Author response: Lower carotid revascularization rates after stroke in racial/ethnic minority-serving US hospitals.

Abstract

We respectfully disagree with Dr. Willey's conclusions about our study.1 The International Classification of Diseases, Ninth Revision, Clinical Modification ( ICD-9-CM ) codes used in our study (433.11 and 433.31) do not apply to asymptomatic internal carotid artery (ICA) stenosis and only capture extracranial disease resulting in infarction.2,3 Therefore, racial differences in the prevalence of intracranial disease or in the use of carotid endarterectomy (CEA) or carotid artery stenting (CAS) in asymptomatic ICA stenosis are an unlikely source of bias. The included ICD-9-CM codes indeed also capture patients with <50% stenosis or complete ICA occlusions, for whom revascularization would not be appropriate. We acknowledge this in the discussion as a well-known limitation of the data source; this has also been commented on elsewhere.4 However, it is important to distinguish differences between white and minority patients from differences between white and minority hospitals. Our results describe the latter and are adjusted for, and therefore independent of, individual patient race and other socioeconomic covariates. In other words, differences in eligibility by individual patient race/ethnicity may in part explain within-hospital differences between white and minority patients but would not explain between-hospital differences among patients of the same race. Figure 3 shows that the between-hospital difference in carotid revascularization is driven by lower odds of CEA/CAS in whites and Hispanics in minority, compared with white, hospitals. Individual-level racial differences in eligibility would not explain why white patients in minority-serving hospitals have lower adjusted odds of CEA/CAS than white patients in white hospitals or why Hispanic patients in minority-serving hospitals have lower adjusted odds of CEA/CAS than Hispanic patients in white hospitals. Last, we did not exclude transferred patients as suggested by Dr. Willey. Rather, we excluded records of patients transferred out to another acute care hospital while including those who were transferred in. Since the unit of observation in the data source is hospitalization, this is common practice to prevent double counting of transferred patients who may have 2 hospitalizations pertaining to the same event; this reduces rather than introduces bias.