HIV Risk and Associations of HIV Infection among men who have sex with men in Peri-Urban Cape Town, South Africa

South Africa has a generalized HIV epidemic with an estimated HIV prevalence of 16.9% among adults 15 and older and 13.1% among men aged 15 and older [1]. HIV transmission has been most commonly linked with high risk heterosexual and vertical transmission and consequently this is where the majority of HIV prevention efforts have been targeted. However, across the African continent, the sexual transmission of HIV between men has been increasingly recognized [2]. HIV prevalence studies among men who have sex with men (MSM) have been completed in numerous countries in Southern and Eastern Africa including Malawi, Namibia, Botswana, Tanzania, Uganda, Kenya, and South Africa  [3‐5].

Notably, the initial presentation of the HIV epidemic in major urban centers of South Africa was among gay men and other MSM [6]. A study of 250 MSM in Johannesburg in 1983 demonstrated an HIV prevalence of 12.8% with 35.6% infected with syphilis, and another 41.2% positive for Hepatitis B virus antibody [7]. The last few years has seen a resurgence of studies focused on MSM, but there was limited epidemiologic research conducted in the interim three decades. In 2006, a study of rural South African men found that approximately 3.6% of men studied (n = 46/1277) reported a history of having had sex with another man [8]. The adjusted odds ratio for prevalent HIV infections was 3.6 (95% CI 1.0-13.0, p = 0.05) for those reporting same sex practices [8]. One of the first studies published that targeted MSM was completed using respondent driven sampling (RDS) by Lane et al. in Soweto, a peri-urban community south-west of Johannesburg, in 2008 [9, 10]. Among 378 MSM, the RDS-weighted HIV prevalence was 13.2% (95% CI 12.4-13.9) with independent associations of prevalent HIV infection being older than 25 (adjusted odds ratio (aOR) 3.8, 95% CI 3.2-4.6), reporting any unprotected receptive anal intercourse (aOR 4.4, 95% CI 3.5-5.7), and having between 6 and 9 male partners in preceding 6 months (aOR 5.7, 95% CI 4.0-8.2). In 2009, an HIV prevalence and risk assessment of 542 MSM attending lesbian, gay, and bisexual (LGBT) venues in urban Cape Town showed an HIV prevalence of 10.4% (56/539). Self-identifying as gay, homosexual, or queer (aOR 4.5, 95% CI 1.0-20.0), and reporting ever having had a sexually transmitted infection (STI) (aOR 4.3, 95% CI: 2.3-8.3) were significantly associated with testing HIV positive [11].

More recently published studies have demonstrated higher HIV prevalence rates including the Johannesburg/eThekwini Men's Study (JEMS) which demonstrated a HIV prevalence rate of 49.5% among 204 MSM in Johannesburg and 27.5% among 81 MSM in Durban [12]. MSM in urban Cape Town also participated in the first multisite randomized controlled trial of antiviral chemoprophylaxis completed [13]. While the HIV prevalence was higher than previous estimates, associations of infection were similar with self-reporting as gay or homosexual (aOR, 8.4; 95% CI, 3.7-19), receptive unprotected anal intercourse (aOR 4.3; 95% CI 2.4-7.6); sex with a person known to be HIV positive (aOR 2.3; 95% CI 1.1-4.9); and a sexually transmitted infection diagnosis (aOR 2.4; 95% CI 1.1-5.2).

South Africa has an HIV epidemic that has disproportionately affected women and girls, especially among those aged 15-29 where prevalence rates have been shown to be significantly higher than their male age counterparts [14]. Fortunately, there is evidence that there have been marginal recent declines in incident HIV infections especially among women and girls resulting in declines in the overall epidemic [15]. Among similarly aged men, HIV prevalence rates have been consistently under 10% HIV prevalence in South Africa [14, 15]. While there is significant variability among the estimates of the burden of HIV among MSM in South Africa, when comparing the rates to age-matched men not reporting same-sex practices, the rates are uniformly and significantly higher. Comparisons of prevalence across sampling designs and time frames are always problematic, but this is particularly challenging for hidden and hard-to-reach populations, where behaviours are stigmatized, and for populations for which size estimates are uncertain. All of these potential sources of measurement error pertain to MSM in African settings. Nevertheless, enough data has emerged to suggest significant HIV disease burdens among these men.

The study presented here represents the first evaluation of HIV prevalence among MSM in the peri-urban townships of Cape Town. The study was designed as an epidemiologic probe of MSM in the townships in terms of socio-demographic characteristics, HIV-risk practices, HIV prevalence and associations with being seropositive, and prevalence of human rights violations. The aims of this project were to characterize this population, assess levels of need, and identify barriers to access to HIV preventive services.

