THERE WERE 5 MILLION NEW HIV infections in 2004, with 57% among women.1 In Africa, 75% of HIV infections among young people are female.1 These statistics reflect the behavioral and possible biologic vulnerability of women, and the limitations of current prevention. Currently, there are over 20 microbicide products under development, with 5 entering phase 2b/3 effectiveness trials.2,3 If vaginal microbicides are shown to be effective against HIV, they would offer women a significant new prevention option.
Different microbicide compounds have different potential mechanisms of reducing the likelihood of HIV transmission with different candidates either killing or immobilizing the virus, creating a barrier between the pathogen and vaginal epithelial cells, or preventing infection from becoming established once it has entered the body.2,3 In vitro and animal models also suggest that most current microbicide candidates may be active against several sexually transmitted infection (STI) pathogens, including Chlamydia trachomatis, herpes simplex II, and Neisseria gonorrhoeae, with some also potentially being active against syphilis and/or chancroid.4,5 Therefore, a microbicide’s HIV impact may result both from its direct activity on HIV and also from its indirect effect on other STIs.6,7 However, until specific products have completed phase III clinical trials, and start to be distributed, it is difficult to know what impact an efficacious microbicide will have on HIV transmission in different settings or the potential importance of a microbicide’s STI efficacy. The impact of a microbicide is likely to differ between settings and be dependent on contextual factors, including HIV/STI prevalence, patterns of sexual behavior and condom use, the microbicide’s HIV and STI efficacy, and the extent to which women can access and use microbicides.
To explore these issues, this article uses mathematical modelling, with site-specific data, to compare the extent to which an efficacious microbicide could reduce HIV transmission in 2 urban African settings: Hillbrow, an inner city area of Johannesburg, South Africa, and Cotonou, Benin. These settings differ in many ways, including having different prevalences of HIV and STIs, and different patterns of sexual behavior and condom use. Comparisons between the 2 settings are used to compare the impact of microbicide use in different partnerships and the importance of STI efficacy in determining impact.
This analysis builds on previous research that used epidemiological, behavioral, and demographic data to fit a dynamic population-based HIV transmission model to HIV and STI data from 1995–1999 for Cotonou and 2000–2001 for Hillbrow.8 The model was used to estimate the impact of 2 female sex worker (FSW)-targeted interventions over these periods. In this analysis, the model simulations from these analyses are used to estimate the HIV impact of widespread microbicide use in each setting.
The methods are described in different parts. First, the study populations are described. After that, a description of the mathematical model and the data used to parameterize it is given. Lastly, the methods used to fit the model and to estimate microbicide impact are detailed.
Hillbrow, South Africa.
Hillbrow is part of inner-city Johannesburg and in 1996 had an adult population of approximately 50,000.9 Few males are circumcised. The Hillbrow population is fairly transient, with individuals living in Hilbrow for 5.5 years on average.8
The main data source for the risk behavior of the Hillbrow general population were Behavioural Sentinel Surveys conducted among male STI clinic clients and female family planning (FP) clients in Hillbrow in August 2000.8 In these surveys, 5% of FP clients and 24% of male STI clinic attendees reported currently having a casual partner, and 26% reported having ever paid for sex.8 Enumeration studies estimated 12% to 20% of women were selling sex, with 31% of their clients being from Hillbrow.8 The apparent discrepancy between only 5% of FP clients reporting currently having a casual partner and 12% to 20% of women selling sex is largely the result of few FSWs attending the FP services and the likely underreporting of casual sex by women attending FP services. Since 2000, a HIV prevention intervention has been operating in Hillbrow (run by the Reproductive Health and HIV Research Unit [RHRU]), providing STI treatment and distributing condoms to FSWs.8 FSWs attending the intervention reported on average 32 clients per month in 2000–2001. Condom use was high with paying clients (83–88% of last sex acts), but much lower with main and casual sexual partners (41% report not using condoms in the last month).8 Likewise, condom use among the general population is quite low (55% of FP clients reported not using condoms in the last month8).
