Yam, Eileen A. PhD*; Mnisi, Zandile BSN†; Sithole, Bheki MPhil‡; Kennedy, Caitlin PhD§; Kerrigan, Deanna L. PhD¶; Tsui, Amy O. PhD∥; Baral, Stefan MD**
Female sex workers (FSWs) are at heightened risk for infection with HIV and other sexually transmitted infections (STIs). Public health interventions have traditionally focused on promoting consistent condom use with clients. However, studies have consistently demonstrated that condom use by FSWs varies by type of sexual partner, with condoms used more frequently with men who pay for sex compared with noncommercial partners.1–10 Furthermore, there is a growing evidence base suggesting that the category of “paying clients” comprises a heterogeneous group of partners, and condom use can vary across clients. Multiple studies have demonstrated that FSWs are less likely to use condoms with more familiar, regular clients compared with new clients.11,12 Moreover, research suggests that noncommercial, steady partners may pose greater HIV risks than clients.13–15
In addition, FSWs may prioritize disease prevention and pregnancy prevention differently from different sexual partners. A tendency to prioritize condom use and other contraceptive methods differently across partners has been reported in non-FSW populations.16 Specifically, a recurrent finding is that women who use other contraception tend to be less likely to use condoms with more steady partners. This inverse relationship between condom use and other contraceptive methods in more stable relationships has largely been documented in studies from high-income settings.16–18
However, in high-HIV-prevalence settings where the likelihood of infection is elevated, all partners may pose considerable risk, irrespective of relationship status. To prevent both unwanted pregnancy and disease, women and their partners can either use condoms consistently (i.e., condoms alone), or they can use condoms consistently along with an effective nonbarrier modern contraceptive method, such as oral contraceptive pills, injectable, or implants (i.e., dual-method use). In either case, condoms must be used at every sex act with every partner, regardless of whether a given partner is perceived to be at elevated risk for HIV/STIs.19 The challenge with this approach is that FSWs often are not in a position to insist on condom use, and the decision to have unprotected sex often is not entirely volitional.11–13 Variation in condom use across partners is likely caused by both individual factors (such as the woman’s desire to experience greater intimacy through unprotected sex) and the partners’ decision-making influence (such as boyfriends who refuse condoms). Although programmatic and policy interventions targeting FSWs generally focus on HIV prevention, the same behaviors that predispose these women to HIV also place them at risk for unintended pregnancy. However, little is known about whether FSWs who use nonbarrier modern contraceptive methods are less likely to use condoms consistently, particularly with more intimate sex partners.
HIV is hyperendemic in Swaziland, with nearly one third of reproductive age adults living with HIV and 62% of new infections occurring among women.20 A 2009 study reported that 68% of new adult HIV infections occur in those older than 25 years, most of whom are in long-term, steady relationships.21 Using a relationship-level analysis among FSWs in Swaziland, we sought to examine the association between consistent condom use and relationship type, as well as the association between consistent condom use and nonbarrier modern contraceptive use. In addition, we assessed whether the association between condom use and nonbarrier modern contraceptive use varies by type of partner. Using data from FSWs in Swaziland, we investigated whether pregnancy prevention and disease prevention behaviors may vary across relationships among FSWs in a high-HIV-prevalence setting. Specifically, we explore whether use of nonbarrier modern methods potentially undermines condom consistency.
MATERIALS AND METHODS
From July to September 2011, study staff recruited FSWs in Swaziland using respondent-driven sampling (RDS), which uses mathematical adjustments to account for the nonrandom nature of this sampling method.22,23 To be eligible for the study, participants had to be 15 years or older, be able to provide informed consent in either English or siSwati, present a valid recruitment coupon, and respond affirmatively to a statement attesting that they exchanged or sold sex for money, favors, or goods in the past 12 months.
After obtaining oral informed consent from participants, interviewers administered an hour-long questionnaire that covered demographic characteristics, sexual and reproductive health knowledge and behaviors, violence, social cohesion, substance use, and stigma. The survey also contained questions asking about condom use in the past month with new clients, regular clients, and noncommercial partners. All interviews took place at a centrally located study clinic. The institutional review board at the Johns Hopkins Bloomberg School of Public Health approved this study protocol, as did the Scientific and Ethics Committee of Swaziland’s Ministry of Health.
