wFs do POLICY RESEARCH WORKING PAPER 2978 The Epidemiological Impact of an HIV Vaccine on the HIV/AIDS Epidemic in Southern India Nico J. D. Nagelkerke Sake J. De Vlas The World Bank Development Research Group k Public Services February 2003 I PoI.cY RESEARCHI WORKING PAPER 2978 Abstract The potential epidemiological impact of preventive HIV Vaccines that convey a high degree of protection in a vaccines on the HIV epidemic in Southern India is share of or all of those immunized and that convey life- examined uSinig a mathenmatical deterministic dynamic long immiiiunity are the most effective in curbing the HIV compartmental model. Various assumptions about the epidemic. Vaccines that convey less than complete degree of protection offered by such a vaccine, the extenit protection may also have substantial public health of immilunological response of those vaccinated, and the impact, but disinhibition can easily undo their effects and duration of protection afforded are explored. Alternative they should be used combined with conventionial targeting strategies for HIV vaccinationi are silulated prevention efforts. Conventional intervenitions that and compared witlh the impact of convenitionial target commercial sex workers and their clients to preventioni interventionis in higlh-risk groups and the increase condomil use can also be highly effective and can general population. The impact of disinhibitioni be implemenited innmediately, before the arrival of (increased risk behavior d(Ic to the presence of a vaccine) vaccines. is also considered. This paper-a product of Public Services, Developmenit Research Group-is part of a research project "The Economics of an HIV/AIDS Vaccine in Developing CouLntries: Potential Impact, Cost-Effectiveness, and Willingness to Pay" sponsored by the European Commnission and the World Bank. The project was launclhed in response to recomimienidationls of the World Bank's AIDS Vaccine Task Force. Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Hedy Sladovich, mail stop MC3-3 11, telephonie 202-473-7698, fax 202-522-1154, email address hslacloviclh @worldbank.org. 'olicy Research Working Papers are also posted on the Web at http:// econ.worldbank.org. The authors may be contacted at nico.niagelkerke@'rivimi.niI or devlas@mgz.fgg.eur.nl. February 2003. (27pages) The l'olicy Research Working Paper Series dissemiiniates the findinigs of iwork in progress to encourage the exchange of ideas about developmient issues Anl obective of tbe series Is to get tbe fidinzgs out quickly, even if iih preseiitations are less than fllvY polished. The pal)ers carry the ,an?ies of tbe autbors and sbotid ibe cited accordingly. Tl)e fidings, initerpretatiois, anid ccolicilisiois expressed in tbis paper are entirely those of the authors. Thev a') do it necessarily represenit the viieiv of the W0orld Bank, its Executive Directors, or the cotrittes they represenit. Produced by the Research Advisory Staff The Epidemiological Impact of an HIV Vaccine on the HIV/AIDS Epidemic in Southern India N. J. D. Nagelkerke Corresponding author: N.J.D.Nagelkerke(lumc.ni S. J. De Vlas devlas®mgz.fgg.eur.nl Address: Department of Public Health, Erasmus MC, University Medical Center Rotterdam, P.O. Box 1738, 3000 DR Rotterdam, Netherlands. The authors thank Prabhat Jha, Martha Ainsworth, and two anonymous reviewers for their invaluable suggestions and comments. This paper is a product of the research project on "The economics of an HIV/AIDS vaccine in developing countries: Potential impact, cost-effectiveness and willingness to pay", sponsored by the European Commission and the Development Research Group of the World Bank (Martha Ainsworth, task manager). The project was launched in response to recommendations of the World Bank's AIDS Vaccine Task Force. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors and do not necessarily represent the views of the World Bank or the European Commission, its Executive Directors, or the countries they represent. Contents 1. Introduction ............................................................ 1 2. Methods .............................................................2 The model ............................................................2 Parameters .............................................................4 Vaccine characteristics .............................................................6 Targeting strategies: ..............................................................6 Conventional interventions for comparison .............................................................7 Disinhibition (increase in risky behavior) .............................................................8 3. Results .............................................................8 Vaccine, condom, and drug requirements ............................................................ 11 4. Discussion ............................................................ 13 References ............................................................ 16 Annex 1. Technical description of the model, including a graphical representation . ........ 19 Tables Table 1. Model parameters (no interventions active) ................................... ..........................5 Table 2. Adult HIV prevalence in 2033 under seven scenarios, with and without disinhibition .9 Table 3. Annual number of vaccinations (millions), by vaccination scenario ................ ............... 13 Figures Figure 1. Structure of the model .............................................................3 Figure 2a. Epidemiological impact of targeting a preventive HIV vaccine to the general population, compared with CSW and STI interventions, South India ............................... 10 Figure 3a. Cumulative number of vaccinations required for targeting the general population . 12 Figure 3b. Cumulative number of vaccinations required for targeting high-risk groups .......................... 12 Acronyms ART Anti-retroviral therapy CSW Commercial sex worker HRG High-risk group IDU Intravenous drug user MSM Men who have sex with men MTCT Mother-to-child transmission NACO National AIDS Control Organisation STI Sexually-transmitted infections 1. Introduction India's current HIV-1 seroprevalence rate of slightly less than 1 percent of adults, or approximately 4 million HIV infected individuals, is bound to increase. In the Southern states of India (Andhra Pradesh, Karnataka, Maharashtra and Tamil Nadu), adult HIV- I prevalence of approximately 2 percent is already observed (National AIDS Control Organisation website). In some districts, it is already over 4 percent. The four states together account for over 75 percent of all Indian HIV infections, even though they have less than 30 percent of the adult population (Government of India 2002). The engine of the Indian epidemic is almost certainly heterosexual transmission from vulnerable groups, chiefly commercial sex workers (CSWs) and their clients. HIV infection in vulnerable groups has grown rapidly in India, where control of HIV and sexually transmitted infections (STI) used to be poor. The conditions for further rapid growth are also in place: paid sex is common, rates of STI are high, male mobility is high, rates of condom use in risky sex are low, and rates of male circumcision-a presumed protective factor-are low. Even an increase to a modest 5 percent infection level in India, the lower end of the African epidemics, in 2025, would represent 25-30 million infected adults and, over the next 25 years, approximately 50 million cumulative HIV- I infections and 40 million cumulative deaths. This is twice the cumulative number of global deaths due to HIV/AIDS over the past two decades. This paper models the potential epidemiological impact of preventive HIV vaccines on the HIV/AIDS epidemic in Southern India, using a mathematical deterministic dynamic compartmental model. In the second section, we describe the basic assumptions and workings of the model and the characteristics of the vaccines and targeting strategies for an HIV vaccination campaign that are modeled. Various assumptions about the degree of protection offered by such a vaccine, the extent of immunological response in those vaccinated, and the duration of protection afforded are explored. In the third section, we present the results, comparing the epidemiological impact of alternative vaccines and targeting strategies to that of conventional prevention strategies aimed at raising condom use among high-risk groups (sex workers and their client) and improving the syndromic treatment of STIs. The impact of disinhibition (increased risk behavior due to the presence of a vaccine) is also 'considered. The final section summarizes the results and points to implications for HIV/AIDS prevention policies. The best long-term hope for control of the HIV epidemic may be a preventive HIV vaccine. Until one is developed, scaling up high impact preventive interventions can reduce the growth of the epidemic. There is an urgent need to develop new candidate vaccines, but also a need to plan the considerable program requirements in introducing new vaccines and in fitting them into other prevention strategies. An AIDS vaccine will greatly help to reduce HIV/AIDS, but it will not be a panacea. Because of the possibility of behavior reversals and an imperfect vaccine (e.g., one that confers only partial protection, or no protection at all in some of those vaccinated), other preventive efforts must be continued, if not expanded. 