Share this article on:

Reduced rates of HIV acquisition during unprotected sex by Kenyan female sex workers predating population declines in HIV prevalence

Kimani, Joshuaa,c,†; Kaul, Ruperta,d,e,f,†; Nagelkerke, Nico JDc,g; Luo, Mac; MacDonald, Kelly Sd,h; Ngugi, Elizabethb; Fowke, Keith Rc; Ball, Blake Tc; Kariri, Anthonya; Ndinya-Achola, Jeckoniaha; Plummer, Francis Ac,j

doi: 10.1097/QAD.0b013e3282f27035
Epidemiology and Social

Objectives: Female sex workers (FSWs) form a core group at high risk of both sexual HIV acquisition and secondary transmission. The magnitude of these risks may vary by sexual risk taking, partner HIV prevalence, host immune factors and genital co-infections. We examined temporal trends in HIV prevalence and per-act incidence, adjusted for behavioral and other variables, in FSWs from Nairobi, Kenya.

Methods: An open cohort of FSWs followed since 1985. Behavioral and clinical data were collected six monthly from 1985 to 2005, and sexually transmitted infection (STI) diagnostics and HIV serology performed. A Cox proportional hazards model with time-dependent covariables was used to estimate infection risk as a function of calendar time.

Results: HIV prevalence in new FSW enrollees peaked at 81% in 1986, and was consistently below 50% after 1997. Initially uninfected FSWs remained at high risk of acquiring HIV throughout the study period, but the rate of HIV acquisition during unprotected sex with a casual client declined by over four-fold. This reduction correlated closely with decreases in gonorrhea prevalence, and predated reductions in the Kenyan HIV population prevalence by over a decade.

Conclusions: The per-act rate of HIV acquisition in high-risk Nairobi FSWs fell dramatically between 1985 and 2005. This decline may represent the impact of improved STI prevention/therapy, immunogenetic shifts in at-risk women, or changes in the proportion of HIV exposures occurring with clients who had acute HIV infection. Declining HIV incidence in high-risk cohorts may predict and/or be causally related to future reductions in population prevalence.

From the aDepartments of Medical Microbiology, Kenya

bCommunity Health (EN), University of Nairobi, Nairobi, Kenya

cDepartment of Medical Microbiology, University of Manitoba, Winnipeg, Manitoba, Canada

dDepartment of Medicine, Canada

eMcLaughlin Institute for Molecular Medicine, University of Toronto, Canada

fDepartment of Medicine, University Health Network, Toronto, Ontario, Canada

gDepartment of Community Medicine, UAE University, P.O. Box 17666, Al Ain, United Arab Emirates

hDepartment of Medical Microbiology, Mount Sinai Hospital, Toronto, Ontario, Canada

jNational Microbiology Laboratory, Health Canada, Winnipeg, Manitoba, Canada.

* These authors contributed equally to the manuscript.

Received 31 January, 2007

Revised 7 September, 2007

Accepted 19 September, 2007

Correspondence to Rupert Kaul, MD, PhD, Clinical Sciences Division, #6356 Medical Sciences Building, 1 King's College Circle, Toronto, Ontario M5S1A8, Canada. E-mail:

Back to Top | Article Outline


HIV-1 (HIV) has already killed more than 25 million people. UNAIDS estimated that 38.6 million people were living with HIV during 2006, with 4.1 million new infections [1]; however, HIV prevalence has recently fallen in some severely affected countries, including Uganda and Kenya [2,3]. Major prevention efforts preceded the decline in HIV prevalence in Uganda [4], although the high AIDS mortality associated with a ‘maturing’ epidemic may also have contributed [3–5]. Additional factors might have played a role, such as a decline in the number of individuals in the early phase of infection, when viremia and infectiousness is highest [6], or a depletion of the most susceptible individuals. A better determination of transmission patterns would be useful for HIV surveillance purposes, since their delayed elucidation confounds our ability to attribute such changes to a specific intervention. While prospective population cohorts might provide a rapid assessment of HIV transmission changes, only a few such cohorts are being monitored in Africa [7,8].

