To the Editor.
—We disagree with the conclusion by Mr Petronis et al1 that Medicare claims data have a 25% "estimated margin of error" in the administration of recombinant human erythropoietin (rHuEPO). The authors considered a group of patients who had records of rHuEPO use in a database abstracted from medical records, the Case Mix/Adequacy Study (CM/AS). They found that 25% of this group did not have corresponding records of rHuEPO use in Medicare's Program Management and Medical Information System (PMMIS), which is based on providers' billing records. The authors' findings are problematic because, among other things, they (1) matched patients' records from different time periods, (2) used incomplete PMMIS files, (3) used a baseline measure that was not validated against its original source, and (4) did not fully account for the effect of Medicare Secondary Payer (MSP) provisions. The Table illustrates how Petronis et al matched patient records
Sadler JH. Race and Access to Erythropoietin by Patients on Hemodialysis. JAMA. 1994;272(23):1818-1819. doi:10.1001/jama.1994.03520230028020