Background: Most prior studies of racial differences in the delivery of cardiac care have focused on potential differences in treatment by individual physicians and hospitals. However, differential use of hospitals with variable practice patterns might also contribute to variations in care.
Methods and results: We compared the use of bioprosthetic valves (BPVs) in 78,154 black and white Medicare beneficiaries > or =65 years of age undergoing aortic valve replacement in 904 US hospitals during 1999 through 2001. Generalized linear mixed models were used to account first for differences in patient characteristics and then for differences in hospitals used by black and white patients. BPV use was lower in black patients relative to white patients after adjustment for patient characteristics (relative risk, 0.93; 95% CI, 0.91 to 0.95; P<0.001). However, black patients were more likely to undergo surgery in hospitals in the lowest quintile of BPV use overall (29% versus 20% of white patients; P<0.001). After hospital-level variability in BPV use was accounted for, the use of BPVs was actually somewhat higher in black patients (relative risk, 1.06; 95% CI, 1.04 to 1.09; P<0.001). Model discrimination as measured by the c statistic was markedly higher after the addition of hospital effects (0.80 versus 0.59 for patient characteristics alone; P<0.001).
Conclusions: Accounting for differences in hospitals preferentially used by black and white patients had a major impact on estimating racial differences in the use of BPVs in patients undergoing aortic valve replacement. Hospital-level effects explained a larger proportion of the variation in BPV use than race and other patient characteristics alone.