PHYSICIAN-LEVEL PERFORMANCE AND RACIAL DISPARITIES IN DIABETES CARE
T.D. SEQUIST1; G. FITZMAURICE1; R. MARSHALL2; S. SHAYKEVICH1; D.G. SAFRAN3; J.Z. AYANIAN1.
1Brigham and Women's Hospital, Boston, Mass.; 2Harvard Vanguard Medical Associates, Newton, Mass.; 3Tufts University, Boston, Mass.
BACKGROUND
Racial disparities in diabetes care are well documented, however little information is available regarding the importance of individual physician
performance. We analyzed variation in disparities in diabetes care to quantify the contributions of patient characteristics and individual physicians to population-level differences in care.
METHODS
We used electronic medical record data to identify primary physicians caring for at least 5 white and 5 black adults with diabetes during 2005 within a large multisite group practice in Massachusetts. We assessed rates of optimal control of HbA1c (<7.0%), LDL cholesterol (<100 mg/dL), and blood pressure (<130/80 mmHg). We fit hierarchical linear regression models to 1) measure population- level disparities in diabetes care (base model); 2) adjust disparities for patient characteristics including age, sex, income, and insurance status (patient model); 3) adjust disparities for patient characteristics and health center and physician effects (physician model); and 4) measure adjusted disparities within individual physician panels.
RESULTS
We identified 85 eligible physicians caring for 5,463 patients (62% white, 38% black) across 13 health centers. The median number of white patients per physician was 38 (interquartile range (IQR) 20 to 53, maximum 124) and of black patients was 13 (IQR 8 to 32, maximum 112). There was substantial clustering of care for black patients, with 39% of physicians caring for 75% of black patients. White patients were significantly more likely than black patients to achieve optimal control of HbA1c (40.9% vs 31.8%), LDL cholesterol (47.5% vs 37.3%), and blood pressure (36.0% vs 29.4%, all p<0.001). Adjustment for patient characteristics had a substantial impact on white-black disparities in these 3 measures, with only minor changes related to additional adjustment for health center and physician effects (Table).
Adjusted white-black differences in control rates varied substantially between physician panels for HbA1c (IQR 5.3% to 9.5%), LDL cholesterol (IQR 2.5% to 8.9%), and blood pressure (IQR 7.7% to 10.0%). There was no association between the magnitude of disparity and number of black patients treated within a physician panel for any of the 3 measures.
CONCLUSIONS
Racial disparities in diabetes care are mainly related to patient characteristics and within-physician differences, with little effect due to between-physician differences and no relation to the number of blacks treated by individual physicians. Therefore, targeting physicians with lower performance or shifting
black patients care to physicians who provide more equal care would have a limited impact on disparities. More systemic efforts to improve care for black patients across all physicians will be required.
