Abstract
We conducted a respondent driven sampling survey to estimate HIV prevalence and risk behavior among female sex workers (FSWs) in Nairobi, Kenya. Women aged 18 years and older who reported selling sex to a man at least once in the past 3 months were eligible to participate. Consenting FSWs completed a behavioral questionnaire and were tested for HIV and sexually transmitted infections (STIs). Adjusted population-based prevalence and 95 % confidence intervals (CI) were estimated using RDS analysis tool. Factors significantly associated with HIV infection were assessed using log-binomial regression analysis. A total of 596 eligible participants were included in the analysis. Overall HIV prevalence was 29.5 % (95 % CI 24.7–34.9). Median age was 30 years (IQR 25–38 years); median duration of sex work was 12 years (IQR 8–17 years). The most frequent client-seeking venues were bars (76.6 %) and roadsides (29.3 %). The median number of clients per week was seven (IQR 4–18 clients). HIV testing was high with 86.6 % reported ever been tested for HIV and, of these, 63.1 % testing within the past 12 months. Of all women, 59.7 % perceived themselves at ‘great risk’ for HIV infection. Of HIV-positive women, 51.0 % were aware of their infection. In multivariable analysis, increasing age, inconsistent condom use with paying clients, and use of a male condom as a method of contraception were independently associated with unrecognized HIV infection. Prevalence among STIs was low, ranging from 0.9 % for syphilis, 1.1 % for gonorrhea, and 3.1 % for Chlamydia. The data suggest high prevalence of HIV among FSWs in Nairobi. Targeted and routine HIV and STI combination prevention strategies need to be scaled up or established to meet the needs of this population.
Keywords: Female sex worker, Kenya, HIV, Sexually transmitted infection, Respondent driven, Sampling
Introduction
Female sex workers (FSWs) continue to bear a high burden of HIV infection in many countries and are an important target population for a public health response to HIV/AIDS [1–5]. However, because behaviors associated with FSWs are considered to be illegal and highly stigmatized in many countries, it is often difficult to conduct probability-based surveys designed to provide representative estimates of biologic and behavioral factors among FSWs [6].
Despite these challenges, numerous studies among FSWs have been conducted in Kenya to estimate the level of HIV risk and to ascertain accurate levels of HIV infection, both prevalent and incident, among cohorts of FSWs [7–10]. These studies which have primarily utilized targeted or selective recruitment methods have provided useful information over the years. To date, however, methods used in previous studies in Nairobi have not measured HIV and STI prevalence among a representative sample of FSW in Nairobi. Sex work is illegal in Kenya, and surveillance of HIV and STI’s of FSW populations requires sampling methods suitable for recruiting hard-to-reach populations. Globally, national HIV surveillance activities have adopted respondent driven sampling (RDS) methods to provide population-based estimates of HIV prevalence and associated risk behaviors for populations at highest risk for HIV infection, such as FSWs [11, 12]. RDS is a probability-based, chain-referral sampling method to reach hidden populations that takes into account the social network size and degree of similarity between participants and other members they recruit into the study that provides population estimates of HIV prevalence and risk behaviors. Despite widespread use of RDS globally, important limitations of the method exists, including high variance of estimates and inflated design effects, making it difficult to identify changes in behavior and disease prevalence in the surveyed populations over time [13]. In 2010, the Government of Kenya established a surveillance system to monitor biologic and behavioral trends among key population groups at highest risk for HIV infection using RDS. In this analysis we present estimates of baseline HIV, sexually transmitted infections (STIs), and behavioral risk among FSWs using data collected from an RDS survey conducted in Nairobi, the capital city of Kenya. This is the first RDS survey to approximate a population-based survey on HIV and STI among FSWs in Nairobi, Kenya.
Methods
Study Design and Eligibility
Between November through December 2010, an RDS survey was conducted among FSWs in Nairobi, Kenya. Details of RDS methodology are described elsewhere [14, 15]. A target sample size of 600 was calculated. This sample size took into consideration the expected RDS-related design effect of 2.0 [16], an estimated HIV prevalence of 33.5 %, and an expected precision of 6 %. Women, 18 years of age or older, who reported selling sex for money, drugs, or goods to a man at least once in the past 3 months, and who lived in Nairobi or adjacent urban areas were eligible for the survey. Women who were mentally impaired due to alcohol or drug use, previously participated in the survey or presented with an invalid study coupon were excluded from participation. Participants were assessed for eligibility and interviewed at a single, centrally located facility in Nairobi close to Kenyatta National Hospital which was easily accessible by public transportation.
Sampling Method
Six initial FSW participants, also known as “seeds,” were selected to initiate coupon-based recruitment. Seeds were selected purposively to represent the geographical, occupational (e.g., brothel vs. street based), social economic and educational diversity of the target populations. Seeds were identified through formative assessments (e.g., focus group discussions, key informant interviews and in-depth interviews) with key study stakeholders and representatives of different key population groups. Eligible FSWs who provided informed consent to participate were administered an interview and asked to provide blood, urine, vaginal, and anal specimens for HIV and STI testing. Participants were provided up to three coupons and instructed to recruit their FSW peers into the study, who in turn were provided up to three coupons to recruit their FSW peers into the study. FSWs were provided a primary incentive of 200 KSH (i.e., $2USD) to participate and a secondary incentive of 200 KSH (i.e., $2USD) for each eligible peer recruited and enrolled into the study. A coupon manager system was used during the study to track coupons and monitor recruitment. The number of coupons distributed to participants was reduced at times when the number of recruits visiting the study site was too large for study staff capacity. Coupon reduction was also employed in order to balance recruitment when the number of participants was skewed towards certain contingencies of Nairobi. Fingerprint recognition software was used to create a unique study identification number for each participant to assist in preventing the same participant from enrolling more than once. Following informed consent participants were interviewed by a nurse counselor with a structured questionnaire in a private room. The questionnaire included questions on participant socio-demographic characteristics, sex work habits, sexual behaviors, condom use with paying clients and nonpaying partners, history of STI symptoms, alcohol consumption, history of drug use, knowledge of HIV transmission risk, self-perceived risk for HIV infection, HIV testing history, and self-reported HIV status. Correct knowledge of HIV status was assessed by comparing the participant’s self-reported HIV status with the survey HIV test result. Participants who were aware of their status were asked if they were currently taking antiretroviral therapy to treat their infection. Participants were requested to provide blood, vaginal, urine, and anal specimens for HIV and STI testing. Anal specimens were collected only for participants that reported anal sex or had anal symptoms.
Laboratory Methods
Rapid HIV testing was conducted on-site on blood samples using a parallel HIV testing algorithm. All specimens were screened with Determine (Inverness Medical, Massachusetts, USA) and Unigold (Biotech PLC, Ireland) rapid test kits. Specimens that were non-reactive by both rapid tests were classified as HIV-negative. Dually reactive specimens were classified as HIV-positive. All discordant specimens were resolved using Bioline rapid test kit (Standard Diagnostics, INC, South Korea). Specimens that were positive on Bioline were classified as HIV-positive. Specimens that were negative on Bioline were classified as HIV-negative. Participants with HIV-positive results were referred to government clinics for confirmatory testing and further HIV management.
