The measurement of adenoma occurrence

Adenomatous polyps are regarded as the dominant precursor lesion of colon cancer. For four decades the occurrence of adenomas has been investigated, both in observational studies and in controlled, randomized cancer-prevention studies. During this period, little consensus has been reached with regard to the data that should be reported in such studies. The emerging consensus on the methods of analysing the results does not have a firm biostatistical basis, and does not deal appropriately with the phenomenon of colonoscopy avoidance. We review a simple biostatistical method that is appropriate, and apply it to a selected recent series of studies. Our finding is that there are common patterns of adenoma occurrence across these studies, with a rate of 0.3-0.4 per year in the first year following polypectomy, dropping to 0.1-0.15 in subsequent years. The method of analysis also serves to highlight exceptions to this pattern.