Abstract
Comprehensive intervention packages are recommended to address multiple sources of HIV risk for adolescent girls and young women (AGYW). DREAMS is a multi-component HIV prevention program designed to reduce HIV incidence among AGYW. We conducted a prospective cohort study among AGYW aged 13-22 years, randomly selected in rural Gem and urban Nairobi informal settlements followed from 2017/2018-2019. AGYW were classified into three groups: (1) invited to DREAMS and received a "complete" package, (2) invited and received a "partial" package, or (3) not invited to DREAMS. We defined the "complete" package as 4-5 primary interventions in Gem and 5 in Nairobi: the "partial" package as 3 specific interventions in Gem and any 3-4 interventions in Nairobi. We used propensity score-adjusted logistic regression to estimate the causal effect of DREAMS on outcomes under three counterfactual scenarios: all AGYW accessed the complete package, all accessed a partial package, or none were invited. In Nairobi, 1081 AGYW were enrolled. By 2019, 26% accessed the complete package and 32% accessed the partial package. Among those receiving the complete package, there was increase in HIV status knowledge(24.8% [95%CI:16.4,32.6]),social support(13.9% [95%CI:3.3,23.6]) and self-efficacy(10.3% [95%CI:0.5,20.4]) and a decrease in the proportion with ≥2 lifetime partners(-8.0% [95%CI:-15.9,0.0]). In Gem, 1171 AGYW were enrolled. By 2019, 24% received the complete package and 21% received the partial package. We found evidence of an increase in HIV status knowledge(10.0% [95%CI:4.5,15.2]), social support(27.2% [95%CI:19.2,35.5]) and a decrease in condomless sex(-9.1% [95%CI:-13.6,-4.1]), and the proportion with ≥2 lifetime partners(-7.6% [95%CI:-12.4,-2.2]) for the complete package. Among those receiving the partial package, there was a decrease in condomless sex(-12.2% [95%CI: -17.0,-6.4]), and an increase in self-efficacy(8.0% [95%CI:0.0,17]). A package of 4-5 primary DREAMS interventions had positive impacts on multiple HIV-related outcomes in both settings. A partial package was effective in Gem, but not in Nairobi, suggesting the need for context-specific intervention strategies.