STI testing was conducted at the University of Nairobi Institute of Tropical and Infectious Diseases Laboratory. Testing for syphilis was performed using rapid plasma reagent (RPR) assays (Roche Amplicor CT/NG test) and Treponema palladium hemagglutination assay (TPHA) for confirmatory testing. Detection of T. vaginalis was performed using the In Pouch™ system. Vaginal cultures were evaluated for Bacterial Vaginosis using Nugent’s scoring criteria and for candidiasis the KOH test. Vaginal and rectal swab specimens and urine were tested for using the Polymerase chain reaction (PCR) (Roche Amplicor, Switzerland) assay for detection of C. trachomatis and N. gonorrhea antigens. Detection of the HSV-2 antibody was conducted using an enzyme-linked immunosorbent assay (Kalon Biological Ltd, United Kingdom). At the end of the visit, participants were provided an appointment card and asked to return to the study site after 2 weeks to receive their STI test results with post-test counseling and secondary incentives for those who recruited a FSW that was later enrolled in the study. Participants with positive STIs and/or STI symptoms were provided free treatment in accordance with the Kenya National STI Treatment Guidelines.
Data Management and Analysis
All interview data were entered into handheld Personal Digital Assistant (PDAs) using customized data entry applications with programmed data entry checks to ensure data quality. Data were analyzed using SAS 9.0 and RDS Analysis Tool (RDSAT) version 6.0.1, an analysis package designed to provide population estimates and their 95 % confidence intervals (CI) accounting for differences in participant recruitment patterns, network size, and homophily [15]. Estimates generated through RDSAT were weighted to compensate for bias due to personal network size or any differential sampling of participants. Personal network size was estimated by asking each participant the question: “Approximately, how many other female sex workers do you know by name, who live in or around Nairobi, and you know how to contact them.” The dual-component estimator within RDSAT was used to measure average network size [17]. Ninety-five percent CIs were determined by bootstrap methods using 15,000 re-samples. The RDS dataset was analyzed using the enhanced data smoothing option. To assess associations with the outcome of unrecognized HIV infection, multivariable analyses were conducted using the relative risk (RR) measure. HIV-positive women who were aware of their infection were excluded from the analysis because reported risk behaviors may have differed among participant’s based on knowledge of HIV-infection. Unrecognized infection was defined as a participant who was HIV-positive based on test results from the study but was unaware of her infection (i.e., either reported her status as HIV-negative or never had been tested before). Relative risk and its corresponding 95 % CIs was estimated using log-binomial regression using the GENMOD procedure in SAS. The REPEATED option was used to calculate robust standard errors for parameter estimates. All analyses were weighted using individual HIV weights exported from RDSAT. Variables (or individual categories within variables) that were found to be significant at p = .20 or lower were included in the multivariable analysis. Prevalence and corresponding 95 % CIs of specific STIs and STI symptoms were also generated by RDSAT. Comparison of STI prevalence and STI symptoms by HIV status was assessed using a z-test.
Ethical Considerations
This study protocol was submitted for review and approved by the Kenyatta National Hospital Ethics and Research Committee (KNH-ERC), the Population Council Institutional Review Board (IRB), and to the Associate Director for Science, Division of Global HIV/AIDS of the US Centers for Disease Control and Prevention.
Results
During the 10 waves of recruitment, a total of 1,219 coupons were issued to peers with 632 returned resulting in a coupon return rate of 52 %. Of the six initial seeds, only one did not recruit additional participants. This seed was HIV-negative, aged 25–29 years, married, reported to be at “great risk” for HIV infection, and was recruited from Makadara constituency—a neighborhood that recruited only 3.4 % of the sample (Fig. 1). Of these 632 women who arrived at the facility with coupons, 596 (94 %) FSWs (non-seeds) were eligible for the survey and consented to participate. Of the 36 women excluded, the reasons for exclusion included eligible but did not provide consent (n = 9), not from the study coverage area (n = 7), under 18 years of age (n = 5), under the influence of substances (n = 4), determined to be disruptive during the screening process (n = 3), did not have a coupon (n = 3), previous participation (n = 2), not a sex worker within the study criteria (n = 2), and mentally impaired (n = 1). The median personal network size of participants was 8 (IQR 4–20). Equilibrium was reached on age-group, marital status, education-level, and duration of sex work before the 10 recruitment waves attained in the sample. Although FSW were recruited from all the eight constituencies of Nairobi (i.e., sub-counties of Nairobi which serve as administrative units for local governance: Makadara, Kamukunji, Starehe, Lang’ata, Dagoretti, Westlands, Kasarani, and Embakasi), recruitment was concentrated in the Kamkunji and Kasarani constituencies of Nairobi. Figure 1 illustrates the recruitment pattern of participants by constituency.
Fig. 1.
Respondent driven sampling recruitment profile of female sex workers by Constituency Nairobi, Kenya, 2010
RDS adjusted population estimates of socio-demographic, behavioral characteristics, and HIV prevalence are shown in Table 1. Ages ranged from 18 to 62 years with the median age of 30 years (unweighted IQR 25–38 years). Almost all women were currently unmarried, and 44.3 % had no formal education or attained only an incomplete primary education. Duration of sex work ranged from 4 to 45 years with a median duration of 12 years (IQR 8–17 years). The majority of women (82.2 %) stated that sex work was their main source of income. The most frequent client-seeking venues were bars (76.6 %) and roadsides (29.3 %), and situations where clients called them (20.1 %). The median number of paying clients in the last 7 days was seven (IQR 4–18 clients). Almost two-thirds (62.6 %) used a condom consistently (“always used”) with their paying clients in the past 30 days, and 86.9 % reported using a condom with their last paying client. Most women (59.8 %) did not have a non-paying partner in the past 30 days. Among those who did, 38.6 % consistently used a condom (always used) within the last 30 days with their non-paying partner(s), and 47.5 % reported using a condom at last sex with their non-paying partner. Less than 20 % reported never consuming alcohol; however, 33.4 % reported consuming alcohol 4 or more times per week. Bhang (marijuana) (36.0 %) and Khat/miraa (40.1 %) were the most frequently used drugs in the past. Very few women (*1 %) ever used heroin or injected drugs. Almost three-quarters (73.9 %) of women used at least one contraception method within the past 30 days, with the most common method being injectable contraception (e.g., depot medroxyprogesterone acetate or DMPA) (27.1 %) followed by male condoms (26.6 %), implants (e.g., Norplant) (12.2 %), oral contraception (the pill) (8.2 %), and female condoms (6.7 %). Just over half (51.6 %) of women stated that they practiced vaginal douching. HIV testing was high with 86.6 % of women who had been tested in the past; 63.1 % of these had been tested within the past 12 months. Of participants who tested HIV-positive during the survey, 51.0 % were aware of their infection before the survey, 25.4 % incorrectly reported they were HIV-negative, and 23.0 % were unaware because they had never tested for HIV. The majority of women (59.7 %) had high self-perceived risk for HIV while 14.9% already knew they were infected.
Table 1.
