Mortality after total pancreatectomy in an observational nationwide cohort study.

[BACKGROUND] As a result of adjuvant therapies for pancreatic cancer including new techniques, such as arterial divestment, and, for instance, as a result of more main-duct intraductal papillary mucinous neoplasms being diagnosed, the number of total pancreatectomies is increasing. Results of these complex operations are only available as centre analyses. The aim of this study was to analyse mortality after total pancreatectomy using real-world data.

[METHODS] An observational retrospective study based on routine data of the nationwide German diagnosis-related group (DRG) statistics was performed. All total pancreatectomies from 2010 to 2020 were analysed and classified into four types. Univariable and multivariable analyses were performed for in-hospital mortality as the endpoint.

[RESULTS] During the study interval, 13 946 patients underwent total pancreatectomy; 7060 patients (50.6%) underwent type 1 standard total pancreatectomy, 1682 patients (12.1%) underwent type 2 total pancreatectomy (included venous procedures), 3921 patients (28.1%) underwent type 3 total pancreatectomy (included multivisceral resections), and 1283 patients (9.2%) underwent type 4 total pancreatectomy (included arterial procedures). Relaparotomy was necessary for 3232 patients (23.2%). Overall, the in-hospital mortality rate was 22.0% (3072 patients), which varied depending on the type of total pancreatectomy (type 1, 15.6% (1102 patients); type 2, 17.2% (290 patients); type 3, 31.5% (1235 patients); and type 4, 34.7% (445 patients); P < 0.001). The worst outcome of 60.7% in-hospital mortality was related to a small subgroup of 56 patients who were operated on in hospitals that performed one to ten type 4 total pancreatectomies in 2010-2020. With increasing caseload, in-hospital mortality significantly decreased (P < 0.001) for all resection types.

[CONCLUSION] The mortality after total pancreatectomy differs significantly between hospitals based on nationwide administrative data across all hospitals. Further in-depth analysis of more granular data to understand the case volume-mortality relationship for total pancreatectomy is warranted.