STATISTICAL ISSUES IN AIDS RESEARCH

University Of Washington, Seattle WA

Investigators

Linked publications & trials

Abstract

This project addresses important issues in the statistical design and analysis of clinical trials of interventions for the treatment or prevention of HTVinfection, and in the analysis of data from HTV/AIDS cohort studies. Research will be continued on the development of useful methods that are required to appropriatelyhandle multivariate outcome data that are subject to complicated censoring mechanismsand that arise in HIV/AIDS clinical trials and cohort studies. Specifically,non- and semi-parametric methods will be developed for analyzing data on recurrent AEDS-defming events, in the presence of death and informativedrop-outs, and for the analysis of outcomes such as quality of life measures that are assessed repeatedly over time in long term studies yieldingincomplete information due to censoring. Semi-parametric methods will also be developed to analyze data arising from HIV/AIDS studies recording the transition of participants among transient disease states reflecting, for example, changes in disease status, quality-of-life,or toxicities. Development and evaluation of statistical methods will be continued for analysis of various types of interval censored, doubly censored and truncated data that arise frequently in HIV/AIDS trials and cohort studies. An array of domestic and internationalfield trials are being designed or are underway in order to evaluate the safety and efficacy of perinatal interventions or vaccines for the prevention of the spread of HTV. Statistical methods will be developed to provide more reliable estimation of the timingof maternal to child transmission of HIV and evaluation of the effect of interventions,taking into account the lower specificityof PCR performed on cord blood samples and the lower sensitivity of PCR testing of filter paper samples. Improved Phase I and II vaccine trial designs will be explored that take into account the multivariate nature of CTL measurements, the informative missingness arising due to the technically challenging nature of the CTL assay, the non-trivial false positive rate of CTL measurements in placebo participants, and the loss of information due to dichotomization of continuous skewed measurements. Research into optimal designs of Phase HE "Intermediate Trials" of candidate HIV vaccines will also be continued.