In sub-Saharan Africa, female sex workers (FSW) have high HIV incidence and prevalence and therefore are in particular need of good access to effective HIV testing, prevention, and treatment services.1 In sub-Saharan Africa, FSW are estimated to have 13.5 times higher odds of HIV infection than in the general population of adult women.1 However, FSW are a marginalized group, sex work is illegal in many countries including Zimbabwe,2 and FSW are often stigmatized by communities and health workers.3,4 Typically, FSW are also highly mobile.5 Designing service-delivery approaches that meet the needs of this population is therefore complex but urgently needed.
There is currently little information about the continuum of care from diagnosis to virological suppression (the HIV care cascade) among FSW with which to guide programing. Previous studies of FSW in sub-Saharan Africa indicate that antiretroviral therapy (ART) can be provided to FSW6 at costs comparable with that of provision in the general population.7 A recent systematic review and meta-analysis of antiretroviral uptake, adherence, and outcomes among FSW found that current ART use among HIV-positive FSW was 39% (95% confidence interval: 29%–48%), but noted a concerning lack of published data available.8 Another review of the provision of sexual and reproductive health services for FSW in Africa found little emphasis among programs on access to antiretroviral treatment and support for adherence.9 Although there is some evidence to guide the design of HIV prevention programs for FSW in Africa, little is known about the best means to improve testing, access, and adherence to ART and effective use of preexposure prophylaxis.10,11
In 2009, in response to a situational analysis conducted among FSW by Zimbabwe's National AIDS Council and partners,12 the “Sisters with a Voice” program was established in 5 sites, and has since expanded to 36 sites covering all the provinces of Zimbabwe. Services provided are based on guidance from the World Health Organization13 and include HIV testing and counseling, sexual and reproductive health services, condom provision, and health education supported by trained peer educators and a program of community mobilization. Results of a respondent-driven sampling (RDS) survey conducted in 3 towns in 2011,14 along with qualitative work,15 suggested that FSW in Zimbabwe were poorly engaged with HIV prevention and care services.
In response to this finding, we launched the Sisters Antiretroviral Programme for Prevention of HIV—an Integrated Response (SAPPH-IRe) trial, a cluster-randomized controlled trial conducted in 14 sites around Zimbabwe (7 matched-pairs). The aim is to determine the effectiveness and cost effectiveness of an enhanced community–based intervention to increase uptake, retention, and adherence to antiretroviral-based prevention and therapy among FSW. Outcomes were assessed at a population level in all 14 communities among FSW recruited to RDS surveys at baseline (December 2013), and will also be assessed at end-line (April 2016–May 2016).
Aiming to contribute to our scant knowledge of the HIV diagnosis and care cascade among FSW in sub-Saharan Africa, this article describes the HIV diagnosis and care cascade at 14 sites around Zimbabwe at the baseline of the SAPPH-IRe trial. Data are presented on sociodemographic characteristics, HIV prevalence, ART coverage, viral suppression, and the proportion of all FSW with unsuppressed HIV viral load: the primary endpoint for the SAPPH-IRe trial.
Study Population and Setting
Fourteen of the 36 sites where the “Sisters” services are provided are included in the SAPPH-IRe trial. These sites were purposively selected to reflect different sex work location types (eg, town, growth point, colliery/army base), were locations of adequate size (85–300 FSW attending clinics annually), and were geographically disparate to minimize contamination during the trial. Women were eligible if they were aged 18 or elder, had exchanged in sex for money or gifts in the preceding 30 days, and had lived at the site for at least the previous 6 months.