Characteristics and estimates of HIV prevalence among female sex workers, Nairobi, Kenya, 2010 (N = 596)
| Characteristic | na | Unadjusted % (95 % CI) |
Adjusted % (95 % CI) |
No. HIV+ | Unadjusted HIV prevalence (95 % CI) |
Adjusted HIV Prevalence (95 % CI) |
|---|---|---|---|---|---|---|
| HIV prevalence (N = 596) | – | – | – | 183 | 30.7 (27.0, 34.4) | 29.5 (24.7, 34.9)f |
| Age group (years) | ||||||
| 18–24 | 149 | 25.0 (21.5, 28.5) | 30.8 (25.0, 36.7) | 20 | 13.4 (7.9, 18.9) | 14.3 (8.1, 24.7) |
| 25–29 | 133 | 22.3 (19.0, 25.7) | 22.5 (18.2, 27.4) | 30 | 22.6 (15.4, 29.7) | 24.3 (15.0, 32.7) |
| 30–34 | 105 | 17.6 (14.6, 20.7) | 18.9 (14.3, 23.6) | 42 | 40.0 (30.6, 49.4) | 42.0 (30.7, 55.9) |
| 35–62 | 209 | 35.1 (31.2, 38.9) | 27.9 (22.3, 33.6) | 91 | 43.5 (36.8, 50.3) | 39.0 (29.4, 50.2) |
| Median (IQR)b | 30 | (25–38) | ||||
| Marital status | ||||||
| Never married | 225 | 37.8 (33.8, 41.7) | 41.7 (35.9, 46.5) | 55 | 24.4 (18.8, 30.1) | 26.5 (18.7, 35.4) |
| Previously married | 365 | 61.2 (57.3, 65.2) | 57.4 (52.6, 63.2) | 125 | 34.2 (29.4, 39.1) | 31.0 (24.7, 37.6) |
| Currently married | 6 | 1.0 (0.20, 1.8) | 1.0 (0.2, 2.0) | 3 | 50.0 (9.9, 90.1) | 43.6 (0.0, 1.0) |
| Education | ||||||
| None/incomplete primary | 278 | 46.6 (42.6, 50.7) | 44.3 (38.5, 50.1) | 105 | 37.8 (32.0, 43.5) | 37.1 (29.9, 45.8) |
| Completed primary | 169 | 28.4 (24.7, 32.0) | 30.5 (25.4, 36.0) | 40 | 23.7 (17.2, 30.1) | 19.6 (13.0, 27.0) |
| Incomplete secondary | 85 | 14.3 (11.4, 17.1) | 12.8 (9.2, 16.3) | 27 | 31.8 (21.8, 41.7) | 30.9 (16.3, 47.4) |
| Completed secondary+ | 64 | 10.7 (8.2, 13.2) | 12.4 (8.5, 17.3) | 11 | 17.2 (7.9, 26.5) | 21.1 (7.0, 38.9) |
| Duration of sex work (years) | ||||||
| 4–9 | 205 | 34.4 (30.6, 38.2) | 39.1 (33.6, 45.1) | 47 | 22.9 (17.2, 28.7) | 23.5 (16.6, 33.0) |
| 10–14 | 170 | 28.5 (24.9, 32.2) | 27.8 (22.8, 33.0) | 50 | 29.4 (22.5, 36.3) | 28.2 (20.7, 38.9) |
| 15–45 | 221 | 37.1 (33.2, 41.0) | 33.1 (27.4, 38.5) | 86 | 38.9 (32.5, 45.4) | 37.2 (28.1, 47.3) |
| Median (IQR)b | 12 | (8–17) | ||||
| Sex work as main source of income | ||||||
| Yes | 485 | 81.4 (78.2, 84.5) | 82.2 (77.6, 86.3) | 147 | 30.3 (23.7, 41.2) | 28.7 (17.9, 46.7) |
| No | 111 | 18.6 (15.5, 21.8) | 17.8 (13.7, 22.4) | 36 | 32.4 (26.2, 34.4) | 32.9 (24.0, 34.7) |
| Client seeking locationsc | ||||||
| Bar | 441 | 74.0 (70.5, 77.5) | 76.6 (70.7, 82.1) | 116 | 26.3 (22.2, 30.4) | 28.7 (23.9, 34.8) |
| Roadside | 174 | 29.2 (25.5, 32.9) | 29.3 (24.6, 34.7) | 47 | 27.0 (20.4, 33.6) | 22.9 (15.5, 31.1) |
| At home | 82 | 13.8 (11.0, 16.5) | 8.8 (5.6, 12.4) | 40 | 48.8 (37.9, 59.6) | 43.6 (28.3, 63.2) |
| Rented stall or shed | 43 | 7.2 (5.1, 9.3) | 4.1 (2.3, 5.9) | 16 | 37.2 (22.7, 51.7) | 35.6 (16.9, 57.7) |
| Clients called participant | 132 | 22.1 (18.8, 25.5) | 20.1 (15.6, 24.7) | 37 | 28.0 (20.3, 35.7) | 33.2 (19.9, 44.8) |
| Other locations | 47 | 7.9 (5.7, 10.1) | 6.2 (3.5, 9.3) | 10 | 21.3 (9.5, 33.0) | 17.9 (6.0, 35.6) |
| Number of paying partners (past 7 days) | ||||||
| 0–10 | 359 | 60.6 (56.7, 64.6) | 67.7 (62.1, 72.4) | 103 | 28.7 (24.0, 33.4) | 28.9 (22.7, 36.1) |
| 11–20 | 105 | 17.7 (14.7, 20.8) | 14.5 (11.1, 18.0) | 30 | 28.6 (19.9, 37.2) | 27.6 (18.0, 41.8) |
| 21–30 | 78 | 13.2 (10.4, 15.9) | 12.3 (8.9, 16.5) | 26 | 33.3 (22.8, 43.8) | 23.2 (13.1, 36.8) |
| 31+ | 50 | 8.4 (6.2, 10.7) | 5.6 (3.5, 8.2) | 22 | 44.0 (30.2, 57.8) | 31.5 (16.4, 48.9) |
| Median number of paying partners past 7 days (IQR)b | 7 | (4–18) | ||||
| Consistent condom use with paying partners (past 30 days) | ||||||
| Always | 383 | 64.6 (60.7, 68.4) | 62.6 (55.9, 68.3) | 115 | 30.0 (25.4, 34.6) | 26.9 (20.9, 33.9) |
| Sometimes/never | 210 | 35.4 (31.6, 39.3) | 37.4 (31.7, 44.2) | 67 | 31.9 (25.6, 38.2) | 33.0 (25.4, 42.6) |
| Condom use with last paying partner | ||||||
| Yes | 513 | 86.2 (83.4, 90.0) | 86.9 (82.8, 90.2) | 161 | 31.4 (27.4, 35.4) | 30.6 (25.2, 36.7) |
| No | 82 | 13.8 (11.0, 16.6) | 13.1 (9.7, 17.3) | 22 | 26.8 (17.2, 36.4) | 24.1 (13.0, 36.6) |
| Number of non-paying sex partner(s) (past 30 days) | ||||||
| None | 336 | 56.4 (52.4, 60.4) | 59.8 (54.6, 65.6) | 112 | 33.3 (28.3, 38.4) | 30.4 (24.0, 37.0) |
| 1 | 181 | 30.4 (26.7, 34.1) | 27.4 (22.1, 32.2) | 48 | 26.5 (20.1, 33.0) | 28.2 (18.0, 39.3) |
| 2+ | 79 | 13.3 (10.5, 16.0) | 12.9 (9.4, 16.6) | 23 | 29.1 (19.1, 39.2) | 27.0 (15.5, 40.6) |
| Consistent condom use with nonpaying partner(s) (past 30 days) (N = 260) | ||||||
| Always | 94 | 36.2 (30.3, 42.0) | 38.6 (21.6, 44.4) | 35 | 37.2 (27.4, 47.1) | 33.6 (22.0, 70.0) |
| Sometimes/never | 166 | 63.8 (58.0, 69.7) | 61.4 (55.6, 78.4) | 36 | 21.7 (15.3, 28.0) | 22.4 (7.8, 27.7) |
| Condom use with last nonpaying partner (N = 260) | ||||||
| Yes | 125 | 48.1 (42.0, 54.2) | 47.5 (29.3, 53.1) | 41 | 32.8 (24.5, 41.1) | 34.0 (25.4, 37.5) |
| No | 135 | 51.9 (45.8, 58.0) | 52.5 (46.9, 70.7) | 30 | 22.2 (15.2, 29.3) | 17.7 (6.1, 27.1) |
| Alcohol consumption | ||||||
| Never | 111 | 18.6 (15.5, 21.8) | 19.4 (14.6, 24.5) | 40 | 36.0 (27.1, 45.0) | 33.1 (24.6, 47.2) |
