Mortality is consistently reported as an outcome metric in pancreatic surgery. Given its heterogeneity, better characterization of it might provide crucial insights for clinical practice. This study aimed to analyze the timeline and sequence of events that lead to death after pancreatoduodenectomy to identify possible distinct pathways of mortality.
All consecutive pancreatoduodenectomy cases from 2010 to 2020 were retrospectively analyzed. A day-to-day appraisal of the postoperative course of each fatality was performed and visualized graphically. The graphical analysis allowed for pattern identification. The respective predictors were explored through logistic regression.
Out of 2065 pancreatoduodenectomy patients, in-hospital mortality was 3.1%. With graphical analysis, 3 patterns were identified. Pattern A deaths (71.4%, n = 45) occurred after a median of 43 days (14-260), following pancreas-specific complications such as postoperative pancreatic fistula, postpancreatectomy hemorrhage, and delayed gastric emptying. Pattern B deaths (15.9%, n = 10) occurred after a median of 18 days (1-55), succeeding a critical status in the early postoperative course, mainly related to elevated surgical complexity. Patients with pattern C (12.7%) died after a median of 8 days, mostly for unknown cause after an uneventful postoperative course. The predictors of each pattern were distinctive.
Mortality after pancreatoduodenectomy occurs through 3 distinct pathways. This knowledge could spawn an additional endpoint of value to clinicians and hospitals, delivering a supplementary tool for comparison between centers and diversified patient populations, and it might facilitate the identification of the best targets for improvement. Further studies are needed to validate this tripartite reclassification.

Copyright © 2022 Elsevier Inc. All rights reserved.