Validation of simplified predictive score for postoperative mortality after pancreaticoduodenectomy

Background: Pancreaticoduodenectomy has long been associated with high rates of morbidity and mortality. The key to a better postoperative outcome is a good patient selection. The aim of this study was to develop a simplified preoperative predictive score for postoperative mortality after pancreaticoduodenectomy. Methods: Patients who underwent elective pancreaticoduodenectomy from 1995 to 2012 were identified from the Division of Digestive Surgery database. Bivariate analysis and multivariate logistic regression analysis identified prediction of morbidity and mortality. ROC curve estimation is use to determined the cut-off value of the predictive score. Results: Of 138 patients who underwent pancreaticoduodenectomy, 27 patients (19.6%) died. The predictor of mortality are serum total bilirubin ≥ 10 mg/dL, serum creatinin ≥ 1.3 mg/dL, hematocrit ≤ 30%, serum albumin ≤ 3.0 g/dL and ASA status ≥ 3, with assign score 1, 1, 2, 1, 1, respectively. The cut-off value was 4 with 96% sensitivity and 91% specificity. The area under the receiver operator characteristic curve was 0.974 (SE 0.011; p 0.001), which demonstrated a reasonable predictive value for the score. Conclusion: A total score of 4 or more is associated with increased postoperative mortality in patients underwent pancreaticoduodenectomy.