Association of Migraine With Aura and Other Risk Factors With Incident Cardiovascular Disease in Women

Key Points

Question  Is there an association between migraine with aura and cardiovascular disease (CVD) incidence rates in women, relative to that of other major vascular risk factors?

Findings  In this cohort study that included 27 858 female health professionals aged at least 45 years, the adjusted incidence rate of major CVD was 3.36 per 1000 person-years for women who reported migraine with aura and 2.11 per 1000 person-years for women who reported migraine without aura or no migraine, a difference that was statistically significant.

Meaning  Among female health professionals aged at least 45 years, self-reported migraine with aura was associated with increased incidence rates of CVD, but the clinical importance of this finding remains to be determined.

Importance  Migraine with aura is known to increase the risk of cardiovascular disease (CVD). The absolute contribution of migraine with aura to CVD incidence in relation to other CVD risk factors remains unclear.

Objective  To estimate the CVD incidence rate for women with migraine with aura relative to women with other major vascular risk factors.

Design, Setting, and Participants  Female health professionals in the US (the Women’s Health Study cohort) with lipid measurements and no CVD at baseline (1992-1995) were followed up through December 31, 2018.

Exposures  Self-reported migraine with aura compared with migraine without aura or no migraine at baseline.

Main Outcomes and Measures  The primary outcome was major CVD (first myocardial infarction, stroke, or CVD death). Generalized modeling procedures were used to calculate multivariable-adjusted incidence rates for major CVD events by risk factor status that included all women in the cohort.

Results  The study population included 27 858 women (mean [SD] age at baseline, 54.7 [7.1] years), among whom 1435 (5.2%) had migraine with aura and 26 423 (94.8%) did not (2177 [7.8%] had migraine without aura and 24 246 [87.0%] had no migraine in the year prior to baseline). During a mean follow-up of 22.6 years (629 353 person-years), 1666 major CVD events occurred. The adjusted incidence rate of major CVD per 1000 person-years was 3.36 (95% CI, 2.72-3.99) for women with migraine with aura vs 2.11 (95% CI, 1.98-2.24) for women with migraine without aura or no migraine (P < .001). The incidence rate for women with migraine with aura was significantly higher than the adjusted incidence rate among women with obesity (2.29 [95% CI, 2.02-2.56]), high triglycerides (2.67 [95% CI, 2.38-2.95]), or low high-density lipoprotein cholesterol (2.63 [95% CI, 2.33-2.94]), but was not significantly different from the rates among those with elevated systolic blood pressure (3.78 [95% CI, 2.76-4.81]), high total cholesterol (2.85 [95% CI, 2.38-3.32]), or family history of myocardial infarction (2.71 [95% CI, 2.38-3.05]). Incidence rates among women with diabetes (5.76 [95% CI, 4.68-6.84]) or who currently smoked (4.29 [95% CI, 3.79-4.79]) were significantly higher than those with migraine with aura. The incremental increase in the incidence rate for migraine with aura ranged from 1.01 additional cases per 1000 person-years when added to obesity to 2.57 additional cases per 1000 person-years when added to diabetes.

Conclusions and Relevance  In this study of female health professionals aged at least 45 years, women with migraine with aura had a higher adjusted incidence rate of CVD compared with women with migraine without aura or no migraine. The clinical importance of these findings, and whether they are generalizable beyond this study population, require further research.

Quiz Ref IDMigraine is a chronic, intermittent primary headache disorder characterized by moderate to severe pain and specific associated features, including nausea/vomiting, photophobia, and phonophobia.¹ About one-third of individuals with migraine also report aura symptoms, which are transient neurological symptoms (most often visual disturbances) prior to headache onset caused by changes in brain activity.² Migraine, particularly migraine with aura, has been consistently associated with increased risk of overall and specific cardiovascular disease (CVD) events on the relative scale,^3,4 and migraine has been included in a risk score for CVD.⁵

Quiz Ref IDPotential reasons for the increased risk of CVD observed among individuals with migraine with aura include impairments of the endovascular function,⁶ genetic predisposition to both migraine and CVD,⁷ inflammatory processes,⁸ and others.⁹ Migraine and, particularly, migraine with aura have also been associated with other vascular risk factors, such as hypertension,¹⁰ smoking,¹¹ lipid levels,¹² and body composition.¹³

When consequences of the association of migraine with CVD on the population level are discussed, it has been argued that the association of migraine with aura on CVD incidence on the absolute scale is small and that other vascular risk factors have much stronger associations.¹⁴ Thus, the aim of this study was to compare the adjusted incidence rate of CVD between women with migraine with aura, women with migraine without aura or no migraine, and women with other major vascular risk factors.

