Letters

Migraine and stroke in young women

Authors' results suggest that all types of migraine are contraindications to oral contraceptives

Prospective study is needed to determine what clinical practice in migraine should be

Authors' reply

Authors' results suggest that all types of migraine are contraindications to oral contraceptives

EDITORChang et al report that the odds ratios for ischaemic stroke in young women with classical migraine (with aura) and simple migraine (without aura) were similar.¹ These findings are at odds with earlier studies, which reported increased risk primarily in migraine with aura.² We note that the 95% confidence interval for ischaemic stroke in migraine without aura included unity (odds ratio 2.97 (95% confidence interval 0.66 to 13.5), unlike that for migraine with aura (3.81 (1.26 to 11.5). Therefore a lesser risk in migraine without aura is not excluded.

Another possible reason for the apparent discrepancy is that the authors failed to identify true aura. They characterised aura by at least one of the following symptoms just before or during the headache: visual disturbances or abnormalities of speech, skin sensation, or muscle power. Although their questionnaire was based on diagnostic criteria proposed by the International Headache Society, their questions were too non-specific.

Crucial characteristics of auras are symptoms and their duration and timing in relation to headache. Most (95%) are visual, typically starting as a flickering, uncoloured zigzag line in the centre of the visual field and gradually progressing laterally to the periphery of one hemifield, usually leaving a scotoma.³ Sensory or motor symptoms, if they occur, are usually also unilateral and rarely without associated visual symptoms.³ Auras typically last under one hour, resolving before the onset of headache. In contrast, the more general prodromal symptoms can precede the headache for several hours. Aura can be identified by asking "Have you ever had visual disturbances lasting 5-60 minutes followed by headache?" ⁴

Why is it important to distinguish between the different types of migraine? On the basis of previous evidence the Faculty of Family Planning and Reproductive Health Care recommends that migraine with focal aura absolutely contraindicates combined oral contraceptives.⁵ On the basis of Chang et al's results, should the contraindication extend to all women with migraine, at least if they smoke or are hypertensive? Migraine is common; hence numerous women could be deprived of a useful contraceptive.

Further studies are necessary before we accept no difference in risk between the two main varieties of migraine. We can reassure most young women with migraine of any type that the absolute risk of ischaemic stroke is minimal. It should still be possible, however, to identify and counsel the minority at specific risk so that they may choose an appropriate contraceptive and avoid other risk factors.

E Anne MacGregor, Senior registrar. 
City of London Migraine Clinic, London EC1M 6DX eamacg{at}aol.com

John Guillebaud, Professor of family planning and reproductive health. 
Department of Gynaecology, University College London, London WC1E 6JF

Prospective study is needed to determine what clinical practice in migraine should be

EDITORChang et al report that migraine with and without aura is an important risk factor for stroke in women aged 20-44.¹ The data should not be taken at face value.

Evaluating the prevalence of migraine in cohorts with stroke presents fundamental problems. For instance, people who have had a stroke know that they have a brain disorder and may therefore have inquired more about their family history of migraine than control patients with no brain disorder. Previous ischaemic episodes may have elicited headaches which were misinterpreted as migraine. Furthermore, individuals with stroke may be more inclined to report their previous headaches and migraine.

Some data reported in the paper do not fit with generally established facts. For example, the prevalence of migraine in the control subjects is much lower than that described for the normal population of female subjects in this age group, ^{2 3} whereas the prevalence of migraine in the stroke group roughly corresponds to the accepted norm. ^{2 3} Likewise, the data on family history are not consistent with known epidemiological facts.³

The terminology used in the paper is confusing because it is a mixture of the internationally accepted terminology (migraine with aura and migraine without aura) and other terminology (simple and classical migraine). In the control group two and a half times more patients seem to have been diagnosed as having classical migraine than as having simple migraine. This is the opposite of the prevalence of these disorders observed in several epidemiological studies. ^{2 3}

The data on headaches within the three days before stroke must be questioned. Many patients with ischaemic events have so called sentinel headache without having migraine.⁴ One must doubt that the precision of the clinical interview was sufficient to distinguish clearly such sentinel headaches from migraine attacks when the interview, as discussed above, was unable to distinguish migraine with aura from migraine without aura.

Chang et al conclude that their paper has some potentially serious consequences for people with migraine. Should we now tell these patients that they have a higher risk of stroke? Can they be insured on normal terms? And should they have prophylaxis against stroke?

I believe that clinical practice should not yet change. An increased risk of stroke in patients who have migraine can be definitively established only by a prospective, longitudinal study of a large cohort of patients with migraine and a matched control group of equal size.

Jes Olesen, Professor. 
Department of Neurology, Glostrup Hospital, University of Copenhagen, DK-2600 Glostrup, Denmark

Authors' reply

EDITORRisk estimates for ischaemic stroke associated with migraine with aura (classical) and without aura (simple) were not significantly different in our study. Odds ratios were higher and significantly increased only for classical migraine, and hence our findings are consistent with those of other studies. ^{1 2}

We recognise that the proportion of patients with classical migraine was high in our study. Our rate was based on patients having had at least one classical attack at any time. The basis for classifying an individual's migraine as classical or simple by the frequency of each type is not evident in similar studies.

Questions to identify common visual phenomena of aura just before or during the headache asked about coloured spots or zigzag lines and about loss or blurring of vision. These questions, and an inquiry about whether the women found bright lights unpleasant, discriminate between aura and migraine. Visual abnormalities are not usual in the non-specific prodrome before migraine; these questions are an unlikely cause of misclassification.

We emphasise that most strokes in young women are haemorrhagic and that the absolute risk of a stroke in women with migraine, with or without low dose oral contraceptives, is low. The addition of smoking or high blood pressure, or both, however, increases risk to worrisome levels. Not wishing to "deprive" women of a useful contraceptive, we re-emphasise that those with migraine should be advised strongly not to smoke and that their blood pressure should be monitored. Our study and other studies ^{1 2} have included too few patients to compare the risk of using low dose oral contraceptives in women with simple or classical migraine; hence we could not evaluate current recommendations for the use of oral contraceptives in such women.³

Olesen's concerns that sentinel headaches associated with non-migrainous ischaemic stroke may have been confused with those of true migrainous stroke were addressed in table 5 of our paper. This showed that headache occurred within three days of the onset of stroke in 70% and 26% of migrainous and non-migrainous women respectively. The occurrence of headache as a prelude to stroke was sought in a separate portion of our questionnaire from that used to determine history of simple or classical migraine.

We agree with Olesen that a cohort study could produce more robust data, but it would have to include thousands of patients followed up for many years. Until cohort data are published, case-control studies such as ours and others ^{1 2} offer the best available assessment of the risk of stroke in women with migraine.

M Donaghy, Reader. 
Department of Clinical Neurology, Radcliffe Infirmary, Oxford OX2 6HE

N R Poulter, Professor. 
C L Chang, Research fellow. 
Cardiovascular Studies Unit, Department of Clinical Pharmacology, Imperial College School of Medicine, London W2 1PG