2. Methods This paper complements a recent paper by Stover and others (2002) in which two HIV/AIDS epidemiological models (the Imperial College Model and IwgAIDS) were used to explore the effects of a potential vaccine in Thailand, Uganda, and Zimbabwe. We extend a mathematical model of HIV- I transmission in Southern India, using methods previously developed for Working Group Five of the Commission on Macroeconomics and Health (Nagelkerke and others 2002). We use the ModelMaker program, version 3.0.3 (AP Benson 1993-97), for implementing our model. The model The model is a dynamic, deterministic compartmental model. The main features of the model that are pertinent to the vaccine exercise are described in Figure 1. The boxes represent compartments, or states, that individuals can be in and the arrows show the flow of individuals between compartments. Each compartment shown has been implemented in duplicate, for men and women separately. Individuals move between gender-specific compartments. For example, for women there are two groupings, CSW and low-risk women. Each of these two groups is split into several groups based on their infection and immunization status: * those uninfected and immunized, * those uninfected and not immunized, and * those who are infected (in three sub-groups, early stage, late state, and AIDS). There is no age structure in the model as used for the purpose of projecting the impact of an HIV vaccine, with the exception that the model only concerns the sexually active adult population.' It assumes that the epidemic is primarily heterosexual, driven by commercial sex, and that unsafe sex work is widespread and contributes substantially to the spread of the infection. CSW and their clients were assigned separate compartments to reflect this assumption. In India, approximately 80 percent of STI are first-generation infections derived from sex work, so this seems reasonable (Rodrigues and others 1995). Early female HIV infections occurred predominantly in CSW; infection in monogamous women is probably linked to their husbands having visited a CSW (Gangakhedkar and others 1997, Pais 1996). 1 The model also incorporates mother-to-child transmission (MTCT, "vertical transmission") and an intervention ("nevirapine") to reduce this. However, as we are mainly interested in adult prevalence, which is not affected by MTCT, this is not considered here. A formal description of an earlier and more complete version of the model is available on the Internet (Nagelkerke and others 2001). 2 Figure 1. Structure of the model Low-risk UnInfected 1 High-risk group not Immunized<, not Immunked group Uninfected . nneceda i mmunized I E mnized Infected Infected Infected Infected late stage Flate stage AIDS FDeath Individuals are "born into" the low-risk uninfected category of their gender and may move to and from high-risk groups (female CSWs and their male clients). In addition to dying from AIDS, they may die from other causes or "age" out of the sexually active age group. High-risk groups may infect low-risk groups (e.g., their current steady partners), including newly acquired low-risk (steady partners), for example, when they get married. The HIV disease process is broken down into three stages: early, late, and AIDS.2 The model includes neither transmission between men who have sex with men (MSM) nor transmission due to intravenous drug use (IDU). Both occur in India, but it is believed that they account for a minority of all transmissions. The role of MSM and IDU transmission and interaction with the heterosexual epidemic may be small, although IDU is an important mode of HIV transmission in the Northeastern state of Manipur (the main exception). Appendix 1 provides a formal technical description of the model, including a graphical representation. 2 The three stages of AIDS in the model facilitate modeling the impact of anti-retroviral therapy (ART) targeted at patients in different stages of HIV disease progression. ART is not among the interventions considered in this study, however. 3 Parameters In setting the parameters of the model, demographic data from South India were used where available; otherwise data from whole of India were used. A recent study conservatively estimates that there are at least 2 million CSW in India, each having on average-very conservatively-two clients per day, and that their clients number approximately 30 million (i.e., slightly over 10 percent of the adult male population) (Venkataramana and Sarada 2001). This suggests that each client may have some 50 CSW contacts annually. Based on the results of the recent nationwide behavioral survey commissioned by the National AIDS Control Organization (NACO), on others' estimates that over 40 percent of all CSWs work in the four Southern states, and on the higher prevalence of HIV in Southern India, we assumed that approximately 20 percent of all adult males are clients of CSW at any one time (NACO 2002). Estimates of the rates of becoming a high-risk individual and transitioning back to the low-risk category are not available. From the fact that most studies find the mean age (and age-range) of sex workers to be low (often around 23 years), it follows that the rate of leaving the profession must be high. We chose 20 percent annually for this rate. We took half this value for the transition rate of from client to non-client. Rates of becoming a client can then be derived from the number of clients and the rate of becoming low-risk. For women we also introduced a demand factor, i.e., low-risk women's rate of becoming a CSW was modeled as a function of the demand for sex work. The average duration from infection to AIDS is the sum of the average duration of the early and late stages; we assumed four years in each of the two stages, resulting in an average duration of HIV infection of eight years. Modeling by the National Intelligence Council (2002) has suggested that the HIV/AIDS epidemic in India, would result in up to 25 million people living with HIV/AIDS by 2010, i.e., an adult HIV prevalence of approximately 5 percent. However, as Southern India appears to be the worst-hit part of the subcontinent, our model corresponds to a scenario in which prevalence grows from its current level of approximately 2 percent of the sexually active population to an equilibrium prevalence of almost 8 percent. Table I gives the (baseline) parameters used for the model: parameters were chosen to reflect conditions in the four Southern Indian states of Andhra Pradesh, Karnataka, Maharashtra, and Tamil Nadu combined. With this choice of parameters, a 2 percent HIV adult prevalence in 2001 was obtained. This prevalence would gradually increase to 7.5 percent in 2033 in the absence of any interventions. While not comparable to the prevalence encountered in many parts of Sub-Saharan Africa, this size of the epidemic would have devastating effects on Indian society and its socio- economic development. 4 Table 1. Model parameters (no interventions active) Parameter Description (where relevant) Value aidsrate Annual rate of developing AIDS from late stage HIV 0.25 Brate Birth rate 0.085 cr_before Contact rate between clients and CSW 50 Cust Rate at which low-risk men become CSW-clients 0.04 Femgr Rate of growth adult female population 0.021 Fmrisk Female to male HIV transmission risk during CSW-client 0.0036 contact HlVprog Rate of HIV progression from early to late stage 0.25 init_frac_cli Initial (1998) fraction of adult men who are CSW clients 0.2 init_frac_csw Initial (1998) fraction of adult women who are CSW 0.013 init_inf cli Initial (1998) clients HIV prevalence 0.07 init_inf_csw Initial (1998) CSW HIV prevalence 0.25 init_inf fem Initial (1998) low-risk adult female HIV prevalence 0.005 init_inf_men Initial (1998) low risk adult males HIV prevalence 0.005 init_pop_female Initial female adult population 70,000,000 init_pop_male Initial male adult population 70,000,000 Leak Transmission parameter high-risk to low-risk 0.075 loss_immun Rate (annual) of loss of vaccine induced immunity 0 Malegr Rate of growth of adult male population 0.022 Marrate Marriage rate 0.058 Mfrisk Male to female HIV transmission risk during CSW-client 0.0052 contact Muaids Death rate AIDS patients I Muhiv_kids Death rate HIV infected children 0.25 Muneg Death rate HIV- adults 0.026 Mupos Death rates (non-HIV) HIV+ adults 0.028 Prof Parameter controlling proclivity of low risk women to become 0.004 Csw stabfactor Parameter on transmission between newly wed discordant 100 couples Startyr_condom_CSW Startyear (+1998)focussed intervention among CSW (use 35 condoms) Startyrstd Startyear (+1998) Mwanza style STI control 35 Startyr_vaccin Startyear vaccine intervention 35 STD_effect Effect of STI control on transmission (I=no effect, 0.7 0=transmission interrupted) Uncust Rate CSW clients become low-risk men 0.1 Unprof Rate CSW become low-risk women 0.2 Unprot_after Level of CSW-client non-use of condoms after focused 0.25 intervention Unprot_before Same. Before intervention 0.5 Vaceff Level of vaccine protection, 0= 100%, 1 = 0% 0 Vactake Proportion of vaccinated who respond 0.5 Vtrate Mother-to-child (i.e. vertical) HIV transmission rate 0.33 wI Relative infectiousness early stage HIV+ I w2 Relative infectiousness late stage HIV+ I Note: The model has additional parameters for estimating the impact of anti-retroviral therapy; these are not presented, as they are not relevant to modeling the impact of an AIDS vaccine. S Vaccine characteristics We compare the epidemiological impact over the period 1998-2033 of four different vaccines, defined by the levels of two parameters: * Level ofprotection. The reduction in HIV susceptibility in those giving an effective immunological response to the vaccine. We consider two levels-50 percent and 100 percent. * Level of immune response. The percentage of those vaccinated who have an immunological response to the vaccine. We consider two levels-50 percent and 95 percent-of those vaccinated. The vaccine has no protective effect on the remaining 50 percent (or 5 percent) receiving the vaccine who have no immune response to it. We use the term "vaccine efficacy" to mean the product of the level of protection and level of immune response-that is, the average protection afforded to an average vaccinated person. Thus, a vaccine of 50 percent average efficacy in the population could be defined as either: (a) a vaccine conveying 100 percent protection to half of those vaccinated (50 percent immune response); or (b) 50 percent protection to everyone who is vaccinated (100 percent immune response) or (c) some other combination of protection and immune response that yields an average efficacy of 50 3 percent. We assume that vaccines would become available in 2008, which is the earliest time vaccines would become available for general use if current development efforts prove successful, and that vaccines would provide (for those conferred any protection) immunity for at least 25 years (or vaccine