TPTh 3.01 Predictable Risk, Preventable Delay: Underprediction of Mortality in ASA 4 Emergency Laparotomy Patients

Abstract

Emergency laparotomy is a central focus of national quality improvement through the National Emergency Laparotomy Audit (NELA). Pre-operative risk prediction informs consent, escalation planning, and critical care allocation. This study aimed to evaluate the discrimination and calibration of NELA and ACS-NSQIP mortality prediction models stratified by ASA grade, and to determine contributory factors to excess mortality in high-risk patients. We conducted a retrospective cohort study of 382 consecutive emergency laparotomies across two hospitals between 2023 and 2025. The primary outcome was 30-day all-cause mortality. Predicted mortality was derived from pre-operative NELA and NSQIP estimates. Model discrimination was assessed using the area under the receiver operating characteristic curve. Calibration was evaluated using observed-to-expected mortality ratios stratified by ASA grade with binomial testing and 95% confidence intervals. Both models demonstrated good discrimination (NELA AUC 0.837; NSQIP AUC 0.811). In ASA 1–3 patients (n=311), observed mortality was 8.0% compared with predicted mortality of 5.8%, indicating acceptable calibration. In contrast, ASA 4 patients (n=70) had a substantially higher observed mortality of 30.0% (95% CI, 20.0–41.4%), exceeding NELA (23.5%) and NSQIP (20.2%) predictions. Among deceased ASA 4 patients (n=21), median decision to knife-to-skin time was 6.8 hours (IQR 2.9–13.8), exceeding the NELA-recommended 6-hour target, compared with 4.0 (IQR 1.8–6.7) in survivors. Current risk models perform well overall but underestimate mortality in the highest-risk emergency laparotomy patients. Excess mortality is associated with failure to meet NELA time-to-surgery standards, identifying this as a modifiable system-level factor that warrants prioritisation.