Statistical Methods For Human Studies

Environmental Health Sciences

Investigators

Linked publications & trials

Abstract

The goal of this project is to develop improved statistical methods for the analysis of data from human studies. In disease screening studies, it is difficult to obtain direct information on the age at onset of disease, the timing of past exposures, and the rate of disease progression. Work has progressed in the development of methods for disease screening studies, motivated by the NIEHS uterine fibroid (leiomyoma) study conducted by Donna Baird. In particular, we developed an approach for modeling of multiple lesion onset and growth from interval censored data. Ongoing work in this area has focused on potential risk factors, including obesity, alcohol, and lutenizing hormone (LH) levels. We have found an association between high levels of drinking and increased incidence of fibroids. In addition, women with high levels of LH have increased incidence and progression rates. To improve power and efficiency, we have also developed methods for incorporating order restrictions in estimating regression curves and assessing exposure effects. These methods were applied to demonstrate an increasing frequency of sleep problems with increasing body mass index. We have also continued to make progress in developing methods for assessing sources of heterogeneity in event time data and in the setting of generalized linear mixed models. Ongoing work focuses on flexible methods, which avoid parametric assumptions, such as normality.