Migraine and Coronary Heart Disease in Women and Men

Transcription

1 Universidade de Sao Paulo From the SelectedWorks of Paulo A Lotufo 2002 Migraine and Coronary Heart Disease in Women and Men Paulo A Lotufo, Universidade de São Paulo Available at:

2 Migraine and Coronary Heart Disease in Women and Men Nancy R. Cook, ScD; Isabela M. Benseñor, MD; Paulo A. Lotufo, MD; I-Min Lee, ScD; Patrick J. Skerrett, MS; Marilyn J. Chown, MPH; Umed A. Ajani, MBBS; JoAnn E. Manson, MD; Julie E. Buring, ScD Objective. We evaluated migraine as an independent risk factor for subsequent coronary heart disease (CHD) events among women in the Women s Health Study (WHS) and men in the Physicians Health Study (PHS). Background. Although several studies have suggested that migraine is associated with increased risk of stroke, there are few and conflicting data on whether migraine predicts risk of future CHD events. Methods. The WHS is an ongoing randomized, double-blind, placebo-controlled trial of low-dose aspirin and vitamin E in the primary prevention of cardiovascular disease and cancer in 39,876 women health professionals aged 45 years in 1993, and the PHS is a completed randomized, double-blind, placebo-controlled trial of aspirin and -carotene in the primary prevention of cardiovascular disease and cancer in 22,071 men physicians aged 40 to 84 years in Primary endpoints were defined as major CHD (nonfatal myocardial infarction [MI] or fatal CHD) and total CHD (major CHD plus angina and coronary revascularization). Results. After adjusting for other CHD risk factors, female health professionals and male physicians reporting migraine were not at increased risk for subsequent major CHD (women: relative risk [RR], 0.83; 95% confidence interval [CI], 0.53 to 1.29; men: RR, 1.02; 95% Cl, 0.79 to 1.31) or total CHD (women: RR, 1.01; 95% Cl, 0.76 to 1.34; men: RR, 0.98; 95% Cl, 0.82 to 1.18). When considered separately, there was also no increase in risk of MI or angina. Conclusion. These prospective data suggest that migraine is not associated with increased risk of subsequent CHD events in women or men. Key words: migraine, headache, coronary heart disease Abbreviations: CHD coronary heart disease, MI myocardial infarction, PHS Physicians Health Study, WHS Women s Health Study, RR relative risk, CI confidence interval, IHS International Headache Society, CABG coronary artery bypass graft, PTCA percutaneous transluminal coronary angioplasty, BMI body mass index (Headache. 2002;42: ) Migraine is a common medical condition affecting more than 10 million people in the United States alone. Women are more commonly affected than men, with 17.6% of women and 5.7% of men reporting at least one migraine attack per year. 1-3 An association From the Division of Preventive Medicine, Department of Medicine, Brigham and Women s Hospital, Harvard Medical School, Boston, MA. Address all correspondence to Dr. Nancy R. Cook, Division of Preventive Medicine, Brigham and Women s Hospital, 900 Commonwealth Avenue East, Boston, MA Accepted for publication June 9, between migraine and risk of subsequent cardiovascular disease has long been hypothesized, beginning with Janeway s classic paper on hypertension and migraine 4 and continuing to the present. Several large studies have suggested that migraine is associated with risk of subsequent stroke. 5-7 A history of migraine has also been associated with increased risk of coronary heart disease (CHD), though these data are more limited and not as consistent. Several case reports have suggested a link between migraine and angina, 8-10 and cross-sectional studies have supported a relationship between migraine and history of vascular disease or hypertension In one of the largest investigations to date, conducted among a cohort of pa- 715

