On the Role of Cause‐Of‐Death Data in the Analysis of Rodent Tumorigenicity Experiments

The cause‐of‐death test provides a valid comparison of dose groups in an animal tumorigenicity experiment as long as a certain representativeness condition holds. We re‐express the representativeness assumption in terms of two easily interpreted tumour lethality functions. This re‐expression clarifies the role of pathologists in assigning cause of death and allows a simple empirical assessment of the validity of the representativeness assumption. The impact of multiple tumours on the representativeness condition is also examined. An investigation using the ED01 study reveals that the presence of multiple tumours increases the degree of non‐representativeness in the data. Analytic and empirical results suggest a simple modification to the cause‐of‐death test which potentially decreases the bias associated with the presence of multiple tumours.