A survey weighted analysis of HPTN 071 (PopART) primary outcome of HIV incidence

Abstract

Introduction: HPTN 071 (PopART) implemented a comprehensive HIV prevention package which aimed to reduce HIV incidence within 21 communities of Zambia and South Africa: Arm A, PopART intervention of universal HIV testing and treatment; Arm B, PopART intervention of universal HIV testing with ART provided according to local guidelines; and Arm C, standard of care. Analyses so far have not accounted for the sampling design of the enrolled cohort. We performed a sample-weighted re-analysis of the primary outcome of the PopART trial to derive a population-based estimate of the intervention effect. Methods: Enrollment used a two-stage sampling design: household and adult participant within each household. We constructed post-stratification weights to match the age and sex distribution of the target population in these communities. Weighted Poisson regression was used to estimate community-level HIV incidence. The PopART intervention effect was estimated using log-transformed community-level incidence estimates in an ANCOVA model. Results: The analysis based on community-level incidence shows a 25% reduction in incidence for Arm B communities compared to standard of care (RR: 0.75, 95% CI: 0.56-1.02, p = 0.06) while Arm A communities show no difference in HIV incidence compared to standard of care (RR: 1.08, 95% CI: 0.81-1.46, p = 0.56). Conclusions: Our re-analysis shows 25% reduction in HIV incidence comparing Arm B to Arm C communities. No effect was observed comparing Arm A communities to Arm C communities. These results align with the primary results of the PopART trial.