For human immunodeficiency virus (HIV) prevention planning, it is important to know in which populations new HIV infections are occurring. Although trends in and current levels of HIV prevalence in Rwanda are fairly well documented, recent information on HIV incidence in the general and high-risk populations in Rwanda is lacking.1 A national HIV seroprevalence survey conducted in Rwanda in 2005 documented a 3% HIV prevalence among adults in the general population.2 Survey data also revealed marked differences in prevalence by region, age, and gender, including higher prevalence in urban areas generally, and 8.0% prevalence among women in Kigali. General population data, however, may be concealing more severe and enduring epidemics among sexually high-risk subpopulations, such as female sex workers (FSW). Studies have documented high HIV prevalence and incidence among FSW in sub-Saharan Africa.3–6
We conducted a prospective observational HIV incidence study among FSW at Projet Ubuzima in Kigali, Rwanda. Projet Ubuzima is a nonprofit organization for medical research located in Kigali, with an onsite research clinic and laboratory. The primary objective of the study was to estimate HIV incidence for the design of future HIV prevention interventions. In this article, we report on the baseline prevalence of HIV and other sexually transmitted infections (STI), HIV incidence rates (IR) over time, participant accrual and retention rates, and incidence of pregnancy and STI.
MATERIALS AND METHODS
Recruitment and Screening
FSW were recruited through community meetings in 3 Kigali districts, after approval from local authorities. Meetings were arranged and advertised by community mobilizers in conjunction with leaders of local associations for women engaged in sex work, and then led by Projet Ubuzima outreach staff with assistance from community mobilizers. The community mobilizers were prominent community members with extensive social networks, some of whom were working as sex workers themselves.
A cross-sectional survey was conducted first, which included face-to-face interviewing, and counseling and testing for HIV (and CD4 count if HIV-positive), herpes simplex virus-type 2 (HSV-2), and pregnancy. Women were eligible for subsequent enrollment in a prospective cohort study if they tested HIV-negative in the cross-sectional survey; were not pregnant or planning a pregnancy in the next year; were over 18 years of age (women aged 18 to 20 obtained parental/guardian consent for participation); were at high risk for sexual exposure to HIV, defined as having exchanged sex for money at least once in the last month and/or currently having sex with multiple partners plus having sex at least twice per week (all enrolled women self-reported sex work); were not participating in another HIV intervention study; and were willing and able to provide written informed consent. The target sample size of 400 women was based on an estimated HIV prevalence of 20% to 25% and target HIV incidence of 2 to 3 per 100 person-years (PY).
All cohort procedures took place at the Projet Ubuzima Clinic. After participating in the cross-sectional survey, women returned to the clinic within 2 to 3 weeks for laboratory test results, counseling and referral for HIV care and treatment, antenatal care, and prevention of mother-to-child transmission (PMTCT) services as appropriate, and assessment of cohort eligibility. If eligible and interested in cohort participation, written informed consent was obtained. Consent forms were translated into the local language (Kinyarwanda). Illiterate women consented in the presence of an independent witness.
Cohort participants returned to the study clinic every 3 months for HIV counseling and testing; STI and pregnancy testing; pregnancy prevention counseling; and face-to-face interviewing to capture changes in physical and reproductive health, relationship status, and to monitor sexual and other HIV-related risk behaviors. The original study protocol was designed to follow the cohort for 12 months, with 4 quarterly study visits (month 3, 6, 9, and 12). However, additional funding midway through the study enabled us to extend cohort follow-up into a second year. Participants returned for a single visit during the second year of follow-up (the “Year 2” visit), which occurred between 3.5 and 16 months after women's month 12 study visit. It also enabled us to retest HIV-negative women (irrespective of pregnancy status) who had participated in the cross-sectional survey but were not enrolled in the cohort because the target sample size of the cohort had been met. Retesting of these women, referred to as the parallel study (Fig. 1), was conducted 6 to 12 months after the cross-sectional survey.
Human Immunodeficiency Virus Testing.