We conducted RDS16 surveys of FSW using identical procedures in each of the 14 sites. We used RDS because it was unfeasible to assemble a sampling frame of the intended target population; it has been recommended for research among hard-to-reach populations17; we successfully conducted similar RDS surveys of FSW in 3 locations in 2011,14 and sex work in these settings is not conducted primarily within brothels or set venues making time-location sampling methods less appropriate. In each site, we first conducted 2–3 days of geographic and social mapping, including informal discussions with trained peer educators, health care staff, and community informants. This formative work informed specific criteria for purposely selected “seed” women to ensure that all subpopulations within the site's sex worker population were represented and helped determine how many of these seeds should be selected.18
In line with RDS methodology, seed participants in each site were interviewed and given 2 recruitment coupons to pass on to their sex worker peers. Women were uniformly advised to recruit other sex workers whose names they knew and who knew theirs, who had not already enrolled in the study, and who met the study eligibility criteria. Interviewers used screening questions to confirm as far as possible that women given coupons met these criteria when they presented for interview. Six seeds were recruited in the smaller sites, whereas in 4 larger sites 8 seeds were recruited. When women receiving the coupons attended for the interview (“recruits”), they were also given 2 coupons to give out to women they knew who worked as FSW in that location. Coupons were coded such that recruiter/recruitee relationships could be tracked and unique IDs recorded. In all 14 sites, a maximum of 5 iterations, or “waves,” of this process were performed (6 waves, including the initial seeds). We aimed to recruit 200 FSW per site to give adequate power to detect the intervention effect at follow-up.19 In line with other RDS surveys, women were reimbursed for participating in the survey ($5) and for recruiting eligible participants ($2 for each recruited). All participants gave informed consent to participate after receiving information about the study from trained interviewers and being given the opportunity to ask questions.
Five teams of trained researchers undertook data collection between November 13, 2013 and December 20, 2013. Interviewer-administered questionnaire data were collected onto tablet computers and directly loaded into a master database using a wireless internet connection in the field. Questionnaires included information on demographics, sex work, sexual behavior and condom use, HIV testing history, ART use, stigma, experience of violence, relationships with other sex workers, and use of sexual, and reproductive health services. We also collected data to determine personal network size, or “degree,” for RDS estimation. In our survey, the degree was the number of FSW a participant reported knowing personally, whose names they knew and who knew theirs, who were at least 18 years old, lived at the site, and whom the participant would consider recruiting to the study.
All women had a finger prick blood sample collected in the form of a dried blood spot (DBS) for detection of HIV antibody [AniLabsytems EIA kit (AniLabsystems Ltd, OyToilette 3, FIN-01720, Finland)]. Blood samples were air-dried on filter papers and stored at room temperature, then transported biweekly to the Flow cytometry Laboratory in Harare. If HIV antibodies were detected then the DBS sample was tested for HIV viral load using NucliSENS EasyQ HIV-1 v2.0, both to confirm HIV-positive status and to quantify the viral load. For samples with a positive HIV antibody test, but an undetectable viral load, a second confirmatory enzyme-linked immunosorbent assay was performed [Enzygnost Anti-HIV 1/2 Plus ELISA (Germany)]. At 2 trial sites, plasma samples were collected in addition to DBS and tested in parallel using NucliSENS EasyQ HIV-1 v2.0, to permit validation of the use of DBS for viral load quantification.20
The Medical Research Council Zimbabwe, University College London, and the London School of Hygiene and Tropical Medicine gave ethical approval for the SAPPH-IRe trial, including the baseline data collection and analysis. The trial was also registered with the Research Council of Zimbabwe, the Pan African Clinical Trials Registry (PACTR201312000722390), and was approved by the Medicines Control Authority of Zimbabwe.
We follow the recommendations of the STROBE-RDS guidelines in reporting our study.21 First, we described the sample recruited. A limitation of RDS is that it is difficult to describe nonparticipation rates because no sample frame is present, and we did not conduct “exit interviews” of women who had distributed coupons to ascertain how many of their peers refused to take part. We calculated cluster-summaries for key sociodemographic characteristics of the sample. We calculated and report the mean of the 14 cluster-level RDS-2 weighted summaries and the range of estimates across clusters (minimum and maximum). Both as a total and summarized across clusters, we described the proportion of participants with suppressed HIV viral load, (<1000 copies/mL, as per World Health Organization guidelines22,23), and steps of the HIV care cascade underlying this: the proportion who were found to be HIV positive; the proportion who reported through questionnaire previously testing positive (ie, knew their status); the proportion who reported being on ART; and the proportion who had a viral load of <1000 copies/mL. We described these estimates both as proportions of the previous step on the cascade and as proportions of the total of women testing HIV positive.
