Development of the PROMOTE model to stratify colorectal cancer risk for prioritization of colonoscopy resource use: a multicenter prospective study

Background Colonoscopy efficacy for colorectal cancer (CRC) prevention is limited by inappropriate or over- prescription. Colonoscopy appropriateness prioritization (CAP) criteria have recently been proposed, but their role in CRC risk stratification remains unclear. The study objective was to derive and validate a predictive model for CRC taking account of CAP criteria, and to assess CRC occurrence in the light of appropriateness of colonoscopies. Method In a prospective observational study across 19 Italian centers, including adults undergoing colonoscopy outside CRC screening programs, three cohorts were analyzed for derivation, temporal validation, and geographic validation of the model. CRC risk was estimated by multivariable logistic regression. Model performance was assessed using the area under the receiver operating characteristic (AUROC), and two risk groups were defined: low-risk (<5%) and high-risk (≥5%). Number-needed-to-scope (NNS) was calculated. Results The derivation and temporal and geographic validation, cohorts included 2059, 1321, and 1924 patients, respectively, with CRC prevalence 3.6%, 3.9%, and 3%, respectively. CRC was more frequent in appropriate versus inappropriate colonoscopies. The PROMOTE model included: ages 50–59 (odds ratio [OR] 1.89, 95% confidence interval [CI] 0.64–5.59), 60–69 (OR 3.87, 95%CI 1.40–10.71), and ≥70 (OR 5.35, 95%CI 2.04–14.06), versus <50; no colonoscopy in previous 10 years (OR 2.92, 95%CI 1.62–5.25); according to CAP criteria, deferrable (OR 3.44, 95%CI 1.42–8.34) and urgent (OR 16.12, 95%CI 7.15–36.36) versus nonurgent. Discrimination was good (AUROC 0.84, 95%CI 0.79–0.89). NNS was 8–9 in the high-risk group and 67–71 in the low-risk group across validation cohorts. Conclusion We developed and validated the PROMOTE model, a simple tool to estimate CRC risk before colonoscopy, to support appropriate referral, optimize prioritization, and improve resource use.