All specimens were first tested by the First Response rapid test (Premier Medical Corporation, Nani Daman, India). Positive specimens were then tested by the Uni-Gold rapid test (Trinity Biotech Plc, Bray, Ireland) for confirmation. Specimens discrepant on First Response and Uni-Gold were tested with the Capillus HIV-1/HIV-2 rapid test (Trinity Biotech Plc, Bray, Ireland) as a tie-breaker. HIV rapid test-positive blood specimens, and baseline specimens from cohort seroconverters, were confirmed by the Murex HIV Ag/Ab Combination ELISA test (Murex Biotech Limited, Dartford, UK), and then tested further for CD4 cell count. HIV rapid test-negative blood specimens from participants in the cross-sectional survey, as well as specimens from antibody-negative cohort participants' month 12 visit, were pooled for testing by HIV-1 ribonucleic acid (RNA) polymerase chain reaction (PCR) to identify acute HIV infections (COBAS TaqMan, PCR Roche Diagnostic Corp, Indianapolis, IN). HIV testing was done at Projet Ubuzima, with the exception of CD4 and HIV-1 RNA PCR testing, which were done at the National Reference Laboratory in Kigali.
Other Laboratory Testing.
Blood specimens were tested for pregnancy at every visit using the Fortress hCG serum pregnancy test (Fortress Diagnostics, Antrim, UK); for HSV-2 with the HerpeSelect 2 ELISA (Focus Diagnostics, Inc, Cypress, CA; an index ≥3.5 indicated positivity) at enrollment and the month 12 visit if negative at enrollment; and for Treponema pallidum at enrollment, month 6, and month 12 with the Spinreact Rapid Plasma test with confirmation by Treponema Pallidum Hemagglutination test (Spinreact, Girona, Spain). Vaginal and cervical specimens collected during pelvic examination at the enrollment, month 6, and month 12 visits (and at other visits if clinically indicated) were assessed for bacterial vaginosis (BV) and candidiasis by wet mount, Trichomonas vaginalis (TV) by InPouch TV culture (Biomed Diagnostics, White City, OR) and wet mount, and Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) by Amplicor CT/NG PCR test (PCR Roche Diagnostic Corp, Indianapolis, IN). All tests were performed at Projet Ubuzima with the exception of CT/NG PCR testing, which was performed by the Institute for Tropical Medicine in Belgium.
HIV/STI Treatment and Referral.
Participants who were diagnosed with a curable STI during follow-up were offered treatment, partner notification, and/or referral for services. Women with STIs detected on wet mount (TV) or presenting with clinical signs (abnormal vaginal or cervical discharge potentially indicative of BV or candidiasis, active HSV-2, acute pelvic inflammatory disease) were treated onsite on the same day of the visit. Women were asked to return to the clinic for treatment of infections for which laboratory results were received after the diagnostic visit (syphilis, NG, CT, TV). Treatment regimens were based on the World Health Organization's 2003 Guidelines for STI/RTI management,7 and on Rwanda's National Guidelines for syndromic management of STIs/RTIs.8 Women who tested HIV-positive were referred to a local health center for care and evaluation for treatment eligibility, as well as psychosocial services. Pregnant women were referred for prenatal care and for PMTCT services if HIV infected. HSV-2 seropositive women were counseled to return to the clinic for care in the event of a herpes outbreak.
Participants received 4000 Rwandan Francs (approximately 7 USD) at scheduled visits as compensation for travel to and time spent at the study clinic. The research protocol, study instruments, and consent forms were approved by the National Ethics Committee and the National HIV/AIDS Committee (CNLS) in Rwanda, and by the Columbia University Medical Center Institutional Review Board in the United States.
Data were double entered and analyzed using SAS version 9.2 (SAS Institute, Cary, NC). Descriptive statistics were used to summarize baseline demographic, behavioral, and clinical characteristics. Categorical variables are expressed as percentages, and continuous data as medians with interquartile ranges (IQR).
HIV, STI, and pregnancy IR were calculated based on a Poisson distribution with person-time of follow-up in the denominator. IRs are expressed as number of cases per 100 PY of follow-up, with 95% confidence intervals (CI). HIV infection was assumed to have occurred at the midpoint between the last available negative HIV test and first positive HIV test. Kaplan-Meier survival curves were generated to display time to HIV infection during follow-up for the cohort as a whole and among subgroups of women stratified by baseline demographic and behavioral characteristics.
