Statistical Methods In Epidemiology--General

Environmental Health Sciences

Investigators

Linked publications, trials & patents

Abstract

Methodologic work is progressing in four areas: (1) We extended Within-Cluster Resampling to permit cluster-specific inference. An affected and an unaffected individual are sampled from each cluster (e.g. family) and compared with regard to covariates by fitting a paired-data logistic regression model. The paired resampling is repeated many times and the separate estimates pooled. When the response to an exposure varies across clusters, due to unmeasured modifiers, usual approaches become invalid, but WCPR remains valid. In two related projects:, we developed a statistical test for unmeasured effect modification, and are also developing an approach that for analysis of the case-crossover study, a design used to study effects of shared time-varying environmental exposures, e.g. air pollution. (2) When studying a continuous marker of health, such as blood pressure, we have shown that one can markedly improve the efficiency of a study (over what would be achieved with random sampling) by our proposed design, which oversamples observations at the extremes, i.e. people with unusually high or low values of the outcome. The analytic strategy is being further developed and applied to studies of neurodevelopmental scores in relation to pesticide exposure. (3) Molecular and genetic markers can be used to subtype cases into etiologically distinct subtypes, which may help to clarify inference in a case-control study. Unfortunately, tissue for carrying out the classification may be incompletely available. We showed that missing data methods can both prevent bias and improve precision of estimation, by exploiting the information from cases who were enrolled but could not be subtyped. (4) A permutation test is being developed for comparing patterns of season of disease onset across clinical subcategories of myositis. Such differences in season of onset would suggest distinct environmentally-related etiologies.