The prevalence of HIV is high in Hillbrow, with approximately 30% of antenatal clinic clients10 and 60% of FSWs attending the RHRU intervention being infected in 2001.11 Other curable STIs are also common: in 1998, 24% of FP clients were infected with chlamydia and/or gonorrhea (GC/CT) (95% confidence interval [CI] = 18.2–30.2%),12 and in 2000, 38% (95% CI = 32.6–43.7%) of FSWs attending the RHRU intervention were infected with GC/CT and 12% were diagnosed with genital ulcer disease (GUD) (95% CI = 8.5–16.1%).12
Cotonou is the largest city of Benin with an adult population of approximately 310,000 in 1998. In contrast to Hillbrow, nearly all men are circumcised13 and the population is fairly stable, with people living in Cotonou for an average of 40 years.14
The main data source for the risk behavior of the Cotonou general population was a multicenter population-based study undertaken in Cotonou in 1998 (4-city study).15,16 According to this study, the general population in Cotonou have fewer sexual partners than in Hillbrow.15 In addition, enumeration studies suggest sex work is less common, with approximately 1% of women currently selling sex, but roughly the same proportion of men currently buying sex (approximately 30%).16–18 Like in Hillbrow, there is a strong HIV prevention intervention in Cotonou run by Projet SIDA-2 (started in 1993 and funded by the Canadian International Development Agency) providing STI treatment for FSWs and their clients, outreach services, and condom promotion.19 FSWs attending the intervention report more clients per month than in Hillbrow, 52 on average in 1998, but condom use is comparable to Hillbrow, with condom use being high with paying clients (81% of last sex acts in 1998) but lower among other sexual partners (80–86% reported never using condoms with their regular partners in 1998).19 Reported condom use among the general population is low (approximately 15% reported using condoms often with nonspousal partners).15
In contrast to Hillbrow, HIV prevalence is lower in Cotonou and decreasing among FSWs: in 1995, 49% of FSWs attending the SIDA-2 intervention were infected, and in 1998, 41% of FSWs attending the intervention and approximately 3.3% of adults from the four-city study were infected.19,20 Other STIs are also less common; in 1995, 33% of FSWs (95% CI = 28.2–38.3%) attending the SIDA-2 intervention were infected with GC/CT, whereas in 1998, less than 3.5% of adults from the 4-city study (n = 1872), 25% of FSWs (95% CI = 21.6–28.8%), and 8% of their clients (95% CI = 5.5–11.0%), both from the SIDA-2 intervention, were infected with GC/CT.19,20 The GUD prevalence in Cotonou is also lower, with 4.5% of adult males (95% CI = 3.3–6.1%)21 and only 2.9% of FSWs (95% CI = 1.3–5.7%)16 from the 4-city study reporting GUD in 1997/1998, and 2% (95% CI = 0.9–3.7%) of the clients of FSWs attending the SIDA-2 intervention having GUD in 1998.19
Model Description and Parameterization
A deterministic, compartmental model (POP 1.0, described in detail in8) was used to simulate the HIV epidemics in Cotonou and Hillbrow, and to estimate the impact of microbicide use on HIV transmission among FSWs, their clients, and the general population in both settings. The model divides the population into subgroups with stratifications by level of sexual behavior, condom use, and HIV/STI status. Individuals of different sexual activity form sexual partnerships with defined durations, coital frequencies, and levels of condom use. The model uses established mathematical techniques to estimate how HIV and other STI spread between the subgroups over time.22 Because of the importance of STIs in facilitating HIV transmission,6 the model incorporates the transmission of different STI. As a result of data limitations, specific STI were not modeled, but symptomatic ulcerative STIs and GC/CT were modeled as 2 separate STIs. In the model, once an individual is infected with GC/CT, they remain infectious for a fixed duration or until treated, whereupon they become susceptible again. Once an individual has a symptomatic ulcerative STI, their ulcers remain for a fixed duration or until treated, except herpes simplex type 2 ulcers are assumed to be untreatable but are assumed to self-cure.23 The model incorporates the effect of STIs on HIV transmission, and also incorporates the higher infectivity associated with primary HIV infection and the lower susceptibility associated with circumcision.24–27
Microbicide use by a woman is assumed to reduce the per-sex-act probability of HIV and STI transmission by specified proportions (the microbicide’s HIV and STI efficacy). The consistency of microbicide use in a particular type of partnership is taken to be the proportion of unprotected sex acts in which a microbicide is used.
The main data sources for the risk behavior of the Hillbrow and Cotonou population were described in the study population sections and are shown in Table 1. The main differences between Hillbrow and Cotonou are highlighted in bold. Unfortunately, no data were available for the male general population in Hillbrow, and so the sexual risk behavior of the male STI clinic clients and the FP clients was used as upper and lower bound for the sexual risk behavior of the male general population. The transient nature of the Hillbrow population was incorporated into the model by assuming individuals only stayed in the area for an average of 5.5 years, after which they exit the model population. Without HIV, the Hillbrow and Cotonou populations were assumed to be of fixed size, and any outflow of people was matched by an inflow of new individuals with the same distribution of sexual behavior and condom use as the initial population. Other inputs such as the HIV and STI transmission probabilities28–36 and STI cofactors6 were obtained from the scientific literature (Table 1). Uncertainty bounds were estimated around all model input parameters (Table 1).