| Once a month or less | 62 | 10.4 (7.9, 12.9) | 8.0 (6.0, 12.8) | 28 | 45.2 (32.7, 57.6) | 37.2 (20.7, 55.9) |
| 2–4 times a month | 56 | 9.4 (7.0, 11.7) | 10.9 (7.4, 14.3) | 18 | 32.1 (19.9, 44.4) | 34.7 (16.8, 50.8) |
| 2–3 times a week | 144 | 24.2 (20.7, 27.6) | 27.4 (22.6, 33.8) | 36 | 25.0 (17.9, 32.1) | 24.0 (17.4, 36.1) |
| 4 or more times a week | 223 | 37.4 (33.5, 41.3) | 33.4 (27.3, 38.4) | 61 | 27.4 (21.4, 33.2) | 28.6 (19.4, 36.9) |
| Drug use (past 12-months)c,d | ||||||
| Marijuana/bhang | 246 | 41.3 (37.3, 45.2) | 36.0 (31.1, 42.1) | 68 | 27.6 (22.0, 33.2) | 24.3 (18.9, 32.2) |
| Khat/miraa | 263 | 44.1 (40.1, 48.1) | 40.1 (34.1, 45.7) | 74 | 28.1 (22.7, 33.6) | 28.7 (21.1, 37.3) |
| Heroin | 9 | 1.5 (0.52, 2.5) | 1.1 (0.4, 2.0) | 3 | 33.3 (2.4, 64.2) | 50.8 (0.0, 81.3) |
| Injection drug use | 9 | 1.5 (0.53, 2.5 | 1.0 (0.3, 1.9) | 1 | 11.1 (0.0, 31.7) | 22.6 (0.0, 54.6) |
| Contraception method (past 30 days)c | ||||||
| Any contraception used | 434 | 72.8 (69.2, 76.4) | 73.9 (69.0, 78.3) | 135 | 31.0 (22.6, 36.7) | 30.2 (24.2, 36.4) |
| Injection | 152 | 25.5 (22.0, 29.0) | 27.1 (21.9, 32.3) | 36 | 23.7 (16.9, 30.5) | 26.6 (18.6, 36.8) |
| Male condom | 167 | 28.0 (24.4, 31.6) | 26.6 (20.1, 31.0) | 68 | 40.7 (33.3, 48.2) | 44.6 (32.7, 54.8) |
| Implant | 63 | 10.6 (8.1, 13.0) | 12.2 (8.1, 16.7) | 19 | 30.2 (18.8, 41.5) | 17.0 (9.1, 29.4) |
| Pill | 57 | 9.6 (7.2, 11.9) | 8.2 (5.9, 11.0) | 10 | 17.5 (7.6, 27.4) | 10.7 (3.5, 19.5) |
| Female condom | 39 | 6.5 (4.6, 8.5) | 6.7 (4.1, 9.6) | 15 | 38.5 (26.3, 34.0) | 41.8 (22.4, 65.5) |
| Practice douching | ||||||
| Yes | 305 | 51.2 (47.1, 55.2) | 51.6 (46.1, 57.1) | 98 | 32.1 (26.9, 37.4) | 31.6 (24.8, 39.2) |
| No | 291 | 48.8 (44.9, 52.9) | 48.4 (42.9, 53.9) | 85 | 29.2 (24.0, 34.4) | 27.4 (20.4, 34.3) |
| Tested for HIV | ||||||
| Yes, tested before | 532 | 89.3 (86.8, 91.8) | 86.6 (81.9, 90.2) | 154 | 28.9 (25.1, 32.8) | 25.9 (20.4, 31.3) |
| Never tested | 64 | 10.7 (8.2, 13.2) | 13.4 (9.8, 18.1) | 29 | 45.3 (33.1, 57.6) | 47.3 (33.1, 65.2) |
| Time since last HIV test (among those tested) (N = 532) | ||||||
| <12 months | 341 | 64.1 (60.0, 68.2) | 63.1 (56.1, 68.4) | 66 | 19.4 (15.1, 23.6) | 17.1 (13.0, 23.6) |
| 12+ months | 181 | 34.0 (30.0, 38.1) | 34.8 (29.7, 41.9) | 84 | 46.4 (39.1, 53.7) | 45.3 (34.5, 58.4) |
| Unknown | 10 | 1.9 (0.72, 3.0) | 2.1 (0.6, 3.6) | 4 | 40.0 (9.5, 70.5) | 32.1 (2.4, 85.9) |
| Correct knowledge of HIV statusd | ||||||
| HIV-positive (N = 181) | ||||||
| Aware, reported HIV-positive | 116 | 64.1 (57.0, 71.1) | 51.0 (40.6, 63.1) | – | – | |
| Unaware of infection, reported negative | 32 | 17.7 (12.1, 23.3) | 25.4 (14.8, 35.3) | – | – | |
| Unaware of infection, never tested | 33 | 18.2 (12.6, 23.9) | 23.0 (14.4, 34.2) | – | – | |
| HIV-negative (N = 409) | ||||||
| Reported HIV-negative | 362 | 88.5 (85.4, 91.6) | 88.1 (84.0, 92.2) | – | – | |
| Unaware of status, never tested | 47 | 11.5 (8.4, 14.6) | 11.9 (7.8, 16.0) | – | – | |
| ARV use by knowledge of HIV-positive status (N = 178)e | ||||||
| Aware, Taking ARVs | 59 | 33.1 (26.2, 40.1) | 30.9 (15.2, 54.5) | – | – | |
| Aware, Not taking ARVs | 54 | 30.3 (23.5, 37.2) | 31.0 (8.4, 41.4) | – | – | |
| Unaware | 65 | 36.5 (29.4, 43.7) | 38.0 (21.8, 66.6) | – | – | |
| Perception of HIV risk | ||||||
| No risk | 26 | 4.4 (2.7, 6.0) | 3.7 (2.2, 5.7) | 3 | 11.5 (0.0, 23.9) | 7.6 (0.0, 17.4) |
| Small or moderate risk | 105 | 17.6 (14.6, 20.7) | 18.1 (13.6, 22.6) | 7 | 6.7 (1.9, 11.5) | 6.5 (1.4, 13.3) |
| Great risk | 333 | 55.9 (51.9, 59.9) | 59.7 (53.8, 62.2) | 53 | 15.9 (12.0, 20.0) | 20.7 (14.1, 27.5) |
| Don’t know risk | 16 | 2.7 (1.4, 4.0) | 3.7 (1.6, 5.6) | 4 | 25.0 (3.7, 46.3) | 27.2 (0.0, 56.7) |
| Already knew HIV infected | 116 | 19.5(16.2, 22.7) | 14.9 (11.3, 19.3) | 116 | 100 (–) | 100 (–) |
The denominator for unadjusted percentages is 596 unless otherwise noted
Median and IQR are unadjusted
Respondents could have selected several client seeking locations
Six observations were excluded because the participant refused to disclose their HIV status and therefore correct knowledge of HIV status could not be classified
Five observations were excluded: three because the participant refused to disclose their HIV status and therefore knowledge of HIV status. could not be classified, and two refused to disclose ARV usage
The design effect (DEFF) for HIV prevalence was 1.94
