Single-Center External Validation and Calibration of the ACS NSQIP Surgical Risk Calculator in Emergency Colorectal Cancer Surgery

Background/Objectives: Emergency colorectal cancer (CRC) surgery combines acute physiological disturbance with heterogeneous operative decisions. We evaluated the discrimination, calibration, and prediction error of the contemporary American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) Surgical Risk Calculator in an institutional emergency CRC cohort. Methods: This retrospective study included 198 patients undergoing clinically indicated emergency CRC resection during 2020–2024. Recorded 30-day outcomes were compared with version 4.0.4 predictions using area under the receiver operating characteristic curve (AUC), observed-to-expected (O/E) ratios, exploratory calibration plots, and Brier scores. Renal-failure analysis included 170 patients, and leak analysis included 118 patients with an anastomosis. Results: Mortality occurred in 25 patients (12.6%); discrimination was strong (AUC 0.882, 95% confidence interval [CI] 0.807–0.958), with an O/E ratio of 1.03 (95% CI 0.67–1.52). Pneumonia and cardiac complications had AUCs of 0.768 and 0.799 but were underpredicted (O/E 3.48 and 3.62). Surgical site infection occurred in 40.4%, with limited discrimination (AUC 0.544, 95% CI 0.462–0.626) and underprediction (O/E 3.15, 95% CI 2.49–3.92). Anastomotic leak occurred in 12 of 118 patients (10.2%). Mean absolute error for hospital stay was 5.46 days. Conclusions: Mortality ranking was strong, while absolute-risk estimation differed by outcome. Larger multicenter validation with consistent outcome definitions is necessary before these probabilities can support reliable individual risk communication in emergency CRC surgery.