Statistical Methods And Applications For Laboratory Anim

Environmental Health Sciences

Investigators

Linked publications & trials

Abstract

This project focuses on developing new statistical methods, and applying new and existing statistical techniques, to analyze data from laboratory animal studies. One avenue of research dealt with tumor incidence, which is the rate at which new tumors arise. We investigated methods that handle complex data structures without relying on unrealistic assumptions. A flexible incidence estimator was developed which accommodates explanatory variables, studies with only one sacrifice time, tumors of unspecified lethality, and tumors that are unobservable before death. Another analysis was developed which, in addition to using explanatory variables from the current experiment, can incorporate information from previous studies and from subject matter experts. This method was extended to permit simultaneous inference about tumors at multiple sites and to account for within-animal correlations among tumor onset times. Another area of research focused on tests for dose-related trends in tumor rates. A general procedure was developed which can be applied to a wide array of problems, including those encountered in National Toxicology Program (NTP) studies. This new test is more powerful than the current test and is robust with respect to the shape of the dose-response curve, the effects of dose on mortality, the spacings between doses, and the size of the background tumor rate. Similar techniques are being developed to more efficiently estimate confidence intervals and to detect trends when the data are correlated. Also, new methods are being developed to analyze data from heterogeneous sources (e.g. different laboratories).