JAIDS Journal of Acquired Immune Deficiency Syndromes:
Epidemiology and Social Science
Associations of Sexual Risk Taking Among Kenyan Female Sex Workers After Enrollment in an HIV-1 Prevention Trial
Yadav, Geeta*; Saskin, Refkin‡; Ngugi, Elizabeth#; Kimani, Joshua¶; Keli, Florence¶; Fonck, Karolien¶**; MacDonald, Kelly S†‡; Bwayo, Job J¶; Temmerman, Marleen**; The Kibera HIV Study Group; Moses, Stephen∥; Kaul, Rupert†§¶
From the *Faculty of Arts and Sciences, University of Toronto, Toronto, Ontario, Canada; †Department of Medicine, University of Toronto, Toronto, Ontario, Canada; ‡Mount Sinai Hospital, Toronto, Ontario, Canada; §University Health Network, Toronto, Ontario, Canada; ∥Department of Community Health Sciences and Medicine, University of Manitoba, Winnipeg, Manitoba, Canada; ¶Department of Medical Microbiology, University of Nairobi, Nairobi, Kenya; #Department of Community Health, University of Nairobi, Nairobi, Kenya; and **International Centre for Reproductive Health, Department of Obstetrics and Gynaecology, University of Ghent, Ghent, Belgium.
Received for publication December 8, 2003; accepted May 6, 2004.
Members of The Kibera HIV Study Group are listed under Acknowledgments.
Supported by The Rockefeller Foundation (2000 HE 025), the European Commission (DG VIII/8, contract no. 7-RPR-28), the Canadian Research Chair Programme (to R.K.), the Ontario HIV Treatment Network (to K.M.; Career Scientist), and the Canadian Institutes of Health Research (to S.M.; Investigator Award).
Reprints: Rupert Kaul, Clinical Science Division, Medical Sciences Building #6356, University of Toronto, Toronto, Ontario, Canada M5S1A8 (e-mail: email@example.com).
Background: Female sex workers (FSWs) often lack the ability to negotiate safer sex and are at high risk for HIV-1 infection and sexually transmitted infections (STIs).
Methods: Seronegative FSWs were enrolled in an STI/HIV-1 prevention trial in Nairobi, Kenya. Demographics and sexual risk taking were assessed every 3 months. Predictors of reduced risk taking were defined using multivariate logistic regression.
Results: Four hundred sixty-six FSWs were enrolled and followed for just over 2 years each. A spectrum of sex work was apparent: FSWs working in night clubs were younger, charged more for sex, and used condoms more frequently; FSWs working from home were older, charged less, and used condoms the least; and those working in bars were intermediate. Increases in reported condom use were most significant and sustained for FSWs working from home and charging less for sex and were poorly maintained for bar-based FSWs. Self-reported lower condom use, higher client numbers, and alcohol use were associated with higher STI rates.
Conclusions: Home-based FSWs and those charging less for sex used condoms the least at baseline but showed the greatest and most sustained improvements over time. Potential response heterogeneity in FSW subgroups should be considered in the design of HIV-1 prevention programs.
Female sex workers (FSWs) play an important role in the dynamics of the HIV-1 epidemic in Africa. Factors including gender inequality, economic disempowerment, and lack of resources and education combine to place these women at an enormously increased risk of acquiring HIV-1 infection.1,2 Infected FSWs may also act as “core groups” in the HIV-1 epidemic, with male clients serving as bridges into the general population.3,4 For these reasons, many HIV-1 prevention programs have focused specifically on FSWs5 and have often been very successful in reducing high-risk sexual behavior and/or rates of sexually transmitted infections (STIs) and HIV-1 infection.6-10 However, there is a broad spectrum of commercial sex work, with varying levels of sexual risk taking and prevalence of HIV-1 infection among different FSW sub-populations.11 It is therefore possible that the uptake or durability of HIV-1 risk reduction interventions will vary, depending on the demographics of the subpopulation targeted. It is important to carefully document and evaluate the efficacy of prevention programs in different situations and populations,12 because resources allocated to HIV-1 prevention fall far short of what is necessary.13
In 1998, a trial of antibiotic prophylaxis for acquisition of both STIs and HIV-1 infection was initiated in a cohort of HIV-1-seronegative sex workers from the Kibera district of Nairobi.14,15 A risk reduction program was provided for all enrolled FSWs, and this was very successful in reducing rates of high-risk behavior and STIs.9 However, self-identified FSWs in this cohort practiced sex work in a number of very different environments, as has been reported for sex worker cohorts elsewhere in Africa.11 Therefore, the purpose of this study was to determine whether baseline factors could be identified that predicted a greater or more sustained response to the risk reduction intervention (ie, increased condom use and/or reduced client numbers), with the aim of improving our understanding of how prevention strategies can be targeted to better address the needs of vulnerable FSWs.
