W. Oliver Tobin^1,4, Joseph G. Hentz², Bentley J. Bobrow³ and Bart M. Demaerschalk⁴

¹Department of Neurology, Adelaide and Meath Hospital, Dublin, incorporating the National Children’s Hospital, Trinity College Dublin, Republic of Ireland. ²Department of Biostatistics, Mayo Clinic Arizona, U.S.A. ³Department of Emergency Medicine, Mayo Clinic Arizona, U.S.A. ⁴Department of Neurology, Mayo Clinic Arizona, U.S.A.

Abstract

Background and Purpose: Previous studies have shown a stroke mimic rate of 9%–31%. We aimed to establish the proportion of stroke mimics amongst suspected acute strokes, to clarify the aetiology of stroke mimic and to develop a prediction model to identify stroke mimics.

Methods: This was a retrospective cohort observational study. Consecutive “stroke alert” patients were identified over nine months in a primary stroke centre. 31 variables were collected. Final diagnosis was defined as “stroke” or “stroke mimic”. Multivariable regression analysis was used to define clinical predictors of stroke mimic.

Results: 206 patients were reviewed. 22% were classified as stroke mimics. Multivariable scoring did not help in identification of stroke mimics. 99.5% of patients had a neurological diagnosis at final diagnosis.

Discussion: 22% of patients with suspected acute stroke had a stroke mimic. The aetiology of stroke mimics was varied, with seizure, encephalopathy, syncope and migraine being commonest. Multivariable scoring for identification of stroke mimics is not feasible. 99.5% of patients had a neurological diagnosis. This strengthens the case for the involvement of stroke neurologists/stroke physicians in acute stroke care.