Predictors of Incident HIV Infection.
Cox proportional hazards regression was used to estimate hazard ratios (HR) for the effect of a range of baseline (fixed) and time-varying behavioral and biologic covariates on risk for HIV infection in the cohort. For unadjusted models, covariates were chosen that are known or hypothesized to have an association with incident HIV infection, including incident STIs and pregnancy. Variables that were associated with HIV infection at P ≤ 0.29 were analyzed in an age-adjusted multivariable model to predict HIV infection during follow-up.
Person-time calculations for curable STIs were adjusted for the timing and duration of treatment. For infections that were diagnosed in real time and treated on the day of the visit (TV detected by wet mount; symptomatic BV, candidiasis, and herpetic lesions), and for those treated in the interval before the next study visit (additional TV detected by InPouch and syphilis), person-time calculations excluded the treatment duration (number of days) from participants' time at risk for a second infection. Results for laboratory testing done outside the study clinic, including for NG and CT, were received with delay (up to 6 months) and so treatment for these infections was not accounted for in person-time for incidence calculations. HIV, HSV-2, NG, CT, and pregnancy incidence analyses allow for only 1 event per participant; women contribute to total person-time until the event and are censored thereafter. Kaplan-Meier curves were also generated for time to incident pregnancy during follow-up, for the cohort as a whole and among subgroups.
Between October 2006 and August 2007, 800 women were screened in the cross-sectional survey (Fig. 1). A total of 192 women tested HIV-positive, for an HIV prevalence of 24% (95% CI: 21.0–27.0). Pregnancy prevalence was 7.6% (95% CI: 5.8–9.5). On the basis of HIV and pregnancy test results, 243 women were excluded from cohort participation; an additional 32 women were ineligible for other reasons, such as age <18 years or not meeting the HIV risk criteria. Of the remaining 525 eligible HIV-negative, nonpregnant women, 397 were enrolled into the cohort. A subgroup of 156 HIV-negative women (irrespective of pregnancy status) who were not enrolled in the cohort were retested for HIV 6 to 12 months after the survey.
Baseline Demographic, Behavioral, and Clinical Characteristics
Participants ranged in age from 18 to 46 years, with a median age of 24.0 years (Table 1). Among them more than one-fifth (22%, 87/397) had no formal schooling, and only 28% (111/397) had completed primary school. Participants' median weekly income was 12,000 Rwandan Francs (about 22 USD). Overall, 91% (362/397) of cohort participants reported currently using any contraceptive method, with 72% (284/397) of women using male condoms, 12% (48/397) an injectable hormonal method, and 6% (24/397) oral contraceptive pills. Women had a median of 2 pregnancies during their lifetimes (IQR: 1–3), and 55% of women were currently breastfeeding. Although 75% (296/397) of women reported that their last sex partner used a male condom, they generally reported inconsistent condom use with both steady partners (defined as a regular sex partner not living with or married to the woman) and casual/paying partners. All women self-identified as sex workers, and nearly all were engaging in regular transactional sex, with 94% (374/397) receiving money/gifts from their last sex partner. Most women (76%, 303/397) had never been married, 14% (55/397) were divorced or separated, and 9% (34/397) were widowed. About one-third (31%, 122/397) of the cohort had a steady partner in addition to casual sex partners and paying clients. Women had worked as sex workers for a median of 3 years (IQR: 2–5), and their median number of clients per week during the last month was 10 (IQR: 5–15). Thirty percent of women (119/397) had never been tested for HIV before being screened for the cohort, and 51% (204/397) had been tested once before. Twelve percent (49/397) had a sex partner whom they knew to be HIV infected. Prevalence of STIs at cohort entry was high (Table 2).
Cohort retention was 96% in year 1 and 85% in year 2 (Fig. 1). Participants attended a total of 2276 study visits, during which they accrued 686.5 PY of follow-up. Median follow-up was 689 days (range: 0–836). Ninety-one PY were accrued in a parallel study of 156 HIV-negative women who were HIV tested 6 to 12 months after screening.