Study participants were from the Women’s Health Study, a large randomized, placebo-controlled trial designed to examine the effects of aspirin and vitamin E on the primary prevention of CVD and cancer. The design, methods, and main results of the trial have been published previously.¹⁵ Briefly, at baseline (1992-1995), 39 876 female health professionals in the US aged 45 years or older without a history of CVD or cancer were randomized to receive low-dose aspirin (vs placebo) or vitamin E (vs placebo). Since the end of the trial in March 2004, women continue to be followed up on an observational basis. The Women’s Health Study was approved by the institutional review board at Brigham and Women’s Hospital and all participants provided written informed consent.

Blood samples were collected in tubes containing ethylenediaminetetraacetic acid prior to randomization and stored in vapor-phase liquid nitrogen (−170 °C). Total cholesterol, high-density lipoprotein (HDL) cholesterol, and triglycerides were measured in a core laboratory certified by the National Heart, Lung, and Blood Institute/Centers for Disease Control and Prevention Lipid Standardization program.¹⁶ Women with missing information on migraine status or whose cholesterol or triglyceride levels could not be measured were excluded.

Women were sent follow-up questionnaires, which included questions about demographics, lifestyle, and health information, twice during the first year of the trial and yearly thereafter. For these analyses, we used data through December 31, 2018. As of that date, follow-up was more than 97% complete.

Quiz Ref IDOn the baseline questionnaire, participants were asked about migraine and associated features. In particular, they were asked “Have you ever had migraine headaches?’’ Women who answered positively were asked “In the past year, have you had migraine headaches?’’ Women who experienced migraine headaches within the past year were asked about the characteristics of their headaches, including aura or any indication that a migraine was coming. Agreement of self-reported migraine with International Classification of Headache Disorders II criteria was excellent.¹⁷ Based on this self-reported information, participating women were classified as either having migraine with aura or not. Women who indicated migraine without aura or a history of migraine but no migraine headaches in the year prior to baseline have not been shown to have increased risk of any CVD events in prior studies using data from the Women’s Health Study.^18,19 Therefore, these women, as well as women with no history of migraine, were included in the reference group. Thus, the reference group was composed of women who reported migraine without aura and those who did not report migraine in the year prior to baseline (including women with and without a history of migraine).

During the follow-up of the Women’s Health Study, participants self-reported cardiovascular events. Medical records were then obtained and reviewed by an end point committee of physicians. Nonfatal stroke was confirmed if the participant had a new focal neurologic deficit of sudden onset that persisted for more than 24 hours, and was classified into its major subtypes (ischemic vs hemorrhagic) based on the available clinical and diagnostic information with an excellent interrater agreement.²⁰ The occurrence of myocardial infarction was confirmed if symptoms met World Health Organization criteria and if the event was associated with abnormal levels of cardiac enzymes or diagnostic electrocardiogram results. Cardiovascular deaths were confirmed by review of autopsy reports, death certificates, medical records, or information obtained from next of kin or family members.

The primary outcome for this study was major CVD incidence, a combined end point defined as the first of any of the following events: nonfatal stroke (ischemic and hemorrhagic), nonfatal myocardial infarction, or death due to CVD. The association with the individual end points of first stroke, first myocardial infarction, and death due to CVD was also evaluated. Because migraine with aura was associated with both ischemic and hemorrhagic stroke in the Women’s Health Study,^18,21 we did not evaluate associations with major stroke subtypes separately in this study.

The mean and frequency of vascular risk factors and main baseline characteristics of women who reported migraine with aura were compared with those of women who did not report migraine with aura (including women who reported migraine without aura and those who reported no migraine in the year prior to baseline).

Crude and adjusted incidence rates per 1000 person-years of the CVD outcomes were calculated for women with migraine with aura and women with other vascular risk factors. Adjusted incidence rates were computed using multivariable regression models including potential confounding factors as independent variables. The vascular risk factors were classified in 2 ways: first as dichotomous and second, if possible, as polychotomous variables. In addition to having migraine with aura vs not (including women who reported migraine without aura and those who reported no migraine in the year prior to baseline), we evaluated the following main vascular risk factors: diabetes (yes vs no), smoking (current vs noncurrent or current, past, never), blood pressure (≥160 mm Hg vs <160 mm Hg or <120, 120-129, 130-139, 140-159, ≥160 mm Hg), total cholesterol (≥280 mg/dL vs <280 mg/dL or <160, 160-199, 200-239, 240-279, ≥280 mg/dL), HDL cholesterol (≤40 mg/dL vs >40 mg/dL or ≥60, 50-59, 40-49, <40 mg/dL), triglycerides (≥194 mg/dL vs <194 mg/dL or <77, 77-104, 105-137, 138-193, ≥194 mg/dL), family history of myocardial infarction prior to age 60 (yes vs no), and body mass index (≥30 vs <30 or <25, 25-29, 30-34, ≥35).