We used “RDS-2” to conduct all analyses, which uses the “Volz-Heckathorn” estimator24 and has been found to be less biased than previous estimators.25 RDS-2 is based on estimating the inclusion probabilities of each survey participant, assuming the recruitment process can be modeled as a “random walk” over the social network of FSW. Within this model, the probability that each participant will be included is approximated as the inverse of the reported degree. Estimates were calculated in Stata 12 using the “rds” analysis package,26 which removes seeds from the proportion estimates.
RDS-2 estimation assumes that recruitment chains progress such that final estimates are no longer dependent on the characteristics of the seeds, that recruitment does not become confined within subgroups of the FSW population (“bottlenecks”), and assumes with-replacement sampling even when women cannot participate more than once in practice.25 We assessed these assumptions and their potential for bias on estimates of HIV prevalence and suppressed viral load for each site, using plots of the convergence of HIV and viral suppression estimates over sample waves (“convergence plots”) and plots of estimate convergence by seed (“bottleneck plots”). We also examined the difference between RDS-2 estimates and estimates produced using the RDS “successive sampling” estimator27 for a range of possible population sizes to assess the bias resulting from assuming with-replacement sampling. These analyses were guided by published advice about RDS diagnostics28 and used the “rds” package for the R statistical language.29 Details of the diagnostic methods and results are given in Appendix 1 (http://links.lww.com/QAI/A952).
RDS Recruitment and Estimation
In total, 2722 participants were recruited over 6 waves in 14 sites. Of these participants, 90 were seeds, of whom 62 (68.9%) were HIV positive and 29 (32.2%) had HIV viral load ≥ 1000 copies/mL. The number of nonseed “recruits” varied from 147 to 212 per site. There were an additional 15 participants from 8 sites who were missing recruiter information and who were treated as seeds and therefore dropped from the estimation.
Estimates for the proportion of FSW with suppressed viral load and for HIV prevalence seemed to converge well by the final sample wave for all sites except one for HIV prevalence and 2 for viral load, and there was little evidence of recruitment becoming confined within subgroups from any site (see Appendix 1).
Characteristics of FSW
Participants were aged between 18 and 65, with a mean age of 31 years (minimum site mean of 29 and maximum of 34). Approximately, one-third of women had no or only primary education, another third had completed forms 1–3, and the final third had completed at least form 4 (Table 1). Very few of the women were married (0.8% overall unweighted, the proportion was too small to calculate RDS weights) and 61.9% (range 46.4%–70.6% across sites) were separated or divorced. Most women (53.5%) reported initiating sex work by 24 years old, with 17.4% (8.5–25.9) reporting having started sex work before they were 18 years old. In total, 8.2% reported having no clients in the past week, 49.9% of women reported having between 1 and 5 clients per week; and 13.2% reported having 16 or more. Just under half of the women in each cluster (45.0%) were food insecure (food insecurity was indicated by any of the following: being unable to eat 2 meals a day; sometimes going to bed hungry; going an entire day without eating in the last week). More than a quarter of women (26.7%) had worked at another geographic location in the previous 12 months, whereas 52.2% had lived in their current location for 6 or more years. Just under two thirds (61.4%) of women reported good or very good relations with other FSW.
Violence from intimate partners was the most common form of interpersonal violence ever experienced (40.3%), followed by violence from clients (27.7%). Violence from police in the previous year was 9.7% overall, although in one location it was 19.5%.
Most participants reported having previously tested for HIV (91.1%), and of those who were HIV negative 70.5% (52.7–88.8) reported having tested for HIV in the previous 6 months.
HIV and the Diagnosis and Care Cascade
The HIV care cascade for HIV-positive FSW is described in Figure 1. HIV prevalence among FSW was estimated to be 57.5%, ranging from 42.8% to 79.2% across sites.