Table 1 also shows HIV prevalence by selected demographic and behavioral characteristics. Overall, the HIV prevalence among FSWs in this study was 29.5 % (95 % CI 24.7, 34.9). HIV prevalence increased with age and duration of time as a sex worker and was highest among women with no or incomplete primary education. Prevalence also varied by client-seeking locations. The highest HIV prevalence was observed among women who received clients in their home [43.6 % (95 % CI 28.3, 63.2)] whereas the lowest prevalence was among women seeking clients at ‘other’ locations such as brothels, hotels, massage parlors [17.9 % (95 % CI 6.0, 35.6)]. Prevalence was similar with overlapping confidence intervals across categories of number of paying clients in past 7 days, consistent condom use with paying clients, and condom use with last paying client. Prevalence was slightly higher for women who consistently used condoms with nonpaying partners (if any) and who used a condom with their last nonpaying partner (if any). Women who used a barrier method (male and/or female condom) as a form of contraception in the past 30 days had high HIV prevalence [male condom: 44.6 % (95 % CI 32.7, 54.8), female condom: 41.8 % (95 % CI 22.4, 65.5)]. Conversely, women who used an oral contraception (“Pill”) as a method of contraception had lower HIV prevalence [10.7 % (95 % CI 3.5, 19.5)].
Table 2 shows prevalence of possible STI symptoms and STI biomarkers, and a comparison of STI biomarkers by HIV test results. The most frequent symptom was abnormal vaginal discharge (23.9 %), followed by burning pain during urination (22.7 %), vaginal ulcer or sore (7.8 %), and anal ulcer or sore (1.0 %). Significantly higher prevalence of abnormal vaginal discharge and vaginal ulcer or sore was observed among HIV-infected women (abnormal vaginal discharge: z-score = 3.02, p = .0025; vaginal ulcer or sore: z-score = 2.39, p = .0167). The prevalence of biomarkers for syphilis, gonorrhea, and chlamydia infection was low (<5 %) among the population as a whole, as well as HIV-infected women. Prevalence of trichomaniasis, bacterial vaginosis, and candidiasis ranged from 10 to 28 % with slightly higher STI rates for HIV-infected women than HIV-uninfected women. Only trichomaniasis was significantly higher among HIV-infected women compared with HIV-uninfected women (z-score: 2.18, p = .0293).
Table 2.
Prevalence of STI symptoms and diagnoses among female sex workers, Nairobi, Kenya, 2010 (N = 596)
| Variable | All participants (N = 596) |
HIV-negative participants (N = 413) |
HIV-positive participants (N = 183) |
p value | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| n | Unadjusted % (95 % CI) |
Adjusted % (95 % CI) |
n | Unadjusted % (95 % CI) |
Adjusted % (95 % CI) |
n | Unadjusted % (95 % CI) |
Adjusted % (95 % CI) |
||
| STI symptoms | ||||||||||
| Abnormal vaginal discharge | 171 | 28.7 (25.0, 32.3) | 23.9 (19.5, 28.8) | 98 | 23.7 (19.6, 27.8) | 18.9 (14.7, 24.1) | 73 | 39.9 (32.8, 47.0) | 36.2 (27.2, 47.6) | .0025 |
| Vaginal ulcer or sore | 52 | 8.7 (6.5, 11.0) | 7.8 (5.1, 10.7) | 27 | 6.5 (4.1, 8.9) | 5.2 (2.9, 8.4) | 25 | 13.7 (8.7, 18.7) | 14.1 (7.4, 20.9) | .0167 |
| Anal ulcer or sore | 7 | 1.2 (0.3, 2.0) | 1.0 (0.2, 2.1) | 4 | 1.0 (0.0, 1.9) | 0.4 (0.0, 10.0) | 3 | 1.6 (0.0, 3.5) | 2.3 (0.0, 5.6) | .5158 |
| Burning pain on urination | 152 | 25.5 (22.0, 29.0) | 22.7 (18.5, 27.3) | 92 | 23.3(0.0, 3.5) | 20.0 (15.3, 25.6) | 60 | 32.8 (26.0, 39.6) | 29.6 (21.5, 40.8) | .0854 |
| Prevalence of STI | ||||||||||
| Syphilis | 5 | 0.8 (0.1, 1.6) | 0.9 (0.2, 1.9) | 2 | 0.5 (0.0, 1.2) | 0.8 (0.0, 2.1) | 3 | 1.6 (0.0, 3.5) | 1.3 (0.1, 3.5) | .5483 |
| Gonorrheaa | 13 | 2.3 (1.0, 3.5) | 1.1 (0.4, 2.1) | 6 | 1.5 (0.31, 2.7) | 0.5 (0.1, 1.0) | 7 | 4.0 (1.1, 6.8) | 2.9 (0.7, 6.0) | .0811 |
| Chlamydiaa | 16 | 2.8 (1.4, 4.1) | 3.1 (1.5, 5.3) | 13 | 3.3 (1.5, 5.0) | 3.9 (1.6, 6.6) | 3 | 1.7 (0.0, 3.6) | 1.1 (0.0, 4.0) | .0851 |
| Trichomaniasisb | 49 | 8.5 (6.2, 10.8) | 10.3 (6.9, 14.5) | 26 | 6.5 (4.1, 9.0) | 7.9 (4.6, 12.8) | 23 | 12.9 (8.0, 17.9) | 17.8 (8.8, 24.6) | .0293 |
| Bacterial vaginosisc | 95 | 16.4 (13.4, 19.5) | 15.1 (11.5, 19.3) | 57 | 14.3 (10.8, 17.7) | 12.3 (8.7, 16.4) | 38 | 21.4 (15.3, 27.4) | 21.4 (13.4, 32.0) | .0764 |
| Candidiasisc | 147 | 25.4 (21.9, 29.0) | 28.4 (23.3, 33.5) | 103 | 25.8 (21.5, 30.0) | 28.2 (22.0, 34.5) | 44 | 24.7 (18.4, 31.1) | 28.5 (19.1, 38.0) | .9586 |
| Positive/reactive for any STI biomarker (except HIV)d |
205 | 34.4 (30.6, 38.2) | 38.0 (33.0, 43.6) | 135 | 32.7 (28.2, 37.2) | 36.4 (29.8, 42.9) | 70 | 38.3 (31.2, 45.3) | 42.9 (32.9, 53.5) | .2966 |
21 participants excluded because of missing STI results
20 participants excluded because of missing STI results
18 participants excluded because of missing STI results
22 respondents excluded because missing one or more STI test results
Table 3 shows results of bivariate and multivariable analysis identifying candidate variables and significant predictors of unrecognized HIV infection. In bivariate analysis, increasing age, receiving clients at home, inconsistent condom use with paying clients, and use of a male condom as a method of contraception were positively associated with unrecognized HIV infection. Conversely, completing primary education, inconsistent condom use with nonpaying clients and use of the ‘implant’ as a method of contraception were negatively associated with unrecognized infection. Multivariable analysis revealed that increasing age, inconsistent condom use with a paying client [adjusted RR = 2.1 (95 % CI 1.4, 3.0), p = .0001], and male condom use as a method of contraception [adjusted RR = 2.5 (95 % CI 1.7, 3.5), p < .0001] were associated with unrecognized HIV infection.
