Effect of hospital volume and experience on in-hospital mortality for pancreaticoduodenectomy.

OBJECTIVE: To examine the relative impact of procedure volume versus years of hospital experience on inpatient death rates after pancreaticoduodenectomy. SUMMARY BACKGROUND DATA: Past studies have identified a significant volume-outcome relation for hospitals performing pancreaticoduodenectomy (the Whipple procedure).
METHODS: Administrative discharge data were examined for 6,652 patients who underwent the procedure between 1988 and 1998 in California and Florida. Patients were divided into approximate quartiles according to each hospital's annual procedure volume: very low (1), low (2 or 3), medium (4-9), and high (10+). Logistic regression analysis was used to examine differences in inpatient mortality among hospitals with different procedure volume and years of experience, while adjusting for patient characteristics.
RESULTS: Medium- and high-volume hospitals had lower rates of inpatient mortality after pancreaticoduodenectomy relative to very-low-volume facilities. Greater years of hospital experience also reduced the odds of inpatient death. Predictions based on the regression estimates indicate that within volume categories, increased hospital experience did not lead to significant reductions in inpatient mortality. However, high-volume hospitals had significantly lower inpatient mortality rates than very-low-volume facilities with the same amount of experience.
CONCLUSIONS: Hospitals with more years of experience with pancreaticoduodenectomy had lower rates of inpatient mortality. However, higher procedure volume has played a larger role than increased experience in reducing inpatient death rates.