Among those who tested HIV positive, an average of 64.0% (51.6–73.7) in each site were aware of their status, ie, they reported a previous positive HIV test. Of those aware of their positive status, 67.7% (53.4–84.1) reported taking ART, which was 43.3% (32.3–54.0) of all those who tested HIV positive in the study. Across sites, an average of 77.8% (64.4–90.8) of women who were on ART had a viral load <1000 copies/mL. Women on ART with viral loads <1000 copies/mL were 33.7% (range 36.5–62.2) of all those testing HIV positive. An additional 15.8% (range 12.6–16.6) of those testing positive had a viral load <1000 copies/mL, despite not reporting being on ART. Of all HIV-positive FSW, 43.3% (32.3–54.0) were on ART and 49.5% (36.5–62.2) had viral loads of <1000 copies/mL.
When considering all FSW as the denominator, there were an estimated 29.3% (18.9–42.3) of women who had an unsuppressed HIV viral load of ≥1000 copies/mL.
We analyzed data from 2722 FSW recruited in 14 sites in Zimbabwe. HIV prevalence was very high (mean 57.5% across sites, ranging 42.8%–79.2%). Although recent HIV testing and access to ART were relatively common, still some 36.0% of HIV-positive FSW did not report that they were positive in the research interview (26.3–48.4). Most women who tested HIV positive and reported being aware of their status reported accessing ART (67.7%) and of those, 77.8% had a viral load <1000 copies/mL. However, overall only 49.5% of all HIV-positive women had a viral load <1000 copies/mL, in part because many were unaware of their status. Significant and rapid progress is needed to reduce HIV infection rates, increase HIV status awareness, and improve overall levels of viral suppression.
We undertook an ambitious field study to collect baseline data and test the feasibility of our proposed approach to the trial end-line data collection. We have shown that it was feasible to rapidly recruit approximately 200 FSW per site in 14 sites across Zimbabwe using RDS methodology. Our findings make an important contribution to the sparse literature on the HIV diagnosis and care cascade among FSW in sub-Saharan Africa.8 We have been able to measure women having unsuppressed viral load as a proportion of all HIV-positive sex workers, not only among those accessing ART, which is important given that approximately one-third of HIV-positive FSW were unaware of their status. Sampling approaches such as ours provide a key means for assessing how close we are to the 90:90:90 targets30 in a given population or setting.
All sampling methods for hard-to-reach populations have limitations, and RDS is no exception. The estimation makes many assumptions about the recruitment process and the social networks of sex workers. Previous mapping of sex workers in Zimbabwe suggests they are well networked.14,31 Critical is that appropriate statistical techniques are used although there remains debate about methods of analysis. We have undertaken a series of in-depth diagnostic tests to determine the validity of RDS among FSW in these surveys, and find little evidence that there were major issues with recruitment. We present diagnostics in Appendix 1. However, as in all applications of RDS in hidden populations, it was not possible for us to empirically verify the extent to which the sample we recruited reflects the characteristics of FSWs working in the 14 sites. A major strength of our study was that we adopted identical field procedures in each of the sites, strengthening our capacity to compare findings across them.
Our estimate of viral load for HIV-positive FSW was based on analysis of DBS samples. Although plasma analysis is normally considered the gold-standard approach, DBS seemed to be an acceptable method for viral load monitoring using the NucliSENS assay, and we estimated high DBS sensitivity compared with plasma “gold-standard” (sensitivity = 87.4% and specificity = 96.8%).20
Coverage of ART among HIV-positive FSW was similar at 43.4% (range 32.3–54.0) to the 40% we had hypothesized before the trial.19 This was slightly higher but in the range of the pooled estimate of 39.3% (27.2%–52.9%) among sex workers from low and middle income in studies found in a recent meta-analysis and systematic review.8 Some 67.7% of those FSW who were aware of their status and reported they were positive also reported taking ART (range 53.4%–84.1%). This was similar to our findings in 3 sites in 2011, when we found 51%–74% HIV-positive FSWs who were aware of their HIV status were also engaged with care.14 However, coverage is well below the target set by UNAIDS (90% of people with HIV are aware of their infection, 90% of people diagnosed with HIV are on ART, and 90% of those on ART adhere and have undetectable levels of HIV in their blood).30 Coverage among the general population of adult women in Zimbabwe is not known.