Recruitment into this HIV/STI prevention trial was mediated through a previously established network of FSW peer educators10,16 from May 1998 to January 2002. The study design and baseline findings have been reported previously.9,14 Sex workers were defined as women who reported receiving money or gifts in exchange for sex during the month before initial screening. FSWs attended the clinic every month and were administered the study drug as directly observed therapy. All study subjects were provided with HIV-1 prevention services that included peer and clinic risk reduction counseling, the provision of free condoms, and prompt treatment of sympomatic STIs. Two standardized 1-hour risk reduction counseling sessions were provided to all women at enrollment, and subsequent clinic-based counseling was provided based on the clients' perceived needs and self-reported risk behavior. Peer-based risk reduction counseling was also provided to all study participants through a series of 3 monthly cohort “barazas” and smaller community meetings. This peer-based counseling followed a previously described model,10 with the addition of counseling regarding the need to negotiate consistent condom use with regular clients and boyfriends in addition to their more commercial clients.11 A behavioral questionnaire was administered at baseline and at 3-month intervals to assess risk-taking behavior. Ethics approvals for the study were obtained from institutional review boards at Kenyatta National Hospital, Nairobi, Kenya, and the University of Manitoba, Winnipeg, Manitoba, Canada.
All women underwent complete physical examination and STI testing and treatment at enrollment, every 6 months, and whenever clinically indicated. Cervical swabs were obtained for Neisseria gonorrhoeae and Chlamydia trachomatis polymerase chain reaction assays (Amplicor PCR Diagnostics, Roche Diagnostic Systems, Ontario, Canada) and for N. gonorrhoeae culture. If a genital ulcer was present, a swab of the ulcer base was taken for polymerase chain reaction (PCR) detection of Haemophilus ducreyi, herpes simplex virus, and Treponema pallidum (Roche Molecular Systems, Ontario, Canada). Trichomonas vaginalis culture was performed using In Pouch TV (Biomed Diagnostics, San Jose, CA), and blood specimens were obtained for HIV-1 and syphilis serology. Any infections identified were treated according to the Kenya National STI Treatment Guidelines. In addition, monthly urine specimens were collected at the time of directly observed study drug administration, stored at −20°C, and tested for N. gonorrhoeae and C. trachomatis by polymerase chain reaction assay after study completion.
All FSWs undergoing HIV-1 counseling and testing completed a baseline clinic questionnaire and had ongoing access to medical care through the clinic, whether they agreed to participate in the randomized trial. For enrolled FSWs, self-reported condom use, weekly client numbers, and hormonal contraceptive use were recorded at baseline and every 3 months. Condom use was reported on a scale of 0 to 5, where 0 represented no condom use and 5 represented condom use with all clients. All other demographic and behavioral data were collected only at the time of enrollment. Women were divided into 3 groups based on their place of work14 as follows: group 1, work only from their own or client's home; group 2, work from a nightclub or disco; group 3, work in a local bar or lodging.
