Statistical Issues in AIDS Research

R56AI140953 (PI: Hughes)

This research addresses a number of important topics in the design and analysis of HIV prevention trials and studies. Use of antiretroviral drugs for pre-exposure prophylaxis (PrEP) of HIV is an important research area for HIV prevention but progress is being hindered by the difficulty in distinguishing drug efficacy from adherence. We directly address these issues in aim 1. 

In addition, understanding the effects of oral PrEP, microbicides and vaccines often requires substudies that use expensive assays. To minimize cost and maximize information from these studies, two-phase sampling plans, pooling and other complex sampling designs are common. In aims 2 and 3 we propose methods that will lead to optimal designs and efficient analyses of such data.

The approaches we propose are diverse and reflect the varied strengths of this highly productive group of investigators. Nonetheless, a common theme that recurs throughout the proposed work is the development of statistical methods for optimal design and efficient nonparametric and semiparametric estimation. In addition to this overall theme, the aims reinforce each other. For example, methods for two-phase sampling may be used in the analyses of adherence in PrEP trials (since objective measures of adherence are typically subsampled) and pooled designs; measures of adherence developed in aims 1 may be used as input into causal estimates of PrEP efficacy; methods for interval censoring may be needed in many contexts, including analysis of two-stage sampling and PrEP efficacy.

We propose to address the following specific aims:

Aim 1. Develop methods for the design and analysis of PrEP studies

Aim 2. Develop optimal design and analysis methods for HIV/AIDS studies with complex sampling and observation schemes

Aim 3. Develop efficient methods for the design and analysis of pooled data