Headache and the Risk of Stroke: A Prospective Observational Cohort Study Among 35 056 Finnish Men and Women

Background  Previous studies have shown an increased risk of stroke among patients with migraine. However, very few data are available on the possible association between chronic unspecified headache and the risk of stroke.

Methods  A prospective cohort study including 35 056 randomly selected Finnish men and women aged 25 to 64 years at baseline who participated in a cardiovascular risk factor survey in 1972, 1977, 1982, or 1987. Self-reported headache, smoking, diabetes, blood pressure, weight, height, serum cholesterol level, and oral contraceptive use were recorded at baseline. During the follow-up, 2167 incident stroke events were ascertained with computer-based record linkage.

Results  Women reported headache twice as often as men (16.7% vs 8.9%). Among men, the headache-associated hazard ratios (95% confidence intervals) for stroke were 4.08 (2.10-7.93), 1.86 (1.33-2.59), and 1.24 (1.05-1.47) during 1, 5, and a maximum of 23 years of follow-up, respectively. Adjustment for the other risk factors decreased the hazard ratios only slightly. Among women, there was also a direct but statistically nonsignificant association between headache and the risk of stroke.

Conclusions  Chronic headache is an independent predictor of stroke among men. Since the association between headache and the risk of stroke was particularly strong during a short follow-up, chronic headache may be a marker of the underlying disease process leading to acute stroke. The sex difference observed in this association may be due to a higher prevalence and a more heterogeneous etiology of headache in women compared with men.

SUBSTANTIAL EVIDENCE shows that migraine is a risk factor for stroke.^1-8 The association between migraine and the risk of stroke has been observed most often among premenopausal women and in some studies also among men. This association has been shown to be particularly strong among smoking women and among women using oral contraceptives.^5-7 Classic migraine with aura may be a more powerful predictor of stroke than simple migraine without aura.^3,5,9

One of 4 stroke events is preceded or followed by acute or subacute headache.^10,11 Acute headache is more often associated with hemorrhagic than ischemic stroke events, and severe headache is a typical symptom of subarachnoid hemorrhage. A relatively large proportion of stroke events among patients with migraine occurs during a migraine attack, and these events are often called a migrainous stroke.¹²

Most of the studies about the association of migraine with the risk of stroke have been retrospective case-control studies, partly owing to difficulties to define migraine in large cohort studies. Prospective data on the headache-associated risk of stroke is scarce, and in particular, there are very few data available on the possible association between chronic unspecified headache and the risk of stroke.

The aim of the present study was to find out whether self-reported chronic headache predicts stroke or a particular type of stroke event in a large prospective cohort. In addition, we wanted to determine whether this association was independent of other cardiovascular risk factors, such as blood pressure, smoking, diabetes, obesity, and serum cholesterol.

Baseline data were collected during 4 independent cardiovascular risk factor surveys carried out in 2 eastern Finnish provinces, North Karelia and Kuopio, in 1972, 1977, 1982, and 1987, and in the Turku-Loimaa region in southwestern Finland in 1982 and 1987.¹³ In 1972 and 1977, a randomly selected sample of 6.6% (13.2% in the city of Joensuu) of the population born between 1913 and 1947 was drawn in North Karelia and Kuopio provinces. In 1977, an additional randomly selected sample of 6.6% of the population born between 1948 and 1952 was drawn in North Karelia. In 1982 and 1987, the sample included the age group 25 to 64 years and was stratified so that in all 3 areas at least 250 subjects were chosen in each sex and 10-year age group according to the international World Health Organization Monitoring of Trends and Determinants in Cardiovascular Disease (WHO-MONICA) Project protocol.¹⁴ The overall study sample comprised 23 327 men and 23 199 women. Practically all of the participants were native Finns (white origin). The average participation rate was 81% among men and 86% among women. The 1746 subjects who participated in more than 1 survey were only included in their first survey cohort. Of the participants, 162 were excluded because of a history of stroke. Another 1728 participants were excluded because of incomplete data on 1 or more risk factors. Thus, 16 992 men and 18 064 women were included in the present analyses.

A self-administered questionnaire, including a set of structured questions (ie, often, occasionally, or never) about the occurrence of headache and other symptoms, was mailed to the subjects in the study sample in advance. Subjects who reported that they often had headaches were classified as chronic headache patients and were compared with those who reported that they never or only occasionally had headaches. Questions regarding smoking and diabetes were also included on the questionnaire, and the use of oral contraceptives was inquired among women.