recipients would be revaccinated sufficiently often to simulate this duration of immunity). In addition, we explored the effects of waning of vaccine efficacy by showing the impact of the "best" vaccine considered here (100 percent protection conferred to 95 percent of those vaccinated), if protection lasted on average 3 years. Following loss of vaccine-induced immunity, vaccine recipients move back to the compartments of susceptibles, from where they may be recruited for vaccination again. Targeting strategies We examine the impact of these vaccines of differing levels of efficacy using two different targeting scenarios with different assumptions on coverage: 3 Stover and others (2002) refer to vaccines that convey 100 percent protection to a share of those vaccinated as "take" vaccines (example (a) in the text) and those that convey partial protection to all who are vaccinated "degree" vaccines (example (b) in the text). They show that the epidemiological impact of a vaccine with a given average efficacy in a population is highly dependent on whether efficacy is achieved through "take" or "degree." This distinction between "take" and "degree" type effectiveness is not relevant for a vaccine with 100 percent efficacy (that is, complete protection of all who are vaccinated), as they are equivalent. Stover and others do not model the impact of vaccines with both partial protection and partial immune response, such as example (c) in the text, though their impact presumably would be somewhere between the impact of "take" and "degree" type vaccines for a given level of effectiveness. 6 High risk group targeting (HRG). Both CSWs and their clients are targeted. Annually, 75 percent of those eligible (i.e. belonging to the target population and not yet immunized) would be vaccinated. This would result in an average coverage rate (proportion of the groups vaccinated) of approximately 90 percent. Population targeting (POP). Every sexually active adult is equally targeted regardless of behavioral risk group. At the time the vaccine becomes available, a 2-year vaccination campaign is launched that succeeds in reaching 25 percent of the target population of susceptibles (HIV negative, not immune) annually. This is followed by an indefinite period during which 5 percent of the target population is vaccinated annually. This leads to an equilibrium situation in which approximately 50 percent of the sexually active population has been vaccinated. The initial 2-year vaccination campaign is included to reach that 50 percent coverage level quickly. Conventional interventions for comparison As an HIV vaccine has not yet been developed and as a point of comparison, we also modeled the epidemiological impact of two conventional HIV prevention interventions: A focused CSW intervention. The objective of this intervention is to increase condom use in CSW-client contacts. Focused interventions have proven to be very effective in increasing condom use in this context. This reduces HIV transmission among sex workers and clients, but also in the general population, because of the ''core" role of these high-risk groups in spreading infection to the rest of the population (Hethcote and York 1984, Jha and others 2001, World Bank 1997). Many peer- mediated CSW intervention programs in India and Africa have shown increases in condom use of 80 percent or more among those reached (Bhave and others 1995, Jana and others 1998, Jana and Singh 1995, Moses and others 1991). We conservatively assumed that the intervention reduces the percentage of unprotected contacts from 50 to 25 percent. We were also conservative in not assuming an additional reduction in the risk of transrnission per CSW-client contact through a reduction in STI prevalence, although this may well be the case. Syndromic treatment of STIs. Epidemiological studies support the hypothesis that STIs are associated with increased HIV susceptibility and infectiousness. However, confounding makes it difficult to reliably estimate these cofactor effects from observational studies (Korenromp and others 2001). Three experimental studies have been carried out to date, one in Tanzania and two in Uganda. STI management, through improved treatment of patients with symptomatic STI infections, has proven to be effective in a controlled community trial in Mwanza, Tanzania, with an approximate 40 percent reduction in HIV transmission (Grosskurth and others 1995). STI management was based on a syndromic approach to symptomatic cases. It was applied to a rural area in a non-targeted way. People with asymptomatic infections were not treated. However, the failure of a similar intervention in a trial in Masaka, Uganda, to replicate this success (reported at the AIDS conference in Barcelona), and the lack of success in 7 Rakai, Uganda, to reduce HIV transmission through a program that offered mass treatment for STIs to the population has sparked debate about the efficacy of such interventions in slowing HIV transmission (Gray and others 1999, Hitchcock and Fransen 1999, Hudson 2001, Korenromp and others 2000, Kvale 1999, Matthys and Boelaert 1999, Nicoll and others 1999). We nevertheless assumed for the purposes of this modeling exercise that HIV transmission would decrease by 30 percent across the board (males, females, high-risk, low-risk). Arguably, this is a strong simplification of reality and requires averaging over partnerships with and without STI. In reality, the average effect of the intervention may also vary among risk categories (e.g., CSW and other women), depending on factors that are largely unknown, such as the uptake of the intervention. No effects of the intervention on sexual behavior were assumed. Note that the way this 30 percent reduction is achieved is irrelevant for our predictions. Increasing condom use in the general population could be equally effective. As both interventions use existing technologies, they were assumed to start in 2003. Disinhibition (increase in risky behavior) We explore whether disinhibition-that is, an increase in risky behavior associated with the availability of an HIV vaccine-could nullify or reverse the impact of the vaccine. Disinhibition has been observed in high-risk gay men in response to the availability of anti-retroviral therapy (Katz and others 2002, Ostrow and others 2002, Stolte and others 2001, Stolte and Coutinho 2002). All models were run in the presence of a strong disinhibition effect, that is, assuming that condom use between CSWs and clients dropped from 50 percent (assumed to have increased already from very low levels in response to other prevention efforts) prior to the availability of vaccines to nil (0 percent). For comparison (adult) HIV prevalence in 2033 (the last year of the simulation) was used. 3. Results Table 2 shows the effect of the different interventions on long run (2033) adult HIV prevalence. Conventional prevention programs begin in 2003 and vaccine interventions in 2008. All vaccine scenarios show a decline in HIV prevalence. Generally, a high degree of protection appears to be more important than a high "take" rate. Disinhibition (i.e. condoms are no longer used during CSW-clients) has the potential of undoing much of the vaccine benefits, and may even aggravate the epidemic. However, our assumed extreme disinhibition effect of total abandonment of condom use, may be unlikely to happen, as condoms also provide protection against conventional sexually transmitted infections, an advantage that CSWs may be keen to keep. Targeting high- risk groups tends to be more effective than targeting the general population (at least at the levels considered). For the best vaccine considered (scenario 4), the impact appears to be similar, perhaps due to "over-vaccination" of high-risk groups (both clients and CSWs have a high vaccine coverage). 8 Table 2. Adult HIV prevalence in 2033 under seven scenarios, with and without disinhibition' Scenario Adult HIV prevalence in 2033 (percent) 0 Baseline 7.5 Conventional Protecton interventions (percent) I CSW condom 75b 1.4 intervention 2 STI syndromic treatment 30c 2.4 Targeted to high-risk groups Targeted to general Vaccine scenarios Efficacy (HRG) population (POP) (percent) No Disinhibitlon No Disnhibition disinhibition I disinhibitlon 3 100% protection, 50% response, 25 years 50 1.0 3.3 1.9 4.8 duration 4 100% protection, 95% response, 25years 95 0.6 1.4 0.6 2.1 duratfon 5 50% protection, 95% response, 25 years 47.5 2.9 9.5 3.2 8.5 duraton 6 50% protection, 50% response, 25 years 25 3.7 10.3 4.6 10.0 duration 7 100% protection, 95% response, 3 years 95 1.5 5.7 5.0 10.5 duraton a. Decline in condom use in commercial sex from 50% to zero. b. Increase in protection from 50% to 75% among CSW-client contacts. c. Reduction in transmission probability. Figures 2a and 2b show the impact of the different interventions (without disinhibition) on HIV prevalence over the period 2003-2033. Note that a highly effective vaccine (scenario 4, 100 percent protection for 95 percent of those vaccinated) appears to be by far the most effective method to bring down HIV prevalence quickly, much faster than conventional prevention programs. The latter, however, have the advantage that their implementation could start immediately. 9 Figure 2a. Epidemiological impact of targeting a preventive HIV vaccine to the general population, compared with CSW and STI interventions, South India - Baseline S 7 tY'. *, .ii ~~~i . -IaSwt CSW _6;: | r i , -<,, , ,grx v S ; X- i . . v2 STI 5 ~~~~ %,. o~~~~-. 3 Pop E 4 . --4 Pop >3- '-a * 5 Pop 2 -.- 6 Pop 7 Pop 0 1998 2005 2012 2019 2026 2033 Year Note: Vaccines target approximately 50 percent of the general population; scenarios defined in Table 2. Figure 2b. Epidemiological impact of targeting a preventive HIV vaccine to the high-risk population, compared with CSW and STI interventions, South India 8 . .. ---*t -.-- Baseline 7 - 1 CSW *6 r i- -2 STI 5 ~~~~~~~~.3 HRG 4- ....,.;. <- 4 HRG >I 3- e- 5 HRG a. ~ ~ ~ ~ ~ Ya 6 -HRG 7 HRG 0 1998 2005 2012 2019 2026 2033 Year Note: Vaccines target approximately 90 percent of CSWs and clients; scenarios defined in Table 2. 