Background: Series from expert centers suggest that pancreas cancer surgery is safe for elderly patients but nationwide data, taking hospital volume into account, are lacking.
Methods: From the Netherlands Cancer Registry, all 3420 patients who underwent pancreatoduodenectomy (PD) for primary pancreatic or periampullary carcinoma in 2005-2013 were selected. Associations between age (<75, ≥75 years), hospital volume (tertiles), and postoperative mortality (30, 90 day) were evaluated by χ (2) tests and logistic regression analyses. Overall survival was investigated by means of Kaplan-Meier and Cox proportional hazard regression analyses.
Results: The proportion of elderly patients (≥75 years) undergoing PD increased from 15 % in 2005-2007 to 20 % in 2011-2013 (p = 0.009). In low (<15 per year), medium (15-28 per year), and high (>28 per year) hospital volume tertiles, the proportion of elderly patients was 16, 20, and 17 %, respectively (p = 0.10). With increasing hospital volume, 30-day postoperative mortality was 6.0-4.5-2.9 % (p = 0.002) and 90-day mortality 9.3-8.0-5.3 % (p = 0.001), respectively. Within each volume tertile, adjusted 30- and 90-day mortality of elderly patients was 1.6-2.5 times higher compared to outcomes of younger patients. Adjusted 30-day mortality in elderly patients was higher in low-volume hospitals (odds ratio = 2.87, 95 % confidence interval 1.15-7.17) compared to high-volume hospitals. Similarly, elderly patients had a worse overall survival in low-volume hospitals (hazard ratio = 1.28, 95 % confidence interval 1.01-1.63). Postoperative mortality of elderly patients in high-volume hospitals was similar to mortality of younger patients in low- and medium-volume hospitals.
Conclusions: Elderly patients benefit from centralization by undergoing PD in high-volume hospitals, both with respect to postoperative mortality and survival. It would seem reasonable to place elderly patients into a high-risk category; they should only undergo surgery in the highest-tertile-volume hospitals.