Performance of bedside stroke recognition tools in discriminating childhood stroke from mimics.

OBJECTIVE: To assess the utility of the Cincinnati Prehospital Stroke Scale (CPSS) and Recognition of Stroke in the Emergency Room (ROSIER) tools in children presenting to the emergency department (ED) with brain attack symptoms.

METHODS: The ROSIER and CPSS tools were retrospectively applied to 101 children with stroke, presenting from 2003 to 2010, and prospectively to 279 children with mimics, presenting from 2009 to 2010. Positive CPSS was defined as ≥1 positive sign (face/asymmetrical arm weakness, speech disturbance). Positive ROSIER was defined as a score of ≥1. Accuracy and interrater agreement between the tools and patients' true status were assessed for combined stroke types and arterial stroke (AIS) and hemorrhagic stroke (HS) subtypes vs mimics.

RESULTS: Stroke subtypes included AIS (55), HS (34), TIA (10), and sinovenous thrombosis (2). Mimic diagnoses included migraine (84), first seizure (45), Bell palsy (29), and conversion disorders (18). Both tools had poor reliability and accuracy for combined strokes vs mimics (CPSS κ 0.36, receiver operator characteristic curve [ROC] 0.66; ROSIER κ 0.32, ROC 0.60) and for AIS vs mimics (CPSS κ 0.37, ROC 0.79; ROSIER κ 0.30, ROC 0.77). Both tools performed inadequately for HS vs mimics (CPSS κ -0.03, ROC 0.51; ROSIER κ -0.02, ROC 0.52).

CONCLUSIONS: Adult stroke recognition tools perform poorly in children and require modification to be useful for pediatric stroke identification.

CLASSIFICATION OF EVIDENCE: This study provides Class II evidence that, for children presenting to the ED with brain attack symptoms, the CPSS and ROSIER tools do not accurately distinguish strokes from mimics.