The following potential confounding factors were included in all multivariable models: age (5-year categories), alcohol use (≥1 drink per week vs <1 drink per week), exercise (≥1 time per week vs <1 time per week), hypertensive treatment (yes vs no), ever use of hormones (yes vs no), and premenopausal status at baseline (yes vs no).

Three different modeling approaches were used. In the first, adjusted incidence rates of the CVD outcomes were calculated for the levels of the dichotomous vascular risk factors. The adjusted incidence rate for each of the individual vascular risk factors was estimated when each of the respective other vascular risk factors and the confounding factors were set to the mean level observed in the entire study population.

In the second approach, the polychotomous categorization of the main vascular risk factors was used. If the main vascular risk factor did not have a polychotomous categorization available, the dichotomous categorization was used (migraine with aura, diabetes, and family history of myocardial infarction prior to age 60 y). The adjusted incidence rates of the CVD outcomes were calculated for each level of the vascular risk factors, while the levels of the respective other vascular risk factors and the confounding factors were set to the mean level observed in the entire study population. Accordingly, the adjusted incidence rates of the dichotomous factors may differ between the 2 models.

In a third approach, adjusted incidence rates of the CVD outcomes were calculated again for the levels of the dichotomous vascular risk factors. However, in this model, each vascular risk factor was analyzed not as a single main factor but with the observable combination of all possible other vascular risk factors and controlling for the confounding factors as before.

For the model with dichotomous vascular risk factors, t tests were used to test whether the adjusted incidence rates of CVD for women who reported migraine with aura differed from those who did not (including women who reported migraine without aura and those who reported no migraine in the year prior to baseline). We also tested whether the adjusted incidence rates of CVD for women with migraine with aura differed from the adjusted incidence rates of CVD for women with the other vascular risk factors. In the model that included polychotomous risk factors, we used P values for trend across risk factor categories, when applicable, or χ² tests.

In secondary analyses, the third modeling approach was used. The adjusted incidence rate of the CVD outcomes for different combinations of all individual vascular risk factors with and without adding migraine with aura were calculated for each vascular risk factor combination observed in the study population. Each combination of risk factors was compared when migraine with aura was added vs not having migraine with aura (including women who reported migraine without aura and those who did not report migraine in the year prior to baseline).

Adjusted incidence rates were defined as the least square mean from the regression model, which assumed that observed CVD events had a Poisson distribution and that the logarithm of the expected number of events can be modeled by a linear combination of the risk factors.^22-24 Using SAS software, version 9.4 (SAS Institute), the GENMOD procedure was used to calculate the least square mean of the Poisson regression model²⁵ and the NLMIXED procedure was used to calculate the P values of the contrast of the adjusted incidence rates.²⁶

Less than 1.7% of the study population had missing information on any variable. Missing information on any of the vascular risk factors was imputed by assigning the median value of that variable. For participants with missing information on other factors, a missing variable indicator was created and included in the multivariable models.

A sensitivity analysis was conducted in which the estimation of the adjusted incidence rates of CVD for migraine with aura and the other vascular risk factors were repeated using multiple imputation to account for missing information. Further details are described in the eAppendix in the Supplement. In another sensitivity analysis, only women who did not report migraine at or prior to baseline were included in the reference group and the primary analysis was repeated.

Two-tailed P values were calculated, and a P value of less than .05 was considered statistically significant. Because of the potential for type I error due to multiple comparisons, findings for analyses of secondary end points and analyses should be interpreted as exploratory.

Blood samples were collected in tubes containing ethylenediaminetetraacetic acid from 28 345 women. Total cholesterol and triglyceride analyses were performed on 27 938 of the blood samples. A total of 79 women with missing information on migraine status on the baseline questionnaire and 1 woman who reported a history of myocardial infarction prior to baseline during follow-up were excluded.

