Statistical Methods for Evaluating and Guiding Implementation of New HIV Prevention Strategies

Fred Hutchinson Cancer Research Center, Seattle WA

Investigators

Linked publications & trials

Abstract

PROJECT SUMMARY The HIV prevention ?eld has advanced dramatically in recent years through the development of antiretroviral- based prevention strategies such as Treatment-as-Prevention and oral pre-exposure prophylaxis (PrEP). Numer- ous challenges remain that impede the implementation of these interventions and HIV remains a major global health problem. We need new preventive interventions to combat the pandemic. The rapid advancements in the ?eld have complicated the statistical design of ef?cacy trials of new interventions. For instance, oral PrEP is now part of the standard HIV prevention package offered to trial participants, which poses challenges in ensur- ing adequate statistical power. Currently, available statistical methods for guiding the implementation of effective interventions are inadequate. Considering this complex HIV prevention context, our goal in Aim 1 is to iden- tify suitable ef?cacy trial designs for evaluating the next generation of HIV prevention tools. We will consider crossover designs, sequential randomization designs, active-arm only designs, and non-inferiority designs with adaptive margins; each of these approaches addresses a key limitation of the prototypical phase 2b/3 trial de- sign currently in use. Through simulation studies and application to candidate interventions, we will investigate the relative statistical performance of the designs. We will also work with leaders in the HIV prevention ?eld, to identify the clinical, ethical, logistical issues and other critical factors that must be considered. We will use these factors to re?ne our design comparisons and to identify the trial design that is most appropriate for each setting. Under Aim 2, we will address speci?c implementation questions in HIV prevention, through the development of improved statistical analysis methods. We will develop methods for evaluating sub-population-speci?c HIV risk and prevention ef?cacy, to identify strategies for implementing interventions, and for bridging HIV incidence and prevention ef?cacy to new settings or populations. Our approach will rely on fewer assumptions than existing methods, use a framework that leverages statistical learning to extract information from multiple predictors of risk and ef?cacy, and accommodate data from multiple sources. We will use important and relevant datasets in HIV prevention and simulation studies designed to mimic their structure and content to gauge the performance of the approaches. Our team members include lead statistical investigators in the major HIV prevention trial networks, with considerable expertise in statistical methods development and clinical trial design, and established collabo- rations with other leaders in the clinical and laboratory science of HIV prevention. Our positions and connections make us uniquely poised to translate the novel methods into practice.