10 Vaccine, condom, and drug requirements Figures 3a and 3b show the cumulative number of vaccinations needed under different strategies, assuming that those who are already HIV-positive are not vaccinated. Targeting high-risk groups typically requires substantially fewer vaccine doses than targeting the general population, at a similar or higher impact. Thus, unless the costs of targeting high risk groups are extremely high, targeting high-risk groups should be typically several times more cost-effective than targeting the general population. Also, to achieve reasonably high vaccination coverage for the general population, we assumed that a very intensive two-year vaccination campaign would "kick-start" vaccination coverage. This would put enormous strains on production facilities and other infrastructure and may be difficult to implement. The annual number of vaccinations implied by each of the targeted and general population strategies are in Table 3-including both the "kick-start" phase and the "maintenance" phase. Costs follow from the costs to set up and maintain the infrastructure and of course the cost per vaccination, at present unknown. Even targeting only high-risk populations may require hundreds of millions of vaccine doses over a 25-year period. However, this is less than the number of childhood vaccinations given over that period and, unless the vaccine is very expensive (over US$ 100, say), this would definitely seem affordable. The number of condoms needed for a focused intervention for high-risk groups is easy to calculate. We assumed that approximately 20 percent of sexually active adult men would be clients, and that they have-on average-50 CSW contacts annually. We (optimistically, but based on NACO's sexual behavior surveys) assumed that approximately half of these contacts are already protected by condoms obtained from other sources (NACO 2002). Thus, to increase condom use to 75 percent, 13 additional condoms per client, or approximately 3 per adult male, would be required annually. The amount of drugs and their costs for an STI control program are hard to predict. In addition to current prevalence and incidence of STIs, knock-on effects in terms of a reduction in transmission and a consequent decline in incidence may yield long-term savings in costs. For Mwanza, the annual per-capita costs of running am STI program were estimated to be $0.39 (Gilson and others 1997). Figure 3a. Cumulative number of vaccinations required for targeting the general population, South India 250 - Baseline °0 2- 0 . . . 7 . 4 .iay:.. m) +03 Pop c E 150 ~~~~~~-e,i, -4 lx5Pop 2 100 -6Pop E_o f 5--*-fi*i- ^- 7 Pop , 0 5; * * 1998 2005 2012 2019 2026 2033 Year Figure 3b. Cumulative number of vaccinations required for targeting high-risk groups, South India 250 - , . .- -L.- Baseline % 200 ,:i : - r; -. -0 -3HRG 7 HRG 50 - 0 1998 2005 2012 2019 2026 2033 Year 12 Table 3. Annual number of vaccinations (millions), by vaccination scenario Year 3Pop 4Pop 5Pop 6Pop 7Pop 3HRG 4HRG SHRG 6HRG 7HRG 2008 0 0 0 0 0 0 0 0 0 0 2009 36.75584 34.83045 34.83009 36.75562 35.24657 9.458652 8.191432 8.190107 9.4577 8.463783 2010 32.95754 28.07567 28.07135 32.95466 30.39358 7.024372 4.732682 4.723124 7.015272 5.880448 2011 6.244858 5.029587 5.026452 6.242568 6.04508 5.64389 3.323605 3.303129 5.61899 5.199646 2012 6.2351 4.984517 4.977201 6.229522 6.485979 4.86343 2.748514 2.718075 4.819448 5.084898 2013 6.231736 4.949949 4.936769 6.221292 6.808229 4.42484 2.514796 2.475336 4.3611 5.139177 2014 6.234492 4.924991 4.904447 6.217654 7.051646 4.181922 2.421921 2.374015 4.09868 5.24384 2015 6.243104 4.908842 4.879604 6.218406 7.242737 4.051666 2.387704 2.331829 3.949434 5.36314 2016 6.257324 4.900801 4.861685 6.223352 7.399158 3.986756 2.378304 2.314969 3.866054 5.486258 2017 6.276922 4.900245 4.850188 6.232322 7.53268 3.96011 2.379766 2.309487 3.821432 5.609946 2018 6.30168 4.906626 4.844663 6.245144 7.651148 3.95619 2.386378 2.309613 3.8 5.733307 2019 6.331402 4.919453 4.844703 6.261668 7.7598 3.966054 2.395833 2.312953 3.792784 5.856183 2020 6.3659 4.938289 4.849935 6.281744 7.862127 3.984578 2.407233 2.318507 3.794608 5.978658 2021 6.40501 4.962744 4.860026 6.305238 7.960466 4.008822 2.42026 2.325846 3.802472 6.100907 2022 6.448568 4.992467 4.874671 6.33202 8.056381 4.037122 2.434812 2.334781 3.814642 6.223128 2023 6.496434 5.027143 4.893588 6.36197 8.150922 4.068544 2.450875 2.345216 3.830102 6.345532 2024 6.548474 5.066488 4.916528 6.394974 8.244798 4.10257 2.468455 2.357095 3.848254 6.468314 2025 6.604564 5.110247 4.94326 6.430926 8.338489 4.138926 2.487564 2.370379 3.868744 6.591671 2026 6.66459 5.158189 4.973572 6.469724 8.432323 4.177472 2.508202 2.385034 3.891352 6.715784 2027 6.728448 5.210106 5.007272 6.511276 8.526526 4.21814 2.530365 2.401028 3.915942 6.840825 2028 6.796044 5.265813 5.044189 6.555494 8.621255 4.260894 2.554045 2.418331 3.942418 6.966954 2029 6.867288 5.325138 5.084162 6.602292 8.716624 4.30573 2.579222 2.436913 3.970718 7.094314 2030 6.942102 5.387928 5.127048 6.6516 8.812714 4.352646 2.605879 2.456742 4.000782 7.223042 2031 7.02041 5.454048 5.172717 6.703342 8.909587 4.401634 2.633994 2.477794 4.032574 7.353264 2032 7.102148 5.523376 5.221045 6.75745 9.007293 4.4527 2.66354 2.500041 4.066054 7.485094 2033 7.18726 5.595799 5.271928 6.813868 9.105866 4.505834 2.694497 2.523456 4.101184 7.618637 4. Discussion The HIV/AIDS epidemic in Southern India is more serious than that in most other parts of India. Current adult HIV prevalence approximates 2 percent, and in the absence of any intervention our model predicts a 7.5 percent adult HIV seroprevalence in 2033. Unfortunately, this prediction cannot be very precise as many model parameters are only approximately known. Our model suggests that all interventions considered are potentially able to substantially dent the HIV epidemic. The effect of both highly effective vaccines and of a focused CSW intervention, based on condom promotion in which unprotected sex is reduced by 50 percent, is impressive. In the presence of a CSW intervention, prevalence would decline to 1.4 percent by 2033, less than the prevalence in 2001. This is consistent with the finding (Gangakhedkar and others, 1997) that infections among monogamous women in Pune (Maharashtra state) arise mostly from their husband's unprotected contact with sex workers. Even in mature epidemics, sex work is a key source of new infections. For example, adult prevalence in Cotonou, Benin, has exceeded 3 percent for the last decade or more. Careful work by Lowndes and others (2002) has concluded that virtually all of the ongoing HIV- I transmission is related to 13 infection of female sex workers, male clients of female sex workers, and the other non- regular sexual partners of those men. Syndromic treatment of STIs would reduce HIV prevalence to 2.4 percent by 2033-not as impressive as a CSW intervention, but still important. It needs to be stressed however, that the empirical basis for the impact of syndromic treatment of STIs in India is less solid or well understood than that for focused CSW interventions, especially since preliminary results from the trial in Masaka, Uganda, have come out. Preventive HIV vaccines could be highly effective in controlling the epidemic. Early understanding of the immunology correlates of HIV- 1 protection, and the genetic variability and rapid mutations of the HIV virus all suggest that a high efficacy vaccine is unlikely at the outset, but could develop with continuous testing (Esparza 2001, Plummer and others 2001). A vaccine that conveys substantially less than full protection to those who are immunized will not prevent sex workers from getting infected, but would delay infection. Thus, targeting vaccines with low protection to high-risk groups is less effective than providing them with highly effective vaccines or condom-based programs. Moreover, sustained condom use among high-risk groups reduces transmission of STIs other than HIV. Given a specified average vaccine efficacy, vaccines would be most effective if providing near 100 percent protection in to those who are immunized, even if not everyone vaccinated has an immunological response-in the terminology of Stover and others (2002), "take"-type efficacy. Vaccines that confer the same average partial level of protection to all vaccine recipients ("degree"-type efficacy) have less of an epidemiological impact. This makes sense, as partial protection may be insufficient to protect individuals with high-risk behaviors, although it would delay their infection. This is consistent with findings by Stover and others (2002). A vaccine that confers 100 percent protection in 95 percent of all vaccine recipients could almost eradicate HIV within 25 years. Irrespective of the targeting strategy (the general population or high-risk groups), adult HIV prevalence would shrink to a mere 0.6 percent in 2033 and would subsequently decline even further. More importantly, this vaccine would still have a substantial impact, even if CSW- client condom use were to drop to zero (disinhibition). If high-risk populations were targeted with this highly effective vaccine and condom use were to drop, adult HIV prevalence in 2033 would be 1.4 percent; if the general population were targeted with this vaccine and with disinhibition, HIV prevalence would reach 2.1 percent. The vaccine that would have the least effect is the vaccine that confers 50 percent protection to 50 percent of recipients-an average efficacy of only 25 percent. Adult HIV prevalence would rise to 3.7 percent in 2033 if high risk groups are targeted, while it would rise to 4.6 percent if the general population is targeted. The effects of this vaccine could be reversed by disinhibition, with adult HIV prevalence in 2033 of 10.3 percent (if high-risk groups are targeted) and 10.0 percent (population targeting), respectively. In other words, in the presence of disinhibition, HIV prevalence in 2033 14 would be 2.5-2.8 percentage points higher than the projected baseline, which reaches 7.5 percent in that time frame. This is also broadly consistent with findings by Stover and others: "a vaccine with low efficacy and low duration could have negative impact on public health if its implementation were accompanied by widespread reversion to riskier sexual behaviors" (p. 29). They conclude that "with low