Abstract TMP82: Rates and Predictors of Stroke Mimics Among In-Hospital Code Stroke Activations

Abstract

Introduction: In-hospital strokes have a different pattern of risk factors, etiologies, and mimics than strokes that are initially evaluated in the Emergency Department. In-hospital stroke patients often have multiple confounding comorbidities which may provide alternative explanations for suspected stroke symptoms. Objective: To characterize strokes vs. stroke mimics among in-hospital Code Stroke activations. Methods: We identified all adults with an in-hospital Code Stroke activation between September 2014 and February 2018 at a single academic medical center (660 beds) and abstracted the reason for Code Stroke activation, comorbidities, primary inpatient service, patient location, NIHSS, and final diagnosis from comprehensive electronic health records. Univariate and multivariate logistic regression were used to assess the association between clinical variables and stroke mimics. Results: Of 461 in-hospital Code Stroke activations, the final diagnoses were 110 (24%) ischemic stroke or TIA, 35 (7.5%) intracranial hemorrhage, and 316 (68.5%) stroke mimics. Of the 85 patients with a final diagnosis of ischemic stroke, 9 (8%) received tPA and 2 underwent embolectomy (1.8%). The most common reason for Code Stroke activation was altered mental status (48%). Altered mental status (OR 0.20, 95% CI 0.10-0.39) and previously-known CNS tumor (OR 0.18, CI 0.05-0.69) were least likely to be associated with a final diagnosis of stroke. The most common stroke mimics were toxic/metabolic encephalopathy, seizure, and delirium. Limb weakness (OR 6.93, CI 3.86-12.7), recent cardiac/vascular procedure (OR 5.03, CI 1.92-8.89) and NIHSS (OR 1.09 per point, CI 1.05-1.13) were predictive of stroke in multivariate analysis. Conclusions: A majority of in-hospital Code Stroke activations are for stroke mimics. Altered mental status was the most common reason for Code Stroke activation, yet was least likely to be associated with a final diagnosis of stroke. While rapid recognition of stroke symptoms and timely Code Stroke activations are critical to maximizing opportunities to apply acute interventions, variability in the reasons for Code Stroke activation particularly for undifferentiated altered mental status may be a target for future optimization.