Article Text

Download PDFPDF

Cohort study
Migraine with aura – indication of an increased risk of hemorrhagic stroke in women (Women's Health Study) and increased risk of overall death in an Icelandic population (Reykjavik Study)
  1. Christina Sjöstrand
  1. Karolinska University Hospital Huddinge and Karolinska Institute
  1. Correspondence to Christina Sjöstrand
    Department of Neurology, Karolinska University Hospital Huddinge, S-14186 Stockholm, Sweden; christina.sjostrand{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Commentary on:




Migraine is a common chronic intermittent neurovascular headache disorder. The prevalence is about the same in different countries and affects about 10–20% of the adult population.1 Women are affected more often than men. Most patients experience their first attacks before the age of 40. Migraine is characterised by a unilateral pulsating pain, aggravated by physical activity, and associated symptoms are nausea, photo- and phonophobia. Up to 30% of patients experience an aura with transient neurological symptoms (eg, scintillary scotoma, speech disturbance or sensory symptoms). About 5% experience a migraine aura without headache. Clinically, migraine forms are defined as migraine with aura (MA) and migraine without aura (MO). Diagnostic criteria for migraine are defined in the International Classification of Headache (ICH) disorders; the first version was edited in 19882 and the revised second edition is from 2004.3

Several studies have shown that patients with migraine have an increased risk for ischemic stroke. Earlier studies indicated an increased risk of ischemic stroke in both MA and MO, confirmed in a systematic review and meta-analysis of observational studies in 2004.4 However, during the last few years, new studies in large cohorts have re-investigated the association between migraine and vascular events, and a new systematic review and meta-analysis, conducted in 2009, established an association between MA and ischemic stroke.5 The increased risk for ischemic stroke seems to be highest among smoking women, <45 years of age, using oral contraceptives.

Two new prospective large cohort studies have recently been performed on migraine. First, an American investigation looking for association between hemorrhagic stroke and migraine using the study population in the Womens Health Study (WHS),6 and second, an Icelandic study on whether migraine in mid-life is associated with mortality from cerebrovascular disease, other causes and all causes. The study population is from the formerly described Reykjavik study,7 a random sample of prospectively followed men and women.


The Harvard group performed a prospective cohort study as part of the formerly described WHS. Twenty-seven thousand eight hundred and sixty women aged >45 years, healthy at baseline, with self-reported information on migraine, aura status and blood lipid values were included in the study. The authors have, in other studies on the same cohort, found that 83% of the persons who reported active migraine fulfilled all but one of the migraine diagnostic criteria from the ICH 1988 edition, and in a subsample of 1675 women, 88% fulfilled criteria for the ICH 2004 edition. Stroke events were collected by vascular neurologists. Clinical information and neuroradiology information were collected from the participants, and death information was taken from hospital records or from next of kin. Bleedings were divided into intracerebral bleedings (ICB) and subarachnoid haemorrages (SAH). No secondary hemorrhagic ischemic strokes were included. The main outcome measure was time to first stroke and subtypes of hemorrhagic stroke. Analysis of covariance was made to adjust continuous measures and direct standardisation to adjust categorical variables' incidence rate for hemorrhagic stroke for age. Cox proportional HR was used to analyse association of MO, MA and hemorrhagic stroke and subtypes. Multivariate adjustment for risk factors, propensity score for potential confounders and logistic regression analysis including age (before), risk factors, menopausal status, hormones, income and more were performed.

The study population in the Icelandic study consisted of a random sample of 18 725 men and women born between 1907 and 1935 from the Reykjavik area. The first examination was performed between 1967 and 1991. Questionnaires and clinical measures were obtained in mid-life, mean age 53 years at study entry. The study information was obtained by nurses, and participants were asked for headache symptoms. Headache was classified and an approximation of the ICH 2004 edition3 was used, the diagnostic criteria was formalised after the Reykjavik Study data were collected (as for the WHS study). Aura distinction was categorised as visual aura or sensory aura or both. Statistics related to the cause of death were obtained from Statistics Iceland. The main end points were deaths from cardiovascular disease, non-cardiovascular disease and all causes. Cox proportional hazard to estimate the relative risk after adjusting for baseline risk factors and demographics was used. An estimation of life expectancy at age 50 by headache status from a Cox model adjusted for age was also performed.