Table 3.
Bivariate and multivariable log-binomial analysis of predictors of unrecognized HIV infection among female sex workers, Nairobi, Kenya, 2010 (N = 480)
| Variable | Unadjusted RR (95 % CI) | p value | Adjusted RR (95 % CI)a | p value |
|---|---|---|---|---|
| Age group (years) | ||||
| 18–24 | Referent | – | Referent | – |
| 25–29 | 1.6 (0.9, 3.1) | .1287 | 1.5 (0.8, 2.9) | .1808 |
| 30–34 | 2.9 (1.6, 5.2) | .0005 | 2.4 (1.3, 4.3) | .0030 |
| 35–62 | 2.3 (1.3, 4.1) | .0045 | 2.1 (1.2, 3.7) | .0124 |
| Marital status | ||||
| Never married | Referent | – | ||
| Previously married | 0.9 (0.6, 1.3) | .4210 | ||
| Currently married | 1.3 (0.3, 6.4) | .7720 | ||
| Education | ||||
| None/incomplete primary | Referent | – | ||
| Completed primary | 0.5 (0.3, 0.8) | .0088 | ||
| Incomplete secondary | 0.8 (0.5, 1.5) | .5638 | ||
| Completed secondary+ | 0.7 (0.4, 1.3) | .2301 | ||
| Duration of sex work (years) | ||||
| 4–9 | Referent | – | ||
| 10–14 | 1.0 (0.6, 1.7) | .9705 | ||
| 15–45 | 1.3 (0.9, 2.1) | .1997 | ||
| Sex as a main source of income | ||||
| No | Referent | – | ||
| Yes | 0.8 (0.5, 1.4) | .4880 | ||
| Client-seeking locationsb,d | ||||
| Bar | 1.0 (0.6, 1.5) | .8992 | ||
| Roadside | 0.7 (0.4, 1.1) | .1448 | ||
| At home | 1.8 (1.0, 2.9) | .0299 | ||
| Rented stall or shed | 1.2 (0.67, 2.1) | .5430 | ||
| Clients called participant | 1.3 (0.8, 2.0) | .3165 | ||
| Other locations | 0.8 (0.4, 1.9) | .6605 | ||
| Number of paying partners (past 7 days) | ||||
| 0–10 | Referent | – | ||
| 11–20 | 0.9 (0.5, 1.6) | .7278 | ||
| 21–30 | 0.6 (0.3, 1.2) | .1514 | ||
| 31+ | 0.2 (0.02, 1.5) | .1119 | ||
| Consistent condom use with paying partners (past 30 days) | ||||
| Always | Referent | – | Referent | – |
| Sometimes/never | 1.9 (1.3, 2.8) | .0015 | 2.1 (1.4, 3.0) | .0001 |
| Condom use with last paying partner | ||||
| Yes | Referent | – | ||
| No | 0.8 (0.4, 1.4) | .3788 | ||
| Number of non-paying sex partner(s) (past 30 days) | ||||
| None | Referent | – | ||
| 1 | 0.6 (0.4, 1.0) | .0613 | ||
| 2+ | 1.0 (0.6, 1.8) | .9442 | ||
| Consistent condom use with nonpaying partner(s) (past 30 days) | ||||
| Always | Referent | – | ||
| Sometimes/Never | 0.7 (0.4, 1.1) | .1380 | ||
| Condom use with last nonpaying partner | ||||
| Yes | Referent | – | ||
| No | 0.5 (0.2, 0.9) | .0287 | ||
| Drug use (past 12 months)c,d | ||||
| Marijuana/bhang | 0.8 (0.5, 1.2) | .1981 | ||
| Khat/miraa | 0.8 (0.6, 1.3) | .3871 | ||
| Heroin | 1.3 (0.2, 6.8) | .7721 | ||
| Injection drug use | 1.2 (0.2, 5.9) | .8899 | ||
| Alcohol consumption | ||||
| Never | Referent | – | ||
| Once a month or less | 1.9 (0.9, 3.9) | .0973 | ||
| 2–4 times a month | 1.1 (0.4, 2.5) | .8879 | ||
| 2–3 times a week | 1.3 (0.7, 2.4) | .4658 | ||
| 4 or more times a week | 1.3 (0.7, 2.4) | .4037 | ||
| Contraception method (past 30 days)c | ||||
| Any contraception used | 1.2 (0.8, 1.9) | .4328 | ||
| Injection | 0.8 (0.5, 1.3) | .3901 | ||
| Male condom | 2.3 (1.6, 3.3) | <.0001 | 2.5 (1.7, 3.5) | <.0001 |
| Implant | 0.5 (0.1, 0.9) | .0322 | ||
| Pill | 0.3 (0.1, 1.0) | .0510 | ||
| Female condom | 0.5 (0.1, 1.8) | .2686 | ||
| Practice douching | ||||
| No | Referent | – | ||
| Yes | 1.1 (0.9, 1.2) | .1315 | ||
Adjusted for all variables or individual categories within variables that were found to be significant at p ≤ .20 or lower in bivariate analysis. Variables found to be significant at the p ≤ .05 are displayed
Respondents could have selected several client-seeking locations
Respondents could have used selected several drugs in the past 12 months
Referent group for multi-response variables were women who did not report the response
Discussion
We confirm high HIV prevalence among FSWs in Nairobi where approximately one-third are infected with HIV. This level of HIV infection is nearly three times that of Nairobi women aged 15–49 years from two national population-based HIV serologic surveys conducted in Kenya in 2007 and 2008/09 [18, 19]. HIV prevalence among FSWs aged 18–24 years was also high at 14 %, highlighting the vulnerability of acquisition and transmission of HIV infection even among young FSWs. The high prevalence among young FSWs requires immediate interventions to encourage young girls and women to adopt early risk reduction strategies to prevent HIV infection and remain uninfected. We found that increasing age was associated with HIV infection, confirming that age increases the cumulative risk for HIV infection in this population. Moreover, inconsistent condom use with paying partners and reported use of male condoms as a contraceptive in the past 30 days were associated with HIV infection. Though condom use is a practical approach for minimizing transmission and acquisition of HIV infection among FSWs, prevention messages around correct and consistent use of condoms during sexual intercourse are critical for ensuring effectiveness of this tool for HIV prevention.