We conducted all analyses using Stata version 11.0.24 We first conducted exploratory data analysis to examine variable frequencies and to assess missing and illogical values. The variables for number of new and regular clients in the past month had the highest proportion of missing values (4.9% and 2.8%, respectively). These missing values were imputed by assigning the average number of each client type reported by other participants who had been sex workers for the same amount of time as the woman with the missing values. Missing values for all other variables were handled by listwise deletion.
For univariate statistics, we report both raw and RDS-adjusted estimates. Respondent-driven sampling adjustment consists of applying variable-specific weights to account for 2 potential biases of the RDS methodology: the tendency for participants to recruit others like themselves (homophily) and the variation in network sizes of different individuals. We used a bootstrap method with 1000 repetitions to estimate standard errors for the RDS-adjusted estimates.22,23,25 Respondent-driven sampling inference was originally conceptualized for univariate inference, so we only present raw estimates for multivariate models because of the current lack of consensus about how and whether to adjust RDS estimates in multivariate analyses.26
Our outcome of interest was consistent condom use in the past month, defined as always using condoms in the past month. The unit of analysis was relationships rather than individual women because each woman reported on condom use for 3 different relationship types: new clients, regular clients, and noncommercial partners. Participants responded to the question, “In the last 30 days, how often did you use condoms when having vaginal sex with new clients?” with response options of “never,” “rarely,” “sometimes,” “most of the time,” or “always.” This question was repeated for regular clients and for noncommercial partners. “Always” responses were considered consistent condom use. Because participants reported on condom use in the past month with up to 3 different types of partners, the 325 FSWs could provide up to 975 observations.
We first used χ2 tests to conduct bivariate analyses examining the unadjusted associations between selected participant characteristics and the outcome of consistent condom use in the past month. To assess whether consistent condom use was associated with relationship type and nonbarrier modern contraceptive use, we conducted multivariate logistic regression analyses to examine the association between the binary outcome of “always used condoms in the past month” and the dependent variables of relationship type and nonbarrier contraception. In the first model (base model), we included women’s background characteristics: seed’s recruitment zone, age, education, income, and number of children. In the next model (full model), we added the following variables: nonbarrier modern contraceptive use, relationship type (new clients, regular clients, or noncommercial partners), number of clients per month, number of noncommercial partners per month, condom failure in past month, and previous HIV-positive diagnosis. For this HIV status variable, a positive value was assigned to those who responded affirmatively to the question, “Have you ever been told by a health care provider that you have HIV?” Lastly, to assess whether the relationship type modifies the effect of nonbarrier contraception on consistent condom use, we introduced a multiplicative interaction term (i.e., contraception * relationship) to determine whether a statistically significant interaction exists between these 2 independent variables. We also tested for effect modification between nonbarrier contraceptive use and previous HIV-positive diagnosis. For all models, we used a generalized estimating equation approach with robust standard errors and unstructured correlation matrix to account for the correlation between the multiple condom use measures reported by each woman (i.e., one measure for each relationship type).
Of the 339 women who presented to the study clinic, 325 met eligibility criteria. Table 1 displays the characteristics of eligible participants, providing both crude estimates and RDS-weighted estimates for univariate statistics. Based on weighted estimates, most participants were between the ages of 21 and 29 years (53.2%), with 30.7% being younger than 21 years and 19.6% being 30 years or older. Nearly all (90.6%) were single and never married, and 70.7% had at least 1 child. In the past month, one third (33.7%) earned less than US$70, 31.6% earned US$70 to US$140, 25.8% earned US$141 to US$281, and 8.8% earned US$282 or more.27 On average, FSWs reported more regular clients (mean [SD], 8.4 [9.5]) than new clients (mean [SD], 5.3 [5.8]). Approximately half (51.1%) reported using nonbarrier modern contraception. More than three fourths (77.8%) had ever been pregnant; of those, 48.7% had ever had an unwanted pregnancy and 11.7% had ever had an abortion. Roughly half (52.1%) had 1 noncommercial partner in the past month, and 39.2% had 2 or more.