3 716 September 2002 tients enrolled in a California health maintenance organization, individuals with migraine were significantly more likely to also report symptoms of chest pain on a baseline preventive medicine examination. 15 The prospective component of the study, however, showed that neither migraine nor chest pain were associated with risk of myocardial infarction (MI) except among a subgroup of women with a family history of MI. An association between migraine and increased risk of CHD is biologically plausible. Basic research has suggested possible mechanisms in the pathophysiology of migraine, including increased vascular reactivity, increased platelet aggregation, 19 and vasoconstrictive effects related to serotonin release, 20,21 that could explain a connection between migraine and CHD. An association between elevated serotonin and increased risk of coronary artery disease and cardiac events has also been observed. 22 Two casecontrol studies of patients with MI found a higher prevalence of migraine specifically among those who had an MI but with normal coronary arteries, 23,24 suggesting a relationship with coronary spasm rather than atherosclerosis. In addition, with regard to the treatment of migraine, both old and new antimigraine drugs have putative vasoconstrictor effects on coronary circulation Several case reports have described patients who experienced an MI following administration of an antimigraine medication, particularly sumatriptan. 26,28,29 In this report, we evaluate the association between self-reported migraine and incident CHD in two large randomized trials, the ongoing Women s Health Study (WHS) and the completed Physicians Health Study (PHS). In both of these trials, information on migraine and other headaches was assessed throughout the study. As cardiovascular disease was a primary outcome in both trials, this presented a unique opportunity to examine these relations prospectively. METHODS Study Populations. Descriptions of the trial methodologies and study populations for both the WHS 30,31 and the PHS 32,33 have been provided in detail elsewhere. Briefly, the WHS is an ongoing randomized, double-blind, placebo-controlled trial designed to evaluate the balance of benefits and risks of lowdose aspirin and vitamin E in the primary prevention of cardiovascular disease and cancer among apparently healthy women. A total of 39,876 American women health professionals aged 45 years or older at baseline in 1993 and without a history of cardiovascular disease (including MI, stroke, angina, and coronary revascularization procedures) or cancer (except nonmelanoma skin cancer) were randomly assigned in a 2-by-2 factorial design to 1 of 4 treatment groups: active aspirin (100 mg on alternate days), active vitamin E (600 IU on alternate days), both active agents, or both placebos. At baseline, information was obtained on a large number of coronary risk factors and lifestyle variables. Each year, participants are mailed a supply of monthly calendar packs containing study medications along with a brief questionnaire asking about compliance with the assigned treatments, side effects, the occurrence of a number of specific medical conditions or diagnoses, and updated or additional information about risk factors. Follow-up from randomization through March 8, 2000 was included in these analyses, with an average of 6.1 years (range 4.1 to 6.9 years). The PHS is a completed randomized, doubleblind, placebo-controlled trial that evaluated the role of aspirin and -carotene in the primary prevention of cardiovascular disease and cancer among apparently healthy men. A total of 22,071 American men physicians aged 40 to 84 years in 1982 and without a history of cardiovascular disease (including MI and stroke) or cancer (except nonmelanoma skin cancer) were randomly assigned in a 2-by-2 factorial design to 1 of 4 treatment groups: active aspirin (325 mg on alternate days), active -carotene (50 mg on alternate days), both active agents, or both placebos. Every 6 months for the first year and annually thereafter, participants were mailed a supply of monthly calendar packs along with a brief questionnaire similar to that of the WHS. In these analyses, follow-up was included from randomization through the termination of the trial on December 31, 1995, with an average of 12.9 years (range years). Participants in both the PHS and WHS provided written informed consent before entry into the trial, and each

