Background and Aims:
Stroke mimics (SM) represent a challenge in the evaluation of acute stroke by Telestroke System (TS), supposing up to one third of all cases. NIHSS scale might not be helpful enough to detect SM by TS. Therefore, it would be useful to have predictive tools to make a correct diagnosis.

Our objective is to evaluate the prevalence and variability of SM in our population and to search for a predictive model of SM when evaluating a patient by TS.

Methods:
We included consecutive patients excluding TIAs. We analyzed demographic variables, stroke-associated variables and diagnosis. To create our scale, we selected the best model from all possible equations, and then we established the SM risk prediction using logistic regression, we calculated the AUC and performed external and internal validation processes (cross-validation).

Results:
215 patients were included. 52 patients (24.19%) were SM, receiving the following diagnosis: Functional: 9(17.31%), Seizure: 8(15.38%), Migraine: 7(13.47%), Non-specific speech disorder: 4(7.69%), Dizziness/Syncope: 2(3.84%), Peripheral facial palsy: 2(3.84%), Encephalopathy: 2(3.84%), Space-occupying lesion: 2(3.84%), Unknown: 16(30.76%). The next variables were included in our scale: age(x 0.1) – epilepsy(3) + NIHSS(x 0.3) + facial palsy(3) + SBP(0.3 x 10 mmHg). Our model AUC was 0.87(95% CI 0.81-0.93). In the internal validation process the pseudo-R2 was 0.337 and in the external validation the AUC was 0.89(95% CI 0.79-0.98). The total score of our scale was associated with a diagnosis of stroke with an OR 1.58(95% CI 1.37-1.82).

Conclusions:
Our scale might be a good predictive tool for SM in the diagnostic evaluation by TS.

DEVELOPMENT OF A STROKE MIMIC PREDICTIVE SCALE IN THE EVALUATION BY TELESTROKE SYSTEM