Rate and predictors of early/missed colorectal cancers after colonoscopy in Manitoba: a population-based study

Am J Gastroenterol. 2010 Dec;105(12):2588-96. doi: 10.1038/ajg.2010.390. Epub 2010 Sep 28.


Objectives: Many of the colorectal cancers (CRCs) diagnosed within 3 years after a colonoscopy are likely because of lesions missed on the initial colonoscopy. In this population-based study, we investigated the rate and predictors of CRCs diagnosed within 3 years of a colonoscopy.

Methods: We identified individuals 50-80 years of age diagnosed with CRC between 1992 and 2008 from the provincewide Manitoba Cancer Registry. Performance of colonoscopy and history of co-morbidities was determined by linkage to the provincial universal health care insurance provider's physician billing claims and hospital discharges databases. CRCs diagnosed within 6 months of a colonoscopy were categorized as detected CRCs and those 6-36 months after a colonoscopy as early/missed CRCs. Logistic regression analysis was performed to identify the patient, endoscopist, colonoscopy, and CRC factors associated with early/missed CRCs.

Results: Of the 4,883 CRCs included in the study, 388 (7.9%) were early/missed CRCs, with a range of 4.5% of rectum/rectosigmoid cancers in men to 14.4% of transverse colon/splenic flexure cancers in women. Independent risk factors associated with early/missed CRCs included prior colonoscopy, performance of index colonoscopy by family physicians, recent year of CRC diagnosis, and proximal site of CRC.

Conclusions: This study suggests that approximately 1 in 13 CRCs may be an early/missed CRC, diagnosed after an index colonoscopy in usual clinical practice. Women are more likely to have early/missed CRC. It is unclear if this relates to differences in procedure difficulty, bowel preparation issues, or tumor biology between men and women.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Aged
  • Aged, 80 and over
  • Chi-Square Distribution
  • Clinical Competence
  • Colonoscopy*
  • Colorectal Neoplasms / diagnosis*
  • Colorectal Neoplasms / epidemiology
  • Colorectal Neoplasms / pathology
  • Comorbidity
  • Diagnostic Errors / statistics & numerical data*
  • False Negative Reactions
  • Female
  • Humans
  • Logistic Models
  • Male
  • Manitoba / epidemiology
  • Mass Screening
  • Middle Aged
  • Population Surveillance
  • Predictive Value of Tests
  • Proportional Hazards Models
  • Registries
  • Risk Factors
  • Time Factors