Value of general surgical risk models for predicting postoperative morbidity and mortality in pancreatic resections for pancreatobiliary carcinomas

BACKGROUND: The present study evaluated the utility of general surgical risk models to predict postoperative morbidity and mortality in the specialty field of pancreatic resections for pancreatobiliary carcinomas. METHODS: We investigated Estimation of Physiologic Ability and Surgical Stress (E-PASS), its modified version (mE-PASS), and Portsmouth Physiologic and Operative Severity Score for the enUmeration of Mortality and morbidity (P-POSSUM) in 231 patients undergoing pancreatoduodenectomy or distal pancreatectomy (Group A). We also analyzed E-PASS and mE-PASS in another cohort of the same procedures (Group B, n = 313). RESULTS: Areas under the receiver operating characteristic curve (AUC) for detecting in-hospital mortality in Group A were moderate at 0.75 for E-PASS, 0.69 for mE-PASS, and 0.69 for P-POSSUM. The predicted mortality rates of the models significantly correlated with severity of postoperative complications (ρ = 0.17, P = 0.011 for E-PASS; ρ = 0.15, and P = 0.027 for P-POSSUM). The AUCs were also moderate in Group B at 0.68 for E-PASS and 0.69 for mE-PASS. The predicted mortality rates significantly correlated with severity of postoperative complications (ρ = 0.18, P = 0.0018 for E-PASS; ρ = 0.17, and P = 0.0022 for mE-PASS). CONCLUSIONS: The present study suggests that the predictive powers of general risk models may be moderate in pancreatic resections. A novel model would be desirable for these procedures.