Endoscopy and Colorectal Cancer Mortality

We feel concerned that the conclusions reached by Müller and Sonnenberg¹ might suffer from a substantial methodological bias. In brief, their data derive from the observation that patients who died from colorectal carcinoma (CC) had undergone significantly fewer colonoscopies than their matched controls during the decade that preceded the diagnosis. This finding remained significant after a number of confounding variables had been adjusted for. However, the authors seem to have excluded from analysis all endoscopic procedures that resulted in a diagnosis of CC, irrespective of whether the endoscopy had been performed because of clinical symptoms or for screening purposes. It seems crucial that screening procedures that resulted in a diagnosis of CC should be included in the analysis. Since roughly 2.5% of patients with CC vs 7.5% of controls had undergone a screening colonoscopy within the previous decade, it appears that if only 5% of cases were detected during