Using statistical and epidemiological methods to address challenges in the design of HIV prevention studies, in the era of highly effective pre- exposure prophylaxis.

Background: Highly effective HIV pre-exposure prophylaxis (PrEP) interventions are available. As a result, trials of experimental PrEP regimens can no longer be designed as placebo-controlled studies, require large sample sizes, and must recruit from populations at very high risk of HIV. Utilising data from PrEPVacc (a trial of new PrEP and vaccine regimens) and its pre-trial registration cohort, I aimed to inform the design and interpretation of future HIV prevention trials.

Methods: The PrEPVacc registration cohort and trial recruited participants in Uganda, Tanzania, South Africa and Mozambique. This PhD research assessed the utility of HIV risk indicators in identifying populations at high risk of HIV acquisition and then developed and assessed the validity of HIV prevalence and incidence prognostic tools. I also conducted a series of analyses to estimate the counterfactual placebo HIV incidence rate for the active-controlled PrEP component of the PrEPVacc trial. These analyses drew on HIV incidence data from the pre-trial registration cohort, post-PrEP component of the PrEPVacc trial, and inference based on the adherence-efficacy relationship of Emtricitabine/Tenofovir (TDF/FTC). Additionally, a review of existing counterfactual placebo HIV incidence estimation approaches is presented, summarising their strengths and limitations.

Results: Self-reported risk data did not reliably predict HIV incidence to allow for the development of a valid incidence risk score. However, I successfully developed a predictive tool for prevalent HIV infection that could be used to increase the efficiency of HIV diagnosis in sub-Saharan Africa. Estimated counterfactual placebo HIV incidence rates for the active-controlled PrEP component of PrEPVacc ranged from 0.7 to 1.9 per 100 person-years. The counterfactual placebo HIV incidence estimation approaches were subject to considerable biases.

Conclusion: Precise estimation of counterfactual placebo HIV incidence in the PrEPVacc trial was challenging. Further development and standardisation of counterfactual placebo HIV incidence estimation approaches will simplify analyses of future HIV PrEP trials.