Objective: To determine if any clinical variables allow early discrimination between stroke and other conditions presenting with a strokelike picture.
Background: New therapeutic modalities for the treatment of acute ischemic stroke are under active investigation. Many of these treatments have potential adverse effects. It is well known that noncerebrovascular conditions can present with a clinical picture mimicking stroke, so early accurate differentiation of such "mimics" from true stroke is essential.
Methods: Consecutive patients who presented to the emergency department with an initial diagnosis of stroke between January 1990 and January 1992 were evaluated. Chart review allowed these patients to be classified into two final diagnostic groups: stroke mimic and true stroke. Logistic regression was used to estimate the effects of predictor variables measured at initial evaluation on the final diagnosis.
Results: There were 411 patients initially diagnosed as having stroke. Of these, 78 patients (19%) were eventually found to have mimics, the majority comprising postictal states, systemic infections, tumors, and toxic-metabolic disturbances. Univariate analysis showed that decreased level of consciousness and normal eye movements increased the odds of mimic, while abnormal visual fields, diastolic blood pressure greater than 90 mm Hg, atrial fibrillation on electrocardiogram, and history of angina decreased the odds of mimic. Multivariate analysis showed that decreased consciousness increased, while history of angina decreased, the odds of mimic. Sensitivity of this model for predicting mimics was only 21% while specificity was 96%.
Conclusion: For the neurologist faced with an immediate decision as to whether to randomize a patient with probable stroke to an acute treatment protocol, these findings have some usefulness but mandate further research into conditions that mimic stroke in the emergency department.