Abstract
Background:
Migraine, seizure and toxic metabolic encephalopathy are common stroke mimics (SM) during code stroke (CS) evaluation and up to 30% of patients treated with IV tPA are found to have SM. An ideal way to differentiate SM from acute ischemic stroke (AIS) is through history and clinical examination in conjunction with MRI brain. Quick identification of a set of clinical characteristics predictive of SM may be helpful to prevent unnecessary IV thrombolysis. This study aimed to develop and evaluate effectiveness of a predictive scoring system to differentiate AIS from SM.
Methods:
The study population included 508 patients who received IV tPA from Jan 2013-Dec 2016 at a large comprehensive stroke center. Sixty clinical characteristics were reviewed and 27 were chosen for prediction analysis based on univariate ANOVA F-test. A prediction model is proposed by multivariate logistic regression with step wise selection of variables with the variable entry and stay threshold of p value 0.2 and 0.05 respectively to develop a predictive score.
Results:
Our scoring system consists of 7 clinical characteristics which are easily assessable during “CS” with yes and no answers. Each clinical variable has score weightage based on strength of statistical association with final diagnosis AIS Vs SM. Score is composed of : seizure at onset (score yes: -1.8, no:0), headache (yes: -2.3, no:0), history of migraine (yes: -1.7, no:0) , encephalopathy (yes: -1.2, no:0), atrial-fibrillation (yes: 1.3, no:0), NIHSS (0.1 per unit) and localizable symptoms (yes : 3.3, no: 0). Area under receiver operative characteristic (ROC) curve was 0.91 and a score >3.4 identified AIS with a sensitivity of 88.3% and specificity of 82.1%.
Conclusion:
Our study found that clinical characteristics of acute presentation were more helpful than conventional vascular risk factors history to differentiate AIS from SM. Our predictive score is reliable to differentiate AIS from SM and is intended to help clinicians identify SM using quick and simple clinical characteristics. Prospective validation of the score is needed.