Predictors of colorectal cancer following a negative in the medicare population

Background: The incidence of colorectal cancer following a normal colonoscopy in the Medicare population is not known. Methods: A 5% national sample of Medicare enrollees from 1996 to 2005 was used to identify patients undergoing complete colonoscopy. A colonoscopy not associated with any procedure (e.g., biopsy, polypectomy or fulguration) was defined as a negative colonoscopy. Patients with history of inflammatory bowel disease, colorectal cancer or death within 12 months of colonoscopy were excluded. A multivariable model was constructed to evaluate the factors associated with a new diagnosis of colorectal cancer in the period from 12 to 120 months following the negative colonoscopy. Results: Among 200,857 patients (mean age 74 years, 61% female, 92% White) with a negative colonoscopy, the incidence of colorectal cancer was 1.8 per 1,000 person-years. The incidence rate for matched follow-up periods decreased from 2.0/1,000 person-years for patients undergoing colonoscopy during 1996-2000 to 1.2/1,000 person years during 2001-2005. Multivariate analysis revealed a significant regional variation in the incidence of colorectal cancer following a negative colonoscopy. The incidence was higher in patients >85 years, males and patients who underwent a negative colonoscopy by a non-gastroenterologist or endoscopist in the lowest volume quartile. On stratified analyses, endoscopist volume was a significant predictor for non-gastroenterologists only. Conclusions: The specialty and experience of the endoscopist are significant predictors of the incidence rate of colorectal cancer in Medicare patients with a negative colonoscopy.