MA is a well-known stroke simulator. Our study showed that one out of every 100 code stroke activations evaluated by a trained vascular neurologist was a MA mistaken for a stroke and was treated with fibrinolysis in the same rate. Age, sex, NIHSS, and fibrinogen levels were the most differential characteristics between MA and ischemic strokes but a diagnostic tool using these variables was not useful for clinical practice. Moreover, after reviewing MA cases we did not find a clinical or radiological pattern that could be tested to improve prediction.
Door-to-needle time has progressively diminished over the last years, provoking a faster decision-making, and consequently a higher prevalence of stroke mimics (including MA) treated with fibrinolysis [7, 10–12]. For this reason, diagnostic predictive models to discriminate between these two entities would allow a quick assessment at the emergency department, reducing misdiagnosis and, consequently, avoiding non-indicated treatments.
We have observed a prevalence of MA that reaches 1.1% of all code strokes after the first assessment by a vascular neurologist. These numbers are almost identical than the previously reported . However, the real prevalence of MA in the code stroke in our population is probably higher since the proportion of stroke mimics activated by the emergency services is around 20% and drops to 7% after the evaluation by a vascular neurologist .
To our knowledge, this is the first study comparing initial characteristics between MA and ischemic stroke codes. Almost 30% of MA were treated with rTPA in a very similar rate than strokes, pointing out the difficulty of distinguishing both diseases at the first evaluation. Younger age and female sex were the most predictive variables associated with MA. Other variables that showed an acceptable discriminatory capacity were NIHSS and fibrinogen levels, which were lower in MA. Although multivariate model with cut-off points showed a good calibration and discrimination capacity, the probability to predict MA was very low, probably due to the low number of MA cases in the cohort. Given the best-case scenario (woman, age ≤ 65, NIHSS ≤ 6, fibrinogen ≤ 400mg/dL), the maximum probability of being a migraine was 25% and therefore the model could not be useful in the clinical practice to rule out ischemic stroke and avoid unnecessary treatments.
Although age is a crucial factor and MA prevalence is higher between 15–49 years [13, 14], half of MA cases in our study were 50 to 75 years old, denoting that MA can happen at any age . In addition, stroke in young people is not infrequent and MA patients carry a higher risk of ischemic stroke . Therefore, although age is very different between the two groups it cannot help to rule out any differential diagnosis. Female sex was two times more frequent in MA as is the prevalence of this disease in the general population .
Initial severity measured by NIHSS was lower in MA and accordingly code stroke was more frequently activated in primary care centers than in the street or home where severe cases are usually evaluated. We did not find significant differences in the vital signs or blood test findings at arrival except a lower level of fibrinogen in the MA compared with ischemic strokes. This result agrees with previous studies, which found that high fibrinogen levels were associated with higher risk of cardiovascular diseases (including stroke)  and a poorer functional outcome from acute ischemic stroke . Further studies would be required to assess the utility of fibrinogen as a biomarker to distinguish MA from stroke. A recent study has shown that CGRP salivary levels are increased in migraine patients compared to controls, especially in ictal phase . Future studies could analyze the utility of this or other migraine biomarkers to rule out stroke mimics in code stroke.
Finally, we described the main clinical and radiological findings of MA in order to generate new hypothesis for future studies. However, we didn’t find any characteristic that could be exclusive of MA. Only half of the patients fulfilled IHS criteria previous to admission. Like in previous studies [5, 20], we found a higher prevalence of sensory symptoms, mainly negative like numbness, making it hard to distinguish from a thalamic stroke. Although visual aura is the most frequent in MA , it is not the most common in cases that raise stroke suspicion. Therefore, although some presentations can be more suggestive of a migraine etiology (positive sensory/visual symptoms, progressive onset, presence of a headache), atypical presentations of MA are frequent and patients and professionals should be aware. In the neuroimaging study two patients had abnormal (increased or decreased) perfusion imaging in the initial CT study performed in the first 60 minutes after symptom onset. Previous case reports and small studies have shown hypoperfusion during the first 120 minutes [21, 22] and hyperperfusion even at 3–6 days [23–26] with high variability between studies. A recent study found 3 patients (12%) with perfusion-CT alterations among 25 cases of MA attended after code stroke activation (two of them with an increased mean transit time (MTT), and one with increased MTT and reduced cerebral blood flow) . Therefore, CT-perfusion seems not useful to rule out ischemic stroke except in a small proportion of patients. MRI with diffusion-weighted imaging (DWI) would be the ideal neuroimaging study to definitely rule out an ischemic stroke but it is not widely used in the code stroke because of time and price costs [27, 28].
Our study has some limitations. The number of MA was low and this could have prevented us from finding other significant associations and obtaining a good model calibration. The proportion of MA was low since code stroke was evaluated by a vascular neurologist. Nevertheless, this is the biggest study of MA in this type of setting. Blood test data were missing in a relative high proportion of patients probably because some test could not be completed to avoid time delays. However, there were no differences between patients with and without missing data. And finally, only a third of MA were studied with CT perfusion and this may have not allowed us to find more radiological findings.