Methods for Complex Longitudinal Data in HIV/AIDS Prevention Research

Fred Hutchinson Cancer Research Center, Seattle WA

Investigators

Linked publications, trials & patents

Abstract

Project Summary/Abstract The advent of biomedical prevention interventions, such as Treatment-as-Prevention (TasP) and Pre-exposure Prophylaxis (PrEP), has reduced the risk of new HIV transmission yet enormous challenges in HIV/AIDS preven- tion remain. Issues with uptake and adherence compromise the full potential of these drugs. The consequences of which include treatment failure and HIV drug resistance (HIVDR). Combination intervention approaches, based on combined multi-components of both biomedical and behavioral interventions, may prevent HIVDR and ensure the success of ART. Although adherence to ART has signi?cant public health implications, the evaluation of ad- herence continues to be challenging both in trials and real-world settings. The complexity of ART uptake and adherence analysis arise due to several reasons. Speci?cally, 1) HIV-infected patients are living longer when treated with ART while the viability of long term viral suppression with ART is not well characterized; 2) less expensive, easy-to-implement instruments with high sensitivity and speci?city are yet to be widely available for measuring ART adherence over time; 3) behavioral interventions are complicated by nature as they are high- dimensional and time-dependent; and 4) social, economic and other structural factors, at multiple-levels work synergistically with the biomedical and behavioral interventions. Without accounting for the complexity involved, data analysis using naive methods could easily lead to severe bias and incorrect inference. In this project, we aim to develop a set of new statistical tools to cope with the complexity involved. Speci?cally, we aim to 1) develop methods for assessing longitudinal pattern of recurrent events over time; 2) develop methods of threshold regres- sion models for longitudinal data subject to interval censoring; 3) develop methods for multi-level event history studies with high-dimensional covariates subject to interval censoring; and 4) develop methods for analyzing data collected from complex sampling design.