BACKGROUND/AIM: Different scoring systems to predict the occurrence of postoperative pancreatic fistula (POPF) after pancreatoduodenectomy have been described, but the considered risk factors often suffer subjective scaling. The aim of this review is to evaluate and compare all published risk metrics predictive of POPF. METHODS: All existing scores were retrieved by literature web search. Inclusion criteria were ISGPF classification of POPF and the development of a risk score metric. RESULTS: From a total of 286 publications, 10 studies were selected. Most of them were retrospective and single center. The models considered a median number of 3 items (range from 2 to 5); in 5 of 10 trials only pre or intraoperative variables were included. The median number of patients/study was 186 (IQR 111.1-229.0). External validation was performed in 6 of 10 studies. The most recurrent items were abdominal fat (4/10), main pancreatic duct diameter (in 4/10), and pancreatic texture (3/10). CONCLUSION: POPF risk estimation should be easy, accurate, and objective. It should consider preoperative patient-related and gland-related features, and intraoperative events. None of the published systems completely adhere to these principles. Large heterogeneous multicentric validations should be endorsed, to account for the case-mix and evaluate the reproducibility of each scoring system.
BACKGROUND/AIM: Different scoring systems to predict the occurrence of postoperative pancreatic fistula (POPF) after pancreatoduodenectomy have been described, but the considered risk factors often suffer subjective scaling. The aim of this review is to evaluate and compare all published risk metrics predictive of POPF. METHODS: All existing scores were retrieved by literature web search. Inclusion criteria were ISGPF classification of POPF and the development of a risk score metric. RESULTS: From a total of 286 publications, 10 studies were selected. Most of them were retrospective and single center. The models considered a median number of 3 items (range from 2 to 5); in 5 of 10 trials only pre or intraoperative variables were included. The median number of patients/study was 186 (IQR 111.1-229.0). External validation was performed in 6 of 10 studies. The most recurrent items were abdominal fat (4/10), main pancreatic duct diameter (in 4/10), and pancreatic texture (3/10). CONCLUSION: POPF risk estimation should be easy, accurate, and objective. It should consider preoperative patient-related and gland-related features, and intraoperative events. None of the published systems completely adhere to these principles. Large heterogeneous multicentric validations should be endorsed, to account for the case-mix and evaluate the reproducibility of each scoring system.
Authors: Sara L Zettervall; Tammy Ju; Jeremy L Holzmacher; Bridget Huysman; Gregor Werba; Anton Sidawy; Paul Lin; Khashayar Vaziri Journal: J Gastrointest Surg Date: 2019-04-03 Impact factor: 3.452