Validation of the NCI colorectal cancer risk assessment tool in the CSP 380 veterans cohort.
Musselwhite, LW; Redding, TS; Sims, KJ; O'Leary, M; Hauser, ER; Hyslop, T; Lieberman, DA; Provenzale, DT
Published in: Journal of Clinical Oncology
e15135 Background: Refining screening to colorectal cancer (CRC) risk may promote screening effectiveness. We applied the National Cancer Institute (NCI) CRC Risk Assessment Tool to estimate 5- and 10-year CRC risk in an average-risk Veterans cohort undergoing screening colonoscopy with follow-up. Methods: This was a prospective evaluation of predicted to actual risk of CRC using the NCI CRC Risk Assessment Tool in male Veterans undergoing screening colonoscopy with a median follow-up of 10 years.Family, medical, dietary and physical activity histories were collected at enrollment and used to calculate absolute 5- and 10-year CRC risk, and to compare tertiles of expected to observed CRC risk. Sensitivity analyses were performed. Results: For 2,934 male Veterans with complete data (average age 62.4 years, 15% minorities), 1.3% (N=30) and 1.7% (N=50) were diagnosed with CRC within 5 and 10 years of survey completion. The area under the curve for predicting CRC was 0.69 (95% CI; 0.61-0.78) at 5 years and 0.67 (95% CI, 0.59-0.75) at 10 years. We calculated the sensitivity (0.60, 95% CI; 0.45-0.73), specificity (0.67, 95% CI; 0.65-0.69) positive predictive value (0.031, 95% CI; 0.02-0.04) and negative predictive value (0.99, 95% CI; 0.98-0.99). Conclusions: The NCI CRC Risk Assessment Tool was well-calibrated at 5 years and overestimated CRC risk at 10 years, had modest discriminatory function, and a high NPV in a cohort of ethnically diverse male Veterans. This tool reliably excludes 10-year CRC in low-scoring individuals and may inform patient-provider decision making when the benefit of screening is uncertain. [Table: see text]
