Background: Identifying and reaching the most-at-risk AGYW with combination HIV prevention programming is critical to reducing HIV incidence. Yet there remains ambiguity on how to identify the most vulnerable AGYW and if multi-sectoral HIV prevention programs can reduce HIV risk for the most vulnerable. We used latent class analysis (LCA) to define classify out-of-school AGYW enrolled in DREAMS programs into vulnerability groups, and assess changes in HIV risk vulnerability group after 12 months of enrollment.
Methods: AGYW enrolled in DREAMS programs aged 15-24 were surveyed Jul-Oct 2017 and Sep-Nov 2018 across two sites in Malawi (n=1,157). Surveys captured knowledge, attitudes, practices, and participation in DREAMS interventions (e.g., life skills, savings and loans). We used LCA models using measures on household & family characteristics, gender attitudes, and HIV knowledge to define vulnerability profiles of AGYW. Multiple logistic regression analyses examined change over time in risk behaviors by vulnerability class.
Results: We identified two distinct HIV vulnerability profiles?higher (56%) and lower (44%). At enrollment, AGYW with a high vulnerability profile had higher odds of engaging in transactional sex, experiencing sexual violence, and having STI symptoms compared to AGYW with a low vulnerability profile. At follow-up, there were no significant differences in program participation or in risk behaviors when comparing high vulnerability profile to low vulnerability profile.
Over time (table 1), AGYW with both high and low vulnerability profiles had significant reductions in multiple sexual partners, STI experiences, and sexual violence from intimate partners and non-partners. There were no significant changes in consistent condom use and transactional sex for either profile. AGYW with low vulnerability profiles had higher odds of having a transactional relationship with a main partner between enrollment and follow-up.
Conclusions: DREAMS programs were successful in reducing HIV-risk behaviors for AGYW with both low and high risk vulnerability profiles, highlighting the need for continued investments in multi-sectoral HIV prevention programs. Research techniques, like LCA, can help programmers classify and research highly vulnerable AGYW.