The analysis of cancer chemoprevention experiments in which there is interval censoring

AbstractIn certain types of cancer chemoprevention experiments animals are examined for the appearance of tumors over the duration of the investigation. If the experimental animals are inspected frequently enough, then it is reasonable to assume that the times to tumor detection accurately represent the times the cancers reach detectable size. If examinations are conducted less frequently, this assumption is not justified. In such experiments, each detectable tumor is associated with a time interval rather than a specific time point. In this paper we introduce a parametric method for the statistical analysis of experiments in which there is interval censoring that explicitly acknowledges the confounding of the number of induced tumors and tumor growth rate