Before using the model to estimate the impact of the hypothetical microbicide intervention, it was fit to available data on HIV, GC/CT, and GUD prevalences among FSWs and the general population from Hillbrow and Cotonou (described in detail in8). For Hillbrow, available HIV prevalence data from 2000–2002 for antenatal clinic attendees,10 females from the Human Sciences Research Council (HSRC) household survey,37 and from the RHRU intervention for FSWs8 were used to fit the model. For Cotonou, available HIV prevalence data from 1995–1999 for adult females from the 4-city study,38 antenatal clinic attendees,18 and from the SIDA-2 HIV prevention intervention18,19 for FSWs and their clients were used to fit the model. The model was also fit to the STI prevalence data described in the study population sections. HIV prevalence data for other years, from both settings,10,12,39–41 were compared against the model projections. A full description of the data used for Hillbrow can be found elsewhere.8 Available HIV prevalence data for both settings can be seen in Figure 1.
To obtain different model fits, random sampling42 was used to obtain parameter sets from the model parameter uncertainty ranges in Table 1. The model simulations that lay within the 95% confidence intervals of the model-fitting data for each setting were selected. The only exception to this requirement was in Hillbrow, as a result of the discrepancy between the HIV prevalence estimates for females attending antenatal clinics10 and from the HSRC household survey.37 Using least squares, 2 simulations that best fit the epidemiologic data from each setting were obtained and can be seen in Figure 1. To make these simulations comparable, they were also constrained to have the same HIV and STI transmission probabilities, HIV high viremia cofactor, and STI cofactors. The figure emphasizes that, although the simulations were only fit to HIV prevalence data from certain years, the best fit simulations also lie within the 95% confidence intervals of most HIV prevalence data from other years.
Although the HIV prevalence among FSWs is fairly comparable in Hillbrow and Cotonou, among their clients and the general population, it is at least 4-fold less in Cotonou. In our model, the 2 main population differences that keep the HIV prevalence lower in Cotonou are the protective effect of males being circumcised and the lower proportion of females who are FSWs (Table 1). Indeed, for the model to fit the Cotonou HIV and STI prevalence data, circumcision had to result in a 48% (20–58%) decrease in the female-to-male HIV transmission probability per sex act. If these parameters are given the same values as in Hillbrow, the best fit model simulation for Cotonou projects a generalized epidemic similar to Hillbrow.
In Cotonou, the observed downturn in the FSW HIV prevalence (Fig. 1) is thought to be partially the result of the impact of the SIDA-2 HIV prevention intervention, which started in 1993.19 To fit the model to HIV prevalence data from Cotonou before this time, it incorporated baseline levels of risk behavior.19
The best fit simulations were used to produce point estimates for the impact of microbicide use and the other model fits were used to produce an absolute uncertainty range around these estimates (shown in brackets after the point estimate).
The objective of this analysis is to use modeling to explore how the impact of microbicide use on HIV transmission may depend on epidemiologic setting. To simplify the analysis and make comparisons easier, one baseline microbicide introduction and use scenario is considered for both settings.
In both settings, the model fits were used to estimate the impact on HIV transmission, over 4 years, of a microbicide that is used by 75% of the population. The high level of accessibility reflects the current availability of condoms in these settings. For specified levels of microbicide efficacy and use, the model simulates the dynamics of HIV and 2 STIs in the presence and absence of microbicides. Comparisons between these are used to estimate infections averted among FSWs and the general population. The impact of the microbicide was obtained by estimating the HIV infections averted per 100,000 adults and the relative decrease in HIV incidence. This impact includes both how a microbicide may directly affect HIV transmission and also how it may indirectly affect HIV transmission by reducing STI transmission. To coincide with behavioral and epidemiologic surveys undertaken by the HIV prevention interventions in both settings, the period over which microbicide use was considered was 1995–1999 for Cotonou and 2000–2004 for Hillbrow.
For the baseline microbicide use scenarios, the microbicide is assumed to reduce the per-sex-act transmission probability of HIV and other STIs by 40% (40% efficacy), approximately half the efficacy of condoms.43,44 Among the 75% of individuals who use microbicides, they are assumed to be used in 50% of unprotected sex acts in all partnerships, and microbicide use is assumed to result in a 5% relative reduction in condom use (to mimic the reductions in condom use observed in N-9/spermicide studies45). This simple pattern of microbicide use was used in the model because there is little data on how they may be used in different partnerships. A sensitivity analysis was not undertaken around this parameter because this was not seen as the main focus of the analysis. In addition, the model fits are used to explore how the projected impact of the microbicide varies between the 2 settings for the microbicide HIV and STI efficacy varying from 0% to 60% and to explore how impact is distributed among different types of sexual partnership.
