An illustration of dangers of ignoring survival differences in carcinogenic data

This paper illustrates the effects of survival differences on the routine analysis of 2-year animal carcinogenesis experiments using quantal response. Not adjusting for decreased survival in the higher dosed groups results in a decrease in the actual significance level for the quantal response trend test, and a corresponding decrease in sensitivity for detecting a true treatment effect. Similar results hold for estimation of carcinogenic risk. Tables of the range of survival differences found in recent National Toxicology Program carcinogenesis studies are presented.