The Nationwide Inpatient Sample may contain inaccurate data for carotid endarterectomy and carotid stenting

Abstract

The Nationwide Inpatient Sample (NIS) contains information from discharge abstracts submitted by hundreds of community hospitals across the United States, and it frequently has been used as a resource for population-based research comparing the safety of carotid artery stenting (CAS) to that of carotid endarterectomy (CEA). However, at least two findings from the NIS dataset seem open to question. First, several NIS studies have indicated that more than 90% of CEAs and CAS procedures now are being done in asymptomatic patients, a figure that substantially exceeds the prevalence of asymptomatic patients that has been reported elsewhere. Second, these studies also have suggested that the periprocedural stroke rate for CEA and CAS is lower at community hospitals contributing to the NIS than it was in the Carotid Revascularization Endarterectomy vs Stenting Trial (CREST), even though the surgeons and interventionalists participating in CREST were stringently selected according to their previous experience and results. Neither of these two findings seems to pass the test of face validity. Furthermore, some unexpectedly low stroke-to-death (STD) ratios are present in the NIS data, especially for CAS. These issues may be related to poor documentation of preprocedural symptoms and periprocedural strokes in the medical records and to subsequent coding errors in the hospital discharge abstracts on which the NIS is based.