Background: Emergency laparotomy is associated with high postoperative morbidity and mortality. Accurate early risk stratification is essential for guiding clinical decision-making and resource allocation. This study aimed to develop and internally validate a robust predictive scoring system for 30-day mortality using routinely available preoperative variables.
Methods: A retrospective cohort study was conducted among 402 adult patients who underwent emergency laparotomy. The dataset was randomly split into training (70%) and testing (30%) subsets. An elastic net logistic regression model (α = 0.5) was developed on the training set with 10-fold cross-validation to optimize model performance and select predictors. Model discrimination was assessed using cross-validated area under the receiver operating characteristic curve (AUROC) and bootstrap-based ROC analysis. Calibration was evaluated using the calibration belt method.
Results: Twelve preoperative predictors were retained in the final model, including ASA grade, cardiovascular disease, serum creatinine, preoperative sepsis, and follow-up surgery. The cross-validated AUROC was 0.7922 (95% CI: 0.7278-0.8440), and the bootstrap AUROC was 0.7895 (95% CI: 0.7365-0.8394), indicating good discriminative ability. The model demonstrated statistically significant fit (LR χ2 = 86.57, p < 0.001) with a pseudo R 2 of 0.1991. A nomogram was constructed to facilitate bedside risk prediction.
Conclusion: The developed scoring system demonstrated good predictive performance in estimating 30-day mortality following emergency laparotomy. Incorporating routine clinical and laboratory parameters, the tool is readily applicable in resource-limited settings. External validation is warranted to assess generalizability and potential for integration into surgical risk assessment workflows.
© 2025 Royal Australasian College of Surgeons.