HIV Incidence During Follow-Up
Thirteen incident HIV infections, including 1 identified by pooled PCR testing, occurred during the first 12 months of follow-up, for a year 1 HIV IR of 3.5 infections per 100 PY (95% CI: 1.6, 5.4) (Table 2). Time to HIV infection varied significantly by lifetime HIV testing history (P = 0.02) and by prevalent gonorrhea infection at baseline (P = 0.03) (Fig. 2). HIV incidence declined nonsignificantly during year 1, with most infections occurring in the first 6 months. HIV IR for the first and second 6 months of the cohort were 4.6/100 PY (1.6, 7.7) and 2.2/100 PY (0.1, 4.4). The IR ratio comparing incidence in the first to the second 6 months was 2.1 (95% CI: 0.7, 7.8). In total, 338 participants attended the year 2 visit. Among the 325 participants who were HIV-negative at the prior study visit, 6 HIV seroconversions were detected for a year 2 IR of 2.1/100 PY (0.4, 3.7). The CI of all HIV IR for the cohort overlap one another (Fig. 3A). The HIV IR in the parallel study was 3.3/100 PY (0, 7.0).
In the multivariable model, risk for incident HIV infection was higher among women who had been HIV tested 2 or more times in their lifetime versus never (HR, 8.0; CI: 0.9, 71.3; P = 0.06), had tested for HIV in the 6 months before enrollment versus those who had not (HR, 2.8; CI: 0.8, 9.5; P = 0.10), initiated sex work within the last year versus 2 to 3 years ago (HR, 6.3; CI: 1.8, 22.2; P < 0.01), had had 2 lifetime pregnancies versus 4 or more (HR, 8.2; CI: 0.9, 71.1; P = 0.06), had prevalent gonorrhea infection at cohort entry (HR, 2.8; CI: 0.9, 9.0; P = 0.08), and had an incident syphilis infection during follow-up (HR, 5.7; CI: 1.3, 24.7; P = 0.02) (Table 3). Compared with women without a history of HIV testing in the prior 6 months, a higher proportion of testers reported condom use (84% vs. 70%; P = 0.03) and having an HIV-positive sex partner (23% vs. 11%; P < 0.01). HIV risk was also higher among women who reported a change in reproductive intentions, including a desire to become pregnant, prevent pregnancy, or a change in their family planning practices (HR, 5.5; CI: 1.2, 26.2; P = 0.03). Finally, HIV risk was lower among women aged 21 to 24 compared with those aged 30 to 34 (HR, 0.3; CI: 0.1, 1.2; P = 0.09).
STI and Pregnancy Incidence During Follow-Up
Of the 182 HSV-2 negative cohort participants at baseline, 13 became HSV-2 infected during the first 12 months of follow-up, for an IR of 8.7/100 PY (4.0, 13.4) (Table 2). Twelve-month IR of trichomoniasis, gonorrhea, chlamydia, and syphilis were 16.9/100 PY (12.7, 21.1), 12.1/100 PY (8.2, 15.9), 8.1/100 PY (5.1, 11.2), and 6.2/100 PY (3.7, 8.7), respectively. Trends over time in STI incidence were nonsignificant, except for syphilis incidence which declined significantly (Fig. 3A).
During the 24-month follow-up period, 139 women (35%) became pregnant for an IR of 26.3/100 PY (21.9, 30.7). Pregnancy incidence declined steadily throughout follow-up (Fig. 3B), and survival analysis indicated an age gradient with younger women experiencing more incident pregnancies than women in the older age groups (data not shown). The rate of pregnancy among women who ever self-reported using a hormonal contraceptive method during follow-up was lower than for the cohort as a whole, at 19.3/100 PY (13.3, 25.2). Open-ended interview data revealed a strong desire among women to become pregnant, as well as significant social pressure from partners and family members.
