EXPERIMENTAL DESIGN AND DATA ANALYSIS METHODOLOGY FOR ANIMAL EXPERIMENTS

点击翻译按钮获取中文摘要

This project is concerned with statistical methodology issues involved in the design, analysis, and interpretation of laboratory animal experiments in general and long-term carcinogenicity studies in particular. During the past year, considerable research was devoted to the derivation and utilization of survival-adjusted methods in the assessment of carcinogenic effects. These include consideration of important issues such as tumor lethality, the availability of cause of death information and mathematical modeling of the underlying response variable. The effect of various confounding variables (e.g., corn oil gavage; body weight differences) on the interpretation of carcinogenicity studies was also studied. Other biological/statistical issues that were the focus of research efforts during the past year were the relative merits of site-specific and overall (all sites) tumor incidence analyses, an evaluation of the frequency in which NTP carcinogenicity studies produce only benign neoplasia (i.e., with no supporting evidence of malignancy), and assessing the degree of synergistic effects in carcinogenicity studies in which two compounds are tested singly and in combination.