Headache at the Chronic Stage of Ischemic Stroke

Mariana Carvalho Dias, MD; Teresa Martins, MD; Goncalo Basilio, MD; Lia Lucas Neto, MD, PhD; Lara Caeiro, Psy, PhD; Jose M. Ferro, MD, PhD; Ana Verdelho, MD, PhD


Headache. 2020;60(3):607-614. 

In This Article



The study was designed as a prospective observational cohort study including acute ischemic stroke patients admitted to a Stroke Unit of a University Hospital over a predefined period of 17 months. The study was performed over 2 time periods (March-July 2014 and September 2014-September 2015) in response to the investigator team availability, in order to have the same team interviewing all patients. No statistical power calculation was conducted prior to the study. The sample size was based on our previous experience with this design.[1,2]

Inclusion criteria were as follows: (1) ischemic stroke admitted in the first week after stroke onset, (2) ability to understand and cooperate, and (3) signed informed consent. Exclusion criteria were: any condition that preclude cooperation such as drowsiness (defined as National Health Institutes Stroke Scale – NIHSS item 1a ≥ 2),[17] moderate to severe aphasia (defined as NIHSS item1b or 1b = 2),[17] previous diagnosis of dementia according to Diagnostic and Statistical Manual of Mental Disorders (DSM-IV R) criteria,[18] delirium (according to DSM-IV R criteria),[18] diagnosis of cerebral infarct associated with cerebral venous thrombosis, subarachnoid hemorrhage and transient ischemic attack. Patients unlikely to attend follow-up interviews after discharge (eg those living outside our hospital referral area or expected to move out of the country) were also excluded.


All patients were interviewed using a predefined structured interview in one of the first 5 days of admission after stroke to collect demographic data, vascular risk factors (including previous vascular events, hypertension, diabetes, hypercholesterolemia and alcohol consumption),[19] symptoms and signs of the current stroke, previous pain complaints, and previous depression (defined as a past depressive episode requiring treatment or hospital admission or formal diagnosis by a clinician) Global cognitive status was accessed with the Mini-Mental State Examination (MMSE)[20,21] in order to ensure ability to cooperate. Severity of stroke was registered using the NIHSS.[17] For each patient functional status was registered using the modified Rankin Scale (mRS)[22] at Stroke Unit discharge. Chronic stage evaluations were conducted 12 months after stroke with personal in site interviews, using the same protocol as in acute stage, including severity (NIHSS) and functional status (mRS) evaluation. Depressive symptoms at the chronic stage were recorded using the 15-item Geriatric Depression Scale (GDS).[23,24] Current medication and complications during follow-up period were also registered. If it was impossible to perform a personal on-site interview at the chronic stage, the patient was contacted by telephone in order to complete the interview.

All patients had a CT scan/MRI performed at admission. Stroke location was based on CT and/or MRI, which were evaluated by two independent neuro-radiologists who reviewed all scans. 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). We used this grouping of stroke locations, because headache at the acute stage of stroke is reported to be more frequent after strokes of posterior location and infrequent in subcortical infarcts.[2,10] Results of ancillary procedures conducted to investigate stroke etiology were collected in standardized form by the team of neurologists working in the Stroke Unit, accordingly to the procedures of the Stroke Unit registry.

Headache Characteristics

Headache characteristics at the acute and at the chronic stage (12 months follow-up) were collected using a previously validated headache questionnaire[25,26] that enables the classification of headache following the ICHD-2 classification.[27,28] We did not consider the diagnostic category "persistent headache attributed to past ischemic stroke" because this study started before the ICHD update (ICHD-3), which included for the first time this type of headache. Pre-stroke headache history was recorded using the same questionnaire and was collected during the interview conducted during the admission due to acute stroke. Frequency of headache was registered and classified as frequent if the patient had more than 1 episode of headache per month, or sporadic if less than 1 episode of headache occurred per month in the previous year.[16,28] Headache onset was defined as sudden (maximal intensity from onset) or progressive. Relationship between headache onset and the onset of other neurological signs was defined as previous, concomitant, or following other neurological signs. Patients classified headache severity using a scale from 0–10, severity of headache was then categorized as mild (1–3/10), moderate (4–7/10), or severe (8–10/10). Use of analgesic therapy was also registered. Headache location was defined as unilateral or bilateral, anterior (defined as located in the orbicular, frontal, and temples regions), posterior (defined as occipital), in the vertex, diffuse or multiple sites. Other locations were also registered. Quality was defined as pressure, throbbing, stabbing, and other qualities (registered using the description of the patient). Pre-stroke headache, headache associated with the acute stage and headache at the chronic stage were classified according to the IHS criteria.[27,28] For each included patient headache classification was performed independently by 2 neurologists using the IHS criteria.[27,28] As far as the classification of headache and the questionnaire were used in the current study there are no relevant differences between the previous[27,28] and the current classification.[15] Discordant classifications were discussed to reach consensus. Classification of stroke associated headache and pre-stroke headache were also made independently, in order to avoid the influence of the diagnosis of stroke associated headache on pre-stroke headache classification and vice-versa.

All patients signed the informed consent at stroke onset, and the study was conducted in accordance with the Declaration of Helsinki. The prospective study was approved by the Centro Académico de Medicina de Lisboa Ethics Committee.

Statistical Analysis

We described the characteristics of headache at the chronic stage and compared them to pre-stroke headache and to acute headache characteristics, in order to evaluate modifications of headache overtime.

Statistical Analysis

Chi-square (Pearson's or Fisher's exact test) with continuity correction when necessary and difference between proportions with 95% confidence intervals (CI) were used to test the bivariate associations between headache (headache vs no headache) and demographic variables (sex), previous conditions (pre-stroke headache and depression history) and stroke characteristics (posterior location). Age, NIHSS, Mrs, and GDS were measured using median and interquartile range (IQR). We used Mann-Whitney U-test to compare medians of nondichotomous variables at the acute (for the total sample) and the chronic stages (for the sample evaluated at 12 months) after stroke (age, NIHSS, mRS, and 15-GDS score) between patients with and without headache. For multivariable analysis of the predictors of headache at the chronic stage after stroke we performed a logistic regression, entering variables with a P < .05 on bivariate analysis (headache previous to stroke and gender) and predictors previously described in the literature[1–11] (age, severity of stroke and location of stroke). The same analysis was repeated controlling additionally for functional status at chronic stage and depressive symptoms at the chronic stage. We calculated the specificity, sensitivity, positive predictive value, and negative predictive value of the model. The independent variables selected in the final model of the logistic regression were submitted to a collinearity analysis to access for a possible correlation. A P value of ≤ .05 was considered statistically significant. Two-tailed testing was used by convention. This was the primary analysis of these data. We used IBM SPSS Statistics 24 version.