Centralization of pancreatoduodenectomy a decade later: Impact of the volume-outcome relationship

Background. The hospital volume outcome relationship for complex procedures has led to the suggestion that care should be centralized. This study was performed to investigate whether centralization is occurring for pancreatoduodenectomy (PD) and to examine its effect on short-term postoperative outcomes. Methods. We queried the New York State Statewide Planning and Research Cooperative System database (n = 6,185, 2002-2011) and the California and Florida State Inpatient Databases (n = 6,766 and 4,810, respectively, 2002-2011) for PD. Hospitals were divided into low 10), medium (11-25), high (25-60), and very high (>= 61) groups depending on annual volume. Hierarchical logistic modeling accounted for patient clustering within hospitals. Results. A migration of cases from low-volume to medium, high, and very high-volume (MHVH) hospitals occurred in these 3 states (P < .01). There was an increase in the number of MHVH hospitals and a decrease in the number of low-volume hospitals performing PD across all states over time, with a large number of hospitals ceasing to perform PD cases entirely. Comorbidities such as congestive heart failure and diabetes were more prevalent in low-volume hospitals. After we adjusted for all predictors, MHVH hospitals had less rates of mortality and morbidity and shorter durations of stay than low volume hospitals (P < .05); 30-day readmission rates were similar across all volume groups. Conclusion. Centralization of PD is occurring in these 3 states and probably across the nation. After PD, MHVH hospitals had statistically better outcomes (mortality, morbidity, and duration of stay) than low-volume hospitals. Readmission rates were not affected by volume.