIDS and WHO 1998a) and United Nations, 1997 (United Nations 1997). Data on the numbers of CSWs, IDUs, and MSM in the world are quite limited. Based on the sources listed in Table 1, we make the following assumppdons: CSWs make up 0.7% of adult women (age 15-49) in every region of the world, IDUs make up 0.10% of adult men (age 15-49) in every region of the world, and MSM make up 3% of adult men (age 15-49) in every region of the world. Because it is doubtful that MSM and IDUs will be effectively targeted in many regions of the less developed world we simply set the effective population estimate for these groups to zero in the regions noted by "NA" in Appenlix Table A2. Sensitivity analyses indicate that because these high risk groups make up a relatively small proportion of the population facing a risk of HIV/AIDS that the amount of vaccine required is not changed significantly even if we set the size of these populations to zero. However regional policy makers will need to form better estimates of these population's sizes because their high risk makes allocating vaccine to them a priority even at high vaccine prices. Appendix Table C: Numbers of People In Each Risk Category By Region Povulation Cateaories Total Children 0-4 Children 5-14 Women 15-49 Men 15-49 Surn of Population I4Included Categories Western Europe 403,603,000 24,010,860 48,021,720 100,565,500 100,565,500 273,1 63,580 North Africa & Middle East 322,211,000 44,465,118 77,975,062 82,129,500 82,129,500 286,699,180 sub-Saharan Africa 593,027,000 100,221,563 160,710,317 134,219,500 134,219,500 529,370,880 South & South-East Asia 1,859,821,000 228,757,983 420,319,546 477,255,000 477,255,000 1,603,567,529 Eastern Europe & Central Asia 373,424,000 29,500,496 60,121,264 96,692,500 96,692,500 283,006,760 East Asia & Pacific excluding C, J, & K 36,614,000 3,331,874 6,114,538 10,475,000 10,475,000 30,396,412 China 1,243,738,000 104,473,992 222,629,102 352,474,500 352,474,500 1,032,052,094 Japan 125,638,000 7,161,366 13,443,266 30,866,500 30,866,500 82,337,632 Republic of Korea 45,717,000 3,428,775 6,583,248 13,462,500 13,462,500 36,937,023 Australia & New Zealand 21,891,000 1,641,825 3,130,413 5,725,000 5,725,000 16,222,238 North America 301,591,000 22,016,143 43,127,513 78,138,500 78,138,500 221,420,656 Carribean 30,932,000 3,185,996 6,124,536 8,184,000 8,184,000 25,678,532 Latin America 455,247,000 48,711,429 95,146,623 120,741,000 120,741,000 385,340,052 World 5,813,454,000 620,907,420 21,163,447,148 1,510,929,000 1,510,929,000 4,806,212,568 Policy Research Working l'aper Series Contact Title Author Date for paper WPS2302 Why Liberalization Alone Has Not Klaus Deininger March 2000 M. Fernandez Improved Agricultural Productivity Pedro Olinto 33766 in Zambia: The Role of Asset Ovvnership and VWorking Capital Constraints WPS2303 Malaria and Growth F. Desmond McCarthy March 2000 H. Sladovich Holger Wolf 37698 Yi Wu WPS2304 Disinflation and the Supply Side Pierre-Richard Agenor March 2000 T. Loftus Lodovico Pizzati 36317 WPS2305 The Impact of Banking Crises on Maria Soledad Martirez March 2000 A. Yaptemco Money Demand and Price Stability Peria 31823 WPS2306 International Contaclion: Implications Roberto Chang March 2000 E. Mekhova for Policy Giovanni Majnoni 85984 WPS2307 Surveying Surveys and Questioning Francesca Racanatihi March 2000 P. Sintim-Aboagye Questions: Learning from World Bank Scott J. Wallsten 37644 EKperience Lixin Colin Xu WPS2308 How Small ShoLld an Economy's Paul Beckerman March 2000 H. Vargas Fiscal Deficit Be? A Monetary 38546 Programming Approach WPS2309 What Drives Private! Saving around Norman Loayza March 2000 E. Khine the World? Klaus Schmidt-Hebbel 37471 Luis Serven WPS2310 Flow Politics and Irstititions Affect Mitchell A. Orenstein March 2000 M. Leenaerts Fension Reform in 1Three 84264 F'ostcommunist Countries WPS2311 The Buenos Aires Water Concession Lorena Alcazar April 2000 P. Sintim-Aboagye Manuel A. Abdala 38526 Mary M. Shirley WPS2312 Measuring Governance, Corruption, Joel S. Hellman April 2000 D. Bouvet and State Capture: How Firms and Geraint Jones 35818 Bureaucrats Shape the Business Daniel Kaufmann Environment in Transition Economies Mark Schankerman WPS2313 How Interest Rates Changed under Patrick Honohan April 2000 A. Yaptenco Financial Liberalization: A Cross- 31823 Country Revievv Policy Research Working Paper Series Contact Title Author Date for paper WPS2314 Technological Leadership and Beata K. Smarzynska April 2000 L. Tabada Foreign Investors' Choice of 36896 Entry Mode WPS2315 Investment in Natural Gas Pipelines Alejandro Jadresic April 2000 M. Salehi in the Southern Cone of Latin America 37157 WPS2316 Distrubutional Outcomes of a Emanuela Galasso April 2000 P. Sader Decentralized Welfare Program Martin Ravallion 33902 WPS2317 Trade Negotiations in the Presence of Keiko Kubota April 2000 L. Tabada Network Externalities 36896 WPS2318 Regulatory Reform, Competition, Mark A. Dutz April 2000 H. Sladovich and Innovation: A Case Study of the Aydin Hayri 37698 Mexican Road Freign Industry Pablo Ibarra WPS2319 Externalities and Production Gunnar S. Eskeland April 2000 H. Sladovich Efficiency 37698 WPS2320 Does More Intense Competition Lead Mark A. Dutz April 2000 H. Sladovich to Higher Growth? Aydin Hayri 37698