Statistical Methods to Study the Impact of HIV Interventions

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Evaluation of novel approaches to prevent and treat HIV remains a pressing public health priority, particularly in sub-Saharan Africa. The ability to conduct such evaluations effectively requires the development of novel statistical methods capable of addressing the unique complexities of the data available to inform HIV research. This includes both data from population-based surveys, and from randomized controlled trials of prevention and treatment interventions. We propose to pursue three avenues of methodological development to advance this research agenda. First, we will develop methods that use population-based survey data to estimate the potential impact of population circumcision interventions, applying these methods to data from Lesotho. Second, we will develop methods for estimation of treatment effects on combined endpoints comprised of events that are monitored irregularly in a clinical trial population and thus subject to informative interval censoring. The techniques will be developed and illustrated using data from the Tshepo trial in Botswana. Third, we will develop and compare methods to estimate the contributions of specific causal pathways to the effects of randomized HIV prevention interventions. We will apply these methods to data from a randomized controlled trial of diaphragm use in Zimbabwe and South Africa to estimate the effect of diaphragms on HIV transmission, controlling for the unintended effect of diaphragms on condom use. Each specific aim is motivated by a concrete public health question, and we anticipate in each case that the results of the proposed research will provide a valuable applied contribution with the potential for considerable impact. At the same time, the larger goal of each aim is to develop a general set of practical methods with broad applicability to a wide range of similar questions in HIV prevention and treatment. In accomplishing these research goals we will build on the diverse skill set, extensive research experience, and established collaborations of the co-investigators, each of whom has a strong record of developing and applying novel statistical methods to research questions in HIV prevention and treatment. !