Relationship between hospital volume and late survival after pancreaticoduodenectomy.

BACKGROUND: Several studies have reported lower perioperative mortality rates with pancreaticoduodenectomy at high-volume hospitals than at low-volume hospitals. We sought to determine whether volume is also related to survival after hospital discharge.
METHODS: Using information from the Medicare claims database, we performed a retrospective cohort study of all 7229 patients over age 65 undergoing pancreaticoduodenectomy in the United States between 1992 and 1995. We divided the study population into approximate quartiles according to their hospital's average annual volume of pancreaticoduodenectomies in Medicare patients: very low (< 1/y), low (1-2/y, medium (2-5/y), and high (5+/y). To adjust for potentially confounding variables, we used a Cox proportional hazards model to examine relationships between hospital volume and mortality, our primary outcome measure.
RESULTS: Overall, 3-year survival was higher at high-volume centers (37%) than at medium- (29%), low- (26%), and very low volume hospitals (25%) (log-rank P < .0001). After excluding perioperative deaths and adjusting for case-mix, patients undergoing surgery at high-volume hospitals remained less likely to experience late mortality than patients at very low volume centers (adjusted hazard ratio 0.69, 95% CI 0.62-0.76). Relationships between hospital volume and survival after discharge were not restricted to patients with cancer diagnoses; patients with benign disease had similar improvements in late survival after surgery at high-volume centers.
CONCLUSIONS: Hospital volume strongly influences both perioperative risk and long-term survival after pancreaticoduodenectomy. Our data suggest that both patient selection and differences in quality of care may underlie better outcomes at high-volume referral centers.