Not Only Caseload but Also Patient Selection Is Predictive of Mortality After Pancreatic Resection

Centralization of pancreatic surgery in high-volume centers is regarded as a key strategy in improving the outcome quality. However, the specific factors, in addition to higher case volumes, that influence inhospital mortality remain unclear.

In this retrospective observational study, the German nationwide diagnostic-related groups statistics were analyzed for 86,073 patients with pancreatic resections. Hospitals performing at least 50 resections per year were identified as high-volume pancreatic centers (HVPCs). Statistical analyses compared crude and adjusted estimates of inhospital mortality for patients treated in HVPCs and non-HVPCs. A generalized mixed model was used for risk adjustment, considering various factors such as age group, sex, diagnosis, and comorbidities (ClinicalTrail.gov, NCT06390891).

A total of 24.2% (n = 20,798) of all pancreatic resections were performed in 23 HVPCs. The crude inhospital mortality for all patients undergoing resection was 9.0%. Crude inhospital mortality in HVPCs was 5.5% compared with 10.1% in non-HVPCs (P < 0.001). HVPCs performed more complex resections including more concomitant procedures. On the other hand, HVPCs treated younger patients and patients with less complicated comorbidities. Statistical adjustment of comorbidities and patient characteristics resulted in a significant increase of inhospital mortality from 5.5% to 8.7% in HVPCs.

HVPCs have significantly lower inhospital mortality than the other hospitals. Nevertheless, the superior quality of outcome can be attributed not only to the enhanced expertise of the centers but also, at least in part, to a healthier patient population on average. However, the extent to which this patient selection is due to active selection by the practitioners or other causes remains unclear.