To explore how the stage of an HIV epidemic might affect the impact of a microbicide, the impact of introducing the microbicide at earlier points in the Hillbrow and Cotonou epidemics (when the HIV prevalence among FSWs and the general population was lower) was also simulated using the best fit model simulations.
Baseline Impact Projections
The model projects the baseline microbicide use scenario would have resulted in a 9.3% (9.0–11.1%) decrease in the HIV incidence in Hillbrow, from 5.67 to 5.14 infections per 100 person-years, with 1566 (1372–1979) HIV infections averted in the district per 100,000 adults over 4 years. In contrast, a 47.6% (40.7–50.0%) decrease in the HIV incidence was predicted for Cotonou, from 0.44 to 0.24 infections per 100 person-years, with 571 (444–875) HIV infections being averted per 100,000 adults over 4 years. If the Hillbrow population had been less transient, fewer HIV infections would have been averted, and the difference between the 2 impact estimates would have been greater. If they stayed in Hillbrow for as long as in Cotonou, this would have decreased the number averted by 15% to 33%.
Impact of Microbicide Use in Different Types of Partnerships
For the baseline microbicide use scenario, Figure 2 shows the distribution of HIV infections averted according to the different types of partnerships in which microbicides are used. In Cotonou, the greatest decrease in HIV incidence and number of HIV infections averted results from microbicide use in commercial sex, whereas in Hillbrow, the greatest impact is from microbicide use in main and casual partnerships.
Impact of Microbicide at Different Stages of the HIV Epidemics
Using the best fit model simulations from each setting, Figure 3 shows how the level and distribution of microbicide impact in different types of partnership changes at different stages of the HIV epidemic in Hillbrow and Cotonou. As is currently predicted for Cotonou, Figure 3 suggests that in the early stages of the HIV epidemic in Hillbrow, a microbicide would have had greatest impact when used in commercial sex partnerships. Similarly, the figure also shows that the reduction in HIV incidence would have been much greater in the early stages of the HIV epidemic in Hillbrow, but the overall HIV infections averted per 100,000 adults would have been much lower. Conversely, for Cotonou, the reduction in HIV incidence remains roughly constant in the early stages of the HIV epidemic, although the HIV infections averted reduces.
However, even when the general HIV prevalence in Hillbrow was 1.5%, the relative reduction in HIV incidence resulting from the microbicide is less than half of what is currently predicted for Cotonou, and less impact is the result of microbicide use by FSWs with their clients. By analyzing the main behavioral differences between the Hillbrow and Cotonou FSW populations, the latter difference was found to be mainly the result of FSWs in Hillbrow reporting less clients per month and greater condom use with these clients than in Cotonou.
Impact of Microbicide Use for Different HIV and Sexually Transmitted Infection Efficacies
The baseline impact projections assumed that microbicides are 40% efficacious against HIV and STI transmission. Figure 4 shows the projected impact of the microbicide for different assumptions about its HIV and STI efficacy, and shows that a 40% HIV efficacious microbicide will have less than 43% of the impact of a 40% HIV and STI efficacious microbicide in these 2 settings. Indeed, only a 4% reduction in incidence will occur in Hillbrow and a 12% reduction in Cotonou. This may seem low but is to be expected considering the current level of microbicide use (75% of the population using a 40% HIV efficacious microbicide in 50% of noncondom-protected acts) only results in, at most, a 15% reduction in the number of unprotected sex acts. However, there are substantial increases in impact associated with increasing the HIV and STI efficacy of the microbicide. The magnitude of these gains is different in each site, with an increase in HIV and STI efficacy from 40% to 60% resulting in a 25% (23–36%) increase in impact in Cotonou and a 72% (67–78%) increase in Hillbrow.
The findings also suggest microbicides could reduce HIV transmission even if it is solely efficacious against the common STIs in each setting. Indeed, in Cotonou, the STI efficacy of the microbicide contributes more to the overall impact of the microbicide than its HIV efficacy. This coincides with the microbicide resulting in a smaller relative reduction in STI incidence in Hillbrow (13.5% for GC/CT) than in Cotonou (24% for GC/CT) and a much smaller relative reduction in STI prevalence in Hillbrow (6% for GC/CT) than in Cotonou (38% for GC/CT) over 4 years.
Microbicide Impact on HIV Transmission
This analysis shows that widespread microbicide use could impact on HIV transmission in 2 contrasting African settings—one where HIV is widespread and another where the HIV epidemic is less generalized. However, the microbicide’s impact differed substantially between the 2 settings, with a 5-fold greater relative decrease in HIV incidence in Cotonou but 3-fold more HIV infections averted (per 100,000 adults) in Hillbrow.