HIV core transmission groups such as female sex workers (FSWs), together with their male clients, may be responsible for much HIV transmission in sub-Saharan Africa [9,10]. This is particularly true during early epidemic stages, but also applies in the context of a mature HIV epidemic [11]. Therefore HIV monitoring within core groups may be important, particularly if transmission changes in these groups presage similar trends in the general population. Certain Kenyan FSW cohorts have a high incidence of HIV [12,13] and sexually transmitted infections (STIs) [12], partly due to low condom use and poor access to STI services [14]. The provision of condoms and peer- and clinic-based risk reduction interventions has been associated with reductions in STI and HIV rates, both in Kenya [14–16] and elsewhere in sub-Saharan Africa [17]. In addition, genetic factors such as HLA type have been associated with reduced HIV susceptibility in FSWs, and their population prevalence has changed over time [18].

Following reports of a reduced adult HIV prevalence in Kenya after 2000 [2], we hypothesized that this might have been preceded by HIV transmission changes within FSW core transmission groups. We therefore examined trends in HIV prevalence and incidence within a group of Nairobi female sex workers over the 20-year period 1985–2005. In particular, we examined trends in (1) HIV prevalence at the time of cohort enrolment, and (2) the rate of HIV acquisition per unprotected sex act. Since sexually-transmitted infections (STIs) may enhance both HIV susceptibility and secondary transmission [19], we also examined STI rates over this period.

Back to Top | Article Outline


Participant enrolment and follow up

FSWs were enrolled through a dedicated clinic in the Pumwani area of Nairobi, Kenya from 1985–2005, regardless of HIV infection status [14]. All enrolled women returned to the clinic for a formal re-survey every 6 months. At enrolment and re-survey a standardized questionnaire was completed, a physical examination and STI/HIV diagnostics were performed, and blood was drawn for HIV serology. Any STIs identified were treated according to Kenyan national guidelines. Self-reported risk taking data included the number of casual clients per day, condom use with casual clients, the number of regular clients (‘boyfriends’), condom use with regular clients, and sexual practices (anal sex, sex during menses). Condom use was reported on a scale of 0–3 ‘never’, ‘seldom’, ‘often’, and ‘always’ (although women reporting ‘always’ condom use did occasionally acquire STIs/HIV, indicating over-reporting or imperfect use). Women also had access to the clinic for outpatient medical services on an ad hoc basis.

Back to Top | Article Outline

Laboratory methods

HIV-1 serology was performed at each visit using a synthetic peptide enzyme immunoassay [EIA; Detect HIV, Biochem ImmunoSystems Inc., Montreal, Canada], and positive tests were confirmed by recombinant antigen EIA [Recombigen HIV-1/2 EIA; Cambridge Biotech Corporation, Galway, Ireland]. Clinic STI protocols during the study period included screening for Neisseria gonorrhoeae, Chlamydia trachomatis, Trichomonas vaginalis, bacterial vaginosis, Haemophilus ducreyii, and Herpes simplex type 2, but there was considerable variability over time in diagnostic methods for several STIs although cervical swabs were consistently obtained for N. gonorrhoeae culture on Thayer-Martin agar from 1985 to 2000 [20,21]. HLA genotyping was performed by high resolution sequence-based typing using banked blood samples [22].

Back to Top | Article Outline

Statistical methods

To explore trends in HIV prevalence at the time of enrolment we calculated HIV seroprevalence at enrolment by year of enrolment. Logistic regression was used to estimate trends in enrolment HIV seroprevalence over time.

Trends in the risk of HIV acquisition per unprotected sex act were analysed prospectively within those women who were HIV uninfected at enrolment. The number of unprotected sex contacts per day was calculated as the product of the number of casual sex partners per day and the ‘fraction unprotected’, estimated as (4 – condom use)/4, based on self reported risk-taking data (see above). Under this formula, women who reported ‘always condom use’ still had unprotected sex partners: this was deemed necessary since several such participants acquired HIV, indicative of over-reporting of ‘always condom use’.

