Background/Aims: Randomized trials of hemoglobin targeting in chronic kidney disease suggest that erythropoiesis-stimulating agent (ESA) dosing increases mortality risk, but dosing intensity is confounded by hemoglobin concentration. Appropriately designed observational studies are needed to clarify the association of ESA dosing with mortality risk. Methods: Using Medicare claims, we conducted a retrospective cohort study of mortality risk associated with epoetin alfa (EPO) dosing in prevalent hemodialysis patients (n = 137,918), 2000-2004. We used marginal structural modeling to account for time-varying confounding attributable to recent history of blood transfusion and catheter insertion for vascular access, hemoglobin, hospital admission and days, and intravenous iron dosing. We stratified mortality analyses according to hemoglobin level (<10, 10-10.9, 11-11.9, and ≥12 g/dl). Results: With adjustment for serial correlation in EPO dosing, hemoglobin, hospital admission and days, and intravenous iron administration were the strongest predictors of outpatient EPO dosing. With hemoglobin <10 g/dl, mean weekly EPO dose in a 3-month period was negatively associated with subsequent mortality risk. With hemoglobin 10-10.9 and 11-11.9 g/dl, EPO dose and mortality risk were associated in a U-shaped form. With hemoglobin ≥12 g/dl, mean weekly EPO dose >20,000 IU was positively associated with mortality risk. Conclusions: ESA dosing may be directly associated with risk of death, but the nature of the association likely varies according to hemoglobin concentration. Small doses with hemoglobin ≤12 g/dl and large doses with hemoglobin ≥10 g/dl may both be associated with poor outcomes.
- Epoetin alfa
- Erythropoiesis-stimulating agent
- Hemoglobin level