What combination of interventions can optimise HIV prevention for adolescent girls and young women? Cohort analysis of DREAMS participation in urban and rural Kenya.

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.