Surgeons' Immediate Postoperative Prediction After Major Emergency Abdominal Surgery

Major emergency abdominal surgery carries high morbidity and mortality, making accurate risk stratification important. Whether surgeons' immediate postoperative clinical judgment predicts outcomes and how it compares with established surgical risk models remains poorly characterized.

Prospective single-center cohort of 411 patients undergoing major emergency abdominal surgery (February 2024-June 2025). Immediately after surgery, the operating surgeon estimated 30-d mortality (ordinal 10% intervals) and the risk of complications (0-10 scale). Discrimination (area under the receiver operating characteristic curve [AUC]) and calibration (observed-to-expected ratio, calibration slope and intercept, Brier score) were assessed. Mortality discrimination was compared with National Emergency Laparotomy Audit and Portsmouth Physiological and Operative Severity Score for the enUmeration of Mortality and Morbidity using the DeLong test.

Thirty-day mortality was 11% (47/411). Surgeon-estimated risk showed good discrimination for mortality (AUC: 0.89; 95% confidence interval: 0.85-0.94), comparable to National Emergency Laparotomy Audit (0.90) and Portsmouth Physiological and Operative Severity Score for the enUmeration of Mortality and Morbidity (0.86; P = 0.83 and P = 0.30). Surgeons systematically overestimated absolute mortality (mean predicted: 23% versus observed: 11%; observed-to-expected: 0.48; intercept: -1.17; slope: 1.20). For complications, discrimination was good for sepsis/multi-organ failure (AUC 0.81) and acceptable for intensive care unit admission (0.78) and cardiac complications (0.78), but poor for reoperation (0.62) and mechanical bowel obstruction (0.62), and absent for prolonged paralytic bowel dysfunction (0.51). Higher estimated risk was associated with increased odds of complications. Predicted discharge destination agreed moderately with observed (accuracy 74%; weighted κ 0.61).

Surgeons' immediate postoperative assessment showed discriminatory performance of a similar magnitude to established models but systematically overestimated absolute risk and performed better for systemic than procedure-specific complications. This bedside estimate may aid early risk stratification but requires external validation before clinical use.