Early Risk Stratification and Mortality Prediction in Gastrointestinal Perforation: A Retrospective Cohort Study for Personalized Surgical Decision Making

Background: Gastrointestinal perforation is a life-threatening surgical emergency associated with high morbidity and mortality despite advances in imaging, perioperative care, and emergency surgical management. Early identification of patients at increased risk of death may improve perioperative risk stratification and support personalized clinical decision-making in emergency settings. Methods: We conducted a retrospective observational cohort study including 166 consecutive adult patients undergoing emergency surgery for gastric, duodenal, ileal, colonic, or intraperitoneal rectal perforation between January 2021 and December 2025. Appendiceal perforations, iatrogenic perforations, and anastomotic leaks were excluded. Univariate analysis was performed using appropriate non-parametric and categorical statistical tests. Variables with p < 0.05 were considered for multivariable logistic regression analysis. Postoperative variables potentially influenced by the outcome were excluded to reduce reverse causation and overadjustment bias. Age was analyzed as a continuous variable in regression analysis and subsequently dichotomized at 75 years for development of a simplified bedside score. Results: Overall in-hospital mortality was 22.3% (37/166). Increasing age (OR 1.08 per year increase, 95% CI 1.04–1.12; p < 0.001), septic shock at emergency department admission (OR 7.06, 95% CI 1.29–38.65; p = 0.024), and intraoperative vasopressor requirement (OR 6.45, 95% CI 1.34–31.10; p = 0.020) were independently associated with mortality. A simplified predictive score based on these variables demonstrated good discrimination, with an area under the receiver operating characteristic curve (AUC) of 0.83. Conclusions: Mortality following gastrointestinal perforation was associated primarily with early physiological derangement and patient frailty rather than anatomical or technical surgical factors alone. Early identification of high-risk patients may support perioperative risk stratification and patient-centered emergency surgical decision-making. The proposed predictive score should be considered preliminary and hypothesis-generating, as neither internal nor external validation was performed.