Abstract
BACKGROUND: Pancreatoduodenectomy is a high-risk operation. National cancer registry data can estimate trends in 90-day outcome measures but are not always included in surgical discussions. We aimed to study national trends in patient selection and postoperative outcomes to better guide shared decision-making between patients and surgeons. METHODS: Patients ≥18 years who underwent pancreatoduodenectomy between 2004 and 2020 were abstracted from the National Cancer Database. Primary outcome was 90-day mortality. Mortality was tabulated by year and stratified by tumor histology: adenocarcinoma, neuroendocrine neoplasm, or other (adenocarcinoma in the setting intraductal papillary mucinous neoplasm, colloid carcinoma, sarcomas, etc.). Multivariable logistic regression determined independent predictors of mortality and survival. RESULTS: A total of 63,283 patients were included. 51.7% were male, with median age of 66 (IQR 59, 73) years. The number of pancreatoduodenectomies per year more than doubled over time from 2219 in 2004 to 4613 in 2020. Tumors were categorized as adenocarcinoma in 91.0%, neuroendocrine in 6.6%, and other malignancies in 2.4%. Change in 90-day mortality over time differed by histology. For patients with pancreatic ductal adenocarcinoma, 90-day mortality decreased from 8.0% in 2004 to 5.3% in 2020 (p < 0.001); for neuroendocrine neoplasm mortality decreased from 8.6% to 3.4% (p < 0.001); for other tumors mortality decreased from 8.7% to 7.2% (N.S). Older age (p < 0.001) and Charlson-Deyo Score ≥3 (p < 0.001) were significant predictors of both 30- and 90-day mortality. CONCLUSIONS: 90-day mortality following pancreatoduodenectomy continues to decrease significantly. Case volume more than doubled over this 16-year period. Continued improvements in patient selection, optimization, and refinement in operative technique may contribute to improving outcomes over time.