The study population included 27 858 women (mean [SD] age at baseline, 54.7 (7.1) years), including 1435 (5.2%) who had migraine with aura and 26 423 who did not (2177 [7.8%] had migraine without aura and 24 246 [87.0%] did not report migraine in the year prior to baseline). The baseline characteristics of select vascular risk factors for women with migraine with aura and women without migraine with aura are summarized in Table 1. Women with migraine with aura were slightly younger, had higher triglycerides levels, and more frequently reported a family history of premature myocardial infarction. They also reported drinking less alcohol and were more likely to have ever used postmenopausal hormones.

During a mean follow-up of 22.6 years (629 353 total person-years), 1666 major CVD events, 887 strokes (fatal and nonfatal), 629 myocardial infarctions (fatal and nonfatal), and 391 deaths due to CVD occurred. The crude overall incidence rate per 1000 person-years was 2.87 for major CVD, 1.52 for total stroke, 1.07 for myocardial infarction, and 0.62 for deaths due to CVD.

With the exception of obesity, all major risk factors were statistically significantly associated with major CVD incidence (eTable 1 in the Supplement). Results for the individual cardiovascular outcomes are summarized in eTables 2, 3, and 4 in the Supplement.

The adjusted incidence rate of major CVD was 3.36 cases per 1000 person-years (95% CI, 2.72-3.99) among women with migraine with aura (adjusted for other dichotomous vascular risk factors and confounding variables) and 2.11 per 1000 person-years (95% CI, 1.98-2.24) for women with migraine without aura or no migraine (P value comparing the adjusted incidence rates of migraine without aura vs migraine with aura <.001) (Table 2). Compared with having migraine with aura, only the presence of 2 other risk factors led to significantly higher adjusted incidence rates of major CVD: diabetes (adjusted incidence rate, 5.76 [95% CI, 4.68-6.84]; P value comparing the incidence rate to that of migraine with aura <.001) and current smoking (adjusted incidence rate, 4.29 [95% CI, 3.79-4.79]; P value comparing the incidence rate with that of migraine with aura = .02). A similar pattern was observed for myocardial infarction and CVD death (Table 2). Women with migraine with aura did not have a significantly different rate of major CVD compared with women who had the highest level of systolic blood pressure and total cholesterol and family history of myocardial infarction. In contrast, the incidence rate of major CVD was significantly higher for women with migraine with aura than the incidence rate among those with obesity (2.29 [95% CI, 2.02-2.56]; P value comparing the incidence rate with that of migraine with aura = .001) or those who had unfavorable lipid profiles (triglycerides ≥194 mg/dL: 2.67 [95% CI, 2.38-2.95]; P value comparing the incidence rate with that of migraine with aura = .043; HDL cholesterol <40 mg/dL: 2.63 [95% CI, 2.33-2.94]; P value comparing the incidence rate with that of migraine with aura = .04) (Table 2).

The adjusted incidence rates of migraine with aura and the polychotomous categorization of the other vascular risk factors is shown in the Figure for major CVD and in eFigures 1, 2, and 3 in the Supplement for the other vascular outcomes.

The adjusted incidence rate of major CVD for individual vascular risk factors with and without adding migraine with aura is shown in Table 3. The adjusted incidence rate after adding migraine with aura to each vascular risk factor was higher than only having the vascular risk factor. The incremental increase in the incidence rate ranged from 1.01 additional cases per 1000 person-years, when migraine with aura was added to body mass index, to 2.57 additional cases per 1000 person-years, when migraine with aura was added to diabetes.

The adjusted incidence rate of major CVD for the maximum combination of vascular risk factors with and without adding migraine with aura is shown in Table 4. With a combination of all vascular risk factors, except for a family history of premature myocardial infarction, the incidence rate of major CVD increased from 39.55 cases per 1000 person-years to 62.96 cases per 1000 person-years when migraine with aura was added.

The results from the sensitivity analysis in which missing values were imputed using multiple imputations is shown in eTable 5 in the Supplement. The results were similar to those of the primary analysis. Results were similar when women who reported migraine without aura or any history of migraine at baseline were excluded from the reference group, indicating that this group did not have different rates of major CVD compared with women who did not report migraine in the year prior to baseline (eTable 6 in the Supplement).

Quiz Ref IDIn this large cohort study of women aged at least 45 years, women with migraine with aura had a higher incidence rate of self-reported major CVD events compared with women without migraine or with migraine without aura, after adjusting for other major vascular risk factors and confounding. Although diabetes and current smoking were associated with higher rates of major CVD, migraine with aura was more strongly associated with the rate of major CVD than obesity or unfavorable lipid levels, and had a similar association as with elevated systolic blood pressure or high total cholesterol. The association patterns were similar for specific CVD events.

