STATISTICAL METHODS IN AIDS EPIDEMIOLOGY

点击翻译按钮获取中文摘要

The overall objective of this grant is to develop and apply statistical methods in AIDS epidemiology. There are three specific objectives of this proposal. The first is to develop and compare different statistical methods and study designs for estimating current HIV incidence. One approach will be to investigate the use of diagnostic tests such as tests for p24 antigen or other markers to identify individuals in cross-sectional surveys who were recently infected. The work will entail a detailed analysis of the sources of uncertainty in estimates of HIV incidence using these approaches. These approaches for estimating HIV incidence will be compared to other approaches including cohort studies and backcalculation methods with respect to sources of uncertainty, precision of the estimates, sample size requirements and sensitivities to model assumptions. The methods will be used to estimate current HIV incidence rates in India, and in an inner city emergency department. The second objective is to develop statistical models and methods for studies of the preantibody (window) period, and in particular to estimate the duration of p24 antigenemia. New statistical methods will be developed and applied to the analysis of data from emergency department patients and plasma donors. The third objective is to generalize the method of backcalculation to account for changes in the AIDS surveillance definition, and in particular to account for the significant revision of the case definition in 1993. This work will include application of the methods to U.S. AIDS incidence data, detailed sensitivity analyses and quantification of the effects of model misspecification.