Does enhanced HIV prevention, diagnosis, and linkage to care reduce hospitalisation in high HIV-burden communities in Zambia and South Africa? findings from the HPTN 071 (PopART) randomised trial

The objective of this study is to explore if a community-based HIV combination prevention intervention reduced inpatient hospitalisations in Zambia and South Africa by diagnosing HIV and TB in earlier stages of disease progression, thereby preventing severe disease and new infections. As part of the HIV Prevention Trials Network (HPTN) 071 trial, hospitalisation data from a cohort of 16 968 consenting randomly sampled adults aged 18–44 years were collected between 28 November 2013 and 16 November 2018 across 21 communities in Zambia and South Africa across three study arms. Arm A included annual visits by Community HIV-care Providers (CHiPs) and universal linkage to care for ART initiation for all PLWH (irrespective of CD4 count); arm B included annual CHiPs visits and ART per local guidelines; control arm C received the standard of care provided at government clinics, including HIV testing and ART offered according to local guidelines. For this study, we used a cluster-level two-stage analysis and adjusted for covariates that were unbalanced across intervention arms. Covariates included in the models were the cluster’s baseline HIV prevalence and hospitalisation rate and data on the respondent’s gender, age, educational attainment, and socio-economic status. Out of the pooled sample of 13 964 responses from the three post-baseline surveys, 439 (3.14%) reported hospitalisation in the past 12 months – 234 (1.68%) when excluding hospital admissions for births or injury. Comparing hospitalisations in the intervention and control arm clusters, the estimated adjusted risk ratio was 1.03 [0.64–1.66] for the full sample and 0.82 [0.39–1.74] for PLWH. We find no compelling evidence of impact of the HPTN071 (PopART) community-wide combination HIV prevention intervention on in-patient hospitalisation among a general population sample.