Patients were consecutively recruited from the stroke unit of city hospital “New Hospital” in Yekaterinburg, Russia. They had first-ever ischemic stroke and were included if they were 18 years and older, had a new infarction on magnetic resonance imaging (MRI) with diffusion-weighted imaging (DWI), agreed to be interviewed and followed up not less than three months after acute headache attributed to stroke. Exclusion criteria were a history of previous stroke/transient ischemic attacks, intracranial/intracerebral haemorrhage, unruptured cerebral artery aneurysm, brain tumour, any brain surgery, traumatic brain injury, multiple sclerosis, epilepsy, encephalitis, meningitis and other severe neurological or somatic diseases that can provoke secondary headache disorders. A patient should be able to give a clear description of previous and current headaches.

We recruited 550 patients with first-ever ischemic stroke out of 2995 patients with ischemic stroke who met the above-described inclusion and exclusion criteria and agreed to participate in the study. All data were collected prospectively. The present study is a continuation of three previous studies about headache in first-ever ischemic stroke which described the prevalence of sentinel headache, headache at onset of first-ever ischemic stroke, their diagnostic criteria and associated factors [2,3,4]. A full description of the inclusion procedure has been published in the previous articles. The present study focuses on patients with persistent headache after stroke, their diagnostic criteria and risk factors for these headaches.

The patients were assessed on the day of stroke and at last 3 months later, with regard to headache attributed to stroke. According to our previous study [3], these headaches included: 1) a new type of headache; 2) headache with altered characteristics. Both occur within 24 h of stroke onset.

Standardized semi-structured interview forms were used to evaluate headaches during face-to-face interviews at stroke onset in 550 patients with first-ever ischemic stroke and telephone interviews at 3 months in 529 follow-up patients after stroke. The first interview was done by two trained neurologists. A third neurologist interviewed patients by telephone at 3 months after stroke. He did not evaluate the patients previously and was blinded to the presence of headache at stroke onset. Therefore, he interviewed all patients who agreed to be follow-up including patients without acute headache attributed to stroke. Patients were asked whether they perceived any headache or not, what are characteristics of these headaches and whether the headache had changed. If the headache disappeared, the date of disappearance was recorded.

We evaluated the following characteristics of headache at onset of stroke and three months after that: 1) the number of attacks per week within 1st month and at 1st, 2nd and 3rd months after stroke; 2) intensity of headache (mild, moderate, severe) during 1st, 2nd and 3rd months after stroke; 3) quality (pulsating, pressing, other); 4) location and side of headache; 5) aggravation by routine physical activity; 6) accompanying symptoms (nausea, vomiting, photo- and phonophobia); 7) the presence of aura and its symptoms (visual, sensory, speech, motor, other); 8) use of drugs for pain relief, preventive treatment of headache and stroke prevention/other drugs, kind and name of a drug, frequency, effect.

Characteristics of headache at stroke onset were compared to headaches at three months. A new type of headache and headache with and without altered characteristics were analysed separately. Type of headache (migraine-like, tension-type-like, etc.) was recorded.

The prevalence of factors possibly associated with persistent headache after stroke was compared to the prevalence in patients without headaches. These factors included: mean age, sex, lack of sleep, a history of hypertension, diabetes, atrial fibrillation, angina, myocardial infarction, peripheral artery disease, large-artery atherosclerosis ( ≥ 50% stenosis or occlusion of arteries on the neck which were verified during triplex ultrasonography or CT-angiography), smoking, alcohol consumption, body mass index > 25, family history of stroke, low physical activity (less than 30 minutes of physical exercises 1 time per week), NIHSS score, type of stroke (TOAST classification), localization and size of the infarct.

Acute headache attributed to stroke was determined using our validated proposed diagnostic criteria [2]. Persistent headache after a first-ever ischemic stroke was defined according to ICHD-3 as an acute headache attributed to ischemic stroke that persisted for 3 months [1]. A new type of headache at stroke onset was defined as a headache which arose for the first time within 24 h after stroke onset. If patients had a pre-existing headache but the headache at stroke onset had changed in characteristics, these headaches were also attributed to stroke. If they had unaltered characteristics, they were not attributed to stroke. If a new type of headache or a headache with altered characteristics had a phenotype of migraine or tension-type headache they were named migraine-like and tension-type-like according to the general rules of ICHD-3.

Medication overuse headache was diagnosed based on the self-reported use of simple analgesics for pain relief as ≥ 15 days per month or triptans and/or combine analgesics ≥ 10 days/month for > 3 months [1].

Stroke location was defined as anterior (medium cerebral artery and anterior cerebral artery), posterior (posterior cerebral artery, cerebellum and brainstem), or subcortical (thalamus, internal capsule lesions).

Data were analyzed with Stata 14.0 software (StataCorp LP, College Station, TX, USA) and Microsoft Excel (2014). Pearson’s 2 test or Fisher’s exact test was conducted for comparison of categorical variables depending on group sizes. T-test and Wilcoxon rank-sum were applied for continuous variables. Odds ratio (OR) was used to estimate associated factors. Two-tailed probability (p) values < 0.05 were considered significant.

When quantitative indicators were evaluated for compliance with a normal distribution, we used the Shapiro–Wilk test (when the number studied was less than 50) or the Kolmogorov–Smirnov test (when the number of investigated was more than 50). In the Chi-square calculation, if the expected number in at least one cell was less than 10, we calculated the χ2 with Yates’s correction.

Multivariate logistic regression analysis was performed to identify independent factors. Each covariate was evaluated individually; those meeting the significance level of p < 0.05 and OR > 1 were then included in multivariate models to identify their independent contributions after adjusting for the presence of all other variables. These factors were analyzed in participants with a headache at stroke onset compared to participants without a headache. All analyses were performed by two statisticians (DVG, NVK).