To explore the effect of calendar year on the risk of HIV infection among initially seronegative women, two approaches were used. First, crude seroconversion rates were calculated for different time periods by dividing the total number of HIV seroconversions by the total number of self-reported unprotected contacts. Second, to adjust for changes in relevant covariables (such as condom use) over time, two separate multivariate analyses were carried out using a Cox proportional hazards model with time dependent covariables. Time dependent covariables in the first model were calendar year (minus 1985), the square of this variable, the number of partners per day, and condom use. ‘Number of sex partners per day’ and ‘condom use’ at any given time were approximated by their values at nearest time points. Specifically, they were calculated as averages weighted inversely by the interval between current time (day) and the times these variables were reported (generally at 6-monthly intervals), thus giving most weight to condom use at near time points. Time at risk was defined as time since enrolment. The fixed covariables used included age at enrolment, and the number of years working as a FSW before enrolment. In the second model, since trends in infection risk might be confounded by time-dependent increases in the enrolment of women with genetic resistance factors, we adjusted for HLA alleles previously shown to be associated with HIV resistance [18]. A ‘genetic resistance score’ was defined as the sum (between 0 and 6) of HLA alleles previously associated with HIV resistance, specifically DRB1*01, A*01, A*2901, B*1801, B*4101, B*151701, B*5702, Cw*070101, and Cw*070401. Data from 1985 to August 2005 were used.

HIV testing was not performed in male clients. To estimate the risk per HIV discordant sex act, a constant HIV prevalence of 30% was assumed in these men, as in previous studies [23,24]. This assumption was based on data from surveys of Kenyan STI clinic attendees, where the median HIV prevalence between 1990 and 1999 was fairly constant (25.6%, 1990; 29%, 1999), with no discernable trends over time [25].

Back to Top | Article Outline


Temporal trends in HIV prevalence at cohort enrolment

As of August 2005, a total of 2283 women had been cumulatively enrolled in the Pumwani FSW cohort. Of these, 878 were initially HIV negative, and had been clinically followed for a mean of 4.3 (± 4.8) years. There has been a clear decrease in HIV prevalence at enrolment for FSWs since at least 1990 (Table 1). Logistic regression yielded an average annual decrease in the odds of infection at enrolment over the whole study period of 6.4% [95% confidence interval (CI), 5.1–7.7%]. Exclusion of participants enrolled during the first 2 years after cohort establishment, who may have had a longer prior duration of HIV exposure prior to enrolment, yielded a similar annual decline (6.7%; 95% CI, 5.0–8.4%).

Table 1

Table 1

Back to Top | Article Outline

Changes in HIV incidence over time

Of the 878 women who were HIV negative at enrolment, 687 (78%) had sufficient follow-up data to be included in the first Cox regression analysis (Table 2(part (a)). Of the 687 eligible participants, 289 (42.1%) subsequently acquired HIV. Although condom use was highly protective against HIV acquisition across all time periods, there was only a weak association with the reported number of sex partners. The estimated time pattern indicated a parabolic relationship between the risk of infection and calendar time, with an increase in infection risk until approximately 1990 after which this risk declined substantially. Repeating the analysis, but treating condom use as a categorical time-dependent variable instead of a continuous one (0–3), yielded a nonsignificant coefficient for the square of calendar time and a coefficient of −0.11 (SE = 0.05) for calendar time itself, suggesting a consistent approximately 10% annual decline in infection risk since the second half of the 1980s.

Table 2

Table 2

For 471 of these women, 178 of whom subsequently acquired HIV, HLA molecular data were also available to allow calculation of a ‘genetic resistance score’; these participants were included in the second Cox regression analysis (Table 2 (part (b)). On average these women had 0.83 (range, 0–4; SD, 0.99) HLA alleles associated with HIV resistance, and this frequency did not change with year of enrolment among those initially HIV seronegative. HLA alleles previously associated with HIV resistance had a clear protective effect. The effect of other variables was very similar to the first Cox regression model. The same was true when the Cox regression was repeated with condom use as a (time-dependent) categorical variable. Results again indicated an approximately 10% annual decrease in infection risk, although this decline was no longer statistically significant (P = 0.11), likely due to the smaller sample size used in the analysis.