A second finding from this study is that HIV testing among FSWs is high, with testing rates more than two times that of persons in the general population [18]. After excluding women with known HIV-positive status, 84 % had a previous HIV test, and more than 50 % of those tested had been tested in the past 12 months. The high level of self-perceived risk among FSWs, together with the expansion of targeted HIV testing services for FSWs in Nairobi, may account for high testing rates in this population. As a direct benefit of HIV testing, over half of HIV-positive FSW were aware of their HIV infection compared with 18 % of HIV-infected women in the general population [19]. Still, despite high awareness of HIV status, a large percentage of HIV-infected FSW remained unaware of their HIV infection, and almost half incorrectly believed they were HIV-negative based on their last HIV test. As a result, FSWs continue to pose high risk of transmission to both their paying and non-paying partners and experience delays in accessing important treatment and care interventions. Some FSWs may have known their HIV-positive status at the time of the survey, but did not feel comfortable in reporting this during the interview. In Kenya, there is an extensive network of service delivery points that target FSWs which provide HIV testing and counseling services [20]. These services include innovative HIV testing and counseling strategies such as mobile testing, door-to-door testing, moonlight testing (i.e., where services are provided during times and in locations convenient to key populations), and provider-initiated testing and counseling services. These programs, while beneficial, should also undergo routine evaluation to assess reach and impact.
A third finding observed in this study is differential condom use patterns among participants. Measures of consistent condom use (“always used”) and condom use at last sex were higher with paying partners (62.3 and 86.8 %, respectively) than with non-paying partners (37.4 and 46.4 %, respectively). Differences in condom use by partner type have been noted in other studies of FSWs and other female populations [8, 10, 21–23]. Of concern is that unprotected sex is likely to occur with non-paying partners, increasing the risk of secondary infection (and other STIs) to themselves if uninfected, or to their sexual partners if infected. More information is needed to determine factors associated with condom use that may explain differences in condom utilization by partner type. It has been noted that behavioral changes that include increased condom use with partners, can be successful among FSWs in Kenya [22, 23], and in general, other low- and middle income countries [24, 25]. Besides increased condom use by partners, other approaches that place more emphasis on female-controlled prevention strategies, such as the use of female condoms, intravaginal products such as microbicides, or oral pre-exposure prophylaxis should be explored [10].
Finally, this study found low prevalence of syphilis %), gonorrhea (1.1 %), and chlamydia (3.1 %) infections, but higher levels of trichomaniasis (10.3 %), bacterial vaginosis (15.1 %), and candidiasis (28.4 %). Our population-based estimate of syphilis prevalence is similar to findings from another study of Nairobi FSWs which found that FSWs had higher syphilis prevalence than the general female population in Kenya (2.5 vs. 1.7 %, respectively) [10, 18]. However, our study found lower prevalence of other STIs than other studies conducted in Kenya and elsewhere [8–10, 25–28]. Low rates of STI among FSWs in Nairobi may highlight the availability and successful access of services through targeted programs for FSWs in the city, including drop-in centers, outreach services, and peed education programs. Programs that target FSWs should continue to be supported and provide a package of prevention interventions, including HIV testing, condom promotion, syndromic management of STIs and education on how to prevent HIV transmission and acquisition.
There were several limitations of this study. Selfreported behaviors may have been mis-reported because of recall or social desirability bias. Although FSWs from all constituencies were eligible to participate, over 50 % of the sample came from only two constituencies: Kamkunji and Kasarani. In addition, we observed high degree of homophily for constituency (i.e. high preference for recruiting in one’s own constituency), suggesting that the sample may not have formed an unbiased social network. Therefore, the sample drawn for this survey may not be generalizable to the broader population of FSWs in Nairobi. In addition, RDS is not a simple random sample, but subject to assumptions, such as quantifying the size of participant networks, which may be difficult to verify. As such, there may be limitations of the RDS method to approximate representative samples of hidden populations.
In spite of these limitations, this study demonstrated that surveillance for FSWs using RDS methodology is feasible in Nairobi, Kenya. Given the high prevalence of HIV infection in this population and the public health concern of potential for HIV transmission among FSWs and their partners, we recommend continued periodic surveillance among FSW in Nairobi, and expansion to other areas in Kenya, to monitor trends of HIV and behavioral risk factors for HIV to appropriately respond to the HIV epidemic in this vulnerable group.
Acknowledgments
The authors acknowledge Nyawira N Njeri, Soloman Kamuti, Gloria Gakii, George Rutherford from the University of California San Francisco, Mary Lou Lindegren from Vanderbilt University, and the study participants for their valuable contributions. The authors also wish to thank the National HIV and STI Control Programme (NASCOP), the National AIDS Control Council, the University of Manitoba’s Sex Worker Outreach Programme (SWOP), Liverpool VCT, the University of Nairobi Institute of Tropical and Infectious Diseases (UNITID) Laboratory, the Mombasa HIV/STD Research Laboratory, the University of Washington, the Gay and Lesbian Coalition of Kenya (GALCK), and the participants for their respective supportive roles in the successful completion of this research. This study was funded by the President’s Emergency Plan for AIDS Relief through The Centers for Disease Control and Prevention (CDC), through the Population Council’s cooperative agreement of Award No. 5U62PS224506.
Footnotes
Disclaimer The findings and conclusions in this paper are those of the authors and do not necessarily represent those of the US Centers for Disease Control and Prevention or the Government of Kenya.
Contributor Information
Helgar Musyoki, Email: helgar@nascop.or.ke, National AIDS and Sexually Transmitted Infection Control Programme, Ministry of Health, P.O. Box 19361-00202, Nairobi, Kenya.
Timothy A. Kellogg, Global Health Sciences, University of California, San Francisco, CA, USA
Scott Geibel, Population Council, Nairobi, Kenya.
Nicholas Muraguri, National AIDS and Sexually Transmitted Infection Control Programme, Ministry of Health, P.O. Box 19361-00202, Nairobi, Kenya.
Jerry Okal, Population Council, Nairobi, Kenya.
Waimar Tun, Population Council, New York, NY, USA.
H. Fisher Raymond, Global Health Sciences, University of California, San Francisco, CA, USA.
Sufia Dadabhai, Global Health Sciences, University of California, San Francisco, CA, USA.
Meredith Sheehy, Population Council, New York, NY, USA.
