Diagnostic Accuracy in Teleneurological Stroke Consultations

Abstract

Background: The accuracy of diagnosing acute cerebrovascular disease via a teleneurology service and the characteristics of misdiagnosed patients are insufficiently known. Methods: A random sample (n = 1500) of all teleneurological consultations conducted between July 2015 and December 2017 was screened. Teleneurological diagnosis and hospital discharge diagnosis were compared. Diagnoses were then grouped into two main categories: cerebrovascular disease (CVD) and noncerebrovascular disease. Test characteristics were calculated. Results: Out of 1078 consultations, 52% (n = 561) had a final diagnosis of CVD. Patients with CVD could be accurately identified via teleneurological consultation (sensitivity 95.2%, 95% CI 93.2–96.8), but we observed a tendency towards false-positive diagnosis (specificity 77.4%, 95% CI 73.6–80.8). Characteristics of patients with a false-negative CVD diagnosis were similar to those of patients with a true-positive diagnosis, but patients with a false-negative CVD diagnosis had ischemic heart disease less frequently. In retrospect, one patient would have been considered a candidate for intravenous thrombolysis (0.2%). Conclusions: Teleneurological consultations are accurate for identifying patients with CVD, and there is a very low rate of missed candidates for thrombolysis. Apart from a lower prevalence of ischemic heart disease, characteristics of “stroke chameleons” were similar to those of correctly identified CVD patients.