Scoring system to predict pancreatic fistula after pancreaticoduodenectomy: A UK multicenter study

Abstract OBJECTIVE: To validate a preoperative predictive score of postoperative pancreatic fistula (POPF). Other risk factors for POPF were sought in an attempt to improve the score. BACKGROUND: POPF is the major contributor to morbidity after pancreaticoduodenectomy (PD). A preoperative score [using body mass index (BMI) and pancreatic duct width] to predict POPF was tested upon a multicenter patient cohort to assess its performance. METHODS: Patients undergoing PD at 8 UK centers were identified. The association between the score and other pre-, intra-, and postoperative variables with POPF was assessed. RESULTS: A total of 630 patients underwent PD with 141 occurrences of POPF (22.4%). BMI, perirenal fat thickness, pancreatic duct width on computed tomography and at operation, bilirubin, pancreatojejunostomy technique, underlying pathology, T stage, N stage, R status, and gland firmness were all significantly associated with POPF. The score predicted POPF (P < 0.001) with a higher predictive score associated with increasing severity of POPF (P < 0.001). Stepwise multivariate analysis of pre-, intra-, and postoperative variables demonstrated that only the score was consistently associated with POPF. A table correlating the risk score to actual risk of POPF was created. CONCLUSIONS: The predictive score performed well and could not be improved. This provides opportunities for individualizing patient consent and selection, and treatment and research applications.