In the Harvard study, participants were divided into no history of migraine, active MA, active MO or previous migraine. Of the 27 860 participating women, 5130 reported any history of migraine, 3612 reported active migraine and of these 1435 were classified as having MA. In the study, 85 confirmed hemorrhagic strokes were found during a mean follow-up of 13.6 years; 44 ICB and 36 SAH and 5 bleedings with unknown aetiology. After adjustment for age, there was a change between women without migraine and women with MA; HR 2.31 and p=0.019. The multivariate adjusted model showed a HR of 2.34, and the propensity score adjusted (risk factors and other probable confounders) HR was 2.25. The increased risk of haemorragic stroke for women with MA was more apparent towards later years of follow-up. Stronger effect estimates were found for women >55 years of age, no history of hypertension, non-active smokers, cholesterol <6.2 mmol, use of postmenopausal hormones, Framingham risk score <5%, randomisation to vitamin E and ASAplacebo. In all, there were a relatively low number of haemorragic strokes and low attributable risk, and the authors recommend taking the results with caution.

The Icelandic study included 9044 men and 9681 women. The mean follow-up was 25.9 years, the average age on follow-up was 75, with 470 990 person-years and 10 358 deaths; 4323 from cardiovascular disease and 6035 from other causes. Six per cent of the men and 15% of women were classified as having migraine, and the proportion of MA was higher than the proportion of MO in both men and women. Participants were divided into no headache and four categories of headache, once or more a month, non-migraine headache, MO and MA. People with migraine were at significantly increased risk of mortality for all causes (adjusted for sex and multivariables), HR 1.21 and mortality from cardiovascular disease with a HR of 1.27 compared with people with no headache, whereas those with MO and non-migraine headache were not. People with MA were at increased risk of mortality from coronary heart disease (HR 1.28) and stroke (HR 1.40). The risk was similar with and without adjustment for cardiovascular risk factors. Women and men were at similarly increased risk of all-cause mortality. Anyhow, women with MA also had an increased risk of mortality from non-cardiovascular disease. The authors stated that MA is an independent risk factor for cardiovascular and all-cause mortality in men and women, but still weaker than conventional risk factors such as smoking, diabetes and hypertension.


Migraine seems to be related to stroke in several ways. It might be a direct effect of the aura and migraineurs also have a higher prevalence of cardiovascular risk factors. There is an increased prevalence of persistent foramen ovale in patients with MA. Furthermore, there is also an increased risk for cervical artery dissection in patients with MA.7 Several studies have also indicated endothelial dysfunction in migraine.

In a subanalysis of a large meta-analysis on migraine and cardiovascular disease by Schürks and colleagues,5no increased risk for hemorrhagic stroke based on three studies was shown. A subanalysis of five studies investigating the association between any migraine and death due to cardiovascular disease did not show an overall association, but two studies showing a protective effect were included. One study with aura specification found an increased risk for death due to cardiovascular disease in MA (RR 2.33).8

Kurth and colleagues have in this new study from WHS followed a large cohort for a mean period of 13.6 years and found an increased risk of hemorrhagic stroke in women with active MA. Women who reported active MO had no increased risk. The strength of this study includes the large homogeneous cohort with quite a long follow-up and the fact that the hemorrhagic strokes were confirmed by stroke physicians. One of the limitations of the study is the self-reported migraine status by participants. The method of asking whether a person has experienced migraine, and only then (following a positive response) about the details of the headache, might give an underestimation of migraine. Whether the participants experienced aura or any other indication that a migraine is coming was the only aura question. Premonitory symptoms such as fatigue, phonophobia and yawning might be misleadingly reported as aura symptoms. An even longer period of follow-up would have added extra information. Further, the study includes women only. Nevertheless, this study adds further important information on the potential relationship between MA and hemorrhagic stroke.

Gudmundsson and colleagues have, in the large cohort from Iceland, shown that patients with MA were at increased risk of all-cause mortality and mortality from cardiovascular disease compared with people with no headache. The strength of the study includes the large random cohort with both men and women with high age at baseline and followed up for many years, including many deaths. Other strengths are information on risk factors for cerebrovascular disease and high ascertainment of mortality due to statistic information based on diagnostic criteria. One limitation of the study is the definition of headache and its characteristics. The prevalence of aura is very high in this population, also mentioned by the authors, and it seems unlikely that the Icelandic population should have such a high proportion of MA. It is thus probable that this might affect the relation between MA and mortality. Further, non-migraine headache is defined as headache with or without one symptom of migraine once or more a month. It is likely that some MO patients would be found in this group. Anyhow, this is important new information on all-cause mortality in a large cohort with headache information.

The connection between migraine and cardiovascular incidents and deaths is likely to be complex. Well-defined cohorts are needed for both epidemiological and experimental studies. It is of greatest importance to identify individuals at highest risk in order to improve their care and treatment.


View Abstract


  • Competing interests CS has received unrestricted grants from Glaxo Wellcome and Astra Zeneca and is the primary investigator in Sweden for the Gore REDUCE Clinical Study. She also received payment from Sanofi Aventis and MSD for educational programmes.