The association of migraine and migraine with aura on overall and specific CVD events has been evaluated using relative risk in many clinic- and population-based studies.^3,4,27,28 Although there was an indication of some heterogeneity across studies, the relative risk estimates ranged from 1.5 for migraine overall to more than 2 for migraine with aura. In a 2018 nationwide population-based study from Denmark, Adelborg and colleagues⁴ estimated the relative hazard of CVD according to migraine status. Migraine with aura was associated with myocardial infarction (hazard ratio, 1.74 [95% CI, 1.44-2.11]), ischemic stroke (hazard ratio, 2.49 [95% CI, 2.16-2.86]), and hemorrhagic stroke (hazard ratio, 1.82 [95% CI, 1.34-2.48]).⁴ In contrast to data from the Women’s Health Study,¹⁸ migraine without aura was also associated with an increased risk of CVD in the study by Adelborg and colleagues,⁴ but the effect size was smaller compared with migraine with aura. A previous study using Women’s Health Study data observed similar incidence rates of major CVD for migraine with aura as seen in this study after adjustment for age.¹⁸ However, to the best of our knowledge, no studies have compared the estimated adjusted incidence rates of CVD by migraine status with the adjusted incidence rates of CVD due to other cardiovascular risk markers.

Quiz Ref IDSeveral biological mechanisms have been proposed to link migraine, particularly migraine with aura, with increased risk of CVD events. These mechanisms have included elevated vascular biomarkers^8,29 leading to endothelial dysfunction, other vascular pathology,³⁰ open patent foramen ovale, genetic predisposition,⁷ or migraine-specific or other pain medication use.³¹ However, with regard to medication use, not only patients with migraine with aura use such treatment. Cortical spreading depolarization, the electrophysical mechanisms involved in migraine aura, can play a role in stroke³² and potentially in other vascular events.³³ Nevertheless, the precise mechanisms linking migraine with aura to CVD incidence are still not well understood.

Strengths of this study include the large number of participants and outcome events, available information on various vascular risk factors, confirmation of CVD events by medical record review, and the homogeneity of the study population, which reduced potential biases, including differences in health care access.

This study has several limitations. First, information on migraine and vascular risk factors were self-reported, and misclassification is possible. Second, migraine with aura and vascular risk factors were assessed at baseline, and changes in these risk factors over time were not taken into account. Third, detailed information on management of migraine or vascular risk factors, which may influence CVD risk, was not available. Fourth, this study was conducted in a cohort of female health professionals and these results may not be generalizable to other settings, particularly to men. Fifth, this study did not explore how changes in migraine frequency over time may modify the association between migraine with aura and CVD incidence. Additionally, no information on when migraine started for the participating women was available.

In this study of female health professionals aged at least 45 years, women with migraine with aura had a higher adjusted incidence rate of CVD compared with women with migraine without aura or no migraine. The clinical importance of these findings, and whether they generalize beyond this study population, require further research.

Corresponding Author: Tobias Kurth, MD, ScD, Institute of Public Health, Charité – Universitätsmedizin Berlin, Charitéplatz 1, 10117 Berlin, Germany (tobias.kurth@charite.de).

Accepted for Publication: April 18, 2020.

Author Contributions: Drs Kurth and Bubes had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Kurth.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Kurth, Kotler, Bubes.

Critical revision of the manuscript for important intellectual content: Kurth, Rist, Ridker, Bubes, Buring.

Statistical analysis: Kurth, Kotler, Bubes.

Administrative, technical, or material support: Ridker, Bubes.

Supervision: Kurth, Ridker.

Conflict of Interest Disclosures: Dr Kurth reported receiving personal fees from Eli Lilly, Newsenselab, CoLucid, Total, Novartis, and Daiichi Sankyo and grants from Amgen during the conduct of the study and personal fees from BMJ outside the submitted work. Dr Rist reported receiving grants from the National Heart, Lung, and Blood Institute, Brigham and Women's Hospital, and Biogen during the conduct of the study and personal fees from the American Heart Association outside the submitted work. Dr Ridker reported receiving grants from Novartis and Kowa and personal fees from Inflazome, Corvidia, Johnson & Johnson, and Amarin outside the submitted work. Dr Buring reported receiving grants from the National Institutes of Health during the conduct of the study. No other disclosures were reported.

Funding/Support: The Women’s Health Study is funded by grants from the National Cancer Institute (CA047988 and UM1 CA182913) and the National Heart, Lung, and Blood Institute (HL043851, HL080467, and HL099355). Dr Rist is supported by a Career Development Award from the National Heart, Lung, and Blood Institute (K01 HL128791).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.