Among the 325 eligible participants, 299 provided responses about consistency of condom use with new clients, 309 reported on consistency of condom use with regular clients, and 284 reported on condom use consistency with noncommercial partners, yielding a total of 892 observations. Overall, consistent condom use was reported for 52.2% of these observations.
In bivariate analyses, women with no children were more likely to report sexual activity protected by consistent condom use (P < 0.05), compared with women with children (58.0% vs. 50.3%, P < 0.05). Consistency of condom use was more frequent in sex acts with new clients (74.2%), compared with regular clients (48.2%) or noncommercial partners (33.5%, P < 0.001). Women who used nonbarrier modern contraceptives were less likely to report consistent condom use than women who did not use such contraceptive methods (48.1% compared with 56.6%, P = 0.011). Those who had experienced condom failure in the past month were less likely to report consistent condom use compared with those who had not experienced condom failure (43.2% compared with 64.3%, P < 0.001). Previous HIV-positive diagnosis was not associated with consistency of condom use (Table 2).
Table 3 displays the adjusted odds ratios (AORs) for consistent condom use in the past month based on 5 different multiple logistic regression models. In the base model controlling for background characteristics alone (i.e., recruitment zone, age, education, income, and number of children), none of these characteristics were significantly associated with consistent condom use in adjusted analysis. After adding the independent variables of interest (nonbarrier modern contraceptive use and relationship type) and control variables (number of clients in the past month, number of noncommercial partners in the past month, recent condom failure, and previous HIV-positive diagnosis), in the full model, consistent condom use was significantly associated with relationship type and condom failure. Compared with new clients, FSWs were less likely to use condoms consistently with regular clients (AOR, 0.30; 95% confidence interval [CI], 0.19–0.47; P < 0.001) or with noncommercial partners (AOR, 0.15; 95% CI, 0.11–0.22; P < 0.001). Those who had experienced condom failure in the past month were less likely to use condoms consistently compared with those who did not report condom failures (AOR, 0.39; 95% CI, 0.27–0.55). Use of nonbarrier modern contraception was not significantly associated with consistent condom use in adjusted analysis, nor was having received a previous HIV-positive diagnosis. There was no significant effect modification between use of nonbarrier methods and any of the 3 relationship types. The association between condom use and nonbarrier contraceptive use also did not vary by awareness of HIV-positive status.
With roughly half of participants reporting that they were HIV positive, Swazi FSWs bear an extremely heavy HIV burden. Condom use did not vary by HIV status, but use was significantly less consistent with more intimate sex partners, with highest reported condom use with new clients and lowest use with noncommercial partners. There are a myriad of potential explanations for this discrepancy, many of which have been described previously.11,12 For example, FSWs may decide whether to use condoms based on an assessment of how risky she perceives a partner to be. Like women in the general population, they may believe that their boyfriends or husbands are not engaging in risky behaviors, and therefore, they opt not to use condoms. Unfortunately, in this hyperendemic context, all sex partners could be considered “risky,” and a partner-specific approach to condom use is not advisable.
However, insisting on condom use can be particularly difficult with noncommercial partners, who may regard condom use as a sign of mistrust or, if they are aware of their partners’ sex work, of being treated like a client.11,12 In addition, although previous studies suggest that noncommercial partners may contribute more to HIV risk than clients,13,14 it is challenging to attempt to change sexual behaviors with intimate relationships. In commercial sex transactions, FSWs may have little or no decision-making power to insist on condom use, particularly if the client offers to pay more for unprotected sex.
The absence of a significant association between condom use and nonbarrier contraceptive use—regardless of the type of relationship with the sex partner—contradicts findings from studies in general female populations.16–18 On one hand, given the relatively high prevalence of nonbarrier modern contraceptive use, it is encouraging that FSWs who are concerned enough about pregnancy prevention to use such methods are not using condoms less consistently. This should help address concerns among some in the HIV prevention community that promoting nonbarrier modern contraception will inevitably lead to reductions in condom use.28 However, owing to the low proportion of women who used condoms consistently in the first place, nearly half the women in our study remain unprotected against HIV and other STIs.
