Assessment of pure red cell aplasia in US dialysis patients: The limits of the Medicare data

Background. A report of neutralizing antibodies to recombinant human erythropoietin (rHuEPO) in European dialysis patients raised concerns that US dialysis patients may be at similar risk. We investigated the frequency of diagnosis codes that include pure red cell aplasia (PRCA) in the Medicare end-stage renal disease data. Methods: Medicare claims data were used to identify incident dialysis patients from 1995 through 1999 aged 67 years or older who did not have a diagnosis code for aplastic anemia (AA; which includes PRCA) or were not administered rHuEPO before dialysis therapy initiation. Patients were assessed for complicating conditions, rHuEPO dose, hematocrit level, blood transfusion, bone marrow testing, and AA diagnosis (maximal follow-up, 13 months). Results: Of 101,782 patients eligible for study, 9,896 patients had diagnosis codes for AA after dialysis therapy initiation, 3,894 patients underwent bone marrow tests (required for PRCA diagnosis), 19 patients had diagnosis codes for AA based on bone marrow examination and no other complicating conditions, and 5 patients were administered blood transfusion in 1 or more months during follow-up, of whom only 1 patient had persistent low hematocrit levels while being administered rHuEPO and blood transfusion. This latter patient was identified from a total of 101,782 patients, or 70,706.75 person-years of rHuEPO exposure. Conclusion: Diagnosis codes in the Medicare data are inadequate for the conclusive study of PRCA in US dialysis patients. Despite limitations of this study, it appears that few cases of rHuEPO-associated PRCA have occurred in US Medicare dialysis patients. Additional investigation is needed to determine whether apparent differences between US complication rates and those elsewhere result from differences in the manufacturing and/or use of erythropoietin products.