Using observational cohort data from Key populations to plan HIV intervention studies

Abstract

Background: Globally, new HIV-infections continue to occur mostly in Sub-Saharan Africa despite the known-to-work HIV-prevention interventions. Suboptimal adherence to the available HIV prevention interventions is cited. Vaccination could help minimise non-adherence to an HIV preventive intervention but does require the completion of the full vaccination schedule. An HIV vaccine would be a useful addition to HIV prevention packages, but vaccines need assessment in efficacy trials and investigators need suitable populations for trials. Key populations including Fisher-folks (FF) and Female Sex Workers (FSW) in Uganda could be useful. However, available data for planning trials in these populations come from observational cohorts and evidence suggests that trial environment and/or participants selection could differ from observational cohorts, which could alter trial targeted outcomes. This difference was investigated using Simulated HIV-vaccine efficacy trials (SiVETs); a trial that mimicked an HIV vaccine efficacy trial using a proxy vaccine. Methods: Two SiVETs were nested within observational cohorts of FF (2012 – 2014) and FSW (2014-2017). The SiVETs screened and enrolled participants from observational cohorts, and administered a licensed Hepatitis B vaccine at 0, 1 and 6 months as a proxy for an HIV-vaccine. Over the 12 month follow-up, SiVETs conducted HIV testing, risk behaviour assessment, and promoted and provided reliable contraceptives to women. Results: In total, there were 3989 [1575 FF & 2414 FSW] participants in the observational cohorts and 572 [282 FF & 290 FSW] of these were enrolled into SiVETs. There were significant differences between characteristics of participants in SiVETs and those in the observational cohorts. At 12-months, HIV incidence and risk behaviours were higher in the observational cohorts than SiVETs while retention was lower. Promotion and provision of reliable contraceptives in SiVETs increased the proportion of women using them from 55% at baseline to >90% at the end of vaccination. Conclusion: Researchers designing HIV efficacy trials using observational data in these and similar populations need to consider potential for changes in the targeted trial outcomes following recruitment into trials and its effect on trial statistical power.