Back to Top | Article Outline

Changes in per-act rates of HIV acquisition over time

In order to further describe trends in HIV incidence over time, the rates of HIV transmission per unprotected sex act were calculated. These were based on self-reported client numbers, self-reported condom use and observed incident HIV infections among at-risk FSWs. Annual per-act rates were calculated, and the results pooled into 3-year blocks for ease of presentation (Fig. 1). The mean rate of HIV acquisition was 0.19/1000 per unprotected sex act during the entire 1985–2005 period. However, per-act rates of HIV acquisition fell more than four-fold during this time, from a high of 0.42/1000 unprotected acts (1985–87 period) to consistently under 0.1/1000 unprotected acts (all periods after 1994). Significant declines in the rates of per-unprotected-act HIV acquisition were already evident in 1988–1990, over 10 years before any reduction was seen in the national HIV prevalence (Fig. 1).

Fig. 1

Fig. 1

Since HIV prevalence among male clients was unknown, the precise rate of HIV transmission per unprotected HIV exposure could not be calculated. Assuming a constant mean HIV prevalence of 30% in male clients (see Statistical Methods, above), however, the pooled HIV acquisition rate per unprotected HIV exposure over the entire 1985–2005 period was 0.63/1000 HIV exposures.

Back to Top | Article Outline

Declines in per-unprotected-act HIV acquisition and sexually transmitted infection prevalence

To further explore associations of the decline in HIV acquisition per unprotected sex act, we calculated the 6-monthly prevalence rates of gonorrhea during the 1985–1999 period based on the results of N. gonorrhoeae cervical culture at resurvey visits. There was a very close correlation between gonorrhea prevalence in the sex worker cohort as a whole, and the rate of HIV acquisition per act of unprotected sex (Pearson correlation coefficient = 0.90; P < 0.0001; Fig. 1). Lack of consistent clinic testing protocols did not allow this analysis to be performed for other STIs, although similar trends were observed informally for rates of C. trachomatis, H. ducreyii and Treponema pallidum (data not shown).

Back to Top | Article Outline


These data demonstrate a clear decline in HIV rates among Nairobi female sex workers (FSWs) during the period 1985–2005. These declines were not only apparent in the decreasing HIV prevalence among cohort enrollees, which may be sensitive to selection bias(es), but also in the decreasing risk of HIV acquisition per unprotected sex act among HIV-uninfected FSWs, with rates during later stages falling more than four-fold from the 1980s. These declines began at least 10 years prior to generalized reductions in HIV prevalence in Kenya, and suggest that reduced HIV incidence within a core HIV transmission group could, at the very least, serve as a useful basis for predicting future trends at a population level.

Core transmission groups such as sex workers may be responsible for a disproportionate amount of heterosexual HIV transmission in the region [9,26,27]. This may be direct, through infection of male clients, or indirect, with infected clients acting as a ‘bridge’ to the general population. Although the role of core transmission groups is clearest during the early phases of an epidemic, the exchange of sex for money remains an important association of HIV infection even in a generalized epidemic [11]. Therefore, declines in FSW incidence, and hence of sex workers practicing sex during the most infectious early stages of HIV infection, may be a cause as well as a predictor of subsequent generalized declines in HIV prevalence.

