Background: Medicare data indicate that black hemodialysis patients receive greater doses of erythropoietin (EPO) than white patients when achieving similar hemoglobin levels. We confirmed and evaluated this observed association between race and EPO dose.
Study design: Cross-sectional cohort study.
Setting & participants: Primary Medicare-insured white (57%) and black (43%) adult long-term hemodialysis patients treated by Fresenius Medical Care who received EPO from January 1 to 31, 2004 (N = 44,721).
Predictor: White/black race.
Outcomes: Average weekly EPO dose.
Measurements: Associations between race and baseline demographic and laboratory variables were evaluated by using logistic and linear regression models. Correlates of log-transformed weekly EPO dose were determined using linear regression models.
Results: Black patients received 12.6% more EPO than white patients (95% limits, 10.9% to 14.3%; P < 0.001). This racial difference in EPO dose was observed across similar hemoglobin levels despite fewer catheters (P < 0.001) and fewer prior hospitalization events in black patients (P = 0.002). Black patients were younger and had larger body size and greater albumin and biointact parathyroid hormone levels, but lower equilibrated Kt/V and white blood cell counts (all P < 0.001). In the 95th percentile of EPO dose (those receiving > 60,000 U/wk), there was a greater proportion of black patients (6% of total black population compared with only 4% in all white patients; P < 0.001). The difference in EPO dose between black and white patients was modified by age and was significant at ages younger than 45 and 65 years or older.
Limitations: Observational study limited to white and black adult Medicare patients only, correlating with EPO doses from a single month, without adjustment for comorbid conditions.
Conclusions: Black patients were administered approximately 12% greater EPO doses than white patients while achieving similar hemoglobin levels. We identified variables that differed across race that may explain this difference, but they were either not actionable or presented limited opportunity for intervention. Additional studies are needed to define a physiological (or pathological) basis for these observations.