Telestroke Assessment With Perfusion CT Improves the Diagnostic Accuracy of Stroke vs. Mimic

Abstract

Background and Purpose: CT perfusion (CTP) has been implemented widely in regional areas of Australia for telestroke assessment. The aim of this study was to determine if, as part of telestroke assessment, CTP provided added benefit to clinical features in distinguishing between strokes and mimic and between transient ischaemic attack (TIA) and mimic. Methods: We retrospectively analysed 1,513 consecutively recruited patients referred to the Northern New South Wales Telestroke service, where CTP is performed as a part of telestroke assessment. Patients were classified based on the final diagnosis of stroke, TIA, or mimic. Multivariate regression models were used to determine factors that could be used to differentiate between stroke and mimic and between TIA and mimic. Results: There were 693 strokes, 97 TIA, and 259 mimics included in the multivariate regression models. For the stroke vs. mimic model using symptoms only, the area under the curve (AUC) on the receiver operator curve (ROC) was 0.71 (95% CI 0.67–0.75). For the stroke vs. mimic model using the absence of ischaemic lesion on CTP in addition to clinical features, the AUC was 0.90 (95% CI 0.88–0.92). The multivariate regression model for predicting mimic from TIA using symptoms produced an AUC of 0.71 (95% CI 0.65–0.76). The addition of absence of an ischaemic lesion on CTP to clinical features for the TIA vs. mimic model had an AUC of 0.78 (95% CI 0.73–0.83) Conclusions: In the telehealth setting, the absence of an ischaemic lesion on CTP adds to the diagnostic accuracy in distinguishing mimic from stroke, above that from clinical features.