RELATION OF PERIOPERATIVE DEATHS TO HOSPITAL VOLUME AMONG PATIENTS UNDERGOING PANCREATIC RESECTION FOR MALIGNANCY

Objective The authors examined the effect of hospital and surgeon volu me on perioperative mortality rates after pancreatic resection for the treatment of pancreatic cancer. Methods Discharge abstracts from 1972 patients who had undergone pancreaticoduodenectomy or total pancreate ctomy for malignancy in New York State between 1984 and 1991 were obta ined from the Statewide Planning and Research Cooperative System. Logi stic regression analysis was used to determine the relationship betwee n hospital and surgeon experience to perioperative outcome. Results Mo re than 75% of patients underwent resection at minimal-volume (fewer t han 10 cases) or low-volume (10-50 cases) centers (defined as hospital s in which a minimal number of resections were performed in a given ye ar), and these hospitals represented 98% of the institutions treating peripancreatic cancer. The two high-volume hospitals (more than 81 cas es) demonstrated a significantly lower perioperative mortality rate (4 .0%) compared with the minimal- (21.8%) and low-volume (12.3%) hospita ls (p < 0.001), The perioperative mortality rate was 15.5% for low-vol ume (fewer than 9 cases) surgeons (defined as surgeons who had perform ed a minimal number of resections in any hospital in a given year) (n = 687) compared with 4.7% for high-volume (more than 41 cases) pancrea tic surgeons (n = 4) (p < 0.001). Logistic regression analysis demonst rated that perioperative death is significantly (p < 0.05) related to hospital volume, but the surgeon's experience is not significantly rel ated to perioperative deaths when hospital volume is controlled. Concl usions These data support a defined minimum hospital experience for el ective pancreatectomy for malignancy to minimize perioperative deaths.