A risk index for advanced neoplasia on the second surveillance colonoscopy in patients with previous adenomatous polyps

摘要

Background Predicting the risk of advanced colorectal neoplasia on the second surveillance colonoscopy could help tailor surveillance. Objective To derive and validate a risk index for advanced neoplasia on the second surveillance colonoscopy. Design Retrospective cohort. Setting Single-specialty practice; Veterans Affairs Medical Center. Patients A total of 965 patients with baseline adenomatous polyps, 2 surveillance colonoscopies, and no reported family history of colorectal cancer; validation cohort of 372. Interventions Multivariable logistic regression including demographics and previous colonoscopy results; derivation and validation of a risk index. Main Outcome Measurements Advanced adenoma (≥1 cm in size, villous histology, or high-grade dysplasia) on the second surveillance colonoscopy. Results Mean age was 57.8 ± 9.8 years, 62% were men, and 36% had an advanced adenoma on the index colonoscopy. Associated with advanced adenoma on the second surveillance colonoscopy were age at index colonoscopy (scored 0 for younger than 55 years of age, 1 for 55-59 years of age, 2 for 60-64 years of age, and 3 for older than 65 years of age) and previous findings (non-neoplastic, nonadvanced, advanced [scored 0, 1, and 2, respectively]) on index colonoscopy and the first surveillance colonoscopy, with scores ranging from 1 to 7. Risks of advanced adenoma on the second surveillance colonoscopy with scores of 5 or less and more than 5 were 4.8% (95% confidence interval, 3.5%-6.4%) and 14.9% (95% confidence interval, 7.4%-25.7%), respectively, comprising 93% and 7%, respectively, of the cohort. Corresponding results in the validation cohort were 5.6% and 19.2%, respectively, comprising 86.1% and 13.9%, respectively, of the cohort. Limitations Retrospective study with potential for selection bias. Conclusion This index stratifies the risk of advanced adenoma on the second surveillance colonoscopy. If validated independently, it may be useful for tailoring surveillance.