In broad terms, a decline in per-unprotected-act rates of HIV acquisition by FSWs might relate to reduced susceptibility among at-risk women, or to reduced infectiousness among their male clients. Focusing on the former, HIV susceptibility may be increased by STIs [28], and it is possible that synergy between multiple STIs might disproportionately increase HIV acquisition. Gonorrhea rates in FSWs fell dramatically over the study period, closely paralleling the reduction in HIV acquisition during unprotected sex. Although gonorrhea prevention in FSWs did not prevent HIV acquisition in a recent trial [29], we also observed declines in other STIs, including the ulcerative bacterial STIs chancroid and syphilis (Kimani J, unpublished data). Ulcerative STIs may enhance HIV susceptibility to a greater degree than nonulcerative STIs [30]. Their decline may therefore have contributed to reduced per-contact HIV acquisition, although the cohort prevalence of Herpes simplex virus type 2 has been extremely high for at least the past decade [31] (and Kaul R, unpublished). Other biological factors have been associated with reduced HIV susceptibility, including genetic factors such as HLA [32] and HIV-specific immune responses [24,33,34]. Therefore changes in these factors over time might affect HIV incidence, and since some genetic associations of HIV susceptibility probably remain to be elucidated, these would be incompletely controlled by our ‘genetic resistance score’.

Changes in sexual behavior are an important mechanism for reduced HIV incidence in FSWs [35,36]. We adjusted our analyses for reported condom use, but could not adjust for the protection this afforded to male clients, and thereby for subsequent transmission from these clients to other FSWs. Furthermore, unmeasured differences in sexual practices and/or condom use with casual clients and regular partners might explain the weak association between partner numbers and HIV acquisition, and the fact that occasional HIV acquisition occurred despite reported condom use with all clients. To better understand HIV transmission dynamics in the FSW context, studying male clients should be made a priority. Reporting bias by FSWs is unlikely to explain our results, since our prevention focus would be expected to lead FSWs to under-report client numbers, if anything, and this would artifactually increase the estimated per-act rates of HIV acquisition over time.

Declines in per-unprotected-act rates of HIV acquisition by FSWs could also relate to reduced HIV infectiousness among male clients, particularly a fall in client HIV prevalence. However, the declines in FSW per-act HIV acquisition substantially predated generalized falls in HIV prevalence in Kenya, where urban HIV prevalence peaked at 16% in 1997 [2]. Furthermore, HIV prevalence in Kenyan STI clinic attendees had remained high when last surveyed in 1999, with no evidence for a decline over the preceding decade [25]. Although ‘localized’ changes in HIV prevalence among male clients remain theoretically possible, this seems improbable. More plausible is that reductions in the per-act transmission relate to lower rates of STI co-infections or acute HIV infection in male clients. Due to high levels of infectious HIV in the blood and semen during the first few months of HIV infection, it is thought that up to half of all secondary sexual transmission takes place during this short window [37,38]. Therefore reductions in HIV incidence in male clients, which might not be immediately apparent in HIV prevalence surveys, could explain the reduced transmission to FSWs. In addition, semen levels of HIV increase up to ten-fold in association with gonococcal urethritis [19,39]. The observed decline in FSW gonorrhea prevalence suggests that rates also fell in male clients, or (less likely) that condom use with gonorrhea-infected clients increased disproportionately. In either case, reduced unprotected sex with HIV-gonorrhea co-infected men would reduce the ‘dose’ of HIV exposure, and might explain the close association between N. gonorrhoeae prevalence and per-act HIV acquisition. Nor is this association likely to be limited to gonorrhea, since other STIs may also affect HIV susceptibility and secondary transmission [19]. However, although we informally observed declines in other STIs, including C. trachomatis, H. ducreyii and T. pallidum, temporal variations in clinic STI testing protocols did not permit a formal analysis.

Access to antiretroviral therapy (ART) has recently increased in Kenya. No sites reported provision of ART in 2003, but this had increased to 250 sites by 2005 [25], with provision of therapy to approximately 15% of those in need [40]. Effective HIV therapy decreases viral levels in both blood and genital secretions, and may significantly reduce the probability of HIV transmission during unprotected sex [41]. Therefore, although the timing of ART rollout in Kenya means that widespread access to therapy is extremely unlikely to have had any impact on HIV incidence in sex workers during the period of our study (1985–2005), this will be an important consideration in future studies.

