Disparities in Health Care Are Driven by Where Minority Patients Seek Care

BACKGROUND: Racial/ethnic disparities in health care are well documented, but less is known about whether disparities occur within or between hospitals for specific inpatient processes of care. We assessed racial/ethnic disparities using the Hospital Quality Alliance Inpatient Quality of Care Indicators. METHODS: We performed an observational study using patient-level data for acute myocardial infarction (5 care measures), congestive heart failure (2 measures), community-acquired pneumonia (2 measures), and patient counseling (4 measures). Data were obtained from 123 hospitals reporting to the University HealthSystem Consortium from the third quarter of 2002 to the first quarter of 2005. A total of 320,970 patients 18 years or older were eligible for at least 1 of the 13 measures. RESULTS: There were consistent unadjusted differences between minority and nonminority patients in the quality of care across 8 of 13 quality measures (from 4.63 and 4.55 percentage points for angiotensin-converting enzyme inhibitors for acute myocardial infarction and congestive heart failure P<.01 to 14.58 percentage points for smoking cessation counseling for pneumonia P=.02). Disparities were most pronounced for counseling measures. In multivariate models adjusted for individual patient characteristics and hospital effect, the magnitude of the disparities decreased substantially, yet remained significant for 3 of the 4 counseling measures; acute myocardial infarction (unadjusted, 9.00 P<.001; adjusted, 3.82 P<.01), congestive heart failure (unadjusted, 8.45 P=.02; adjusted, 3.54 P=.02), and community-acquired pneumonia (unadjusted, 14.58 P=.02; adjusted, 4.96 P=.01). CONCLUSIONS: Disparities in clinical process of care measures are largely the result of differences in where minority and nonminority patients seek care. However, disparities in services requiring counseling exist within hospitals after controlling for site of care. Policies to reduce disparities should consider the underlying reasons for the disparities.