At the survey site, specially trained research staff checked and collected the questionnaires and measured blood pressure, weight, and height using a standardized protocol. Body mass index (BMI) was used as a measure of obesity. After the measurements, a venous blood specimen was taken. Serum cholesterol level was determined using the Lieberman-Burchard reaction in 1972 and 1977. In 1982 and 1987, an enzymatic assay method was used (CHOD-PAP, Monotest, Boehringer Mannheim, Mannheim, Germany). In our laboratory, the enzymatic assay method gave 2.4% lower values then the Lieberman-Burchard method. Cholesterol values from 1972 and 1977 were corrected by this percentage.

Mortality data were obtained from the Central Statistical Office of Finland. Data on nonfatal cerebrovascular events were received from the National Hospital Discharge Register. The National Causes-of-Deaths register captures all deaths in Finland and also most of the deaths of Finnish citizens outside the country. The hospital discharge registration covers all hospitals and inpatient hospitalizations in Finland. Mortality data and hospital discharge register data were linked to the risk factor data using the identification numbers assigned to every resident of Finland through computer-based linkage, assuring a complete follow-up. The eighth revision of the International Classification of Diseases, Injuries, and Causes of Death (ICD-8) was used in Finland from 1969 to 1986, and the ninth revision (ICD-9) was adopted in 1987. Subarachnoid hemorrhage event was classified by ICD-8 code 430, intracerebral hemorrhage event by ICD-8 code 431, intracerebral infarctions by ICD-8 codes 432 to 438, and any stroke event by ICD-8 codes 430 to 438; ICD-9 codes were otherwise similar, but code 432 was classified as intracerebral infarction. The accuracy of routine registered data has been compared with the specific FINMONICA (The Finnish part of the WHO-MONICA Project) stroke registry data.¹⁵ The results showed that the national death and hospital discharge registers are a reliable source for identifying incident stroke events.

The end point during the follow-up was the incident stroke event, which was defined as either first nonfatal stroke event or stroke death without a preceding nonfatal event. The follow-up of each subject in our present analyses continued through the end of 1995 or until the date of the end point or death from causes other than stroke. The number of incident stroke events among the cohort members was 2167, 382 of which occurred during the first 5 years of follow-up.

Standard t and χ² tests were used to assess the cardiovascular risk factor distribution at baseline. Multivariate analyses were performed using Cox proportional hazards model.¹⁶ The association of chronic headache with the risk of stroke was analyzed for 1, 5, and a maximum of 23 years of follow-up. The statistical analyses were performed using the SAS statisticalprograms.¹⁷

Chronic headache was more common among women (16.7%) than in men (8.9%) (P<.001). In both sexes, subjects with chronic headache were somewhat older than those without headache. Men with headache were more often smokers and had higher serum cholesterol level and blood pressure compared with men without headache (Table 1). Women with headache were more often diabetic and had higher blood pressure, BMI, and serum cholesterol level compared with women without headache.

During the first year of the follow-up, men with headache had a 4-times higher risk of stroke compared with men without headache (Table 2). The association of headache with the risk of stroke markedly attenuated when the follow-up time was extended. During the 5-year and 23-year follow-up, the age-adjusted hazard ratios were 1.86 and 1.24, respectively. Adjustment for smoking, systolic blood pressure, BMI, diabetes, and serum cholesterol level decreased the hazard ratios only slightly. Among women, the headache-associated hazard ratios of stroke also tended to be increased, although the observed associations were not statistically significant. The use of oral contraceptives at baseline was not associated with the subsequent risk of stroke among women (data not shown). In addition to any stroke events, headache also predicted the risk of ischemic stroke during a 5-year follow-up among men (Table 3). The association between headache and hemorrhagic stroke was also direct but remained statistically not significant.

Among men, the headache-associated hazard ratios of stroke tended to be higher in the age group 25 to 49 years than in the age group 50 to 64 years at baseline, higher in smokers than in nonsmokers, and higher in obese than in nonobese men (Table 4). However, none of the interactions between headache and the other risk factors were statistically significant at the P<.05 level.