4 Headache 717 study was conducted after approval by the institutional review board of Brigham and Women s Hospital. Assessment of Migraine. In the WHS, information on migraine was elicited on the baseline questionnaire. Women were asked whether they had ever had one of a list of specified diseases or conditions, including migraine. For those answering yes to migraine, a separate question asked whether they had had 1 or more migraine attacks in the past year and, if so, the approximate frequency of these attacks (daily, weekly, monthly, every other month, or less than 6 times per year), as well as information regarding the characteristics of their migraines. The women were asked to indicate which characteristics from a specified list applied to their migraines, including aura or any other indication of an impending migraine, unilateral location of pain, pulsating character of pain, nausea and vomiting, sensitivity to light, sensitivity to sound, tingling or numbness, dizziness or vertigo, unilateral weakness in the face, arms, or legs, difficulty with speech, inhibition of daily activities, aggravation by physical activity, and migraine duration. A total of 77 women with missing data on migraine at baseline were excluded from these analyses. To ascertain the validity of self-reported migraine, we analyzed data on self-reported migraine and descriptions of migraine characteristics among a subsample of women who reported at baseline having had 1 or more migraine attacks in the past year. The information these women provided was compared to the criteria of the International Headache Society (IHS) for categorizing migraine. 34 Because we did not ascertain the number of migraine attacks each woman experienced prior to randomization in the WHS, we could not establish one of the IHS requirements of at least 5 attacks of migraine without aura or 2 attacks of migraine with aura during one s lifetime. Similar modified IHS criteria have been used in other investigations. 1,3 On each annual follow-up questionnaire, women were asked whether in the past year they had had any of a list of specified diagnoses or conditions, including migraine and nonmigraine headache. Women who said no to migraine but yes to headache in one or more of the follow-up questionnaires were classified as having nonmigraine headaches. Women who reported no migraine on the baseline questionnaire and no headaches on follow-up questionnaires were classified as having no headaches and they formed the comparison group for secondary analyses against whom women with migraine and women with nonmigraine headaches were compared. In the PHS, information on both migraine and headache was not obtained at baseline, but was first elicited on the 6-month follow-up questionnaire. Physicians were asked at that time, and on each subsequent annual follow-up questionnaire, whether they had experienced migraine or, in a separate question, headache since returning the previous questionnaire. A physician was classified as having migraine if he reported having a migraine attack on at least one follow-up questionnaire and was classified as not having migraine if he never reported migraine. Because information regarding migraine was not elicited on the baseline PHS questionnaire, a random sample of physicians who reported migraine on a follow-up questionnaire was contacted to ascertain whether or not a history of migraine was present at baseline. In virtually every case, the physician indicated that the first attack had occurred long before he was randomized into the trial. This finding is compatible with earlier reports that new cases of migraine are rare after age 40 2 and supports our assumption in these analyses that in this age group, migraines reported at follow-up are indicative of a history of migraine rather than the new onset of migraine. In secondary analyses, physicians who said no to migraine and yes to headaches were classified as having nonmigraine headaches. Physicians who said no to both migraine and headaches were classified as having no headaches and they formed the comparison group for these secondary analyses. Endpoint Ascertainment. The primary endpoint of interest was the first occurrence of nonfatal MI or fatal CHD, a combined endpoint denoted here as major CHD. We also considered a broader endpoint, denoted total CHD, which includes nonfatal MI, fatal CHD, angina, and coronary revascularization procedures (either coronary artery bypass graft [CABG] or percutaneous transluminal coronary angioplasty [PTCA]). In addition, in secondary analyses, we sep-

