Abstract W P183: Testing the Utility of an Emergency Room-Based Stroke Evaluation Protocol

Background and purpose: A fundamental goal in diagnostic stroke evaluation is to identify the underlying etiology. We sought to determine the yield of an emergency department-based diagnostic evaluation protocol for identifying the etiology of stroke. Methods: We determined etiologic stroke subtypes using the automated Causative Classification System (CCS, available at https://ccs.mgh.harvard.edu) in 2422 consecutive patients with ischemic stroke at admission and discharge. Admission assessment was based on information from clinical evaluation, ECG, brain imaging (CT or MRI), and vascular imaging (CTA/MRA). Discharge CCS was performed blinded to the admission CCS subtype using information from additional tests such as echocardiography, cardiac monitoring, and special blood and CSF tests. Results: Table 1 shows the distribution of CCS subtypes. Overall, admission and discharge CCS subtypes were different in 29% of the patients. The size of “undetermined” category decreased from 37% at admission to 12% at discharge. The shift from “undetermined” to a known etiology was primarily due to detection of cardiac sources with low or uncertain risk of stroke (94%). The yield of investigations performed after admission in identifying a major known subtype was only 4.1% (p=0.008). Conclusions: A careful clinical evaluation and first-line diagnostic testing including brain and vascular imaging in the emergency department identify > 90% of those with a major stroke etiology. The low yield of additional testing suggests a need for developing cost-effective evaluation strategies in suspected patients.