Stephen J.LurieMD, PhD, Senior EditorIndividualAuthorJody W.ZylkeMD, Contributing EditorIndividualAuthor
To the Editor: Dr Asch and colleagues attempted to measure underuse of necessary care among Medicare beneficiaries.1 To improve the interpretability of this approach, we suggest 2 areas for additional investigation.
First, their approach requires more rigorous validation before it can be used to make comparisons between providers. The authors suggest that the lower use rates of many indicators among typically underserved populations, compared with the general population, demonstrate the validity of their method. A plausible alternative explanation, however, is that these differences in rates reflect variations in data quality and coding strategy. For example, the low rates of electrocardiogram use noted in the study (eg, 54.9% of all subjects received an electrocardiogram within 2 days of the initial diagnosis of a transient ischemic attack) may be more suggestive of undercoding than of true underuse of care. Similarly, differences in the use rate of this indicator (eg, 63.0% among African American subjects and 48.2% among subjects living in a poverty area) may reflect differences in coding rather than in receipt of care. Indeed, without validation of each of the specific indicators in Medicare Part A and B claims data, it may be impossible to disentangle these effects. We also would encourage further refinement of indicator selection process, especially given the greater variability in expert panel recommendations that may occur when the level of evidence is low.2
Hsu J, Go A, Selby J. Overuse of Administrative Data to Measure Underuse of Care. JAMA. 2001;285(6):735-737. doi:10-1001/pubs.JAMA-ISSN-0098-7484-285-6-jlt0214