We observed a high prevalence and incidence of HIV, other STIs, and pregnancy in FSW in Kigali, Rwanda. FSW are a most-at-risk-population in many countries. HIV prevalence and incidence were therefore expected to be high, but were not known in the Rwandan context. In our cohort, there was a nonsignificant downward trend in HIV incidence during the first year of follow-up. This decline could have been due to frequent study visits and risk-reduction interventions in the cohort during this period, or because the highest-risk participants became HIV infected early in the study period. HIV incidence was high among nonenrolled women in the parallel study, in which there were no interim study visits or risk-reduction interventions. There was, however, no rebound in the IR between the month 12 and year 2 visits when study visits were also infrequent. None of the time trends, however, was statistically significant due to limited statistical power. HIV incidence reductions over time have also been observed in other cohorts of high-risk women.10–14
Several factors emerged as predictors of incident HIV infection in this cohort. The positive association between HIV testing history and risk for incident HIV infection likely reflects risk-based testing; a higher proportion of repeat testers reported condom use and having an HIV-positive sex partner. The association between older age and higher risk for incident HIV could be due to a greater proportion of women aged 30 to 34 years being divorced, widowed, and frequent HIV testers, and fewer reporting condom use at last sex, compared to younger women. The associations between gonorrhea and syphilis infection and incident HIV infection are consistent with a body of literature supporting a synergistic relationship between other STIs and risk for HIV.3,15,16 However, the lack of associations with other STIs, and particularly HSV-2, was unexpected, particularly given the endemicity of these infections in the population at cohort entry.17–19 The lack of such associations in our study is likely due to limited statistical power from a small number of HIV events.
In general, the burden of STIs in this population was substantial. For example, with the exception of chlamydia, pooled baseline STI prevalence rates from a multicenter feasibility study conducted at 4 African sites20 were consistently, and in certain cases (e.g., gonorrhea and trichomoniasis) substantially, lower than those in our population. Furthermore, the incidence of STIs during follow-up was high even in the presence of STI control efforts built into our study. Treatment delays and limited uptake of partner notification and treatment might have dampened the effect of our STI screening and treatment services on STI incidence.21,22 STI screening and treatment services are somewhat available in Kigali, but are not widely accessed. Better services and education about the importance of STI screening and treatment are needed, especially for high-risk groups such as FSW.
The pregnancy IR was very high but comparable to rates reported for other cohorts of women at risk for HIV in sub-Saharan Africa.20,23,24 Even though we excluded women who reported that they were planning a pregnancy within the next year from cohort enrollment, one-third of the cohort became pregnant during follow-up. Condom use was for the most part inconsistent (especially with primary partners) and few women were using other contraceptive methods. The low rates of condom use for family planning in this group are comparable to rates described for the general population in Rwanda.2 However, the high risk for HIV combined with substantial unmet need for feasible and effective contraceptive methods in this group underscores the potential effect of integrated HIV/STI prevention and family planning services. Integration of such programs for FSW in Rwanda is underway.25
This study has a number of strengths, including rapid accrual of an at-risk, hard-to-reach population; excellent cohort retention; documentation of population HIV/STI prevalence rates during screening; and collection of in-depth demographic and behavioral information about an understudied population in Rwanda. Limitations are also noted. Most importantly, few HIV events during follow-up limited our statistical power to detect all but the strongest predictors of HIV incidence. Sexual behavior and contraceptive use data are based on self-report, and therefore subject to bias. Participant recruitment methods, including eligibility criteria, and comprehensive and frequent STI testing and treatment during follow-up may reduce the generalizability of HIV/STI IRs.
FSW in Rwanda are clearly in need of HIV/STI prevention services. Intense counseling, condom promotion, and other risk-reduction strategies, and potentially structural interventions related to empowerment or education that may allow women to leave sex work for other employment opportunities should they so desire, should be considered in future interventions in this population.26
1.Kayirangwa E, Hanson J, Munyakazi L, et al. Current trends in Rwanda's HIV/AIDS epidemic. Sex Transm Infect 2006; 82(suppl 1):i27–i31.
2.Institut National de la Statistique du Rwanda. Rwanda Demographic and Health Survey-Plus, 2005.
Calverton, MD: INSR and ORC Macro, 2006.
3.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; 291:2555–2562.
4.Nagot N, Ouedraogo A, Ouangre A, et al. Is sexually transmitted infection management among sex workers still able to mitigate the spread of HIV infection in West Africa? J Acquir Immun Defic Syndr 2005; 39:454–458.
5.Cowan FM, Hargrove JW, Langhaug LF, et al. The appropriateness of core group interventions using presumptive periodic treatment among rural Zimbabwean women who exchange sex for gifts or money. J Acquir Immun Defic Syndr 2005; 38:202–207.
