Colonoscopic Miss Rates for Colorectal Cancer: A Population Based Analysis

Abstract

Background: The gold standard for the diagnosis of colorectal cancer (CRC) is colonoscopy (COL). However, COL does contain an inherent miss rate for CRC. Previous reports evaluated miss rates in patients seen in academic centers or units with endoscopists known for their expertise. The CRC miss rate of COL performed in the course of usual clinical practice is unknown. Objective: To determine the proportion CRCs missed during COL. Methods: Using electronic data from the Canadian Institute for Health Information and the Ontario Health Insurance Program (OHIP), we identified all individuals older than 18 years old, with a new diagnosis of CRC, in the province of Ontario from 1/4/1997 to 31/3/2001. We excluded all individuals with synchronous, site unspecified, or location coded as other CRC. We excluded those with CRC in the descending colon and splenic flexure because we are unable to determine the depth of insertion of the instrument for those locations. We also excluded those who did not have a COL (or flexible sigmoidoscopy in those with rectal or sigmoid CRC) within 3 years prior to their diagnosis. The remaining individuals comprised our study cohort. We separated our cohort into three groups: right-sided (CRC in cecum or ascending colon), transverse CRC, and rectal or sigmoid CRC. We divided each group into two categories. The detected cancers category consisted of individuals who had lower gastrointestinal (GI) endoscopy within 6 months prior to the diagnosis (in this category we assumed the endoscopic procedure identified the cancer); the missed cancers category consisted of those who had lower GI endoscopy 6-36 months prior to the diagnosis (in this category we assumed the endoscopic procedure missed the cancer). Results: We identified 31,553 patients with a new diagnosis of CRC and excluded 21,366 because they were diagnosed with synchronous, site unspecified, location coded as other CRC, descending colon, or splenic flexure CRC, or did not have a COL within 3 years of their CRC diagnosis. The remaining 10,187 patients comprised our study cohort, of whom 2,580 had right-sided CRC, 702 had transverse CRC, and 6,905 had rectal or sigmoid CRC. The proportions of missed cancers were: 157 patients (6%) with right-sided CRC, 29 patients (4%) with transverse CRC, and 207 patients (3%) with rectal or sigmoid CRC. Conclusion: The proportion of CRCs missed by colonoscopy in usual clinical practice varies between 3-6% depending on the location of the CRC. When consent is obtained for colonoscopy patients need to be informed of the risk of missing CRC.