National estimates of mortality rates for radical pancreaticoduodenectomy in 25,000 patients

Background: Recent publications suggest an inverse relationship between mortality rates in the Whipple procedure for periampullary cancer and hospital volume/teaching status. Methods: The Nationwide Inpatient Sample database from 1988 to 1995, containing 24,926 patients undergoing pancreatectomy for periampullary cancer, was used. Results: The mean number of procedures per hospital per year was 1.5, and the overall mortality was 14%. The volume of procedures per year increased from the rural to the urban nonteaching hospitals to the urban teaching hospitals (.6, 1.1, and 2.7, respectively), with a steady decrease in mortality among the three hospital types (18%, 15 %, and 11 %). A multiple logistic regression model with mortality odds ratios (ORs) showed that male sex (OR, 1.3), increasing age (OR, 1.6 to 6.7 in decades from 50 to greater than or equal to80 vs. <50 years), emergency admission (OR, 1.5), and hospital volume (less than one vs. one or more cases per year; OR, 1.5) were significantly predictive for increased in-hospital mortality. Conclusions: In-hospital mortality in the low-volume hospital setting is prohibitive, and review of each institution's mortality rates must occur before these procedures are performed in those institutions. In addition, patients over the age of 60 years, male patients, and those with an urgent admission are at a significant risk of in-hospital death, and consideration should be given toward transfer to an experienced institution.