Meta-analysis of Pregnancy Events in Biomedical HIV Prevention Trials in Sub-Saharan Africa: Implications for Gender...

Historically, pregnant and lactating populations (PLP) have been excluded or disenrolled from biomedical HIV prevention trials, despite being more likely to acquire HIV during pregnancy and the post-partum period. We conducted a meta-analysis of pregnancy events in biomedical HIV prevention trials in sub-Saharan Africa to support trialists moving toward more inclusive clinical and implementation studies. We searched peer-reviewed literature reporting pregnancy events and contraceptive requirements in HIV prevention trials between 2001 and 2022. We hypothesized four variables to explain variation: contraceptive requirements, study start year, study product, and sub-region. We fit a meta-analytic model to estimate individual effect sizes and sampling variances, then conducted sub-group analyses to assess moderating effects. We identified 38 references for inclusion, across which the proportion of pregnancy events was 8% (95% confidence interval [CI]: 6-10%) with high heterogeneity (I² = 99%). Studies not requiring contraceptives (21%, 95%CI: 7-48%) reported a significantly higher proportion of pregnancy events than studies requiring two methods (5%, 95%CI: 2-10%). Studies launched between 2001 and 2007 (11%, 95%CI: 8-16%), microbicide gel trials (12%, 95%CI: 8-18%), and studies conducted in Western Africa (28%, 95%CI: 13-51%) reported higher proportions of pregnancy events than reference groups. Together, these variables have a moderating effect on pregnancy events (p < 0.0001), explaining 63% of heterogeneity in trials. Results describe how, over time, more stringent contraceptive requirements reduced pregnancy events, which ensured necessary statistical power but limited reproductive choice by participants. With the move toward continuing PLP on experimental products, trialists can utilize estimated pregnancy events reported here to inform strategies that accommodate participants' changing fertility preferences.