The findings suggest that a microbicide’s impact will vary substantially by epidemiologic setting, with widespread microbicide use having a greater relative impact on HIV incidence in less generalized HIV epidemics such as in Cotonou or the earlier stages of the epidemic that occurred in Hillbrow. This is the result of the HIV incidence in Hillbrow being much greater than in Cotonou, resulting in a smaller relative reduction in incidence (for the same level of use) because individuals are more frequently exposed to infection.46 These results agree with other modeling studies that have illustrated the overall impact of core group HIV prevention interventions would decrease as an epidemic progresses,47,48 and the impact of other HIV prevention interventions would be less in settings with more generalized epidemics.49–51 The findings illustrate the potential importance of microbicides but highlight the potential problems in generalizing the results of microbicide trials to other epidemiologic settings.
Microbicide Impact in Different Sexual Partnerships
The relative impact of microbicide use in different types of partnership was found to differ between the 2 settings, with most impact being the result of microbicide use in commercial sex in Cotonou, and microbicide use in main and casual partnerships in Hillbrow. Although our results suggest that this is partly the result of FSWs reporting less clients and higher levels of condom use with their clients in Hillbrow than in Cotonou, they also suggest that when an HIV epidemic is more generalized, a greater proportion of a microbicide’s impact is likely to arise from microbicide use in primary and casual partnerships because this is where most HIV transmission is occurring.
As shown for Hillbrow but not Cotonou, the relative impact of microbicide use in different types of partnership also changes as an epidemic progresses. This suggests that early on in a generalized epidemic such as has occurred in Hillbrow, most impact is attained through microbicide use in commercial sex, whereas later on, most impact comes from casual and primary partnerships. This agrees with a previous modeling analysis that illustrated how the impact of core group interventions will reduce as epidemics progress.47,48 This highlights the importance of tailoring microbicide promotion or distribution strategies to specific settings, taking into account the risk behavior of the population and the type or stage of the HIV epidemic.
Importance of Sexually Transmitted Infection Efficacy for Microbicide Impact
Our analysis also shows that the STI efficacy of a microbicide is likely to be important in determining the overall impact of a microbicide on HIV transmission. This is illustrated in our projections by the fact that a 40% HIV efficacious microbicide had less than half the impact of a 40% HIV and STI efficacious microbicide. Indeed, a sole STI efficacious microbicide can result in a substantial decrease in HIV transmission and in certain settings such as Cotonou, the STI efficacy of a microbicide could potentially be more important in determining impact than its HIV efficacy. However, the degree to which STI efficacy contributes to the impact of a microbicide is highly variable, with our results suggesting that it depends on the degree to which the STI efficacy of the product results in a decrease in STI prevalence. In our analysis, the STI efficacy of the microbicide contributed more to impact in Cotonou, the lower STI prevalence setting, because the microbicide resulted in a 6-fold greater relative decrease in STI prevalence than in Hillbrow, where STIs were more prevalent and so harder to affect because individuals are more frequently exposed to infection.46
The differences in the relative importance of a microbicide’s STI efficacy highlight that different candidate microbicide products may have different levels of impact in different settings. Indeed, eventually, it is likely that the greatest efficacy may be achieved using combinations of active ingredients3 that maximize both a product’s HIV and STI efficacy.
The analysis has several limitations. We only consider 2 scenarios with contrasting HIV/STI epidemics, and so the generalizability of the findings to other settings needs to be explored. The model projections have been conducted for different time periods and are dependent on the reliability of available epidemiologic data, reported behavioral data, and HIV/STI transmission data. There is also uncertainty in the estimates of the STI cofactors,6,52 STI/HIV transmission probabilities,28–30 and the protective effect of circumcision.53 However, despite this uncertainty, the model was fit and validated against available HIV and STI prevalence data from both settings. Lastly, ulcerative STIs were modeled simply as a whole and so the analysis did not fully take into account the possible effect of asymptomatic herpes simplex type 2 viral shedding on HIV transmission. Because of this, the effect of herpes simplex type 2 on HIV transmission may be more important than was assumed in this model, and so the impact of microbicide use may be greater than was projected if the microbicide reduces herpes simplex type 2 transmission.
