Background: Previous studies that reported the incidence of perforation from a colonoscopy are limited by small sample sizes, restricted age groups, or single-center data.
Objective: To determine the incidence and risk factors of colonic perforation from a colonoscopy in a large population cohort.
Design: Retrospective, population-based, cohort study, followed by a nested case-control study.
Setting: California Medicaid program claims database.
Patients: A total of 277,434 patients (aged 18 years and older) who underwent a colonoscopy during 1995 to 2005, age, sex, and time matched to 4 unique general-population controls.
Main outcome measurements: Perforation incidence in the 7 days after colonoscopy (or matched index date for controls) with odds ratio (OR); multivariate logistic regression to calculate adjusted ORs for subsequent analysis of risk factors.
Results: A total of 228 perforations were diagnosed after 277,434 colonoscopies, which corresponded to a cumulative 7-day incidence of 0.082%. The OR of getting a perforation from a colonoscopy compared with matched controls (n = 1,072,723) who did not undergo a colonoscopy was 27.6 (95% CI, 19.04-39.92), P < .001. On multivariate analysis, when comparing the group that had a perforation after a colonoscopy (n = 216) with those who did not (n = 269,496), increasing age, significant comorbidity, obstruction as an indication for the colonoscopy, and performance of invasive interventions during colonoscopy were significant positive predictors. Performance of biopsy or polypectomy did not affect the perforation risk. The rate of perforation did not change significantly over time.
Limitations: Validity of coding and capturing of all perforation diagnoses may possibly be deficient.
Conclusion: The risk of perforation from a colonoscopy is low, but, despite increased experience with the procedure, it remains unchanged over time.