efficacy vaccines it will be very important to support the vaccination program with efforts to combat any reversal to riskier sex. If efforts to maintain safer sex behaviors are not successful, then behavioral reversals could eliminate most of the benefits of the vaccine. In some cases the effect could be to increase HIV incidence" (p. 29). Whether disinhibition is a likely scenario is unknown. It seems to be largely based on the experience with anti-retroviral therapy. While a vaccine may have the same effect, a vaccination campaign may also raise HIV awareness in the population and increase a sense of vulnerability in unvaccinated individuals. A sense of invulnerability in vaccinated individuals would only be a problem in partially effective vaccines. Aside from effectiveness, there is the issue of the cost and feasibility of interventions. Interventions share infrastructure costs (e.g., surveillance costs would be used for both types of program). Large population laboratories are needed to support new generations of vaccine testing and newer intervention research on interventions for high-risk groups. These costs are often of the nature of joint costs. Costs for preventing HIV growth have to be integrated with costs of other interventions. For example, outreach campaigns for vaccines would probably aim to deliver several vaccines, including those for childhood vaccine preventable diseases. The assumed vaccination coverage rates, while not 100 percent, even if integrated in existing structures, would still require substantial efforts and costs, with tens to hundreds of millions of vaccines administered over a 25-year period. A major advantage of a preventive HIV vaccine, which it shares with CSW and STI interventions, is that a potential recipient is not required to take an HIV test as a prerequisite for receiving a vaccine. While vaccines given to HIV-positive adults are clearly wasted (in the case of high-risk-group targeting this can be substantial), a policy of non-testing may be more efficient than one in which individuals are tested and counseled. Nevertheless, our estimates of the required number of vaccines only include vaccines for those who are HIV-negative. For population targeting, the wastage of vaccinating everybody is small. For high-risk populations, with a higher HIV prevalence, the wastage may be more substantial, at least in relative terms. Targeting high-risk groups is much more cost-effective than targeting the entire adult population. Using approximately one third of the number of vaccines, a higher reduction in prevalence is achieved. Although we did not explore this scenario, highly effective vaccines could be targeted to CSW only (i.e., excluding their clients), as in the long run this would be almost as effective as protecting both CSW and their clients. Vaccinating both CSWs and their clients would in the long run lead to substantial redundancy in 15 prevention efforts. By contrast, for vaccines conferring partial protection or with a low "take" rate it would seem sensible to vaccinate both CSWs and their clients. Conventional HIV prevention programs, especially those targeting CSWs (focused interventions) and using existing low-tech methods, may achieve results that are similar to reasonably effective vaccines and are probably less sensitive to disinhibition effects. It would therefore seem wise not to wait for the arrival of a vaccine, but to implement and expand focused CSW prevention programs as early and vigorously as possible. This will also create the infrastructure for effectively introducing HIV vaccines into these groups as soon as vaccines become available, and for scaling up vaccination campaigns. Such programs, however, require the political will to initiate and sustain them. Political support for vaccination campaigns, even for partially effective vaccines, may come easy, perhaps more so than for programs seemingly focusing on marginal groups such as CSWs. In sum, for the next few years expanding coverage of vulnerable group interventions while accelerating vaccine research and strengthening capacity for both with surveillance, human resource development, and operations research are the best strategies to contain the Indian HIV- 1 epidemic. When vaccines become available and particularly if efficacy or coverage is not perfect (most likely they are not!), then "other prevention programs should continue in conjunction with vaccination programs in order to reduce HIV infections to the lowest possible levels and maintain the other health benefits, such as prevention of sexually transmitted diseases" (Stover and others 2002, p. 30). References AP Benson, Inc. 1993-97. ModelMaker, version 3.0.3. United Kingdom: Old Beaconsfield, Buckinghamshire. [Website: http://www.modelkinetix.com/modelmaker/index.htm.] Bhave, G., C.P. Lindan, E.S. Hudes, S. Desai, U. Wagle, S.P. Tripathi, and J. S. Mandel. 1995. "Impact of an intervention on HIV, sexually transmitted diseases, and condom use among CSW in Bombay, India." AIDS 9 Suppl 1: S21-30. Esparza, J. 2001. "An HIV vaccine: how and when?" Bulletin of the World Health Organization 79: 1133-37. Government of India (GOI), Office of the Registrar General and Census Commissioner. 2002. Census of India website. (Website: http://www.censusindia.net). Gangakhedkar, R.R., M.E. Bentley, A.D. Divekar, D. Gadkari, S.M. Mehendale, M.E. Shepherd, R.C. Bollinger, and T.C. Quinn. 1997. "Spread of HIV infection in married monogamous women in India." Journal of the American Medical Association 278(23): 2090-92. Gilson, L., R. Mkanje, H. Grosskurth, F. Mosha, J. Picard, A. Gavyole, J. Todd, P. Mayaud, R. Swai, L. Fransen, D. Mabey, A. Mills, and R. Hayes. 1997. "Cost-effectiveness of improved treatment services for sexually transmitted diseases in preventing HIV- 1 infection in Mwanza Region, Tanzania." Lancet 350(9094): 1805-09. Gray, R.H., M.J. Wawer, N.K. Sewankambo, D. Serwadda, C. Li, L.H. Moulton, T. Lutalo, F. Wabwire-Mangen, M.P. Meehan, S. Ahmed, L.A. Paxton, N. Kiwanuka, F. Nalugoda, E.L. Korenromp, and T.C. Quinn. 1999. "Relative risks and population attributable fraction of incident HIV associated with symptoms of sexually transmitted diseases and treatable symptomatic sexually transmitted diseases in Rakai District, Uganda." AIDS 13: 2113-23. 16 Grosskurth, H., F. Mosha, J. Todd, E. Mwijarubi, A. Klokke, K. Senkoro, P. Mayaud, J. Changalucha, A. Nicoll, G. ka-Gina, et al. 1995. "Impact of improved treatment of sexually transmitted diseases on HIV infection in rural Tanzania: a randomised controlled trial." Lancet 346(8974): 530-36. Hethcote, H.W., and J.A.Yorke. 1984. Gonorrhea transmission dynamics and control. Berlin: Springer Lecture Notes in Biomathematics 56. Hitchcock, P., and L. Fransen. 1999. "Preventing HIV infection: lessons from Mwanza and Rakai." Lancet 353: 513-15 Hudson, C.P. 2001. "Community-based trials of sexually transmitted disease treatment: repercussions for epidemiology and disease prevention." Bulletin of the World Health Organization 79: 48-60. Jana, S., N. Bandyopadhyay, S. Mukherjee, N. Dutta, 1. Basu, and A. Saha. 1998. "STD/HIV Intervention with Sex Workers in West Bengal, India." AIDS 12 Suppl B: S 01 -S 108. Jana, S., and S. Singh. 1995. "Beyond the medical model of STD intervention--lessons from Sonagachi." Indian Journal of Public Health 39(3): 125-31. Jha, P., N.J. Nagelkerke, E.N. Ngugi, J.V. Prasasa Rao, B. Willbond, S. Moses, and F.A. Plummer. 2001. "Reducing HIV transmission in developing countries." Science 292: 224- 25. Katz, M.H., S.K. Schwarcz, T.A. Kellogg, J.D. Klausner, J.W. Dilley, S. Gibson, and W. McFarland. 2002. "Impact of highly active antiretroviral treatment on HIV seroincidence among men who have sex with men: San Francisco." American Journal of Public Health 92(3): 388-94. Korenromp, E.L., S.J. de Vlas, N.J. Nagelkerke, and J.D. Habbema. 2001. "Estimating the magnitude of STD cofactor effects on HIV transmission- how well can it be done?" Sexually Transmitted Diseases 28(11): 613-21. Korenromp, E.L., C. Van Vliet, H. Grosskurth, A. Gavyole, C.P. Van der Ploeg, L. Fransen, R.J. Hayes, and J.D. Habbema. 2000. "Model-based evaluation of single-round mass treatment of sexually transmitted diseases for HIV control in a rural African population." AIDS 14: 573-93 Kvale, G. 1999. "Preventing HIV-1: lessons from Mwanza and Rakai." Lancet 353: 1522-24. Lowndes, C.M., M. Alary, H. Meda, C.A. Gnintoungbe, L. Mukenge-Tshibaka, C. Adjovi, A. Buve, L. Morison, M. Laourou, L. Kanhonou, and S. Anagonou. 2002. "Role of core and bridging groups in the transmission dynamics of HIV and STIs in Cotonou, Benin, West Africa." Sexually Transmitted Infection 78 Suppl 1: i69-77. Matthys, F., and M. Boelaert. 1999. "Preventing HIV-1: lessons from Mwanza and Rakai." Lancet 353: 1523-24. Moses, S., F.A. Plummer, E.N. Ngugi, N.J. Nagelkerke, A.O. Anzala, and J.O. Ndinya-Achola. 1991. "Controlling HIV in Africa: effectiveness and cost of an intervention in a high-frequency STD transmitter core group." AIDS 5: 407-11. Nagelkerke, N.J., P. Jha, S.J. de Vlas, E.L. Korenromp, S. Moses, J.F. Blanchard, and F.A. Plummer. 2002. "Modelling HIV/AIDS epidemics in Botswana and India: impact of interventions to prevent transmission." Bulletin of the World Health Organization 80(2): 89-96. Nagelkerke, N.J., P. Jha, S. de Vlas, E. Korenromp, S. Moses, J. Blanchard, and F. Plummer. 2001. "Modeling the HIV/AIDS epidemic in India and Botswana: the effects of interventions" (Internet communication, 23 November 2001, at http://www.cmhealth.org/docs/wg5 paper4.pdf.) National AIDS Control Organisation (NACO). 2002. Website: httv://naco.nic.in National AIDS Control Organisation (NACO). 2002. National Baseline General Population Behavioural Surveillance Survey Report, 2001. Delhi: Ministry of Health and Family Welfare, Government of India. 17 National Intelligence Council. 2002. "The Next Wave of HIV/AIDS: Nigeria, Ethiopia, Russia, India, and China." ICA 2002-04D, September. http://www.fas.org/irp/nic/hiv-aids.html Nicoll, A., A.M. Johnson, M.W. Adler, and M. Laga. 1999. "Preventing HIV-1: lessons from Mwanza and Rakai." Lancet 353: 1522. Ostrow, D.E., K.J. Fox, J.S. Chmiel, A. Silvestre, B.R. Visscher, P.A. Vanable, L.P. Jacobson, and S.A. Strathdee. 