Although it is difficult to determine how generalizable our results are to other settings, it is likely that they are fairly applicable to other communities with concentrated HIV epidemics, e.g., India and China, or generalized HIV epidemics, e.g., Zimbabwe and Botswana. The results of our analysis show that a partially effective microbicide can have a substantial impact on HIV transmission but that the magnitude will be highly dependent on the STI efficacy and will be highly site-specific, depending not only on the risk behavior of the population, but also on the HIV and STI distribution in the population. As previously discussed for other forms of HIV prevention,54 the relative impact of microbicide use on a population’s HIV incidence is likely to be less in settings with a more generalized HIV epidemic and higher HIV incidence. This suggests that much more needs to be done to control HIV transmission in these settings such as using high-efficacy microbicides, achieving high coverage, and initiating a range of prevention interventions. In addition, the results have repercussions for the planning of microbicide trials. Because a microbicide’s overall impact on HIV incidence will depend both on its HIV and STI efficacy, it is important that phase III microbicide trials collect sufficient biologic data to enable estimates of a microbicide’s efficacy against STI to be estimated.
Our results also suggest that the impact of microbicide use in different partnership types is highly variable, and primarily dependent on the stage and type of HIV epidemic. This should be considered in devising microbicide promotion interventions to ensure the most efficient use of limited resources.
1. UNAIDS. AIDS Epidemic Update. Geneva, 2004.
2. Stone A. Microbicides: A new approach to preventing HIV and other sexually transmitted infections. Nat Rev Drug Discov 2002; 1:977–985.
3. Harrison PF, Rosenberg Z, Bowcut J. Topical microbicides for disease prevention: Status and challenges. Clin Infect Dis 2003; 36:1290–1294.
4. Zeitlin L, Whaley KJ. Microbicides for preventing transmission of genital herpes. Herpes 2002; 9:4–9.
6. Rottingen J, Cameron DW, Garnett GP. A systematic review of the epidemiologic interactions between classic sexually transmitted diseases and HIV. Sex Transm Dis 2001; 28:579–595.
7. Grosskurth H, Mosha F, Todd J, et al. Impact of improved treatment of sexually transmitted diseases on HIV infection in rural Tanzania: Randomised controlled trial. Lancet 1995; 346:530–536.
8. Vickerman P, Terris-Prestholt F, Delaney S, et al. Are targeted HIV prevention activities still cost-effective in high prevalence settings? Results from an STI treatment intervention for sex-workers in Hillbrow, South Africa. Sex Transm Dis. In press.
10. Department of Health. Summary Report: National HIV and Syphilis Sero-Prevalence Survey of Women Attending Public Antenatal Clinics in South Africa. 2001.
11. Delany S. Summary of Behavioural and Epidemiological Data From Hillbrow Sex Worker Intervention. Reproductive Health Research Unit, 2001.
12. Fehler HG, Htun Y, Radebe F, et al. Patterns of Genital Tract Infection Among Family Planning Clinic Attenders and Sex Workers in South Africa (STDP 48). Lesedi STD Conference, 1999.
13. Auvert B, Buve A, Lagarde E, et al. Male circumcision and HIV infection in four cities in sub-Saharan Africa. AIDS 2001; 15(suppl 4):S31–40.
14. Alary M. HIV and STDs in Cotonou, Benin (West Africa): Importance of preventive interventions targeted at core groups for the control of the epidemic. CAHR, 2001.
15. Ferry B, Carael M, Buve A, et al. Comparison of key parameters of sexual behaviour in four African urban populations with different levels of HIV infection. AIDS 2001; 15(suppl 4):S41–50.
16. Morison L, Weiss HA, Buve A, et al. Commercial sex and the spread of HIV in four cities in sub-Saharan Africa. AIDS 2001; 15(suppl 4):S61–69.
17. Lowndes C, Alary M, Labbe AC, et al. Male Clients of Female Sex Workers in Cotonou, Benin (West Africa): Contributions to the HIV Epidemic and Effect of Targeted Interventions [Abstract 0729]. 15th Annual Congress of the International Society for Sexually Transmitted Disease Research, Ottawa, 2003.
18. Lowndes CM, Alary M, Meda H, et al. Role of core and bridging groups in the transmission dynamics of HIV and STIs in Cotonou, Benin, West Africa. Sex Transm Infect 2002; 78(suppl 1):i69–77.
19. Alary M, Mukenge Tshibaka L, Bernier F, et al. Decline in the prevalence of HIV and sexually transmitted diseases among female sex workers in Cotonou, Benin, 1993–1999. AIDS 2002; 16:463–470.
20. Buve A, Carael M, Hayes RJ, et al. The multicentre study on factors determining the differential spread of HIV in four African cities: Summary and conclusions. AIDS 2001; 15(suppl 4):S127–131.
21. Auvert B, Buve A, Ferry B, et al. Ecological and individual level analysis of risk factors for HIV infection in four urban populations in sub-Saharan Africa with different levels of HIV infection. AIDS 2001; 15(suppl 4):S15–S30.