Baseline (enrollment) associations of sexual risk taking, prevalent STIs, and HIV-1 infection were examined using a table for either 1-way analysis of variance (for continuous variables) or χ2 test (for dichotomous variables) in SPSS version 10.0 (SPSS, Chicago, IL). The impact of baseline demographic and behavioral factors on subsequent changes in risk taking and STI rates was then prospectively modeled using multivariate Poisson regression and logistic regression models for correlated data (generalized estimating equations; PROC GENMOD, SAS for Windows version 8.1, SAS Institute, Cary, NC). Variables included in the model were those previously associated with increased risk taking in other FSW cohorts and those associated with differences in risk taking in the baseline analysis (P < 0.05). These were place of work, charge per sex act (dichotomized into charge for sex more or less the cohort average), daily alcohol consumption, ever smoking, and age at enrollment. Rates of condom use were analyzed only at visits where women reported at least 1 weekly client (ie, they were still active in sex work).
Cohort Characteristics and Follow-up
Four hundred sixty-six HIV-1-seronegative FSWs were enrolled in the trial from May 1998 to January 2002. Participants were encouraged to remain in the trial for at least 2 years, after which time they were free to choose to continue in the trial or to exit the trial and attend the clinic as needed for routine medical care. The mean duration of follow-up was 760 days, for a total of 969.6 person years of follow-up, and the mean number of visits was 23.9.
Risk-Taking Behavior at Study Enrollment
When grouped according to place of work, significant differences were observed between FSW subgroups with respect to several behavioral and biologic variables at baseline (Table 1). Women working out of their homes or their clients' homes (home based) were older, charged the least for sex, and used condoms the least. Home-based FSW were also less likely to drink alcohol or smoke every day than were FSWs working in bars or nightclubs. FSWs working in nightclubs (nightclub based) were the youngest, charged the most, and used condoms the most frequently. Women based in a local bar or lodging (bar based) were generally similar to those in nightclubs but charged less and used condoms less frequently. There were no differences by place of work in rates of anal sex, sex during menses, intravenous drug use, or STI prevalence. Duration of follow-up varied significantly with place of work (home-based FSWs, 853.9 days; nightclub-based FSWs, 870.4 days; and bar-based FSWs, 644.4 days; P < 0.001). There were no differences in preenrollment client numbers or condom use between FSWs who were followed for over 2 years and those followed for a shorter time (15.7 vs. 15.2 clients per week [P = 0.6] and 2.4 vs. 2.6 on the semiquantitative condom use scale of 0 to 5 [P = 0.2], respectively).
FSWs who drank alcohol daily were more likely than other FSWs to be based in a bar or nightclub (Table 1) and were more likely to report ever having practiced anal intercourse (19.8% vs. 12.8%, respectively; P = 0.05), sex during menses (26.5% vs. 15.8%, respectively; P = 0.006), and intravenous drug use (8.6% vs. 1.7%, respectively; P < 0.001). These women were also less likely than other FSWs to use a condom with all clients (11.7% vs. 20.1%, respectively; P = 0.002). A very similar pattern of increased risk taking was seen for FSWs reporting several (≥3) drinks per day. Neither daily use of any alcohol nor taking several alcoholic drinks per day was associated with a higher prevalence of any STIs at enrollment. Reported use of a condom with all clients was associated with a lower enrollment prevalence of C. trachomatis infection (2.3% vs. 18.6%; P < 0.001) but not of other STIs.
Prospective Changes in Self-Reported Sexual Risk Taking
Condom use increased dramatically after enrollment in all groups (Fig. 1A) and client numbers decreased (Fig. 1B), as has been previously reported.9 Home-based FSWs showed the greatest improvements in condom use over time (relative risk [RR], 1.8; 95% confidence interval [CI], 1.2-2.7; multivariate Poisson regression for correlated data; Table 2). Although condom use by home-based FSWs was lower than that at enrollment by FSWs with other places of work, they had improved to the same level by 6 months and sustained these improvements for the duration of the study. In contrast, the early improvements in condom use reported by bar-based FSWs were not sustained over time. After 1 year of follow-up, their condom use was significantly lower than that by home-based FSWs, and after 2 years, it had fallen close to preintervention levels (Fig. 1A). In multivariate analysis, the only baseline factors independently associated with improved condom use over the study period were working from home and charging less than the cohort mean for sex (Table 2).
