Evaluation of Possum Scoring System in Patients Undergoing Laparotomy for Risk Assessment

Abstract

Accurate perioperative risk stratification is essential for consent, triage, and audit in general surgery. The Physiological and Operative Severity Score for the Enumeration of Mortality and Morbidity (POSSUM) and its Portsmouth modification (P-POSSUM) are widely used, but external performance varies by setting. Objective: To evaluate discrimination and calibration of POSSUM (mortality and morbidity) and P-POSSUM (mortality) in patients undergoing midline laparotomy, overall and within key subgroups. Methods: We conducted a prospective observational study of consecutive adults undergoing elective or emergency midline laparotomy at Surgical Unit 5, Civil Hospital Karachi, from 1st August 2024 to 31st January 2025. POSSUM physiological (PS) and operative (OS) scores were computed; predicted risks were derived using standard logistic equations. Primary outcomes were 30-day mortality and 30-day morbidity (any postoperative complication ≥Clavien–Dindo II). Discrimination (AUC) and calibration (Hosmer–Lemeshow, observed: expected [O: E] ratios, decile plots) were assessed, including subgroup analyses by urgency and ASA class. Results: Among 150 patients (60% male; mean age 56.6±15.9 years), 65% underwent emergency surgery. Thirty-day outcomes were: complications 55%, ICU admission 32%, mortality 9%; mean length of stay 14.6±9.1 days. Mortality discrimination was moderate for P-POSSUM (AUC 0.652, 95% CI 0.484–0.820) and POSSUM (AUC 0.629, 95% CI 0.460–0.798); morbidity discrimination was acceptable for POSSUM (AUC 0.704, 95% CI 0.622–0.786). Calibration indicated overall over-prediction: O: E 0.888 for P-POSSUM mortality (13 observed/14.64 expected), 0.421 for POSSUM mortality (13/30.89), and 0.895 for POSSUM morbidity (82/91.57); Hosmer–Lemeshow was significant for mortality and morbidity (p≤0.005). Subgroups showed under-prediction in elective mortality (O: E 1.405) and over-prediction in emergency mortality (O: E 0.763); ASA I–II and III–V strata were closer to unity. Conclusions: POSSUM/P-POSSUM provided moderate discrimination but clinically relevant miscalibration, most pronounced at risk extremes. For decision support and benchmarking, local intercept/slope recalibration and, where appropriate, context-specific models (e.g., NELA or CR-POSSUM) are recommended to improve accuracy.