22. Weinstein MC, Graham JD, Siegel JE, Fineberg HV. Cost-effectiveness analysis of AIDS prevention programs: Concepts, complications and illustrations. In: Turner CF, Miller HG, Moses LE, eds. Confronting AIDS: Sexual behaviour and intravenous drug use. Washington, DC: National Academy Press, 1989:471–499.
23. Corey L, Wald A. Genital herpes. In: Holmes KK, Mardh P, Sparling PF, et al., eds. Sexually Transmitted Diseases. New York: McGraw-Hill, 1999:285–312.
24. Fideli US, Allen SA, Musonda R, et al. Virologic and immunologic determinants of heterosexual transmission of human immunodeficiency virus type 1 in Africa. AIDS Res Hum Retroviruses 2001; 17:901–910.
25. Chakraborty H, Sen PK, Helms RW, et al. Viral burden in genital secretions determines male-to-female sexual transmission of HIV-1: A probabilistic empiric model. AIDS 2001; 15:621–627.
26. Pilcher CD, Tien H, Vernazza PL, et al. Semen Viral Dynamics in Acute HIV Infection: Implications for Sexual Transmission [Abstract ThOrC1489]. 14th International AIDS Conference, Barcelona, Spain, 2002.
27. Auvert B, Taljaard D, Lagarde E, et al. Randomized, controlled intervention trial of male circumcision for reduction of HIV infection risk: The ANRS 1265 Trial. PLoS Med 2005; 2:e298.
28. Quinn TC, Wawer MJ, Sewankambo N, et al. Viral load and heterosexual transmission of human immunodeficiency virus type 1. Rakai Project Study Group. N Engl J Med 2000; 342:921–929.
29. Royce RA, Sena A, Cates W Jr, Cohen MS. Sexual transmission of HIV. N Engl J Med 1997; 336:1072–1078.
30. European Study Group on Heterosexual Transmission of HIV. Comparison of female to male and male to female transmission of HIV in 563 stable couples. European Study Group on Heterosexual Transmission of HIV. BMJ 1992; 304:809–813.
31. Holmes KK, Johnson DW, Trostle HJ. An estimate of the risk of men acquiring gonorrhea by sexual contact with infected females. Am J Epidemiol 1970; 91:170–174.
32. Lycke E, Lowhagen GB, Hallhagen G, et al. The risk of transmission of genital Chlamydia trachomatis
infection is less than that of genital Neisseria gonorrhoeae
infection. Sex Transm Dis 1980; 7:6–10.
33. Kretzschmar M, Welte R, van den Hoek A, Postma MJ. Comparative model-based analysis of screening programs for Chlamydia trachomatis
infections. Am J Epidemiol 2001; 153:90–101.
34. Hooper RR, Reynolds GH, Jones OG, et al. Cohort study of venereal disease. 1: The risk of gonorrhoea transmission from infected women to men. Am J Epidemiol 1978; 108:136–144.
35. Ruijs GJ, Schut IK, Schirm J, Schroder FP. Prevalence, incidence, and risk of acquiring urogenital gonococcal or chlamydial infection in prostitutes working in brothels. Genitourinary Medicine 1988; 64:49–51.
36. Quinn SC, Gaydos C, Shepherd ME, et al. Epidemiologic and microbiologic correlates of Chlamydia trachomatis
infection in sexual partnerships. JAMA 1996; 276:1737–1742.
37. Shisana O, Simbayi L. Nelson Mandela/HSRC Study of HIV/AIDS. South African National HIV Prevalence, Behavioural Risks and Mass Media. Cape Town: The Human Sciences Research Council, 2002.
38. Buve A, Weiss HA, Laga M, et al. The epidemiology of gonorrhoea, chlamydial infection and syphilis in four African cities. AIDS 2001; 15(suppl 4):S79–88.
39. Rees H, Beksinska M, Dickson-Tetteh K, et al. Commercial sex workers in Johannesburg: Risk behaviour and HIV status. South African Journal of Science 2000; 96:283–284.
40. Ballard R. HIV/RPR Surveillance Report Among STD Clinic Attenders. Johannesburg: South African Institute of Medical Research, 2000.
41. US Bureau of the Census. HIV/AIDS Surveillance Data Base. Washington, DC: Population Division, International Programs Center, 2000.
42. Blower SM, Dowlatabadi H. Sensitivity and uncertainty analysis of complex models of disease transmission: An HIV model, as an example. International Statistical Review 1994; 62:229–243.