5 718 September 2002 arately examined the individual endpoints of nonfatal Ml, fatal CHD, and angina. In all analyses, we excluded the small number of participants who provided postrandomization reports of CHD that had actually occurred prior to randomization. In the WHS, there were 4 such reports, leading to a total of 39,795 participants included in these analyses after further excluding 77 women with missing information on migraine. In addition, although women who reported angina and coronary revascularization procedures at baseline were excluded from the trial, 159 women subsequently reported these conditions with a date prior to randomization. These women were excluded from analyses of total CHD. In the PHS, because migraine was first assessed at 6 months, the 60 reported occurrences of MI prior to this time were excluded from all analyses, leaving a total sample size of 22,011. In analyses of total CHD, the 413 men who reported angina or a coronary revascularization procedure prior to 6 months were also excluded. In both the WHS and PHS, when participants selfreported a relevant cardiovascular endpoint, written consent for the review of their medical records was obtained and the pertinent records were requested from hospitals and treating physicians. Reported diagnoses of CHD or death from CHD were considered to have been confirmed only after examination of all available information by an Endpoints Committee of physicians who were blinded to the assigned treatment. Self-reported diagnoses of nonfatal MI were confirmed based on World Health Organization criteria. 35 Fatal CHD was documented by convincing evidence of a cardiovascular mechanism from all available sources, including death certificates, hospital records, and, for deaths occurring outside the hospital, observers impressions. In the WHS, self-reported angina was confirmed by a positive coronary angiogram ( 70% stenosis), subsequent occurrence of MI, or subsequent coronary revascularization procedure. In the PHS, physicians who reported angina were mailed a supplementary questionnaire. Self-reports of angina were considered confirmed if the participant also reported on this supplementary questionnaire a positive diagnostic test (including exercise tolerance test, exercise thallium scan, exercise radionuclide ventriculogram, or 50% stenosis of coronary arteries on angiogram). As in the WHS, self-reported angina was also considered confirmed by the subsequent occurrence of MI or a subsequent coronary revascularization procedure. In the WHS, self-reported coronary revascularization procedures were confirmed with hospital records; in the PHS, self-reports were accepted without requesting hospital records. Statistical Methods. In both the WHS and PHS, we first examined the distribution of age-adjusted baseline coronary risk factors by reported migraine status. These included body mass index (BMI), cigarette smoking, alcohol use at least once per week, vigorous exercise at least once per week, history of hypertension, history of elevated cholesterol, history of diabetes, parental history of MI before age 60, and randomized treatment assignment. Information was also compared on menopausal status and current hormone use in the WHS, and history of angina in the PHS. In the WHS, hypertension was defined as systolic/diastolic blood pressure of 140/90 mmhg or physician diagnosis of hypertension; in the PHS, it was defined as systolic/diastolic blood pressure of 140/90 mmhg or current treatment with antihypertensive medication. In the WHS, elevated cholesterol was defined as a reported cholesterol level of 240 mg/dl or higher, a physician diagnosis of high cholesterol, or current treatment with cholesterol-lowering medication, whereas in the PHS it was defined as a reported level of 240 mg/dl or higher or current treatment with cholesterol-lowering medication. Within each study, we computed relative risk (RR) estimates of each CHD endpoint associated with migraine status using Cox regression analysis adjusting for age and randomized treatment assignment. 36 For each relative risk, the 95% confidence interval (CI) was calculated. Cox regression was also used to control simultaneously for differences in baseline coronary risk factors between the groups. The main analysis in the WHS defined exposure as migraine status provided at baseline. The presence of migraine was defined as ever suffering from migraine headache. Additional analyses were performed separating those experiencing a migraine in the past year and those with a migraine history only. In the PHS, because migraine status was first assessed at the

6 Headache month follow-up, we used a time-varying Cox regression analysis to reduce bias due to differences in exposure opportunity. All events occurring prior to the first assessment of migraine at 6 months in the PHS were excluded. Beginning at the 6-month follow-up, person-time in each exposure category was used as the denominator for comparison. Men whose first report of migraine occurred following an MI were treated as unexposed throughout. We performed a similar analysis using reports of migraine on followup questionnaires in the WHS, and compared results with those from the baseline exposure model. We also explored the association with nonmigraine headache. In both the WHS and PHS, questions on headache were asked beginning with the first followup questionnaire. For these analyses, time-varying exposures were computed, classifying individuals as having migraine if they ever reported migraine up to that time, as having headache if they ever reported headache up to that time, and otherwise classifying them in the no headache group. These analyses were also adjusted for coronary risk factors. Finally, we evaluated the possibility that baseline coronary risk factors, particularly those suggested from other studies such as age, family history of premature MI, 15 and smoking, 37,38 may have modified the effect of migraine on CHD through subgroup analyses. Tests of effect modification were conducted using interaction terms in the Cox regression model while simultaneously controlling for other coronary risk factors. RESULTS Baseline Characteristics. In the WHS, 7,334 (18.4%) of the 39,795 participants who were initially free of CHD and who provided migraine information reported a history of migraine on the baseline questionnaire. Table 1 presents the age-adjusted prevalence of various coronary risk factors reported at baseline for women who reported migraine versus no migraine. Women reporting migraine in this cohort Table 1. Age-Adjusted Baseline Coronary Risk Factors by Reported Migraine Status among Female Health Professionals in the WHS Characteristic Migraine (n 7,334) No Migraine (n 32,461) P Age (years)* Body mass index (kg/m 2 )* Cigarette smoking Current (%) Past (%) Never (%) Alcohol use 1/week (%) Vigorous exercise 1/week (%) History of hypertension (%) History of elevated cholesterol (%) History of diabetes (%) Parental history of MI age 60 (%) Menopausal status (%) Postmenopausal (%) Premenopausal (%) Unknown (%) Current hormone use among postmenopausal women (%) Randomized aspirin (%) Randomized vitamin E (%) *Mean SE. Hypertension is defined as self-reported systolic/diastolic blood pressure of 140/90 mmhg or higher, or physician diagnosis of hypertension. Elevated cholesterol is defined as a reported level of 240 mg/dl (6.2 mmol/l) or higher, a self-reported physician diagnosis, or current treatment with cholesterol-lowering medication.