Overall, our results have two major implications. First, the decline in HIV incidence in this core transmitter cohort predated generalized declines in HIV prevalence by at least 10 years, suggesting that monitoring HIV incidence in high-risk cohorts may be an important means to predict future population trends. If reductions in FSW incidence were the cause of subsequent declines in population HIV prevalence, something that seems plausible but that our data cannot prove, then this would make such monitoring even more critical. Second, the close association between the per-act rates of HIV acquisition and gonorrhea prevalence in FSWs suggests that STI control may be an important component of HIV prevention programs, irrespective of whether STIs enhance HIV transmission by increasing susceptibility in vulnerable FSWs, or by increasing the infectiousness of their HIV-STI co-infected clients.

Back to Top | Article Outline


The authors are grateful to the women of the Pumwani cohort for their support and enthusiasm.

Sponsorship: This study was supported by grants from the NIH (FAP, grant RO1-A156980); the Canadian Institutes of Health Research (RK), grant HOP-75350; (KF); the Ontario HIV Treatment Network (KM), grant and career scientist); and the Canada Research Chair Programme (FAP and RK salary support).

Back to Top | Article Outline


1. UNAIDS/World Health Organization (WHO). AIDS epidemic update: December 2006. In: World Health Organization HIV/AIDS strategic information reports, 2006. Geneva: UNAIDS/WHO; 2006.
2. Cheluget B, Baltazar G, Orege P, Ibrahim M, Marum LH, Stover J. Evidence for population level declines in adult HIV prevalence in Kenya. Sex Transm Infect 2006; 82(Suppl 1):i21–i26.
3. Stoneburner RL, Low-Beer D. Population-level HIV declines and behavioral risk avoidance in Uganda. Science 2004; 304:714–718.
4. Cohen J. ABC in Uganda: success or subterfuge? HIV AIDS Policy Law Rev 2005; 10:23–24.
5. Singh S, Darroch JE, Bankole A. A, B and C in Uganda: the roles of abstinence, monogamy and condom use in HIV decline. Reprod Health Matters 2004; 12:129–131.
6. Wawer MJ, Gray RH, Sewankambo NK, Serwadda D, Li X, Laeyendecker O, et al. Rates of HIV-1 transmission per coital act, by stage of HIV-1 infection, in Rakai, Uganda. J Infect Dis 2005; 191:1403–1409.
7. Mekonnen Y, Sanders E, Messele T, Wolday D, Dorigo-Zestma W, Schaap A, et al. Prevalence and incidence of, and risk factors for, HIV-1 infection among factory workers in Ethiopia, 1997–2001. J Health Popul Nutr 2005; 23:358–368.
8. Mwaluko G, Urassa M, Isingo R, Zaba B, Boerma JT. Trends in HIV and sexual behaviour in a longitudinal study in a rural population in Tanzania, 1994–2000. AIDS 2003; 17:2645–2651.
9. Cote AM, Sobela F, Dzokoto A, Nzambi K, Asamoah-Adu C, Labbe AC, et al. Transactional sex is the driving force in the dynamics of HIV in Accra, Ghana. AIDS 2004; 18:917–925.
10. Plummer FA, Nagelkerke NJ, Moses S, Ndinya-Achola JO, Bwayo J, Ngugi E. The importance of core groups in the epidemiology and control of HIV-1 infection. AIDS 1991; 5(Suppl 1):S169–S176.
11. Chen L, Jha P, Stirling S, Sgair SK, Millson P, Daid T, et al.Sexual risk factors for HIV infection in early and advanced HIV epidemics in sub-Saharan Africa: systematic overview of 68 epidemiological studies.PloS ONE 2007; 2:e1001.