43. Pinkerton SD, Abramson PR. Effectiveness of condoms in preventing HIV transmission. Soc Sci Med 1997; 44:1303–1312.
44. Holmes KK, Levine R, Weaver M. Effectiveness of condoms in preventing sexually transmitted infections. Bull World Health Organ 2004; 82:454–461.
45. Foss A, Vickerman P, Heise L, Watts CH. Shifts in condom use following microbicide introduction: Should we be concerned? AIDS 2003; 17:1227–1237.
46. Mann JR, Stine CC, Vessey J. The role of disease-specific infectivity and number of disease exposures on long-term effectiveness of the latex condom. Sex Transm Dis 2002; 29:344–349.
47. Robinson NJ, Mulder D, Auvert B, et al. Type of partnership and heterosexual spread of HIV infection in rural Uganda: Results from simulation modelling. Int J STD AIDS 1999; 10:718–725.
48. Boily MC, Lowndes C, Alary M. The impact of HIV epidemic phases on the effectiveness of core group interventions: Insights from mathematical models. Sex Transm Infect 2002; 78(suppl 1):i78–90.
49. Nagelkerke NJ, Jha P, de Vlas SJ, et al. Modelling HIV/AIDS epidemics in Botswana and India: Impact of interventions to prevent transmission. Bull World Health Organ 2002; 80:89–96.
50. Korenromp EL, White RG, Orroth KK, et al. Determinants of the impact of sexually transmitted infection treatment on prevention of HIV infection: A synthesis of evidence from the Mwanza, Rakai, and Masaka intervention trials. J Infect Dis 2005; 191:S168–178.
51. White RG, Orroth KK, Korenromp EL, et al. Can population differences explain the contrasting results of the Mwanza, Rakai, and Masaka HIV/sexually transmitted disease intervention trials. J Acquir Immun Defic Syndr 2004; 37:1500–1513.
52. Korenromp EL, Sake JDV, Nagelkerke N, Habbema JD. Estimating the magnitude of STD cofactor effects on HIV transmission. Sex Transm Dis 2001; 28:613–621.
53. Weiss HA, Quigley MA, Hayes RJ. Male circumcision and risk of HIV infection in sub-Saharan Africa: A systematic review and meta-analysis. AIDS 2000; 14:2361–2370.
54. Garnett GP, Anderson RM. Strategies for limiting the spread of HIV in developing countries: Conclusions based on studies of the transmission dynamics of the virus. J Acquir Immun Defic Syndr Hum Retrovirol 1995; 9:500–513.
55. Korenromp EL, Sudaryo MK, de Vlas SJ, et al. What proportion of episodes of gonorrhoea and chlamydia becomes symptomatic? Int J STD AIDS 2002; 13:91–101.
56. Hanschell HM. Sulphanilamide in the treatment of chancroid. Lancet 1938; 1:886–888.
57. Garnett GP, Aral SO, Hoyle DV, et al. The natural history of syphilis. Implications for the transmission dynamics and control of infection. Sex Transm Dis 1997; 24:185–200.
58. Morgan D, Mahe C, Mayanja B, et al. HIV-1 infection in rural Africa: Is there a difference in median time to AIDS and survival compared with that in industrialized countries? AIDS 2002; 16:597–603.
59. Morgan D, Mahe C, Mayanja B, Whitworth JAG. Progression to symptomatic disease in people infected with HIV-1 in rural Uganda: Prospective cohort study. BMJ 2002; 324:193–196.
60. Post FA, Badri M, Wood R, Maartens G. AIDS in Africa—Survival according to AIDS-defining illness. S Afr Med J 2001; 91:583–586.
61. Maartens G, Wood R, O’Keefe E, Byrne C. Independent epidemics of heterosexual and homosexual HIV infection in South Africa—Survival differences. QJM 1997; 90:449–454.
62. Pinkerton SD, Abramson PR, Turk ME. Updated estimates of condom effectiveness. J Assoc Nurses AIDS Care 1998; 9:88–89.
63. Seven Buildings Project. Hillbrow Community Profile. Johannesburg: Reproductive Health Research Unit, 1994.
64. Population Reference Bureau. World Population Datasheet. Washington, DC: Population Reference Bureau, 2002.
65. Alary M, Lowndes M, Gnintoungbe C, et al. Male Clients of Female Prostitutes in Benin, West Africa: Factors Associated With Condom Use. 13th International AIDS Conference, Durban, South Africa, 2000.
66. Lowndes CM, Alary M, Gnintoungbe CA, et al. Management of sexually transmitted diseases and HIV prevention in men at high risk: Targeting clients and non-paying sexual partners of female sex workers in Benin. AIDS 2000; 14:2523–2534.