There were rapid and significant reductions in the reported number of weekly clients in the cohort as a whole. Over the duration of the study, women charging more than the average price for sex (≥137 Kenyan shillings) reported having fewer clients than women charging less (RR, 1.2; 95% CI, 1.0-1.4), and there was also a trend for greater reductions from baseline in client numbers over time in this group (RR, 0.1; 95% CI, 0.1-1.2). However, no significant differences in client numbers were seen based on FSW workplace (Fig. 1B). Neither drinking alcohol daily nor taking several (≥3) drinks per day was independently associated with prospective differences in reported condom use or client numbers in multivariate analysis.
Self-Reported Risk Taking and STI Incidence
To assess the accuracy of self-reported risk taking, multivariate logistic regression was performed with STIs (infection by N. gonorrhoeae, T. pallidum, T. vaginalis, or C. trachomatis) as the response variable (Table 3). Incident STIs during the study period were associated with lower reported condom use (P = 0.03) and higher reported client numbers (P = 0.02). This confirms the importance of interventions aiming to improve sexual risk taking and validates the use of self-reported risk taking as a study end point. In addition, younger age was associated with higher STI rates (RR, 0.97; 95% CI, 0.94-0.98; Table 3). Although alcohol use was not associated with differences in reported risk taking (as mentioned above), it was an independent risk factor for acquiring an STI (RR, 1.4; 95% CI, 1.0-1.9; Table 3). Total numbers of STIs throughout the study period were as follows: 82 episodes of N. gonorrhoeae infection, 98 episodes of C. trachomatis infection, 153 episodes of T. vaginalis infection, and 37 episodes of syphilis.
It is well established that FSWs in Kenya are at a very high risk of HIV-1 infection. The term “female sex worker” is applied to a heterogeneous group of women who work in a variety of situations and all exchange money or goods for sex. However, the circumstances in which sexual transactions take place vary enormously. This may lead to differences in the ability of FSWs to negotiate safer sex, or their motivation to do so, in both the short and long term. A better understanding of these factors is important to plan and implement interventions aiming to reduce FSW sexual risk taking. In this study, we examined the association of baseline sociodemographic variables with the uptake and durability of an HIV-1 behavioral risk reduction program in high-risk seronegative Kenyan FSWs. To have as wide a cross section of FSWs as possible, we relied on the expertise of a network of community “peer leaders” to access FSWs in a variety of work settings.
In general, women could be divided into 3 broad groups at study enrollment based on their place of work. Home-based FSWs were older, used condoms less, and charged less for sex. Nightclub-based FSWs were younger, used condoms more, and charged considerably more for sex, while bar-based FSWs fell between these extremes. Therefore, we hypothesized that home-based FSWs were either placing themselves at higher risk due to an inability to negotiate condom use with clients or a lack of access to condoms, which might either not be available or not be affordable. Condom use increased dramatically in the cohort after study enrollment, as previously described,9 and this improvement was greatest in women working from their own homes (Fig. 1A). Although condom use in all groups improved to a similar level after enrollment, preenrollment condom use by home-based FSWs had been low, so that the relative improvement was greatest in this group. This suggested that the low baseline levels of condom use in home-based FSW had likely reflected a lack of access to condoms, or an inability to afford them, rather than an inability to negotiate their use with clients. As far as we are aware, home-based FSWs are not considered “safer” by male clients, although HIV-1 knowledge and beliefs have not been formally assessed for these men. It is unlikely that increases in condom use represent regression to the mean. FSWs were not enrolled based on risky behavior, and so there is no reason to expect that their condom use varies from the true population mean of FSW condom use in Kibera. This is supported by the fact that similar or lower condom use has been seen in a number of observational Kenyan FSW cohorts.2,4,10
Although early reductions in FSW sexual risk taking are important to document, the durability of this response is key to the success of an intervention. This varied significantly in our study depending on the FSW place of work (Fig. 1). The initial reductions in sexual risk taking by home-based sex workers were maintained throughout the study, but this was not true for bar-based FSWs, where condom use fell to near-baseline levels despite continuous access to free condoms. The only factors independently associated with sustained improvements in condom use were charging less than the cohort mean for sex and working from home, further supporting the hypothesis that low preenrollment condom use in home-based FSWs was related to a lack of condom access, rather than an inability to negotiate their use. Only approximately one half the cohort was still in follow-up after 2 years; therefore, it is possible that preferential follow-up of more “compliant” FSWs might give the false impression of improved risk taking over time. However, risk taking (client numbers and condom use) at enrollment was very comparable between FSWs who were followed for >2 years and those who were followed for a shorter time. Therefore, this is not likely to be a source of study bias.