2002. "Attitudes towards highly active antiretroviral therapy are associated with sexual risk taking among HIV-infected and uninfected homosexual men." AIDS 16(5): 775-80. Pais, P. 1996. "HIV and India: looking into the abyss." Tropical Medicine and International Health 1(3): 295-304. Plumnmer, F.A., N. Nagelkerke, J.V.R. Prasada Rao, B. Willbond, E. Ngugi, S. Moses, G. John, R. Nduati, K.S. MacDonald, and S. Berkley. 2001. "The evidence base for interventions to prevent HIV infection in low and middle-income countries." Commission on Macroeconomics and Health Working Paper Series, no. WG5-2. August. [Retrieved on December 23, 2002 from http://www.cmhealth.org/docs/wg5 vaper2.pdf.] Rodrigues, J.J., S.M. Mehendale, M.E. Shepherd, A.D. Divekar, R.R. Gangakhedkar, T.C. Quinn, R.S. Paranjape, A.R. Risbud, R.S. Brookmeyer, D.A. Gadkari, et al. 1995. "Risk factors for HIV infection in people attending clinics for sexually transmitted diseases in India." British Medical Journal 311(7000): 283-86. Stolte, I.G., N.H. Dukers, J.B. de Wit, J.S. Fennema, and R.A. Coutinho. 2001. "Increase in sexually transmitted infections among homosexual men in Amsterdam in relation to HAART." Sexually Transmitted Infection 77(3): 184-86. Stolte, I.G., and R.A. Coutinho. 2002. "Risk behaviour and sexually transmitted diseases are on the rise in gay men, but what is happening with HIV?" Current Opinion in Infectious Diseases 15(1): 37-41. Stover, J., G. Garnett, S. Seitz, and S. Forsythe. 2002. "The epidemiological impact of an HIV/AIDS vaccine in developing countries." World Bank Policy Research Discussion Paper, no. 2811. Development Research Department, World Bank, Washington, D.C. March. http://www.econ.worldbank.org/files/ 13172 wps281 l.pdf Venkataramana, C.B., and P.V. Sarada. 2001. "Extent and speed of spread of HIV infection in India through the commercial sex networks: a perspective." Tropical Medicine and International Health 6(12): 1040-61. World Bank. 1997. Confronting AIDS: Public priorities in a global epidemic. Washington, D.C.: The World Bank. 18 Annex 1. Description of the model, including a graphical representation. MalFemles |Low Risk X ihRs |Hg ik||Lo-w risk| usolimu l 117w'ds.XS , s mmun fe unmu me cstm j nidmmun_csw mmu csw No mu_e 'Immn e,n nimmun_cli\ immnf_csw; s unimmun fem -nmnimm_n Sslnustomo . "7'inf csw_immun s prdfO } \EDz/ mmn in Hi I un. tw immunn_-un /BY/Z//XEĆ i prccustomSpro~_cli_s ft>> Y-4/z ;PtunproW3 s p roje ZZZ | RP777777Xo,ncustom43 7 X7 voX prof4 3 ;/>. art_Mennunus ri prog hhdinhdnptg FE<_emp E % % t w~~~uouart fem uno ut pro m*_r _ _ / / ~~~~~prq_cIi-r pr_sfr \ PI9' - / prD-hem-r t sG_cli_~~~~out) Sismcl MPr \I n3s_menl0ut / \/ \/ ~~~~~~~~res_cswTout / \ 57Austomn8 / t.f>,->Xzz>unpol --z n D Zi~~~~~~~ds di a - to c r1@id_r h Ndy i n g_fem alidssfem r 19 COMPARTMENTS csw_early_s dcsw-early_ s/dt = -mupos'csw_early_s+prof2- Differential equations and initial unprof2+inf_csw_s-pro_csw_s+inf_csw_immun_s values Initial Valueu frac_cswini_inf_cSw*(l. AIDS female hivprog/(aidsrate+hivprog)*hivprog/(hivprog+unprof)) dAIDS_female/dt= +aids c- es~~~~~~~~cw_Immun +aids_csw_s+aids_csw_rl+aids_femr_rl+aids_fem_s- dcsw_immun/dt = -muneg*csw immun- hivdying_females- unprofO+profO+immun csw-inf_csw immun r- mupos*AIDS female+aids csw_r2+aids_fem_r2 inf_csw_immun_s-unimmun_csw Initial Value = 0.0 Initial Value = 0.0 AEDS_male csw late_ri dAIDS_male/dt = - dcsw_late_rl/dt = .mupos*csw late rl+prf7- hivdying_males+aids_cli_rl+aids_men_rl+aids_men s+ai unpowf7l+ro_csw_r-aids_csw_rl ds_cli_s-mupos*AIDS_male+aids_men_r2+aids_cli_r2 Initial Value = 0.0 Initial Value = 0.0 csw_late_r2 ell_early r dcsw late r2/dt = -mupos'csw late_r2+prof8- dcli_early_r/dt=-mupos*cli_early_r+custom3- unprof8+res cswout+res c w_prog-aidcswcr2 uncustom3+inf cli_r-pro_cli_r+inf_cli_immun r Initial Value = 0.0 Initial Value = 0.0 csw_late_s elI_earlys adcsw_late_s/dt=-mupos*csw late s+prof4- dcli_early s/dt=-mupos*cli earlys+custom2- unprof4+procsws-artcswprog+artcswunprog- uncustom2+inf cli s-pro_cli_s+inf_cli immun_s art_cw_out+art _csw_unout-aids_csw_ Initial Value = init_pop_male*init_frac_ci*init_inf cli*(l- Initial Value = hivprog/(aidsrate+hivprog)*hivprog/(hivprog+uncust)) initpopjfemale*init_frac-csw*mnit inf csw*hivprogl(aids cil immnun rate+hivprog)*hivprog/(hivprog+unprof dcl i_immun/dt= -munegcch_immun+custom0- csw_out uncustomO+immun_cli-inf Ci_immun_r-inf cli_immun_s- dcsw_out/dt = -mupos*csw_out+prof6- unimmun_cli unprof6+art_csw_out-art_csw unout-res csw out Initial Value = 0.0 Initial Value = 0.0 cIl late rl csw_prog dcli_late.rl/dt = -muposscliilate_rl+custom7- dcsw_prog/dt = -mupos*csw_prog+prof5- uncustom7+pro__cli_r-aids_cli_rI unprof5+artncsw_prog-art_csw_unprog-res_csw_prog Initial Value = 0.0 Initial Value = 0.0 cil late r2 csw_uninf dcli_late r2/dt = -mupos*cli_late_r2+custom8- dcsw_unmnf/dt=-muneg*csw uninf+profl-unprofl- uncustom8+res_cliprog+res_cli_out-aids_clihr2 inf csw_r-inf_csw_s-immun csw+unimmun_csw Initial Value = 0.0 Initial Value =- initpop female'init_frac_csw*(l- initinf csw) ciI late_s dcli_late_s/dt = -mupos*cli_late_s+custom4- cum_incidence uncustom4+pro cli s-art cli_prog+art_cli_unprog- dcum incidence/dt = incidence art cli out+art_cli_unout-aids_cli_s Initial Value = 0.0 Initial Value = init_pop_male*init_frac_cli*init_inf cli*hivprog/(aidsrate+ fem early r hivprog)*hivprog/(hivprog+uncust) dfem_early_r/dt=-mupos'fem earlyr- prof3+unprof3+inf fem r-pro_fem_rr+inf fem_immun_r ciI_out Initial Value = 0.0 dcli out/dt = -mupos*cli_out+custom6- uncustom6+art_cli_out-art_cli_unout-res_cli_out fem_early_s Initial Value = 0.0 dfem early_s/dt = -mupos*fernearly_s- prof+2unprof2+inf fern s-pro femrs+inf_ferr_immun_s c_lprog Initial Value= init_pop_female*(l- dcli_prog/dt =-mupos*cli_prog+custom5- init frac csw)*init inf fem'aidsrate/(aidsrate+hivprog) uncustom5+artcli_prog-art_cliunprog-res_cli_prog Initial Value = 0.0 fem-immun dfernimrnun/dt=-muneg*fem immun+unprofO- cIl_unlnf profO+immun_fem-inf_fern_immun r-inf fernimmun a- dcli_uninf/dt = -muneg*cli_uninf+customrl-uncustomrl- unimmun_femr inf cli_r-inf cli_s-imnun_cli+unimmun_cli Initial Value = 0.0 Initial Value = init_pop maleinit_frac_cli'(l-int_inf_cl) fem_late rI csw_early_r dfem_late_rl/dt = -mupos*fernmlate_rl- dcsw_early_r/dt = -mupos*csw_early r+prof3- prof7+unprof7+pro_fern_r-aids_fem_rl unprof3+inf csw_r-pro_csw_r+inf_csw_immun_r Initial Value = 0.0 Initial Value = 0.0 fem_late_r2 20 dfem_late_r2/dt = -mupos'fem_late_r2- prof8+unprof8+res_fem_prog+res_fem_out-aids_fern_r2 men-out Initial Value = 0.0 dmen out/dt = -mupos*men_out- custom6+uncustom6+art_men_out-art_men_unout- femr_late_s res_men_out dfem_late_s/dt = -muposfem_late_s- Initial Value = 0.0 prof4+unprof4+pro_fem_s-art_fem_prog+art_fem_unprog- art_femr_out+art_femr_unout-aids_fem_s men-prog Initial Value = init_pop_female*(l- dmen_prog/dt = -mupos'men_prog- init_frac_csw)*init_inf femrhivprog/(aidsrate+hivprog) custom5+uncustom5+art_men_prog-art_men_unprog- res_men_prog fem out Initial Value = 0.0 dfem out/dt = -mupos*fern_out- prof6+unprof6+art fem_out-art_fem_unout-res_fem_out men_uninf Initial Value = 0.0 dmen_uninf/dt = -muneg'men uninf-customl+uncustoml- inf_men_r-inf men_s+population*malegr- femprog immun_men+unimmun_men dfem_prog/dt = -mupos'fem_prog- Initial Value = init_pop_male(l-init_fracrcli)*(l- prof5+unprof5+art_fem_prog-art_fem_unprog- init_inf_men) res_fem_prog Initial Value = 0.0 prog_recr dprog_recr/dt= fem uninf art_men_prog+art_cli_prog+art_csw_prog+art_fem_prog dfem_uninf/dt = -muneg*fem_uninf-profl+unprofl- Initial Value = 0.0 inf fem_r-inf_femrs+population*femgr- immun fem+unimmun_fem wild recr Initial Value = init_pop_female*(l-init frac_csw)*(l- dwild_recr/dt = init inf fem) art_men out+art_cli_out+art_csw_out+art_fem_out Initial Value = 0.0 hivdeaths dhivdeaths/dt FLOWS +hivdying_males+hivdying_females+hivdyingkids Initial Value = O.O Movements between compartments Inf kids aids_cli_rl dinf kids/dt = +inf births-hivdying_kids Flow from cli_late_ri to AIDS_male Initial Value =0.0 aids_cli_rl = aidsrate * cli_late_rI men early_r aids ell r2 dmen early r/dt = -mupos*men early_r- Flow from cli_late_r2 to AIDS male custom3+uncustom3+inf men r- aids cl i r2 = aidsrate * cl i_late_r2 pro men_r+inf men_immun_r Initial Value = 0.0 aids cil s Flow from cli late s to AIDS_male men_early_fs aids_cli_s = aidsrate * clihlate s dmen early_s/dt =-mupos*men early s- custom2+uncustom2+inf men_s- aids_csw_rl pro men_s+inf_men_immun_s Flow from CSW_late_rl to AIDS female Initial Value init_pop_male'(l- aidsacsw_rI = aidsrate * csw_late_rI init_frac_clh)'init_inf_men*aidsrate/(aidsrate+hivprog) aidsacsw-r2 men _lmmun Flow from CSW late r2 to AIDS female dmen immun/dt = -muneg*men immun- aids_csw_r2 = aidsrate * csw_late_r2 customO+uncustomO+immun_men-inf men_immun_r- inf men immun_s-unimmun men aids csw_s Initial Value = 0.0 Flow from CSW late s to AIDS female aids_csw_s = aidsrate ' csw late s men_late_rl dmen late rl/dt=-mupos*men late rl- aids fem rl custom7+uncustom7+pro men r-aids men rl Flow from femrlate rI to AIDS female Initial Value= 0.0 aids fem rI = aidsrate * fem late rl men late r2 aIds_fern_r2 dmen late r2/dt = -mupos'men late_r2- Flow from fem-late-r2 to AIDS_female custom8+uncustom8+res_men_out+res_men_prog- aids_fem_r2 = aidsrate ' fem_later2 aids_men r2 Initial Value = 0.0 aids fem s Flow from fem late s to AIDS_female men late a aids_femrs = aidsrate * femrlate_s dmen_late s/dt = -muposrmen late s- custom4+uncustom4+pro_men s- aids men rl arnmen_prog+art_men_unprog- Flow from men_late_rI to AIDS_male art_men out+art_men_unout-aids_men_s aids_men_rI = aidsrate * men_late_rI Initial Value = init_pop_male'( I - init_frac_cli)'init_inf men'hivprog/(aidsrate+hivprog) aids_men_r2 21 Flow from men_late_r2 to AIDS_male customl aids_men_r2 = a idsrate * men_late_r2 Flow from men_uninf to cli_uninf customl = cust' men_uninf aids_menas Flow from men_late_s to AIDS_male custom2 aids_men_s = aidsrate * men_late_s Flow from men_early_s to cli_early_ custom2 = cust' men early_s art_cil out Flow from cli_late_s to cli_out custom3 art_cli_out = recr_cli_out * cli_late_s Flow from men_early_r to cli_early_r custom3 = cust * men_early r art_cli_prog Flow from