Sexual partners play a critical role in HIV transmission, but efforts to study and target sexual partner types for HIV prevention have been stymied by current measurement approaches, which have not shown clear associations with HIV acquisition, and do not provide clear guidance on the design of specific, targeted interventions to prevent HIV acquisition across different sexual partner types and contexts.
This secondary analysis examined the sexual partners of 1034 sexually active, adolescent girls and young women (AGYW), ages 13-20, enrolled in a 3-year randomized controlled trial of cash transfers for HIV prevention (HPTN 068) in Mpumalanga Provence, South Africa. AGYW were tested for HIV infection annually and reported the following 10 indicators for each sexual partner (up to three partners per visit): partner age, school enrollment, children with AGYW, children with other women, cohabit with AGYW, sex only one time, always use condoms, partner HIV-status, partner concurrency status, and transactional sex with partner. We used these indicators to identify sexual partner types using Latent Class Analysis (LCA), and estimated risk ratios (RR) and 95% confidence intervals (CI) for the association between LCA-identified sexual partner type and incident HIV-infection using generalized estimating equations with a robust variance estimator and exchangeable correlation matrix, controlling for confounders.
Over the course of 2140 AGYW-visits, 1034 AGYW reported 2968 sexual partners, and 63 AGYW become HIV infected. We identified six, distinct sexual partner types: only one partner type was older, two were not enrolled in school, condom use was low across all but one type, while transactional sex was present in all but one type (Table 1). Compared to AGYW with ‘monogamous’ partners, AGYW with ‘older’ partners had more than 3 times the risk of HIV infection (RR: 3.35, 95% CI: 1.43, 7.85), while AGYW with ‘unprotected’ partners had more than 2 times the risk of HIV infection (RR: 2.45, 95% CI: 1.11, 5.44). AGYW with ‘casual protected’ or ‘anonymous’ partners were at increased risk of infection, while AGYW with ‘cohabiting’ partners were at decreased risk compared to AGYW with ‘monogamous’ partners (results not statistically significant).
Partner types based on explicit, reported partner characteristics predict incident HIV-infection among AGYW, and offer an alternative approach for measuring and targeting specific partner types for HIV research and intervention.