Overall, 77.8% of those reporting taking ART had a viral load <1000 copies/mL, as did 15.8% of HIV-positive women who did not report being on ART. That such a large proportion of women not on ART had a suppressed viral load was not anticipated; one explanation is that women under-reported their knowledge of HIV status and ART usage. However, there have been other surveys with similar findings: the 2012 Kenya AIDS Indicator Survey found that 30% of individuals who reported not being on ART were virally suppressed32 and among men who have sex with men in the United States reporting to be unaware of their status and therefore not on ART in 2004–2011, 2/11 to 3/7 were found to be virally suppressed.33 We plan to investigate this further.
These data point to an urgent need to improve HIV prevention and care for FSW in Africa using comprehensive community–based sex worker–led interventions.10 There is good evidence of their cost effectiveness34,35 but in few countries has comprehensive care as recommended by WHO11 been taken to scale. Interventions that work across the cascade are likely to be more scalable and cost effective than those that work on only one aspect of the cascade.36
In conclusion, our findings have contributed to knowledge of the HIV care cascade among sex workers in Southern Africa. They confirm the urgent need for HIV prevention and care services in this population that address each step of the care cascade. We hope that the SAPPH-IRe trial will contribute to our understanding of how best to serve the needs of FSW in the region.
1. 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;12:538–549.
2. Gruskin S, Ferguson L, Alfven T, et al. Identifying structural barriers to an effective HIV response: using the National Composite Policy Index data to evaluate the human rights, legal and policy environment. J Int AIDS Soc. 2013;16:18000.
3. Pulerwitz J, Bongaarts J. Tackling stigma: fundamental to an AIDS-free future. Lancet Glob Health. 2014;2:e311–e312.
4. Scambler G, Paoli F. Health work, female sex workers
and HIV/AIDS: global and local dimensions of stigma and deviance as barriers to effective interventions. Soc Sci Med. 2008;66:1848–1862.
5. Ramesh S, Ganju D, Mahapatra B, et al. Relationship between mobility, violence and HIV/STI among female sex workers
in Andhra Pradesh, India. BMC Public Health. 2012;12:764.
6. Huet C, Ouedraogo A, Konate I, et al. Long-term virological, immunological and mortality outcomes in a cohort of HIV-infected female sex workers
treated with highly active antiretroviral therapy
. BMC Public Health. 2011;11:700.
7. Cianci F, Sweeney S, Konate I, et al. The cost of providing combined prevention and treatment services, including ART, to female sex workers
in Burkina Faso. PLoS One. 2014;9:e100107.
8. Mountain E, Mishra S, Vickerman P, et al. Antiretroviral therapy
uptake, attrition, adherence and outcomes among HIV-infected female sex workers
: a systematic review and meta-analysis. PLoS One. 2014;9:e105645.
9. Dhana A, Luchters S, Moore L, et al. Systematic review of facility-based sexual and reproductive health services for female sex workers
. Global Health. 2014;10:46.
10. Chersich MF, Luchters S, Ntaganira I, et al. Priority interventions to reduce HIV transmission in sex work settings in sub-Saharan Africa
and delivery of these services. J Int AIDS Soc. 2013;16:17980.
11. WHO. Consolidated Guidelines on HIV Prevention, Diagnosis, Treatment and Care for Key Populations. Geneva, Switzerland: World Health Organization; 2014.
12. Sex work and HIV and AIDS in Zimbabwe: analysis of current settings, policies and interventions. In: National AIDS Council, International Organization for Migration, Joint United Nations Team on HIV and AIDS: Harare, Zimbabwe: United Nations Population Fund; 2008.
13. WHO. Prevention and Treatment of HIV and Other Sexually Transmitted Infections for Sex Workers
in Low- and Middle-Income Countries: Recommendations for a Public Health Approach. Geneva, Switzerland: World Health Organization; 2012.
14. Cowan FM, Mtetwa S, Davey C, et al. Engagement with HIV prevention treatment and care among female sex workers
in Zimbabwe: a respondent driven sampling survey. PLoS One. 2013;8:e77080.
15. Mtetwa S, Busza J, Chidiya S, et al. “You are wasting our drugs”: health service barriers to HIV treatment for sex workers
in Zimbabwe. BMC Public Health. 2013;13:698.