cli_late_s to cli_prog custom4 art_cli_prog = recr_cli_prog * cli_late s Flow from men_late_s to cli_late_s custom4 = cust * men_late_s art cli_unout Flow from cli_out to cli_late_s custom5 art_cli_unout = outloss * cli_out Flow from men_prog to cli_prog custom5 = cust * men_prog art_cli_unprog Flow from ch_prog to cli_late_s custom6 art_ch_unprog = progloss cli_prog Flow from men_out to cli_out custom6 = cust * men_out art csw out Flow from csw_late_s to csw_out custom7 art csw_out = recr_CSW_out * csw late_s Flow from men_late_rl to cli_late_rl custom7 = cust' men_late_rI art_csw_prog Flow from csw_late_s to csw_prog custom8 art_csw_prog = recr_csw_prog * csw_late s Flow from men_late_r2 to cli_late_r2 custom8 = cust * men_late_r2 art_csw_unout Flow from CSW_out to csw_late_s hivdying_females art_csw_unout = outloss * csw_out Flow from AIDS_female to hivdeaths hivdying_females = round(muaids * AIDS_female) art_csw_unprog Flow from csw_prog to csw_late_s bivdying_kids art_csw_unprog = progloss ' csw_prog Flow from lnf_kids to hivdeaths hivdying_kids = round(muhiv_kids ' inf_kids) art_fem_out Flow from fem_late_s to fem_out hivdying_males art_fem_out = recr_fem_out * fem_late_s Flow from AIDS_male to hivdeaths hivdying_males = round(muaids' AIDS-male) art_fem_prog Flow from fem_late_s to fem_prog Immun_cli art fem_prog = recr_fem_prog * fem_late s Flow from cli_uninf to cli_immun immun_cli = vactake'vacrate_cli'cli_uninf art_fem_unout Flow from fem_out to fem_late_s Immun_csw art_fem_unout = outloss * fem_out Flow from csw_uninf to csw immun immun_csw = vactake'vacrate_csw'csw_uninf art_fem unprog Flow from fem_prog to fem_late_s Immun fem art_fem_unprog = progloss ' fem_prog Flow from femr_uninf to fem_immun immun_fem = vactake'vacrate_fem' fem_uninf art_men_out Flow from men_late_s to men_out Immun men art_men_out = recr_men_out ' men_late_s Flow from men_uninf to men_immun immun_men = vactake'vacrate_men'men_uninf art_men_prog Flow from men_late_a to men_prog lnf_citlImmun_r art_men_prog = recr_men_prog * men_late_s Flow from cli_immun to cli_early_r inf cli_immun_r = vaceff'(cli_immun'cr*fmrisk'unprot ' art_men_unout (wI'csw_earlyr+ w2'csw_late_rl + w2'counsel_csw'(1- Flow from men_out to men_late_s sustrans)'csw_late_r2) / csw) art_men_unout = outloss * men_out Inf clilImmun s art_men_unprog Flow from cli_immun to cli early_s Flow from men_prog to men_late_s inf_cli_immun_s = vaceff*(cli immun*cr*fmrisk*unprot ' art_men_unprog = progloss ' men_prog (wI'csw_early_s + w2'csw_late_s + w2'counsel_csw'sustrans'csw_late_r2 + customO w2'resid_infect'csw out) / csw) Flow from men_immun to cli_immun customO = cust ' men immun Inf_cit_r Flow from cli_uninfto cli_early_r 22 inf cli r= STD_control'(cli uninf'cr'finrisk'unprot' w2'resid infect'cli_out + wl'men early s + (wl'csw_earlyr + w2'csw_latejrl + w2'counsel_csw'(l- w2'men_late_s + w2'counsel_men'sustrans*men_late_r2 sustrans)'csw late r2) / csw) + w2'resid infect'men out) ' fem uninf/fem +fem uninP'stabfactor'mfrisk'marrate2' Inf_cll_s (wl'mnen early_s + w2'men_late_s + Flow from cli_uninf to cli early s w2'counsel_men'sustrans'men late_r2 + inf cli s = STD_control'(cli uninf'cr'finrisk'unprot' w2*resid_infect'men out) / men)) (wl'csw early_s + w2'csw late_s + w2'counsel csw'sustrans'csw late r2 + Inf_men_immun_r w2'resid infect'csw out) / csw) Flow from men_immun to men early_r inf men immun r = vaceffP( (leak'(wl'csw early r+ lnf_csw_immun_r w2*csw late rl +w2'counsel csw'(l- Flow from csw immun to csw_early_r sustrans)'csw_late_r2 + wl'fem early r + inf csw_immun_r = vaceff'(csw_immun w2'fem_late_rl+ w2'counsel_fem'(l- 'annualCSWcontacts'mfrisk'unprot * (wl 'cli earlyr + sustrans)'fem late r2) ' men_immun/men+ w2'cli late rl +w2'counsel cli'(l-sustrans)'cli_late r2)/ men_immun'stabfactor'finrisk'marrate * (wl'fem early r clients) + w2'fem late rI + w2'counsel_fem'(l- sustrans)'fem late r2) / femr)) lnf csw Immun s Flow from csw_immun to csw_early s Inf men_immunas inf csw immun_a = vaceffP(cswimmun Flow from men immun to men early a 'annualCSWcontacts'mfrisk'unprot * (wl'cli early_s + inf men immun s = vaceffP (leak'(wl 'csw_early s + w2'cli late s+w2'counsel_cli'sustrans'cli_late_r2 + w2'csw_late_s + w2'counsel csw'sustrans'csw late r2 + w2'resid_infect'cli out)/ clients) w2'resid-infect'csw_out+ wl'fem early s+ w2*fem_late_s+w2'counsel femrsustrans*fem_late_r2 + inf esw r w2'resid infect'fem out) * men_immun/men+ Flow from csw uninf to csw early_r men_immun'stabfactor'fmrisk'marrate' inf csw_r = STD_control*(csw_uninf (wl'fem early-s + w2'fem late_s 'annualCSWcontacts'mfrisk'unprot * (wl'cli early r + +w2*counsel-fem'sustrans*fem_late_r2 + w2'cli late rl +w2*counselcli*(l-sustrans)'cli_late r2)/ w2'resid_infect'fem out) / fem) clients) Inf_men_r inf caw s Flow from men uninf to men_early_r Flow from csw_uninf to csw early s inf men_r= STD control'( (leak'(wl'csw early_r+ inf csw_s = STD_control'(csw_uninf w2'csw late rl + w2'counsel_csw'(l- 'annualCSWcontacts'mfrisk'unprot * (wI 'cli early s+ sustrans)*csw_late_2 + wI 'fem early r+ w2'cli late_s +w2*counsel_cli'sustrans'cli_late_r2 + w2*fem_late_rl+ w2*counsel_fem*(l- w2'resid infect'cli_out) / clients) sustrans)*feml_ate r2) * men_uninf/men+ men_uninf'stabfactor'fmrisk*marrate * (w I 'fem early r Inf_femr_immun_r + w2*fem_late rl + w2'counsel_fem'(l- Flow from fem_immun to fem_early_r sustrans)'fem late r2) / femr)) inf fem immun_r = vaceff* (leak'(wl 'cli_early r + w2'cli late_rl + w2'counsel_cli'(1-sustrans)'cli_late_r2 + inf_men_s w1lmen earlyr + w2*men_late_rl+ w2*counsel_men'(l- Flow from men_uninf to men early s sustrans)*men later2)* fem immun/fem + inf men_ s=STD control*((leak*(wl*csw early s+ fem immun*stabfactormfrisk'smarrate2' w2'csw late s + w2*counsel_csw'sustrans'csw_late_r2 + (wl'men early +w2'men_late_rl+ w2'resid_infect'csw_out+ wl'fem early + w2'counselrmen'(l-sustrans)'men_late r2) / men) w2'femrlate_s+w2'counsel-fem*sustrans*fem_late_r2 + w2*resid infect*fem out) * men uninf/men+ Inf fem rimmun_s men_uninf'stabfactor'fmrisk*marrate * (wlIfem early s Flow from fem_immun to fem_early_s + w2'fem_late_a +w2'counsel_fem'sustrana'fem_late_r2 inf fem immun s=vaceff'(leak'(wl'cli early s+ + w2'resid_infect'fem out) / fem)) w2*cli late s + w2'counsel cli'sustranscli late r2 + w2'resid infect*cli out+wl'men_early s+ pro cl ir w2*men late s + w2*counselmen*sustransmen late_r2 Flow from cli early r to cli late rl + w2'resid infect'men out) * femr immun/fem pro cir = hivprog * cli early r +fem_immun'stabfactor'mfrisk'marrate2' (wl'men early_s + w2'men_late_s + pr_ecll_s w2*counsel men'sustrans*men late r2 + Flow from cli early s to cli late s w2*resid_infect'men out) / men-) pro_cli_s = HlVprog * cli early s lnf_femr_r pro_awrj Flow from femruninf to femr early r Flow from CSW early rto CSW_late_rI inf fem r= STD control'( (leak'(wl 'cli early r + pro_cswr = hivprog 'cswearly_r w2'cli late_ri + w2'counsel cli'(l-sustrans)'cli_late_r2 + wl'men_earlyr + w2'men_late_rl+ w2'counsel_men'(l- pro csw a sustrans)'men iate r2) * femn_uninf/fem + Flow from CSW early s to CSW_late_s fem uninf'stabfactor'mfrisk'marrate2 ' (wl 'men early_r pro csw s = HlVprog 'csw early s + w2'men late_rl+ w2'counsel_men*(l- sustrans)*men late r2) / men)) pro_fem_r Flow from fem early_r to femrlate_rI Inf fem rs pro fem r= hivprog * fem early_r Flow from fem uninf to fem early s inf fem s = STD_control'((leak*(wl'cli early s+ pro ferns w2'cli_late_s + w2*counsel_cli'sustrans*cli_late_r2 + Flow from fem_early_s to fem_late_s 23 pro_fem_s = HlVprog * fem_early_s res_men_out pro men_r Flow from men_out to men_late_r2 Flow from men_early r to men_late_ri resamen_out = outRDR * men out pro men_r = hivprog * men_early_r resmen_prog pro_men_s Flow from men_prog to men late r2 Flow from men_early_s to men_late_s res men prog = progRDR * men_prog pro men_s = HlVprog * men_early_s uncustomO profO Flow from cli_immun to men_immun Flow from femrimmun to csw_immun uncustomO = uncust * cli-immun profO = prof* exp(annualCSWcontacts/1000-1) * fem_immun uncustoml Flow from cli uninf to men_uninf pronl uncustoml = uncust* cli_uninf Flow from femruninf to csw_uninf profl = prof femruninfexp(annualCSWcontacts/1000-1) uncustom2 Flow from cli_earlyYs to men_early_a prof2 uncustom2 = uncust * cli_early a Flow from femrearly_s to csw_early_s prof2 = prof' fem_early_*exp(annualCSWcontacts/l OOO- uncustom3 I) Flow from cli_early_r to men_early_r uncustom3 = uncust * cli_early r prof3 Flow from femrearly r to CSW_early_r uncustom4 prof3 =prof* fem_early_r*exp(annualCSWcontacts/1000- Flow from cli_late_a to men lates I) uncustom4 = uncust * cli_late-s prof4 uncustom5 Flow from femrlate_s to CSW_late_s Flow from cli prog to men_prog prof4 = prof* exp(annualCSWcontacts/1000-1) * uncustom5 = uncust * cliprog fem_late_s uncustom6 prof6 Flow from cli out to men_out Flow from fem_prog to CSW_prog uncustom6 = uncust * cli_out prof5 = profexp(annualCSWcontacts/1000-1) * fem_prog uncustom7 prof6 Flow from cli late_ri to men late rl Flow from femrout to CSW_out uncustom7 = uncust * cli_late_ri prof6 = prof'exp(annualCSWcontacts/1000-1) * fem_out uncustom8 prof7 Flow from cli late_r2 to men late_r2 Flow from femrlate_rI to CSW_late_rI uncustom8 = uncust * chllate_r2 prof7 = prof*exp(annualCSWcontacts/1000-1) * fem_late_rl unimmun_cli Flow from cli_immun to cli_uninf prof8 unimmun_cli = losstimmun * cliiimmun Flow from femrlate r2 to CSW_late_r2 prof8 = prof*exp(annualCSWcontacts/l OOO-I) * unimmun_esw fem_late_r2 Flow from csw_immun to csw_uninf unimmun_csw = loss_immun* csw_immun res_cil_out Flow from cli_out to cli_late_r2 unimmun fem res_cli_out = outRDR * cli_out Flow from fern_immun to fem uninf unimmun_femr loss_immun * femrimmun res_cll_prog Flow from cli_prog to cli_late_r2 unimmun_men res_clijprog = progRDR * cli prog Flow from men_immun to men uninf unimmun_men - loss_immun * men_immun res_esw_out Flow from csw_out to csw_late_r2 unprofO res_csw_out = outRDR * csw_out Flow from csw_immun to fem immun unprofO = unprof * csw_immun rescw_prog Flow from csw_prog to csw_late_r2 unprofi res_csw_prog = progRDR * csw_prog Flow from csw_uninf to fem uninf unprofl = unprof* csw_uninf res_fern_out Flow from fern_out to fem_late_r2 unpro12 res_femrout = outRDR * fem_out Flow from CSW_early_s to fem earlys unprof2 = unprof* csw_early_s res_femprog Flow from fem_prog to femr_late_r2 unproB3 res_fem_prog = progRDR * fem prog Flow from CSW_early_r to femrearly_r 24 unprof3 = unprof ' csw_early_r female_prev_res = (esw_late_rl+csw_late_r2+fem_late_rI +fem_late_r2+csw_ unprof4 early r+fem_early_r)/females Flow from CSW late_s to fem_late_s unpromf4 = unprof ' cswlate_s female_prevalence female_prevalence = (females-femruninf-csw_uninf- unprof6 femrimmun-csw_immun)/females Flow from CSW_prog to fem_pmg unprof5 = unprof * csw_prog females females = fem+csw unprof6 Flow from CSW out to fem out In outl unprof6 = unprof csw_out In_outl = fem_out+men_out+csw_out+cli_out unprof7 In-out2 Flow from CSW late_ri to fem_late_ri In_out2 = femrout+men_out+csw_out+cli_out+(l- unprof = unprof ' csw_late_ri old_femnpp)*fem_late_r2+( I- old men_pp)'men_late r2+(I- unprofg old csw_pp)'cswjlate_r2+(1-old_cli_pp)'cli_late_r2 Flow from CSW_late_r2 to femrlate_r2 unprof8 = unprof * csw_late_r2 In_progl Inprogl = fem_pmg+men_pmg+csw_prog+cli_prog VARIABLES In_prog2 Variables defined in terms of fem-progmen_pmg+csw_pmg+cli_pmg+old2fem_pp*fe compartments, flows etc in the model m_late_r2+old_men_pp*men_late_r2+old_csw_pp*csw_lat e_r2+old_cli_pp*cli_ate_r2 aids_dead aids dead = Incidence hivdying-females+hivdying_males+hivdying kids incidence= round(inf cli_immun_s+inf csw_immun_s+inf_men_imm annualCSWcontacts un_s+inf_fem_immun_s+inf_cli_immun_r+inf_csw_immu annualCSWcontacts = cr'clients/csw i_rmnf_men_immun_r+inf fern_immun_r-i- inf cli_s+inf csw_s+inf men_s+inf fem_s+inf cli_r+inf c cll_prop_prog sw_r+inf men_r+inf femrr+inf births) cli_prop_prog = res_cli_prog/(res_cli_prog+res_cli_out+O.OOOI) Inf births inf births = round(brate*vtrate*( clients w I csw_early_r+w I fem early_r+(wl *csw_early_s+w2*c clients= sw_late_s+wl *femearlys+w2*fem late_s)*(nevirapine_r cli_immun+cli_early_s+cli late s+cli uninf+cli_prog+cli I ate*nevirapine effect+(l- nevirapine rate))+(csw_prog+femjges_ut+e_ut ate rl+cli late r2+cli out+cli earlyr )wtnevirapine_re))+(cs w latem+fem_ lout+femr ot - - - - ~~~- -I )'w2'nevirapine_effect+w2*(csw_1atc_rl +fem_late_rl +cs counsel_cil w_late_r2+femJ_ate_r2))) counsel cli = (I-old_cli_pp)+old cli_ppecounseI maie_prevres counsel csw male_prevres= counsel csw= (-old csw_p)+old cswppeounselcsw (cli late rl+cli late_r2+men_late rl+men_late_r2+cli_earl y_r+men_early_r)/males counsel_rem counsel fem=(1 -old fempp)+old fem_ppcounsel male_prevalence - -P M-PP male_prevalence = (males-cli uninf-men_uninf. counsel_men men_immun-cli immun)/males counsel men = (I -old men_pp)+old men_pp'counsel males asw males = men+clients caW csw = csw_immun+csw early_s+csw late_s+csw uninf+csw_pm marrmte2 g+csw_late_rl +csw_late_r2+csw_out+csw_early_r marrate2marratemen/fem men csw_prop_prog men csw_propJ pmg=men rescswprog(rescsw_rog+res csw out+O.OOOI) men_immun+men early_s+men_late_s+men_uninf+men_I ate_rl+men_late r2+men_prog+men_early_r+men out fem fem = men_prop_prog fem_immun+fem_early s+fem_late_s+fem_uninf+-fem late men_propprog= rl+fem late r2+fem_prog+fem early r+fem out resmen_prog/(res_men_prog+res_men_out+O.OOO I) fem_prop_prog milpop fem_prop_prog = milpop = population/1000000 res_femprog/(res_fem_prog+res_fem_out+O.OOO I) non_vaccnated female_prev_res 25 non_vaccinated = 1- (men immun+cli_immun+csw_immun+femrnimun)/(men _immun+cli immun+csw_immrun+fem_immun+men_uninf DELAYS +cli_uninf+csw_uninf+fem uninf) Time-lagged variables population population = round(males+females) oid_ ei_pp Delay =1.5 prev CSW Initial Value = 0 prev CSW = (csw-csw_uninf-csw_immun)/csw Maximum Delay = 1.7 preval_res old_esw_pp preval_res = Delay = 1.5 (males*male_prev_res+females*female_prevres)/(males+f Initial Value = 0 emales) Maximum Delay = 1.7 prevalence old_fem_pp prevalence = Delay= 1.5 (males*male_prevalence+females*female_prevalence)/(mal Initial Value = 0 es+females) Maximum Delay = 1.7 prim resistant old_men_pp prim resistant = Delay = 1.5 (csw early_r+csw_late_rl+fem_early_r+fem_late_rl+clihe Initial Value = 0 arly_r+cli_late_rl+men_early_r+men_late_rl)/(population* Maximum Delay = 1.7 prevalence) prop_elient DEFINE VALUES prop_client = clients/mates Variables influenced by interventions cr propesw cr= prop_csw = csw/females cr = cr_before prop_inf births nevirapine rate prop inf births = inf_births/(brate*females) nevnrapinmerate = 0 prop.males reer_cli_out prop_males = males/population recr_clibout = 0 resistant recr._ecl_prog resistant = recr_cli_prog = 0 (csw early_r+csw_late_rl+csw_late_r2+fem_early_r+fem late_rl+fem_late_r2+cli_early_r+cli_late_rl+cli_late_r2+m recr_CSW out en_early_r+men_late_rl+men late_r2)/(population*prevale recr_CSW_out = 0 nce) recr_csw_prog vacrate cli Conditional recrcsw_prog = 0 vacrate_cli = 0.75 for t>startyr_vaccin+2 recr_fem_out 0.75 for t>startyr_vaccin recr_fem_out = O 0 by default recr_fem_prog vacrate_csw Conditional recr_fem_prog = 0 vacrate_csw = 0.75 for t>startyr_vaccin+2 recr_men_out 0.75 for t>startyr_vaccin recr_men_out = 0 0 by default reer_men_prog vacrate fem Conditional recr_men_prog = 0 vacrate_fem = 0 for t>startyr_vaccin+2 STDlcontrol 0 for t>startyr_vaccin STD_control = I 0 by default unprot vacrate men Conditional unprot = unprot_before vacrate_men = 0 for t>startyr_vaccin+2 0 for t>startyr_vaccin INDEPENDENT EVENTS 0 by default Interventions STD_.prog year Non-periodic triggers at: year = t+ 1998 startyr_std Actions: STD_control = STD effect; 26 Introd_art Non-penrodic triggers at: startyr art intro Actions: recr_men_out = art_out_effect; recr_fern_out = art_out_effect; recr_csw_out = art_out_effect; recr_cli_out = art_out_effect; Inter_AART_pop Non-periodic triggers at: startyr_art_pop Actions: recr_cli_prog = recr_cli_prog_effect; recr_femn_prog=recr_fem_prog_effect; recrmen_prog = recr_men_prog_effect; recr_men_out gen_pro_eff*ar_out_effect; recr_fem_out = gen_prog_efftart_out_effect; recr cli out = gen_prog_eff*art_out_effect; cr-cr_after; Inter HAART_CSW Non-periodic triggers at: startyr art csw Actions: recr csw prog = recr csw_rog_effect; recr_csw_out = gen_prog_eff*art_out_effect; Inter_CSW Non-periodic triggers at: startyr_condom CSW Actions: unprot = unprot after; Inter MCT Non-periodic triggers at: startyr_MCT Actions: nevirapine_rate = nevirapine_prop; 27 Policy Research Working Paper Series Contact Title Author Date for paper WPS2952 The Effects of a Fee-Waiver Program Nazmul Chaudhury January 2003 N. Chaudhury on Health Care Utilization among the Jeffrey Hammer 84230 Poor: Evidence from Armenia Edmundo Murrugarra WPS2953 Health Facility Surveys: An Magnus Lindelow January 2003 H. Sladovich Introduction Adam Wagstaff 37698 WPS2954 Never Too Late to Get Together Bartlomiej Kaminski January 2003 P. Flewitt Again: Turning the Czech and Slovak Beata Smarzynska 32724 Customs Union into a Stepping Stone to EU Integration WPS2955 The Perversity of Preferences: The Caglar Ozden January 2003 P. Flewitt Generalized System of Preferences Eric Reinhardt 32724 and Developing Country Trade Policies, 1976-2000 WPS2956 Survey Compliance and the Johan A. Mistiaen January 2003 P. Sader Distribution of Income Martin Ravallion 33902 WPS2957 Mexico: In-Firm Training for the Hong Tan January 2003 H. Tan Knowledge Economy Gladys Lopez-Acevedo 33206 WPS2958 Globalization and Workers in Martin Rama January 2003 H. Sladovich Developing Countries 37698 WPS2959 Wage Differentials and State- Michael M. Lokshin January 2003 P. Sader Private Sector Employment Choice Branko Jovanovic 33902 in the Federal Republic of Yugoslavia WPS2960 The Poverty/Environment Nexus in Susmita Dasgupta January 2003 Y. D'Souza Cambodia and Lao People's Uwe Deichmann 31449 Democratic Republic Craig Meisner David Wheeler WPS2961 Strategic Planning for Poverty Rob Swinkels January 2003 N. Lopez Reduction in Vietnam: Progress and Carrie Turk 88032 Challenges for Meeting the Localized Millennium Development Goals WPS2962 High Consumption Volatility: Philippe Auffret January 2003 K. Tomlinson The Impact of Natural Disasters? 39763 WPS2963 Catastrophe Insurance Market in the Philippe Auffret January 2003 K. Tomlinson Caribbean Region: Market Failures 39763 and Recommendations for Public Sector Interventions WPS2964 Wages and Productivity in Mexican Gladys L6pez-Acevedo January 2003 M. Geller Manufacturing 85155 WPS2965 Informality Revisited William F. Maloney January 2003 P. Soto 37892 WPS2966 Health and Poverty in Guatemala Michele Gragnolati January 2003 M. Gragnolati Alessandra Marini 85287 WPS2967 Malnutrition and Poverty in Alessandra Marini January 2003 M. Gragnolati Guatemala Michele Gragnolati 85287 Policy Research Working Paper Series Contact Title Author Date for paper WPS2968 Refining Policy with the Poor: Local Edwin Shanks January 2003 N. Lopez Consultations on the Draft Carrie Turk 88032 Comprehensive Poverty Reduction and Growth Strategy in Vietnam WPS2969 Fostering Community-Driven Monica Das Gupta January 2003 M. Das Gupta Development: What Role for the Helene Grandvoinnet 31983 State? Mattia Romani WPS2970 The Social Impact of Social Funds Vijayendra Rao February 2003 P. Sader in Jamaica: A Mixed-Methods Ana Maria lbanez 33902 Analysis of Participation, Targeting, and Collective Action in Community- Driven Development WPS2971 Short but not Sweet: New Evidence Jishnu Das February 2003 H. Sladovich on Short Duration Morbidities Carolina Sanchez-Paramo 37698 from India WPS2972 Economic Growth, Inequality, and Richard H. Adams, Jr. February 2003 N. Obias Poverty: Findings from a New Data Set 31986 WPS2973 Intellectual Property Rights, Guifang Yang February 2003 P. Flewitt Licensing, and Innovation Keith E. Maskus 32724 WPS2974 From Knowledge to Wealth: Alfred Watkins February 2003 A. Watkins Transforming Russian Science and 37277 Technology for a Modern Knowledge Economy WPS2975 Policy Options for Meeting the Francisco H G. Ferreira February 2003 P. Sader Millennium Development Goals in Phillippe G. Leite 33902 Brazil: Can Micro-Simulations Help9 WPS2976 Rural Extension Services Jock R. Anderson February 2003 P. Kokila Gershon Feder 33716 WPS2977 The Strategic Use and Potential Christopher Desmond February 2003 H. Sladovich for an HIV Vaccine in Southern Africa Robert Greener 37698