Background and purpose
	Stroke-mimics (SM) are usually difficult to diagnose in emergency setting (up to 20% of potential strokes). Thus, some SM are misdiagnosed and treated with intravenous thrombolysis. Our aim was to evaluate TM-score (TeleStroke mimic score) and clinical features of NIHSS to improve discrimination of SM from cerebrovascular diseases (CVD).
Material and methods
	We included 165 patients (82 SM, 83 haemorrhagic/ischaemic CVD, paired by initial NIHSS-score from a sample of 453 code-stroke) attended in a tertiary referral hospital. Data was recorded prospectively from May/2014 to February/2017. TM-score and NIHSS items were evaluated and logistic regression was made to find the best predictors of SM.
Results
	CVD were older and had higher rates of vascular risk factors. Mean NIHSS in SM was 4.3. Thrombolysis was administered in 20.5% of CVD vs 6% SM. Forty-two percent of SM had <15points in TM-score, while 78% of CVD had >15. ROC curve of TM-score was 0.66 (0.58-0.75, p<0.001). Between NIHSS items, disturbance in level of consciousness (LOC) and LOC questions were significantly associated to SM (OR 6.39 [1.1-36.7], OR 5.63[2.0-15.3]). Contrarily, visual loss and facial palsy were negatively associated to SM (OR 0.31[0.1-0.94], OR 0.28[0.1-0.6]). With this NIHSS model, ROC curve was 0.73 (0.66-0.81, p<0.001). Motor symptoms and aphasia did not improve discrimination of SM. 
Conclusion
	TM-score helps to identify stroke mimics, but clinical findings in NIHSS items such as decrease of LOC and alteration in orientation questions can also be used in SM diagnosis, improving selection for acute stroke treatment.

ESOC-2019