Risk Evaluation of the NCD Risk Calculator for Open Pancreaticoduodenectomy in Elderly Patients: A Validation Study

There is no clear indication for surgery in pancreaticoduodenectomy (PD) for the elderly patients. The aim of this study was to use real-world data to investigate the usefulness of preoperative risk assessment with the risk calculator available in the National Clinical Database (NCD) in Japan.

A retrospective analysis of 311 patients aged ≥ 65 years who underwent PD was performed. In addition to background factors, preoperative predicted incidence rates calculated with the risk calculator, as well as visceral fat analysis items, were analyzed. Patients with (1) serious postoperative complications, (2) a decline in postoperative activities of daily living (ADL), or (3) discharge to a place other than home were defined as having a poor postoperative course. All patients were randomly assigned to the training cohort (n = 209) or validation cohort (n = 102).

Comparisons of patient characteristics revealed no differences between the training and validation cohorts. In the training cohort, multivariate analysis revealed that "Predicted incidence of postoperative ADL decline" of ≥ 44.8% (OR 4.68; p = 0.031) and "Predicted incidence of Clavien-Dindo grade IV or higher" of ≥ 9.2% (OR 5.92; p = 0.025) among those calculated with the risk calculator were independent predictors of a poor postoperative course. Among patients with 2, 1, and none of these factors, 100%, 47.4%, and 15.7%, respectively, had a poor postoperative course.

A "Predicted incidence of postoperative ADL decline" and "Predicted incidence of Clavien-Dindo grade IV or higher" in the NCD risk calculator were useful predictors of a poor postoperative course after PD.