Both colonoscopy and flexible sigmoidoscopy are accepted procedures for colorectal cancer (CRC) screening in the United States.

To compare risk of CRC after negative findings on screening colonoscopy versus sigmoidoscopy and to evaluate racial/ethnic disparities in postscreening CRC.

Retrospective, comparative cohort study.

Integrated community-based health-care system.

Average-risk patients 50 to 75 years of age with negative findings on an initial endoscopic screening examination from January 2000 to December 2010.

Colonoscopy versus sigmoidoscopy as the initial screening procedure.

Incident cases of CRC identified via a prospective internal cancer registry, risk of CRC determined by Cox regression adjusted for age, sex, race/ethnicity, and comorbidity.

The study cohort included 138,297 patients (42,938 patients with negative findings on colonoscopy and 95,359 with negative findings on sigmoidoscopy). The median age was 57.9 years (interquartile range 53.0-64.1 years). Women comprised 51.8% of the cohort with 42.2% non-Hispanic white patients, 24.1% Hispanic patients, 10.7% non-Hispanic black patients, 9.7% Asian patients, and 13.3% other/unknown. A total of 241 cases of CRC was detected during 553,543 person-years of follow-up. The adjusted hazard ratio (HR) of postscreening CRC was 0.42 (95% confidence interval [CI], 0.28-0.64; P < .0001) for colonoscopy versus sigmoidoscopy. Risk reduction was primarily among proximal tumors (adjusted HR 0.30; 95% CI, 0.16-0.57). Non-Hispanic black patients were at higher risk of postscreening CRC compared with non-Hispanic white patients (adjusted HR 1.71; 95% CI, 1.20-2.42); however, this disparity was noted only in the sigmoidoscopy cohort.

Retrospective study with potential selection bias and residual confounding.

Negative screening colonoscopy was associated with decreased incidence of subsequent CRC and a decrease in disparities compared with negative sigmoidoscopy findings in this large, community-based setting.