Chronic unspecified headache was a significant predictor of stroke among men. During the first year of follow-up, men with chronic headache had a 4-times higher risk of stroke compared with men without headache. Moreover, headache was associated with the risk of total and ischemic stroke. The association between headache and the risk of hemorrhagic stroke was also direct, even though statistically not significant owing to a relatively small number of events. The association between headache and the risk of stroke attenuated markedly by the duration of follow-up. Among women, there was a weak direct association between headache and the risk of stroke.

The association of migraine with the risk of stroke has been observed in case reports, several case-control studies, and in a few prospective observational studies.^1-8 Such an association has been found particularly among premenopausal women and, in a few studies, also among men. Migraine with aura may be a stronger predictor of stroke than migraine without aura.^8,9 The association of migraine with the risk of stroke may be particularly strong among women who are smokers or use oral contraceptives.^3,5,9 It has also been suggested that migraine should be regarded as a possible risk factor for stroke among subjects with low levels of other major risk factors of stroke.

Several possible pathophysiological mechanisms have been suggested to explain the increased stroke risk among patients with migraine. These putative mechanisms include a patent foramen ovale, antiphospholipid antibodies, mitochondrial DNA mutations, and prothrombic genetic abnormalities.^18-21 All of these phenomena, however, are rare and can only explain a small proportion of stroke events.

There is only 1 previous study in which the association between chronic unspecified headache and the risk of stroke has been reported. Merikangas and colleagues⁶ analyzed the prospective data from the first US National Health and Nutrition Examination Survey (NHANES) and found that both migraine and severe headache were associated with the risk of stroke. In accordance with our results, the risk of stroke associated with headache also decreased with increasing age in their study. In contrast to our results, however, they found an association between headache and the risk of stroke also among women. Even though the NHANES study population was fairly similar to ours, there were some differences in research methods. In the NHANES, chronic headache was defined based on drug use, whereas we asked the occurrence of headache directly from the study subjects. In their report, the NHANES researchers raise the question that it is possible that the headache-stroke association was attributable to the medications used to treat headache.⁶

We can assume that the mechanisms behind the association of chronic unspecified headache and the risk of stroke are at least partly different than the specific mechanisms suggested for the association between migraine and the risk of stroke.^18-21 It is possible that the vascular damage caused by arteriosclerosis, which usually causes arterial occlusion or intracerebral bleeding, also irritates sensitive nerves of the brain vascular bed through inflammation or other mechanism. Our observation that the association between headache and the risk of stroke markedly weakened with increasing follow-up time may indicate that the pathophysiological processes involved in this association are of a subacute nature.

Some studies have reported an association between headache and hypertension.²² In our study also, the people with headache had somewhat higher blood pressure than the people without headache. However, the difference was very small, and the association between headache and the risk of stroke remained significant after adjustment for blood pressure.

The association between chronic headache and the risk of stroke was stronger among men than among women. This may be explained by different origin of headache in men and women. Among men, headache was less common than in women, and it is possible that headache had a vascular or other intracranial origin more often in men than in women. In women, not only migraine but also tension neck and other musculoskeletal types of headache are more common than in men.^23,24 This most probably reduces the sensitivity of detecting an association between headache and the risk of stroke among women.

A limitation of our study is that we did not ask about the symptoms of migraine, but only unspecified symptoms of headache at baseline. Thus, we cannot estimate the proportion of stroke risk observed with headache that might be due to migraine. On the other hand, it is not easy to get reliable information about the cause of headache in population surveys and even in clinical studies. The estimated prevalence of migraine among stroke patients and their controls has varied markedly in previous case-control studies. In a French case-control study including young women, the prevalence of migraine was as high as 60% among stroke patients and 30% among controls.³ In a large European multicenter study, the corresponding figures were 25% and 13%, respectively.⁷ According to recent survey data, 12.1% of Finnish women and 4.4% of Finnish men reported having migraine (Arpo Aromaa, MD, PhD, oral communication, July 2002).

In conclusion, the results from the present study suggest that chronic unspecified headache is an independent predictor of stroke among men. Since the association between headache and the risk of stroke was particularly strong during a short follow-up, chronic headache may be a marker of the underlying disease process leading to acute stroke. The sex difference observed in this association may be due to a higher prevalence and a more heterogeneous etiology of headache in women compared with men.

Corresponding author and reprints: Pekka Jousilahti, MD, PhD, National Public Health Institute, Department of Epidemiology and Health Promotion, Mannerheimintie 166, FIN-00300 Helsinki, Finland (e-mail: pekka.jousilahti@ktl.fi).

Box Section Ref ID