Abstract WP99: Validating a Prediction Rule for Identifying Stroke Mimics Evaluated Over Telestroke in a Population of Face-to-Face Acute Stroke Evaluations

Abstract

Introduction: We recently published a risk prediction score for telestroke (TS) encounters to differentiate stroke mimics (SM) from ischemic cerebrovascular disease, derived and validated at multiple telestroke sites across the US and Europe. In this study, we assessed if it could be applied to a comprehensive stroke center, non-telemedicine, stroke code registry. Method: In this IRB approved analysis, we performed ROC curve analysis on retrospectively assessed prospectively collected data from acute stroke code registry database for patients from 10/2004 to 7/2018. We tested only characteristics previously shown to be associated with SM. The TM score = (Age multiplied by 0.2) + 6 (if Hx of atrial fib) + 3 (if Hx of HTN) + 9 (if facial weakness) + 5 (if NIHSS > 14) - 6 (if Hx of seizure)). Result: Based on final diagnosis, SM accounted for 1,978/4,185 (47.2%) of patients. Age, NIHSS > 14, facial weakness, atrial fibrillation, hypertension, and seizure were all significantly associated with diagnosis of SM. The TM Score performed well on ROC curve analysis with AUC of 0.704 (p<0.001). Conclusion: This non-TS, face to face, validation cohort performed similarly to our validations at prior centers (0.70 vs 0.72). This finding not only continues to validate the TM score as an effective tool in assessing the ability to predict SMs, but also broadens its potential for use in non-TS populations. It is promising that it performed well despite a substantially higher proportion of SMs than in TS encounters where some may have already been screened out. Tools like the TM score may help highlight key clinical differences between mimics and stroke patients during complex, time-critical acute evaluations.