Evaluating the PULP Score's Resilience in Surgical Risk Stratification During Ramadan and Non- Fasting Periods: A 20-Year Retrospective Study on Peptic Ulcer Perforation

Background The holy month of Ramadan brings profound physiological changes for millions of fasting Muslims worldwide, yet evidence-based tools to stratify critical surgical risks during this period remain scarce. We investigated whether the PULP score, a proven predictor of mortality in peptic ulcer perforation, maintains its life-saving precision during Ramadan's unique metabolic challenges. Methods We conducted a 20-year retrospective cohort study (2004–2024) at a tertiary hospital in Ibb, Yemen. We analyzed 203 patients who underwent emergency surgery for PUP, 31 of whom presented during Ramadan. The predictive performance of four risk scores (PULP, Boey, Mannheim, and Jabalpur/Acute Physiologic and Chronic Health Evaluation) was rigorously validated against a composite primary outcome of in-hospital mortality or major complications. Analysis included multivariable regression and receiver operating characteristic (ROC) curves. Results The PULP score demonstrated exceptional discriminative ability (AUC = 0.747) that remained robust during both Ramadan and non-Ramadan periods. At the optimal cutoff (≥ 12), it identified high-risk patients with 80% sensitivity and 98% negative predictive value, substantially outperforming established alternatives. Strikingly, very high-risk patients (PULP ≥ 12) faced dramatically worse outcomes: 9.0% poor outcome rate versus 0–2% in lower-risk categories (p < 0.001), significantly longer hospital stays (9.0 ± 3.2 vs 7.9 ± 1.4 days, p = 0.014), and higher complication rates (19.1% vs 13.2%). Despite theoretical concerns, Ramadan presentation did not independently predict poor outcomes (OR = 0.83, p = 0.855) nor diminish PULP score performance. Conclusion The PULP score provides a robust, Ramadan-resilient tool for emergency risk stratification that could transform surgical triage in resource-limited settings. Its implementation offers a tangible pathway to optimize ICU resource allocation while safeguarding patient safety during periods of heightened surgical demand. This validation in a unique physiological context strengthens its global applicability and immediate clinical utility. Additionally, this proof-of-concept framework highlights the potential of explainable artificial intelligence (XAI) to generate clinical hypotheses and enhance model interpretability, though it requires external validation before clinical application. This work holds promise as a foundation for future tools aimed at supporting risk-stratified treatment decisions in gastric cancer management.