Reasons why data from the Nationwide Inpatient Sample can be misleading for carotid endarterectomy and carotid stenting

Abstract

The Nationwide Inpatient Sample (NIS) is often used for population-based research comparing the safety of carotid artery stenting (CAS) to that of carotid endarterectomy (CEA) in the United States. At least two findings from the NIS dataset seem questionable, however. First, several NIS studies indicate that >90% of CEAs and CAS procedures are currently being performed for asymptomatic carotid stenosis, which considerably exceeds the prevalence of asymptomatic patients reported elsewhere. Second, these studies also suggest that periprocedural stroke rates for CEA and CAS are collectively lower at hundreds of community hospitals contributing data to the NIS than they were in the Carotid Revascularization Endarterectomy Versus Stenting Trial (CREST), even though the participating surgeons and interventionalists in CREST were vetted on the basis of their previous experience and results. In addition, some unexpectedly low stroke to death ratios are present in NIS studies, implying that not all iatrogenic strokes have been entered into the NIS dataset. These issues might be related to inadequate documentation of preprocedural symptoms and periprocedural strokes in the medical records, leading to subsequent coding errors in the hospital discharge abstracts from which NIS data are extracted. The clinical limitations of the NIS and other administrative datasets have been pointed out in the past, but they appear to be particularly relevant to carotid interventions and must be recognized.