Should all pancreatic surgery be centralized regardless of patients' comorbidity?

Background: It remains to be established whether centralization to high volume centers is essential for all patients undergoing pancreatic surgery. The aims of this study were to identify the optimal cut-off volume to optimize patient outcomes and to determine if patient comorbidity affected the volume-outcome relationship.

Methods: Patients undergoing pancreatectomy from 2012 to 2015 were retrospectively identified (n = 12 333) in the French nationwide database. The 90-day Post-Operative Mortality (POM) was analyzed according to hospital volume of pancreatectomy (very low:<10, Low:10-19, High:20-49 and very high:≥50 resections/year) and Charlson Comorbidity Index (ChCI).

Results: The overall POM was 6.9%. The cut-off of 20 pancreatic resections per year was identified as predictor of POM. Compared to high volume centers, POM was significantly higher in low and very low volume centers whatever the ChCl. Regarding surgical procedures, there was a significant decrease in POM with increasing hospital volume only after pancreaticoduodenectomy regardless of the ChCl. On multivariable analysis, low and very low volume centers were independently associated with increased mortality rates.

Conclusion: The optimal cut-off of annual caseload was 20 pancreatic resections. POM following pancreaticoduodenectomy is high in low and very low volume centers independently of ChCl, suggesting that this procedure should be centralized.