This study presents the first data describing burden of HIV and associations of prevalent HIV infection among MSM in the Peri-urban townships surrounding Cape Town. As with the majority of studies of MSM in Africa, the study sample tended to be young, with a mean age of 25. However, the study demonstrated a significant and disproportionate burden of disease among these young men [15]. While analytic comparisons using direct or indirect age standardization have not yet been done to the knowledge of the authors, these data along with other recent descriptions of predominantly young MSM suggest that HIV risks and rates are consistently high in Sub-Saharan Africa [2].

The estimates of HIV prevalence observed here were significantly higher than a recent HIV prevalence study of MSM in urban Cape Town [11]. In that study, HIV prevalence was found to be 10.6% (n = 57/538) as compared to 25.5% (n = 51/200) observed in this study (p < 0.05) (23). While prevalence in the townships was higher, there were not notable differences observed in levels of sexual risk practices. This apparent contradiction may be due to varying structural drivers of HIV-related risk, including social determinants of health and human rights violations. Network level factors, including the background population HIV prevalence, may also play a role.

Human rights violations in this sample of MSM were common, with nearly a quarter of the participants reporting at least one rights violation. This study demonstrated the public health outcomes of these rights violations, including a higher prevalence of HIV among those MSM who reported blackmail, both in bivariate and multivariate analyses. The experience of these men suggests that there is a disconnect between legally mandated protections in the constitution and daily life in for MSM in the townships studied here [18, 19]. The implications for combination HIV prevention interventions are significant as these results suggest that human rights violations would interfere with the ability to identify, accrue, and prospectively retain MSM.

This study found lower rates of active bisexuality and bisexual concurrency among MSM than in other countries in the region. This may be related to better social and legal contexts for gay men in South Africa as compared to other South African Development Community (SADC) countries (3;14). Understanding the associations of bisexual practices and bisexually concurrent partnerships can be help inform programs targeting behavior change and partner reduction that traditionally target exclusively heterosexual multiple concurrent partnerships. While qualitative studies are needed to further characterize drivers of bisexual practices, it is conceivable that men who feel more comfortable in same-sex relationships will be less likely to have female partners. Encouragingly, Black and Coloured South African MSM practicing active bisexuality reported higher rates of condom use and were less likely to be infected with HIV then were men who reported only sex with men. The observation of lower HIV risks among bisexually active MSM has been seen in a number of studies of MSM in the African context (10;14;15;26).

There were numerous limitations with the study design limiting the generalizability of the results to all MSM in South Africa. Primarily, this study was cross-sectional which prevents any characterization of causality between the associations described here and incident HIV infections. In addition, recruitment occurred at social venues for gay men and other MSM in townships which limits the generalizability of the findings to MSM who frequent these social venues. While the men were accrued using venue-based sampling, there was no sampling frame of all venues characterized highlighting that these results are not probability estimates of risk and burden of disease. Strategies that may be feasible to obtain probability estimates or unbiased non-probability estimates include venue-day-time sampling or respondent driven sampling (RDS). Gay venues in the townships tend not be brick and mortar bars, as seen in urban Cape Town; rather they are more clandestine venues that shift location. While this would limit time-location sampling such as has been done successfully with MSM in Thailand [20], RDS is an option to generate unbiased parameter estimates approximating population-based estimates of disease burden and predictors of disease among MSM in these contexts [21]. Population-based HIV surveillance strategies such as demographic health surveys are not well suited for MSM in Africa given that these tend to be household based surveys. If administered in a household, same-sex practices may be underreported. The aforementioned active and targeted recruitment methods will likely be required to accrue MSM in settings where these practices remain stigmatized and even criminalized.

HIV risk-related practices among these men were very common, with relatively low reported rates of both always wearing condoms with all sexual partners, and using water-based lubricants during sex. Indeed, 0 men reported regularly practicing safe sex with either men or women as defined above. There were numerous individual-level associations with HIV risk that were expected based on previous studies including not wearing condoms, higher numbers of male partners, and having received money for transactional sex [3, 22, 23]. However, as described there were also structural drivers of HIV including rape and blackmail. In response to these multilevel risks for HIV infection among MSM, it is vital to explore and evaluate combination HIV prevention interventions (CHPI) for MSM including behavioural, biomedical, and structural interventions [24].

While there were limitations in the methodology of the study, it is appropriate to conclude that there are MSM living in peri-urban Cape Town townships that are at significant risk of HIV infection. With the emergence of new biomedical strategies such as oral and topical antiviral chemoprophylaxis to prevent HIV infection, these men should be considered as a high risk population that population that are appropriate for inclusion into large scale prevention trials. However, to ensure appropriate risk identification accrual and retention of these men, structural interventions such as health sector interventions and building social capital should be included in the package of combination HIV preventive interventions.