16. Heckathorn D. Respondent-driven sampling: a new approach to the study of hidden populations. Social Probl. 1997;44:174–199.
17. Sabin KM, Johnston LG. Epidemiological challenges to the assessment of HIV burdens among key populations: respondent-driven sampling, time-location sampling and demographic and health surveys. Curr Opin HIV AIDS. 2014;9:101–106.
18. Johnston LG, Whitehead S, Simic-Lawson M, et al. Formative research to optimize respondent-driven sampling surveys among hard-to-reach populations in HIV behavioral and biological surveillance: lessons learned from four case studies. AIDS Care. 2010;22:784–792.
19. Cowan F. Antiretrovirals for HIV prevention and treatment among Zimbabwean sex workers
. Trial Protoc. Harare, Zimbabwe: Available on request; 2013.
20. Mavedzenge SN, Davey C, Chirenje T, et al. Finger prick dried blood spots for HIV viral load
measurement in field conditions in Zimbabwe. PLoS One. 2015;10:e0126878.
21. White RG, Hakim AJ, Salganik MJ, et al. Strengthening the reporting of observational studies in epidemiology for respondent-driven sampling studies: “STROBE-RDS” statement. J Clin Epidemiol. 2015;68:1463–1471.
22. WHO. Consolidated Guidelines on the Use of Antiretroviral Drugs for Treating and Preventing HIV Infection: Recommendations for a Public Health Approach. Geneva, Switzerland: World Health Organization; 2013.
23. WHO. Consolidated Strategic Information Guidelines for HIV in the Health Sector. Geneva, Switzerland: World Health Organization; 2015.
24. Volz E, Heckathorn DD. Probability based estimation theory for respondent driven sampling. J Official Stat. 2008;24:79–97.
25. Gile KJ, Handcock MS. Respondent-driven sampling: an assessment of current methodology. Sociol Methodol. 2010;40:285–327.
26. Schonlau M, Liebau E. Respondent-driven sampling. Stata J. 2012;12:72–93.
27. Gile KJ. Improved inference for respondent-driven sampling data with application to HIV prevalence estimation. J Am Stat Assoc. 2011;106:135–146.
28. Gile KJ, Johnston LG, Salganik MJ. Diagnostics for respondent-driven sampling. J R Stat Soc Ser A. 2015;178:241–269.
29. Handcock MS, Fellows IE, Gile KJ. RDS: Respondent-driven Sampling, Version 0.6; 2012. Available at: Project home page at http://hpmrg.org
, URL http://CRAN.R-project.org/package=RDS
. Accessed June 1, 2015.
30. UNAIDS. 90-90-90: An Ambitious Treatment Target to Help End the AIDS Epidemic. Geneva, Switzerland: UNAIDS; 2014.
31. Mtetwa S, Busza J, Davey C, et al. Competition is not necessarily a barrier to community mobilisation among sex workers
: an intervention planning assessment from Zimbabwe. BMC Public Health. 2015;15:787.
32. Wafula R, Masyuko S, Ng'ang'a L, et al. Engagement in HIV care among Kenyan adults and adolescents: results from a national population-based survey. J Acquir Immune Defic Syndr. 2014;66(suppl 1):S98–S105.
33. Das M, Raymond HF, Chu P, et al. Measuring the unknown: calculating community viral load among HIV-infected MSM unaware of their HIV status in San Francisco from National HIV Behavioral Surveillance, 2004–2011. J Acquir Immune Defic Syndr. 2013;63:e84–e86.
34. Steen R, Wheeler T, Gorgens M, et al. Feasible, efficient and necessary, without exception—working with sex workers
interrupts HIV/STI transmission and brings treatment to many in need. PLoS One. 2015;10:e0121145.
35. Wilson D. HIV Programs for Sex Workers
: Lessons and Challenges for Developing and Delivering Programs. PLoS Med. 2015;12:e1001808.
36. Govindasamy D, Meghij J, Negussi EK, et al. Interventions to improve or facilitate linkage to or retention in pre-ART (HIV) care and initiation of ART in low- and middle-income settings—a systematic review. J Int AIDS Soc. 2014;17:19032.