6.Kimani J, Kaul R, Nagelkerke NJ, et al. Reduced rates of HIV acquisition during unprotected sex by Kenyan female sex workers predating population declines in HIV prevalence. AIDS 2008; 22:131–137.
7.WHO. Management of Sexually Transmitted Disease
. Geneva, Switzerland: WHO/UNAIDS, 2003.
8.Ministry of Health, Rwanda. Guide National de Prise en Charge Syndromique des Infections Sexuellement Transmissibles.
Kigali: Ministry of Health, 2002.
9.Greenland S. Modeling and variable selection in epidemiologic analysis. Am J Public Health 1989; 79:340–349.
10.Baeten JM, Richardson BA, Martin HL Jr, et al. Trends in HIV-1 incidence in a cohort of prostitutes in Kenya: Implications for HIV-1 vaccine efficacy trials. J Acquir Immun Defic Syndr 2000; 24:458–464.
11.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 Immun Defic Syndr 2002; 30:69–72.
12.van Loggerenberg F, Mlisana K, Williamson C, et al. Establishing a cohort at high risk of HIV infection in South Africa: Challenges and experiences of the CAPRISA 002 acute infection study. PLoS One 2008; 3:e1954.
13.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 2001; 15:1421–1431.
14.Laga M, Alary M, Nzila N, et al. Condom promotion, sexually transmitted diseases treatment, and declining incidence of HIV-1 infection in female Zairian sex workers. Lancet 1994; 344:246–248.
15.Wasserheit JN. Epidemiological synergy. Interrelationships between human immunodeficiency virus infection and other sexually transmitted diseases. Sex Transm Dis 1992; 19:61–77.
16.van de Wijgert JH, Morrison CS, Brown J, et al. Disentangling contributions of reproductive tract infections to HIV acquisition in African women. Sex Transm Dis 2009; 36:357–364.
17.Brown JM, Wald A, Hubbard A, et al. Incident and prevalent herpes simplex virus type 2 infection increases risk of HIV acquisition among women in Uganda and Zimbabwe. AIDS 2007; 21:1515–1523.
18.Van Der Pol B, Kwok C, Pierre-Louis B, et al. Trichomonas vaginalis
infection and human immunodeficiency virus acquisition in African women. J Infect Dis 2008; 197:548–554.
19.Kapiga SH, Sam NE, Bang H, et al. The role of herpes simplex virus type 2 and other genital infections in the acquisition of HIV-1 among high-risk women in northern Tanzania. J Infect Dis 2007; 195:1260–1269.
20.Ramjee G, Kapiga S, Weiss S, et al. The value of site preparedness studies for future implementation of phase 2/IIb/III HIV prevention trials: Experience from the HPTN 055 study. J Acquir Immun Defic Syndr 2008; 47:93–100.
21.Golden MR, Whittington WL, Handsfield HH, et al. Effect of expedited treatment of sex partners on recurrent or persistent gonorrhea or chlamydial infection. N Engl J Med 2005; 352:676–685.
22.Schillinger JA, Kissinger P, Calvet H, et al. Patient-delivered partner treatment with azithromycin to prevent repeated Chlamydia trachomatis infection among women: A randomized, controlled trial. Sex Transm Dis 2003; 30:49–56.
23.Feldblum PJ, Adeiga A, Bakare R, et al. SAVVY vaginal gel (C31G) for prevention of HIV infection: A randomized controlled trial in Nigeria. PLoS One 2008; 3:e1474.
24.Bosire W. A Prospective Microbicides Preparedness Study for Estimation of HIV-1 Incidence among Female Sex Workers in Mombasa, Kenya. Abstract No. BP28-364, New Delhi, India: International Microbicides Conference, 2008.
25.UNFPA. HIV programming with sex workers in Rwanda: Overview of current efforts: UNFPA, Kigali, Rwanda, 2009.
26.Shahmanesh M, Patel V, Mabey D, et al. Effectiveness of interventions for the prevention of HIV and other sexually transmitted infections in female sex workers in resource poor setting: A systematic review. Trop Med Int Health 2008; 13:659–679.