Our study found an estimated HIV prevalence of 8.3% among study participants and a RDS-weighted HIV prevalence of 3.6% (95% CI: 1.3–6.0%) among TSMSM in Nairobi. The weighted HIV prevalence observed among TSMSM in our study (96% aged 18–24 years) was six times higher than that of young Kenyan men in the general population aged 18–24 years, which is 0.6%[6]. Accordingly, the recent call to have programming that caters for the HIV prevention needs of YMSM in Africa [30], is quite relevant for TSMSM in Nairobi. The weighted HIV prevalence observed is comparable to the 4.1% HIV prevalence found among TSMSM in both China [10] and South Africa [11]. Notably, with only 77.3% of study participants knowing their HIV status to be seropositive before the study, there is need to scale up HIV testing in this population so as to reach the first 95 of the UNAIDS fast-track 95-95-95 [31]. HIV self-testing is readily available and acceptable in Kenya [32], and should thus be harnessed to increase testing rates in this key population.

Higher HIV prevalence was observed among TSMSM studying in private institutions and staying at home with their families. We hypothesize that most TSMSM who study in private institutions and stay with their families have predominantly lived most of their lives in the city, which is more liberal than the rural areas of the country. Therefore, it is likely that these TSMSM could have started experimenting with anal sex at an earlier age than their rural counterparts who due to costs, mostly tend to join the public institutions and stay either in campus or rented accommodation outside their campuses. For these reasons, there could be an increased exposure period for the former, leading to the elevated risk of HIV infection. Indeed, even higher HIV prevalence (30.3%) than we found, has been observed among YMSM in urban Indonesia [33], a country where it is not illegal to be homosexual.

Meeting sex partners online was significantly associated with HIV infection. This finding is consistent with that of a previous study in Kenya which showed that MSM who sought sex partners online had four-fold greater odds of having HIV infection compared to those who sought sex partners from physical sites [34]. Additional research in sub-Saharan Africa has demonstrated that MSM who seek sex partners online have higher prevalent HIV infection and are more likely to be younger [35], like our study participants. Young MSM, including TSMSM who use the internet to find sex partners are a critical population to target for HIV interventions, since they are inclined to have increased levels of sexual risk behavior, and the internet itself may be a promising mechanism to deliver interventions [36]. Like in our study where 89.6% owned a smart phone, another study showed that 95.6% of tertiary students in Kenya own smart phones [37], and use the phones to access the internet. Future YMSM-focused risk reduction interventions should leverage the internet to host campaigns and safe sex messaging efforts to directly and discretely reach TSMSM who predominantly seek sex partners online. This is necessary since TSMSM—as a result of criminalization of homosexuality, societal stigma and discrimination, may be afraid to seek services from programs that largely depend on physical outreach, thus resulting in missed opportunities for HIV prevention.

Another significant risk factor identified for HIV infection was having reported the last sex partner being > 25 years. This is consistent with the findings of a study in the USA which showed that among YMSM, the odds of HIV infection were significantly elevated as the age of sexual partners increased [38]. This could be explained by the fact that risk of HIV infection generally increases with age, thus increasing the risk of HIV infection among YMSM who have sex with older male partners. In addition, practices that increase the risk of HIV infection such as condomless anal sex have been shown to be common in age-disparate sexual relationships among MSM, and are significantly associated with unrecognized HIV infection [39]. Previous research has shown that YMSM are motivated to seek older MSM partners for economic and social support [40]. Subsequently, the economic power imbalance between YMSM and older MSM may diminish the former’s ability to negotiate for safer sex [41], with YMSM previously reporting sexual coercion by older MSM [42]. Since the YMSM—older MSM partnerships are more likely to persist than wane, HIV prevention programs for YMSM should include information about the HIV risk associated with older MSM partners and skills for negotiating for safer sex, particularly the correct and consistent use of condoms. Additionally, TSMSM should be provided with more tools for HIV prevention, including oral pre-exposure prophylaxis (PrEP) which is available in Kenya [43], and offers YMSM more control on usage as compared to condoms which need negotiation between sex partners. As well, since YMSM may choose more pleasurable (condomless sex) over safer sex, PrEP would offer them the opportunity to enjoy sex without fear of HIV infection, the risk of acquiring other STIs notwithstanding [44].

A notable independent risk factor for HIV infection that was observed in our study was testing positive for NG. This coupled with the high prevalence of CT observed (58.7%) has implications for HIV control and prevention. The causal relationship between prevalent CT/NG and incident HIV has previously been established, with CT/NG shown to increase both the risk of HIV acquisition and transmission [45]. Among MSM, rectal CT/NG re-infection has been associated with increased risk of HIV seroconversion [46], with modelling studies demonstrating that 10.4% of incident HIV infections are attributable to CT/NG infection [47]. TSMSM should therefore be encouraged to have diagnostic STI screening often, depending on sexual activity and risk-taking, and thereafter receive appropriate care. This would help mitigate the deleterious synergy between STI and HIV, as well as forestall morbidities such as infertility that occur secondary to some STI.

Although not independently associated with HIV infection in our study, other behaviors that increase the risk of HIV infection were common. These include engaging in condomless sex, group sex, transactional sex, alcohol and drug use during sex, as well as having multiple sex partners. These findings are consistent with what has been observed among TSMSM in China [10] and South Africa [13]. Behavioral change is therefore urgently needed in this population to augment the biomedical tools available for HIV prevention.

To our knowledge, this study was the first one to assess the prevalence and associated factors of HIV among TSMSM in Kenya, and possibly in sub-Saharan Africa using the RDS method. These findings should be viewed in light of some limitations. Firstly, RDS is not a classical probabilistic sampling method. To offset this limitation, when calculating our sample size, we applied a design effect of 3 to account for the clustering that occurs due to homophily and minimize the traditional bias associated with snowball sampling. We also excluded from analysis the 6 seeds who were purposely selected to start off recruitment. Secondly, since we asked about what happened in the past year, there was a possibility of recall bias. To mitigate this, we aimed to include in our questionnaire either single or multiple choice questions and have very few instances where participants needed to type out answers. We also limited the number of questions about events that happened more than 12 months before the study. Thirdly, we asked questions about sexual behavior which may be affected by social desirability bias. To minimize this, participants self-administered the survey on an online platform accessed through tablet computers, since this has been shown to be more effective in offsetting this kind of bias, as compared to face-to-face interviews [48]. Fourthly, the cross-sectional nature of the study limits causal inference of the independent risk factors identified for HIV infection. Finally, though we used a comprehensive set of variables in the logistic regression analysis, we cannot rule out the possibility of residual confounding from other variables such as experiences of stigma, discrimination and violence. Despite these limitations, this survey offers valuable lessons on the burden of HIV and associated factors among TSMSM, and hopefully portends the development of tailored interventions for the HIV response in this key yet understudied population.