Chronic Disease Population Research Issues and Strategies

点击翻译按钮获取中文摘要

This project will address methodology development needs that arise in disease prevention trials and epidemiologic cohort studies. Our continuing work on failure time data methods will include sub-aims on multivariate survivor function estimation, on cohort and case-control estimation under a semiparametric normal transformation model, on attributable risk estimation for a preventive intervention, and on case-only estimation methods in a randomized controlled trial context. Our continuing work, motivated by dietary and physical activity epidemiology, on covariate measurement error methods will develop and compare estimation procedures based on biomarker data on subsets of a cohort, and self-report data on the entire cohort. Both recovery-type biomarkers, corresponding to the expenditure of a nutrient, and concentrationtype biomarkers, reflecting the concentration of a nutrient blood or another body compartment, will be considered. Our work on population science research issues and strategies will continue to contrast randomized controlled trial and observational study data, toward identifying sources of bias, with emphasis on both postmenopausal hormone therapy and dietary intervention, and with motivation and data derived from the Women's Health Initiative (WHI) clinical trial and cohort study. Efforts to elucidate postmenopausal hormone therapy effects in the WHI have led to a number of case-control studies using the WHI specimen repository, including genome-wide single nucleotide polymorphism (SNP) association studies of diseases that were adversely affected by estrogen plus progestin use. Aspects of the design and analysis of highdimensional SNP association studies is an additional Project 1 research aim. These aims will be addressed by using statistical models for disease risk, non-standard exposure measurement models, standard genetic models, asymptotic distribution theory development, computer simulations, and applications to important chronic disease data sets.