Sex, drugs and intersecting risks : HIV among people who belong to more than one key population in the United States

Abstract

Background: The HIV epidemic in the United States (US) is mainly concentrated in so-called ‘key populations’ including men who have sex with men (MSM) and people who inject drugs (PWID). In addition, other groups such as people of low socioeconomic status and people belonging to some ethnic minorities have a higher prevalence of HIV. Globally, people who exchange sex for money or drugs are recognised as another key population with high HIV prevalence, but there is limited recent data from rigorous studies in the United States on HIV prevalence among people who exchange sex. High prevalence among certain groups may be a combination of individual risk behaviours such as condomless sex, drug use and unsafe injection practices, and structural factors such as poverty, violence and residential segregation that can impact HIV risk indirectly. This thesis examines whether, among populations known to be at high risk for HIV, prevalence is higher among those who belong to more than one key population or vulnerable group and for whom several risk factors intersect. Furthermore, it examines sociodemographic factors and sexual and drug-use risk behaviours in these populations and how they may be relevant to HIV acquisition and transmission.

Methods. Data: I worked with the US National HIV Behavioral Surveillance System (NHBS), which is a surveillance system that collects data from three populations at high risk for HIV in annual rotating cycles: men who have sex with men (MSM), people who inject drugs and heterosexuals at increased risk of HIV (referred to as ‘IDU’ and ‘HET’, respectively). We recruited participants for all three cycles in around 20 large US cities on an annual rotating schedule. For the IDU and HET cycles, we use respondent-driven sampling (RDS), which is a sampling method specifically designed to reach hidden populations and approximate a random sample and where participants recruit each other using coupons. In 2016 I led a pilot data collection focused specifically on women in five cities who exchange sex, also using RDS. During the MSM cycle we recruited participants through venue-based sampling which allows random sampling of venues in a city, time-slots within venues and individual men attending the venue. In all cycles participants take an interviewer-administered survey asking about demographic characteristics, sexual and drug-use risk behaviours and access to services such as frequency of HIV testing and use of health care and preventive services. Participants are also offered a rapid HIV test and receive an incentive for taking the survey and the HIV test.

Analysis: For Papers 1 and 2, I looked at women who inject drugs and MSM to examine whether people who exchange sex are more likely to be HIV-infected, including being HIV-positive but unaware of one’s positive status (HIV-positive–unaware), compared with those who do not exchange sex belonging to the same populations. Paper 3 estimates the HIV prevalence among women who exchange sex and compares it with the prevalence among women of low socioeconomic status who do not exchange sex from the same cities three years earlier. In these papers, exchange sex is defined as having had oral, vaginal or anal sex with a male partner in the past 12 months. In Paper 4 I used a subset of the IDU survey data to examine whether MSM who inject drugs (MSM–IDU) are more likely to be HIV infected if they report methamphetamine as their primary drug compared with other drugs. For bivariable and multivariable analyses in Papers 1, 2 and 4 I used generalised estimating equations (GEE), in PROC GENMOD in SAS v. 9.2 or 9.3. The GEE method enables analyses of clustered data where observations in a cluster are thought to be more similar to each other than to other observations. For Paper 3 we used RDSAT to estimate the prevalence of HIV and risk behaviours among women who exchange sex.

Results: In Paper 1, 10% of women who injected drugs and exchange sex were HIV infected. There was no statistically significant difference in HIV prevalence between those who exchanged sex and those who did not (10.0% vs 7.4%, P = 0.33). However, those who exchanged sex were more likely to be HIV-positive but unaware of their positive status (HIV-positive–unaware) compared with those who did not exchange sex: 5.0% vs 2.6% (P = 0.01). This difference remained significant in multivariable analysis with an adjusted prevalence ratio (aPR) of 1.97 (95% CI 1.31–2.97). In Paper 2, HIV prevalence among MSM who exchanged sex was higher than among MSM who did not (29.1% vs 17.7%, P < 0.001). However, this difference became non-significant in multivariable analysis. MSM who exchanged sex were also more likely to be HIV-positive–unaware than those who did not exchange sex (13.2% vs 5.6%, P ≤ 0.001) and this difference remained in multivariable analysis (aPR 1.34, 95% CI 1.05–1.69). In Paper 3, the prevalence of HIV among women who exchanged sex was 4.9%, approximately three times as high as the prevalence of HIV among women of low socioeconomic status who did not exchange sex (1.6%), and almost nine times as high compared to women in the general population (0.55%). In Paper 4, MSM who primarily injected methamphetamine were significantly more likely to be HIV-positive (29.3%) than MSM who primarily injected other drugs (15.5%, aPR 1.48, 95% CI 1.08–2.03). This association was mediated by sexual risk behaviours, but not drug-use risk behaviours. Among people who exchange sex, and among MSM who inject methamphetamine, sexual and drug-use risk behaviours were common, putting people at risk for HIV acquisition as well as onward transmission. Exchange sex was furthermore common among people living in poverty, homelessness or with other markers of low socioeconomic status.

Conclusions: The populations studied in this thesis are already known to be at high risk for HIV. This thesis demonstrates that HIV prevalence – in particular the prevalence of being HIV positive but unaware of one’s status – is high among people who belong to more than one key population or vulnerable group. Sexual and drug-use risk behaviours are common. To address the risk for HIV acquisition and onward transmission it is essential to consider a variety of services around prevention, including harm reduction, and testing and linkage to care and treatment. However, narrowly targeted services focusing on a single key population such as people who inject drugs or MSM may not adequately address the needs of those who belong to more than one key population. Additionally, the higher-order structural factors that put individuals and communities at risk for HIV must be addressed.