Background: Information about colonoscopy complications, particularly postpolypectomy bleeding, is limited.

Objective: To quantify the magnitude and severity of colonoscopy complications.

Design: Retrospective cohort.

Setting: Kaiser Permanente of Northern California.

Patients: 16, 318 members 40 years of age or older undergoing colonoscopy between January 1994 and July 2002.

Measurements: Electronic records reviewed for serious complications, including hospital admission within 30 days of colonoscopy for colonic perforation, colonic bleeding, diverticulitis, the postpolypectomy syndrome, or other serious illnesses directly related to colonoscopy.

Results: 82 serious complications occurred (5.0 per 1000 colonoscopies [95% CI, 4.0 to 6.2 per 1000 colonoscopies]). Serious complications occurred in 0.8 per 1000 colonoscopies without biopsy or polypectomy and in 7.0 per 1000 colonoscopies with biopsy or polypectomy. Perforations occurred in 0.9 per 1000 colonoscopies (CI, 0.5 to 1.5 per 1000 colonoscopies) (0.6 per 1000 without biopsy or polypectomy and 1.1 per 1000 with biopsy or polypectomy). Postbiopsy or postpolypectomy bleeding occurred in 4.8 per 1000 colonoscopies with biopsy (CI, 3.6 to 6.2 per 1000 colonoscopies). Biopsy or polypectomy was associated with an increased risk for any serious complication (rate ratio, 9.2 [CI, 2.9 to 29.0] vs. colonoscopy without biopsy). Ten deaths (1 attributable to colonoscopy) occurred within 30 days of the colonoscopy.

Limitations: 99.3% (16 204) of colonoscopies were nonscreening examinations. The rate of complications may be lower in a primary screening sample. The small number of observed adverse events limited power to detect risk factors for complications.

Conclusions: Colonoscopy with biopsy or polypectomy is associated with increased risk for complications. Perforation may also occur during colonoscopies without biopsies.