Background: The optimal hemoglobin target and possible toxicity of epoetin therapy in hemodialysis patients are controversial. Previous studies suggest that African American patients use higher doses of epoetin and have better survival compared with white hemodialysis patients.
Study design: Retrospective longitudinal cohort.
Setting & participants: Epoetin-exposed incident hemodialysis patients (N = 12,733; African Americans, n = 4,801; white, n = 7,386) treated in Dialysis Clinic Inc facilities during 2000 to 2006.
Predictors: Hemoglobin, epoetin, iron.
Outcomes: Mortality, hospitalization.
Measurements: Proportional hazards models with time-varying covariates.
Results: Hemoglobin concentrations less than 10 g/dL in whites and less than 11 g/dL in African Americans were associated with increased mortality and hospitalization versus the referent hemoglobin level of 11 to 11.9 g/dL. Hemoglobin levels of 13 g/dL or greater in whites were associated with decreased noncardiovascular mortality. Six-month cumulative epoetin doses of 20,000 U/wk or greater were associated with increased mortality and hospitalization versus the referent group (8,000 to 12,499 U/wk). Epoetin doses less than 8,000 U/wk were associated with decreased risk. Higher epoetin doses were associated with increased mortality at hemoglobin concentrations of 10 to 12.9 g/dL and with increased hospitalization at all hemoglobin concentrations of 10 g/dL or greater. Higher epoetin doses were associated with increased mortality and hospitalization within each tertile of serum albumin concentration. These patterns did not differ by race.
Limitations: Treatment-by-indication bias and unidentified confounders cannot be excluded. Small sample sizes in the highest and lowest hemoglobin strata decrease statistical power.
Conclusions: Relationships between hemoglobin concentration and mortality differed between African Americans and whites. Additionally, the relationship of lower mortality with greater achieved hemoglobin concentration seen in white patients was observed for all-cause, but not cardiovascular, mortality. A higher cumulative epoetin dose was associated with worse outcomes, even in patients with albumin levels greater than 4 g/dL. There were no statistically significant interactions between race and epoetin dose. Further studies are needed to confirm and to define the mechanism of these findings.