Andrea A. Kim, Division of Global HIV/AIDS (DGHA), Center for Global Health, US Centers for Disease Control and Prevention, Nairobi, Kenya
References
- 1.Joint United Nations Programme on HIV/AIDS (UNAIDS) UNAIDS guidance note on HIV and sex work. UNA-IDS; Geneva: 2012. [Google Scholar]
- 2.UNAIDS . Global HIV/AIDS response: epidemic update and health sector progress towards universal access. UNA-IDS; Geneva: 2011. [Google Scholar]
- 3.UNAIDS . AIDS epidemic update: December 2009. Vol. 20. UNAIDS; Genevar: 2009. 2012. http://data.unaids.org/pub/Report/2009/JC1700/Epi/Update/2009.en.pdf. Accessed on January. [Google Scholar]
- 4.Baral S, Beyrer C, Muessig K, et al. Burden of HIV among female sex workers in low-income and middle-income countries: a systematic review and meta-analysis. Lancet Infect Dis. 2012;12:538–49. doi: 10.1016/S1473-3099(12)70066-X. [DOI] [PubMed] [Google Scholar]
- 5.Mathers BM, Degenhardt L, Phillips B, et al. Global epidemiology of injecting drug use and HIV among people who inject drugs: a systematic review. The Lancet. 2008;372(9651):1733–45. doi: 10.1016/S0140-6736(08)61311-2. [DOI] [PubMed] [Google Scholar]
- 6.Kimani J, Kaul R, Nagelkerke NJD, Luo M, et al. Reduced rates of HIV acquisition during unprotected sex by Kenyan female sex workers predating population declines in HIV prevalence. AIDS. 2008;22:131–7. doi: 10.1097/QAD.0b013e3282f27035. [DOI] [PubMed] [Google Scholar]
- 7.Fonck K, Kaul R, Keli F, et al. Sexual transmitted infections and vaginal douching in a population of female sex workers in Nairobi. Kenya. Sex Transm Infect. 2001;77:271–5. doi: 10.1136/sti.77.4.271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Hawken MP, Melis RDJ, Ngombo DT, Mandaliya K, et al. Part time female sex workers in a suburban community in Kenya: a vulnerable hidden population. Sex Transm Infect. 2002;78:271–3. doi: 10.1136/sti.78.4.271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Priddy FH, Wakasiaka S, Hoang TD, et al. Anal sex, vaginal practices, and HIV Incidence in female sex Workers in Urban Kenya: implications for the development of intravaginal HIV prevention methods. AIDS Res Hum Retroviruses. 2011;27(10):1067–72. doi: 10.1089/aid.2010.0362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Magnani R, Sabin K, Saidel T, Heckathorn D. Review of sampling hard-to-reach and hidden populations for HIV surveillance. AIDS. 2005;19(Suppl 2):S67–72. doi: 10.1097/01.aids.0000172879.20628.e1. [DOI] [PubMed] [Google Scholar]
- 11.Johnston LG, Malekinejad M, Rifkin MR, Rutherford GW, Kendall C. Implementation challenges to using respondent-driven sampling methodology for HIV biological and behavioral surveillance: field experiences in international settings. AIDS Behav. 2009;12(Suppl 1):131–41. doi: 10.1007/s10461-008-9413-1. [DOI] [PubMed] [Google Scholar]
- 12.Heckathorn D. Respondent-driven sampling: a new approach to the study of hidden populations. Soc Probl. 1997;44(2):174–99. [Google Scholar]
- 13.Goel S, Salganik MJ. Assessing respondent-driven sampling. Proc Natl Acad Sci USA. 2010;107(15):6743–7. doi: 10.1073/pnas.1000261107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Heckathorn D. Respondent-driven sampling II: deriving valid population estimates from chain-referral samples of hidden populations. Soc Probl. 2002;49(14):11–34. [Google Scholar]
- 15.Heckathorn DD. Extensions of respondent-driven sampling: analyzing continuous variables and controlling for differential recruitment. Sociol Methodol. 2007;37(1):151–207. [Google Scholar]
- 16.Salganik MJ. Variance estimation, design effects, and sample size calculations for respondent-driven sampling. J Urban Health. 2006;83(6S):i98–112. doi: 10.1007/s11524-006-9106-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.National AIDS/STI Control Programme (NASCOP) Kenya AIDS indicator survey: final report. NASCOP; Nairobi: 2007. 2009. [Google Scholar]
- 18.Kenya National Bureau of Statistics (KNBS) and ICF Macro . Kenya demographic and health survey 2008–09. KNBS and ICF Macro; Calverton, MA: 2010. [Google Scholar]
- 19.NASCOP . National guidelines for HIV/STI programs for sex workers. NASCOP; Nairobi: 2010. [Google Scholar]
- 20.McClelland L, Wanje G, Kashonga F, et al. Understanding the context of HIV risk behavior among HIV-positive and HIV-negative female sex workers and male bar clients following antiretroviral therapy rollout in Mombasa, Kenya. AIDS Educ Prev. 2011;23(4):299–312. doi: 10.1521/aeap.2011.23.4.299. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Ngugi EN, Chakkalackal M, Sharma A, et al. Sustained changes in sexual behavior by female sex workers after completion of a randomized HIV prevention trial. JAIDS. 2007;45:588–94. doi: 10.1097/QAI.0b013e3180f616db. [DOI] [PubMed] [Google Scholar]
- 22.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 enrollment in an HIV-1 prevention trial. J Acquir Immune Defic Syndr. 2005;38:329–34. [PubMed] [Google Scholar]
- 23.Wariki WMV, Ota E, Mori R, et al. Behavioral interventions to reduce the transmission of HIV infection among sex workers and their clients in low- and middle-income countries. Cochrane Database Syst Rev. 2012 doi: 10.1002/14651858.CD005272.pub3. doi:10.1002/14651858.CD005272.pub3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Foss AM, Hossain M, Vickerman PT, Watts CH. A systematic review of published evidence on intervention impact on condom use in sub-Saharan Africa and Asia. Sex Transm Infect. 2007;83:510–6. doi: 10.1136/sti.2007.027144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Fonck K, Kidula N, Kirui P, et al. Pattern of sexually transmitted diseases and risk factors among women attending an STD referral clinic in Nairobi, Kenya. Sex Transm Dis. 2000;7:417–23. doi: 10.1097/00007435-200008000-00007. [DOI] [PubMed] [Google Scholar]
- 26.McClelland RS, Richradson BA, Hassan WM, Chohan V, et al. Improvement of vaginal health for Kenyan women at risk for acquisition of human immunodeficiency virus type 1: results of a randomized trial. JID. 2008;197:1361–8. doi: 10.1086/587490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Luchters S, Chersich MF, Rinyiru A, et al. Impact of five years of peer-mediated interventions on sexual behavior and sexually transmitted infections among female sex workers in Mombasa, Kenya. BMC Public Health. 2008;8:143. doi: 10.1186/1471-2458-8-143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Kaul R, Kimani J, Nagelkerke NJ, 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(21):2555–62. doi: 10.1001/jama.291.21.2555. [DOI] [PubMed] [Google Scholar]