12. Simonsen JN, Plummer FA, Ngugi EN, Black C, Kreiss JK, Gakinya MN, et al. HIV infection among lower socioeconomic strata prostitutes in Nairobi. AIDS 1990; 4:139–144.
13. Martin HL Jr, Jackson DJ, Mandaliya K, Bwayo J, Rakwar JP, Nyange P, et al. Preparation for AIDS vaccine evaluation in Mombasa, Kenya: establishment of seronegative cohorts of commercial sex workers and trucking company employees. AIDS Res Hum Retroviruses 1994; 10(Suppl 2):S235–S237.
14. Ngugi EN, Wilson D, Sebstad J, Plummer FA, Moses S. Focused peer-mediated educational programs among female sex workers to reduce sexually transmitted disease and human immunodeficiency virus transmission in Kenya and Zimbabwe. J Infect Dis 1996; 174(Suppl 2):S240–S247.
15. Moses S, Plummer FA, Ngugi EN, Nagelkerke NJ, Anzala AO, Ndinya-Achola JO. Controlling HIV in Africa: effectiveness and cost of an intervention in a high-frequency STD transmitter core group. AIDS 1991; 5:407–411.
16. Yadav G, Saskin R, Ngugi E, Kimani J, Keli F, Fonck K, et al. Associations of sexual risk taking among Kenyan female sex workers after enrolment in an HIV-1 prevention trial. J Acquir Immune Defic Syndr 2005; 38:329–334.
17. Ghys PD, Diallo MO, Ettiegne-Traore V, Kale K, Tawil O, Carael M, et al. Increase in condom use and decline in HIV and sexually transmitted diseases among female sex workers in Abidjan, Cote d'Ivoire, 1991–1998. AIDS 2002; 16:251–258.
18. Luo M, Kimani J, Nagelkerke N, Ball TB, MacDonald KS, Ndinya-Achola J, et al.Selection for HLA alleles that protect against HIV-1 infection correlates with the declining incidence of HIV-1 in an East African sex worker population. In: Keystone Symposium on Molecular and Cellular Determinants of HIV Pathogenesis. Whistler, British Columbia, 2007 [Abstract #315].
19. Cohen MS. HIV and sexually transmitted diseases: lethal synergy. Top HIV Med 2004; 12:104–107.
20. Kaul R, Rowland-Jones SL, Gillespie G, Kimani J, Dong T, Kiama P, et al. Gonococcal cervicitis is associated with reduced systemic CD8+ T cell responses in human immunodeficiency virus type 1-infected and exposed, uninfected sex workers. J Infect Dis 2002; 185:1525–1529.
21. Kimani J, Maclean IW, Bwayo JJ, MacDonald K, Oyugi J, Maitha GM, et al. Risk factors for Chlamydia trachomatis pelvic inflammatory disease among sex workers in Nairobi, Kenya. J Infect Dis 1996; 173:1437–1444.
22. Luo M, Mao X, Plummer FA. Identification of four novel HLA-B alleles, B*1590, B*1591, B*2726, and B*4705, from an East African population by high-resolution sequence-based typing. Tissue Antigens 2005; 65:187–191.
23. Kaul R, Rowland-Jones SL, Kimani J, Dong T, Yang HB, Kiama P, et al. Late seroconversion in HIV-resistant Nairobi prostitutes despite pre existing HIV-specific CD8(+) responses. J Clin Invest 2001; 107:341–349.
24. Kaul R, Rutherford J, Rowland-Jones SL, Kimani J, Onyango JI, Fowke K, et al. HIV-1 Env-specific cytotoxic T-lymphocyte responses in exposed, uninfected Kenyan sex workers: a prospective analysis. AIDS 2004; 18:2087–2089.
25. Joint United Nations Program on HIV/AIDS (UNAIDS)/World Health Organization (WHO). Kenya. In: Epidemiological fact sheets on HIV/AIDS and sexually transmitted infections: 2006 update. Geneva: UNAIDS/WHO; 2006. Accessed 27 October 2007.
26. Alary M, Lowndes CM. The central role of clients of female sex workers in the dynamics of heterosexual HIV transmission in sub-Saharan Africa. AIDS 2004; 18:945–947.
