The Impact of Hospital and Surgeon Volume on In-Hospital Mortality of Ventricular Assist Device Recipients

Background: The use of left ventricular assist devices (LVADs) in the United States has increased since the Food and Drug Administration approved the 1st device in 1994. Despite a rapid increase in the number of LVADs implanted per year, there are substantial variations in procedure volume among hospitals and surgeons. This study evaluated the association between hospital and surgeon volumes of LVAD procedures and in-hospital mortality. Methods and Results: We conducted a retrospective cross-sectional analysis of all patient discharges after an LVAD implantation from University HealthSystem Consortium (UHC) academic medical center members from January 2007 through June 2012. With the use of International Classification of Diseases-9th Edition, Clinical Modification, procedure code 37.66, we identified 7714 patients who received an LVAD from 581 surgeons across 88 hospitals. The primary outcome was all-cause in-hospital mortality. Annual hospital and surgeon LVAD procedure volumes were evaluated as both continuous variables and quintiles. Hierarchical binary logistic regression models were fitted to test the association of in-hospital mortality with hospital and surgeon volume, controlling for hospital and patient characteristics. Hospital volume was not associated with lower in-hospital mortality. Highest annual surgeon volume quintile was a significant predictor of lower in-hospital mortality (odds ratio 1.69; P < .001); this model had the highest predictive accuracy, with area under the receiver operating characteristic curve of 0.79. Conclusions: Surgeons' LVAD procedure volume, not annual hospital procedure volume, was associated with in-hospital mortality.