Background: In the elderly, the increased prevalence of colorectal neoplasia and the protective effect of colonoscopy may be offset by advancing age and comorbidity.
Objective: To describe and quantify the endoscopic findings, survival, and predictors of mortality of elderly persons after colonoscopy.
Design, setting, and patients: Retrospective cohort study of persons aged>or=75 years who underwent colonoscopy in 1999 and 2000 at a U.S. Veterans Affairs facility and urban county hospital.
Main outcome measures: Advanced neoplasms were defined as colorectal cancer (CRC), polyp with high-grade dysplasia, villous histologic features, or tubular adenoma>or=1 cm. Comorbidity was measured with the Charlson comorbidity index. Subjects were followed until death or study closure.
Results: Of 469 eligible subjects, 65 were excluded and 404 were included in the study. Fifty-nine of 404 (15%) had an advanced neoplasm, including 8 (2%) with CRC. There were 167 deaths (41%); the mean overall survival was 4.1+/-0.1 years (median 5.95 years). A symptomatic indication for colonoscopy was not predictive of death. Mortality was predicted by age (hazard ratio 1.16 for each year increase beyond age 75 years, 95% CI 1.07-1.3, P=.0003) and Charlson score (hazard ratio 8.3 for each point increase, 95% CI 1.4-48.5, P=.02). The median survival of patients aged 75 to 79 years was >5 years if the Charlson score was <or=4. Among patients aged>or=80 years, the median survival was <5 years regardless of Charlson score.
Limitations: Retrospective design.
Conclusions: In this cohort of elders, age and comorbidity were predictors of death. The protective effect of younger age lessens as comorbidity increases. These findings may have important implications for CRC screening and surveillance in elders.