Regardless of whether an FSW is using a nonbarrier modern contraceptive method, health promoters should emphasize the importance of correct, consistent condom use to prevent HIV transmission with all sex partners. There is a need to direct prevention activities not only at FSWs but also at their sexual partners. Male clients and noncommercial partners often have greater control over whether condoms are worn, yet HIV prevention activities rarely target these men, who can be even harder for public health programs to reach than FSWs. In Swaziland, there is no organized sex work infrastructure such as brothels, making it extremely difficult to identify and reach male clients, and the limited existing FSW prevention activities are directed at the women themselves. Focusing prevention activities solely on FSWs may inadvertently contribute further to these women’s stigmatization, solidifying the belief that the spread of disease is their fault and responsibility.14
Another worrisome finding that warrants further investigation is the frequency of condom failure and its negative association with consistent condom use. Incorrect condom use and poor quality condoms may be the underlying causes of the high prevalence of recent condom failure. There are currently few outreach programs targeting sex workers in Swaziland, freely distributed condoms are not lubricated, and there seems to be a need for increased education about correct use of condoms and condom-compatible lubricants.
This study has several limitations that are common to all research conducted on sensitive topics among hidden populations. Despite the demonstrated use of RDS sampling in facilitating efficient recruitment of FSWs, there may be limited generalizability to the general population of FSWs in Swaziland.29,30 Condom use reporting may be biased by social desirability concerns because FSWs may overreport condom use in an effort to appease interviewers. Similarly, the association between inconsistent condom use and condom failure could partly be attributable to a biased response pattern, whereby women who first acknowledge inconsistent condom use falsely claim condom failure as a less blameworthy explanation for inconsistent use.
Studies have previously characterized the relationship between condom use and use of other contraceptive methods in other populations of women, but the results of this study are novel because they examine these associations among sex workers in a country where HIV is endemic. These data highlight the need to provide condoms and condom-compatible lubricants and targeted education programs for sex workers to encourage the consistent use of these commodities with all sex partners, irrespective of the use of other contraceptive methods.
* Swazi FSWs use condoms less consistently with more intimate sex partners.
* Swazi FSWs who use nonbarrier modern contraception do not use condoms less consistently than FSWs who do not use these methods.
* Condom promotion interventions should target male sex partners of Swazi FSWs.
S.B., C.K., D.K., and Z.M. collaborated in the design and implementation of the survey. E.Y. performed the statistical analyses and wrote the initial draft of the manuscript. B.S. and A.T. supported study design and implementation and critically reviewed the manuscript. S.B., C.K., D.K., and Z.M. provided critical inputs and review of the manuscript. All authors had full access to the data, and all authors can take responsibility for the integrity and accuracy of the analysis.
1. Albert AE, Warner DL, Hatcher RA. Facilitating condom use with clients during commercial sex in Nevada’s legal brothels. Am J Public Health 1998; 88: 643–646.
2. Barrientos JE, Bozon M, Ortiz E, et al.. HIV prevalence, AIDS knowledge, and condom use among female sex workers in Santiago, Chile. Cad Saude Publica 2007; 23: 1777–1784.
3. Borquez A, Hallett TB, Gomez GB, et al.. Condom use by female sex workers and their clients in Mexico: Who decides and does it matter? Sex Transm Infect 2011; 87: 254–256.
4. Bruce E, Bauai L, Sapuri M, et al.. HIV knowledge, risk perception, and safer sex practices among female sex workers in Port Moresby, Papua New Guinea. Int J Womens Health 2011; 3: 53–61.
5. Centers for Disease Control and Prevention. Epidemiological notes and reports: Antibody to human immunodeficiency virus in female prostitutes. MMWR Morb Mortal Wkly Rep 1987; 36: 157–161.
6. Dandona R, Dandona L, Gutierrez JP, et al.. High risk of HIV in non-brothel based female sex workers in India. BMC Public Health 2005; 5: 87.
