Clinical validation of the risk scoring systems of postoperative pancreatic fistula after laparoscopic pancreatoduodenectomy in Chinese cohorts: A single-center retrospective study

Abstract

BackgroundAlthough several prediction models for the occurrence of clinically relevant postoperative pancreatic fistula after laparoscopic pancreatoduodenectomy exist, most were established using Western cohorts. The utility of these models using a Chinese cohort has not been validated widely. The aim of this study was to validate the original Fistula Risk Score, the alternative Fistula Risk Score, and the updated alternative Fistula Risk Score for patients undergoing laparoscopic pancreatoduodenectomy in a large-scale Chinese cohort externally. MethodsThree clinically relevant postoperative pancreatic fistula risk models were selected for external validation with our database. Primary outcome was grade B/C postoperative pancreatic fistula (clinically relevant postoperative pancreatic fistula). Performance was measured based on sensitivity, specificity, positive and negative predictive value, positive and negative likelihood ratio, and area under the curve. The original Fistula Risk Score was also compared with the alternative Fistula Risk Score and the updated alternative Fistula Risk Score. ResultsOf the 400 patients who underwent laparoscopic pancreatoduodenectomy, 60 (15.00%) developed clinically relevant postoperative pancreatic fistula. For the original Fistula Risk Score, the alternative Fistula Risk Score, and the updated alternative Fistula Risk Score, the sensitivity was 65.00%, 90.00%, and 90.00%; the specificity was 43.53%, 44.12%, and 37.65%; the positive predictive value was 16.88%, 22.13%, and 20.30%; the negative predictive value was 87.57%, 96.15%, and 95.52%; positive likelihood ratio was 1.151, 1.611, and 1.443; negative likelihood ratio was 0.804, 0.227, and 0.266, respectively. The area under the curve values were 0.608 (95% confidence interval 0.573–0.649), 0.733 (95% confidence interval 0.692–0.797), and 0.720 (95% confidence interval 0.688–0.763) on the original Fistula Risk Score, the alternative Fistula Risk Score, and the updated alternative Fistula Risk Score (P < .05). ConclusionThe alternative Fistula Risk Score and the updated alternative Fistula Risk Score had similarly good predictive utility. The original Fistula Risk Score performed less well. We recommended to use the alternative Fistula Risk Score and the updated alternative Fistula Risk Score to predict occurrence of clinically relevant postoperative pancreatic fistula after laparoscopic pancreatoduodenectomy when applied to a Chinese cohort.