Design and analysis of studies to reduce the incidence of colon polyps.

Decisions regarding the design and analysis of a phase III study to reduce the incidence of colorectal polyps must take into account two complicating factors: the possibility that polyps are missed during screening exams, and variable lengths of follow-up. In this paper we investigate the effects due to misclassification on the power of statistical tests to detect a change in polyp recurrence rates. We also use Monte Carlo studies to examine the relative efficiency of different methods of adjusting for variable times of follow-up.