7. Hong Y, Li X. Behavioral studies of female sex workers in China: A literature review and recommendation for future research. AIDS Behav 2008; 12: 623–636.
8. Hooykaas C, van der Pligt J, van Doornum GJ, et al.. Heterosexuals at risk for HIV: Differences between private and commercial partners in sexual behaviour and condom use. AIDS 1989; 3: 525–532.
9. Kayembe PK, Mapatano MA, Busangu AF, et al.. Determinants of consistent condom use among female commercial sex workers in the Democratic Republic of Congo: Implications for interventions. Sex Transm Infect 2008; 84: 202–206.
10. Pickering H, Quigley M, Hayes RJ, et al.. Determinants of condom use in 24,000 prostitute/client contacts in the Gambia. AIDS 1993; 7: 1093–1098.
11. Morris M, Pramualratana A, Podhisita C, et al.. The relational determinants of condom use with commercial sex partners in Thailand. AIDS 1995; 9: 507–515.
12. Kerrigan D, Ellen JM, Moreno L, et al.. Environmental-structural factors significantly associated with consistent condom use among female sex workers in the Dominican Republic. AIDS 2003; 17: 415–423.
13. Syvertsen JL, Robertson AM, Abramovitz D, et al.. Study protocol for the recruitment of female sex workers and their non-commercial partners into couple-based HIV research. BMC Public Health 2012; 12: 136.
14. Lowndes CM, Alary M, Gnintoungbe CA, et al.. Management of sexually transmitted diseases and HIV prevention in men at high risk: Targeting clients and non-paying sexual partners of female sex workers in Benin. AIDS 2000; 14: 2523–2534.
15. Cote AM, Sobela F, Dzokoto A, et al.. Transactional sex is the driving force in the dynamics of HIV in Accra, Ghana. AIDS 2004; 18: 917–925.
16. Cooper ML, Agocha VB, Powers AM. Motivations for condom use: Do pregnancy prevention goals undermine disease prevention among heterosexual young adults? Health Psychol 1999; 18: 464–474.
17. Roye CF. Condom use by Hispanic and African-American adolescent girls who use hormonal contraception. J Adolesc Health 1998; 23: 205–211.
18. Sangi-Haghpeykar H, Horth F, Poindexter AN 3rd. Condom use among sterilized and nonsterilized Hispanic women. Sex Transm Dis 2001; 28: 546–551.
19. Cates W, Steiner MJ. Dual protection against unintended pregnancy and sexually transmitted infections: What is the best contraceptive approach? Sex Transm Dis 2002; 20: 168–174.
20. Reed JB, Justman J, Bicego G, et al. Estimating national HIV incidence from directly observed seroconversions in the Swaziland HIV Incidence Measurement Survey (SHIMS) longitudinal cohort. Proceedings of the XIX International AIDS Conference; July 22–27, 2012; Washington, DC, USA.
21. National Emergency Response Council on HIV/AIDS (NERCHA). Swaziland HIV Prevention Response and Modes of Transmission Analysis. Mbabane: NERCHA, 2009.
22. Heckathorn DD. Respondent-driven sampling: A new approach to the study of hidden populations. Soc Probl 1997; 44: 174–199.
23. Salganik MJ, Heckathorn DD. Sampling and estimation in hidden populations using respondent-driven sampling. Sociol Methodol 2004; 34: 193–239.
24. StataCorp. Stata Statistical Software: Release 11 [computer program]. Version 11. College Station, TX: 2009.
25. Salganik MJ. Variance estimation, design effects, and sample size calculations for respondent-driven sampling. J Urban Health 2006; 83: 98–112.
26. Schonlau M, Liebau E. Respondent-driven sampling. Stata J 2012; 12: 72–93.
28. Askew I, Berer M. The contribution of sexual and reproductive health services to the fight against HIV/AIDS: A review. Reprod Health Matters 2003; 11: 51–73.
29. McCreesh N, Frost SD, Seeley J, et al.. Evaluation of respondent-driven sampling. Epidemiology 2012; 23: 138–147.
30. Salganik MJ. Commentary: Respondent-driven sampling in the real world. Epidemiology 2012; 23: 148–150.