Validation of a clinical score in predicting pancreatic fistula after pancreaticoduodenectomy.

ABSTRACT

BACKGROUND: Postoperative pancreatic fistula (POPF) is one of the most severe complications after cephalic pancreaticoduodenectomy, with mortality as high as 30%. Risk scores may help predict the risk of POPF. Multiple external validations substantially improve generalized clinical acceptability of a scoring system. AIM: The aim of this study was to externally validate previously described fistula risk score in the prediction of clinically relevant POPF. METHODS: All patients who underwent pancreaticoduodenectomy for any indication during a 5-year period were prospectively analyzed. A total of 132 patients were analyzed. RESULTS: Of the 132 patients, 44 (33.3%) developed pancreatic fistula, including 12.9% biochemical leaks, 7.6% grade B fistula, and 12.9% grade C fistula. Cut-off point of 4.5 was determined to best separate patients who developed clinically relevant POPF with area under curve of 78% (p = .00003). Sensitivity and specificity for the prediction of clinically relevant POPF with the cut-off value of 4.5 was 70.4 and 74.3%, respectively. Positive predictive value with cut-off value 4.5 was 57.8%, and negative predictive value was 83.4%. CONCLUSION: Fistula risk score identified low risk patients with false negative rate of 16.6%. Further external validation studies on large cohorts of patients and with wide case-mix may enable additional refinements of the score model.