27. Lowndes CM, Alary M, Meda H, Gnintoungbe CA, Mukenge-Tshibaka L, Adjovi C, 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–i77.
28. Plummer FA. Heterosexual transmission of human immunodeficiency virus type 1 (HIV): interactions of conventional sexually transmitted diseases, hormonal contraception and HIV-1. AIDS Res Hum Retroviruses 1998; 14(Suppl 1):S5–S10.
29. Kaul R, Kimani J, Nagelkerke NJ, Fonck K, Ngugi EN, Keli F, et al. Monthly antibiotic chemoprophylaxis and incidence of sexually transmitted infections and HIV-1 infection in Kenyan sex workers: a randomized controlled trial. JAMA 2004; 291:2555–2562.
30. Orroth KK, White RG, Korenromp EL, Bakker R, Changalucha J, Habbema JD, et al. Empirical observations underestimate the proportion of human immunodeficiency virus infections attributable to sexually transmitted diseases in the Mwanza and Rakai sexually transmitted disease treatment trials: Simulation results. Sex Transm Dis 2006; 33:536–544.
31. Rebbapragada A, Wachihi C, Pettengell C, Sunderji S, Huibner S, Jaoko W, et al. Negative mucosal synergy between Herpes simplex type 2 and HIV in the female genital tract (In Press). AIDS 2007; 21:589–598.
32. Shrestha S, Strathdee SA, Galai N, Oleksyk T, Fallin MD, Mehta S, et al. Behavioral risk exposure and host genetics of susceptibility to HIV-1 infection. J Infect Dis 2006; 193:16–26.
33. Broliden K, Hinkula J, Devito C, Kiama P, Kimani J, Trabbatoni D, et al. Functional HIV-1 specific IgA antibodies in HIV-1 exposed, persistently IgG seronegative female sex workers. Immunol Lett 2001; 79:29–36.
34. Kaul R, Dong T, Plummer FA, Kimani J, Rostron T, Kiama P, et al. CD8(+) lymphocytes respond to different HIV epitopes in seronegative and infected subjects. J Clin Invest 2001; 107:1303–1310.
35. Ngugi EN, Chakkalackal M, Sharma A, Bukusi E, Njoroge B, Kimani J, et al. Sustained changes in sexual behavior by female sex workers after completion of a randomized HIV prevention trial. J Acquir Immune Defic Syndr 2007; 45:588–594.
36. Kaul R, Kimani J, Nagelkerke NJ, Fonck K, Keli F, MacDonald KS, et al. Reduced HIV risk-taking and low HIV incidence after enrollment and risk- reduction counseling in a sexually transmitted disease prevention trial in Nairobi, Kenya. J Acquir Immune Defic Syndr 2002; 30:69–72.
37. Pilcher CD, Tien HC, Eron JJ Jr, Vernazza PL, Leu SY, Stewart PW, et al. Brief but efficient: acute HIV infection and the sexual transmission of HIV. J Infect Dis 2004; 189:1785–1792.
38. Cohen MS, Pilcher CD. Amplified HIV transmission and new approaches to HIV prevention. J Infect Dis 2005; 191:1391–1393.
39. Cohen MS, Hoffman IF, Royce RA, Kazembe P, Dyer JR, Daly CC, et al. Reduction of concentration of HIV-1 in semen after treatment of urethritis: implications for prevention of sexual transmission of HIV-1. AIDSCAP Malawi Research Group. Lancet 1997; 349:1868–1873.
40. Joint United Nations Program on HIV/AIDS (UNAIDS)/World Health Organization (WHO). Progress on global access to HIV antiretroviral therapy. An update on 3 by 5, June 2005. Geneva: UNAIDS/WHO; 2005. ( Accessed: 27 October 2007.
41. Montaner JS, Hogg R, Wood E, Kerr T, Tyndall M, Levy AR, et al. The case for expanding access to highly active antiretroviral therapy to curb the growth of the HIV epidemic. Lancet 2006; 368:531–536.

female sex worker; HIV; incidence; prospective cohort; sexually transmitted infections

© 2008 Lippincott Williams & Wilkins, Inc.