Alcohol use, defined as either taking alcohol every day or taking several drinks (≥3) per day, was associated with risk taking at enrollment, as has been seen in other FSW cohorts,17 and was more common in bar-based and nightclub-based FSWs. Although no direct association was seen between alcohol intake and the gradual resumption of risk-taking behavior in bar-based FSWs, alcohol use did correlate with higher STI incidence during follow-up. This suggests that in practice condom use was lower in FSWs drinking alcohol but may not have been detected due to misreporting in the monthly questionnaires.
The fall in client numbers after recruitment is likely to be multifactorial. Mobilization and education of FSWs across a community empower women from this area to standardize their charges, forcing their clients to conform. In this way, FSWs are able to charge more for each sexual encounter and to avoid the economic embarrassment that would otherwise result from a reduction in partner numbers. Although this strategy formed part of the peer-based counseling provided through the study, prospective data regarding charges for sex were not collected, and so we cannot confirm this hypothesis. An alternative explanation for reduced client numbers is regression to the mean. However, the only independent association of sustained reductions in client numbers was with a higher charge for sex at enrollment, suggesting that these FSWs may have been more economically empowered to reduce client numbers.
FSWs met monthly with clinic staff and were counseled regarding condom use and sexual risk taking if necessary. There was a concern therefore that there may have been a bias toward overreporting reductions in risk behavior. To better understand this issue, we examined the association between both reported condom use and weekly client numbers and laboratory-confirmed STIs due to N. gonorrhoeae or C. trachomatis. In keeping with an interim analysis,9 STIs were associated with lower reported condom use and higher reported client numbers in multivariate regression, confirming the validity of participant responses. The strong independent association of older age with lower STI rates may relate to acquired immunity to N. gonorrhoeae and C. trachomatis after repeated exposure,18-20 as has been demonstrated in other Kenyan FSW cohorts,21,22 although our study was not designed to test this hypothesis. In addition, it should be noted that overall STI rates may well have been artificially low in the context of this clinical trial, because FSWs had been randomized 1:1 to receive monthly treatment with azithromycin or placebo. Azithromycin has substantial activity against N. gonorrhoeae, C. trachomatis, and H. ducreyi and some activity against T. pallidum.
Overall, our findings suggest that more intensive follow-up counseling should be targeted at bar-based FSWs in risk reduction campaigns and that programs in this population should include specific counseling regarding alcohol use and increases in sexual risk taking. Global HIV-1 prevention programs are significantly underfunded,13 and so there is a great need for an evidence-based focus of resources to maximize the impact of those programs that do receive funding. Our results suggest that there is significant heterogeneity in the HIV-1 prevention needs and the uptake and sustainability of risk reduction interventions in different FSW subpopulations; therefore, HIV-1 prevention programs in FSW populations must be prepared to adapt accordingly.
The authors acknowledge the generous support of the Nairobi City Council and give special thanks to all the women who participated in this study for their enthusiasm and dedication.
Members of The Kibera HIV Study Group were as follows: Dr. Florence Keli, Grace Kamunyo, Ruth Wanguru, Rachel Mwakisha, Grace Waithira, Daniel Nganga, Cornelius Nyambogo, John Ombette, Jane Njeri, Isaiah Onyango, Dr. Isaac Malonza, Dr. Francis Mwangi, Judy Strauss, and Dr. Samuel Kariuki.
1. Kreiss JK, Koech D, Plummer FA, et al. AIDS virus infection in Nairobi prostitutes. Spread of the epidemic to East Africa. N Engl J Med
. 1986; 314:414-418.
2. Martin HL Jr, Jackson DJ, Mandaliya K, 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.
3. 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-i77.
4. Plummer FA, Nagelkerke NJ, Moses S, et al. The importance of core groups in the epidemiology and control of HIV-1 infection. AIDS
. 1991;5 (Suppl 1):S169-S176.
5. Alary M. More community-based trials of STD control or more appropriate interventions: which is the priority for preventing HIV-1 infection in developing countries? Bull World Health Organ
6. 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
7. Ghys PD, Diallo MO, Ettiegne-Traore V, 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
8. Ghys PD, Diallo MO, Ettiegne-Traore V, et al. Effect of interventions to control sexually transmitted disease on the incidence of HIV infection in female sex workers. AIDS
9. Kaul R, Kimani J, Nagelkerke NJ, 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
10. Ngugi EN, Wilson D, Sebstad J, et al. 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.
11. Nagot N, Ouangre A, Ouedraogo A, et al. Spectrum of commercial sex activity in Burkina Faso: classification model and risk of exposure to HIV. J Acquir Immune Defic Syndr
12. Parker RG, Easton D, Klein CH. Structural barriers and facilitators in HIV prevention: a review of international research. AIDS
. 2000;14(Suppl 1):S22-S32.
14. Fonck K, Kaul R, Kimani J, et al. A randomized, placebo-controlled trial of monthly azithromycin prophylaxis to prevent sexually transmitted infections and HIV-1 in Kenyan sex workers: study design and baseline findings. Int J STD AIDS
15. Kaul R, Kimani J, Nagelkerke NJD, et al. Monthly antibiotic chemoprophylaxis and incidence of sexually transmitted infections and HIV-1 infection in Kenyan sex workers: a randomized controlled trial. JAMA
16. Walden VM, Mwangulube K, Makhumula-Nkhoma P. Measuring the impact of a behaviour change intervention for commercial sex workers and their potential clients in Malawi. Health Educ Res
17. Wilson D, Chiroro P, Lavelle S, et al. Sex worker, client sex behaviour and condom use in Harare, Zimbabwe. AIDS Care
18. Loomis WP, Starnbach MN. T cell responses to Chlamydia trachomatis. Curr Opin Microbiol
19. Kim SK, Angevine M, Demick K, et al. Induction of HLA class I-restricted CD8+ CTLs specific for the major outer membrane protein of Chlamydia trachomatis
in human genital tract infections. J Immunol
20. Fox KK, Thomas JC, Weiner DH, et al. Longitudinal evaluation of serovar-specific immunity to Neisseria gonorrhoeae. Am J Epidemiol
21. Brunham RC, Nagelkerke NJ, Plummer FA, et al. Estimating the basic reproductive rates of Neisseria gonorrhoeae
and Chlamydia trachomatis
: the implications of acquired immunity. Sex Transm Dis
22. Plummer FA, Chubb H, Simonsen JN, et al. Antibodies to opacity proteins (Opa) correlate with a reduced risk of gonococcal salpingitis. J Clin Invest
female sex workers; male condom; prospective; Africa; workplace
© 2005 Lippincott Williams & Wilkins, Inc.
What does "Remember me" mean?
By checking this box, you'll stay logged in until you logout. You'll get easier access to your articles, collections,
media, and all your other content, even if you close your browser or shut down your
To protect your most sensitive data and activities (like changing your password),
we'll ask you to re-enter your password when you access these services.
What if I'm on a computer that I share with others?
If you're using a public computer or you share this computer with others, we recommend
that you uncheck the "Remember me" box.
Highlight selected keywords in the article text.
Data is temporarily unavailable. Please try again soon.