7 720 September 2002 tended to be younger by 1.3 years. After adjusting for this difference in age, women with migraine had a slightly higher BMI and were statistically significantly less likely to be current smokers, to drink alcohol at least once per week, and to exercise vigorously at least once per week. They were also significantly more likely to report a history of hypertension, a history of elevated cholesterol, and a parental history of MI, and were less likely to report a history of diabetes. Women with migraine were less likely to be premenopausal and, if postmenopausal, were more likely to be currently using hormone replacement therapy. In the PHS, 2,665 (12.1%) of the 22,011 men who were free of CHD at 6 months reported at least 1 episode of migraine at some point during the 12 years of postrandomization follow-up. Table 2 presents the age-adjusted prevalence of various coronary risk factors reported at baseline for men who reported migraine at anytime during the study versus no migraine. As in the WHS, men reporting migraine in the PHS tended to be younger, with a difference of 2.4 years. After adjustment for age, those who reported migraine had a slightly lower BMI and were significantly less likely to be current smokers and weekly drinkers of alcohol, but were similar with respect to weekly exercise. History of hypertension, angina, coronary revascularization procedures, and parental history of MI were similar in the 2 groups, but those reporting migraine were somewhat more likely to have a history of elevated cholesterol and less likely to have a history of diabetes than those reporting no migraine. Risk of CHD. After an average of 6.1 years of follow-up in the WHS, among the 39,795 women free of CHD at baseline, 183 women experienced major CHD events (MI or fatal CHD), including 171 cases of nonfatal Ml and 12 deaths due to fatal CHD (Table 3). Of these, 26 events (59.1 per 100,000 personyears) occurred among the 7,334 women reporting migraine, and 157 (80.4 per 100,000 person-years) occurred among the 32,461 women reporting no migraine. A total of 395 women experienced total CHD events (major CHD or angina or a coronary revascularization procedure). Of these, 64 events (146.5 per 100,000 person-years) occurred among women re- Table 2. Age-Adjusted Baseline Coronary Risk Factors by Reported Migraine Status among Male Physicians in the PHS Characteristic Migraine (n 2,665) No Migraine (n 19,346) P Age (years)* Body mass index (kg/m 2 ) Cigarette smoking Current (%) Past (%) Never (%) Alcohol use 1/week (%) Vigorous exercise 1/week (%) History of hypertension (%) History of elevated cholesterol (%) History of diabetes (%) History of angina (%) History of coronary revascularization (%) Parental history of MI age 60 (%) Randomized aspirin (%) Randomized beta-carotene (%) *Mean SE. Hypertension is defined as self-reported systolic/diastolic blood pressure of 140/90 mmhg or higher, or current treatment with anti-hypertensive medication. Elevated cholesterol is defined as a reported level of 240 mg/dl (6.2 mmol/l) or higher, or current treatment with cholesterol-lowering medication. Ever report of migraine during the follow-up.

8 Headache 721 Table 3. RR and 95% CI for Coronary Heart Disease According to Migraine Status among Women in the WHS and Men in the PHS* Women Men Migraine No Migraine Migraine No Migraine Major CHD Cases Age-adjusted RR % CI (ref) (ref) MV-adjusted RR % CI (ref) (ref) Total CHD Cases ,956 Age-adjusted RR % CI (ref) (ref) MV-adjusted RR % CI (ref) (ref) *MV, multivariable; ref, reference group. Major CHD is defined as fatal CHD or nonfatal MI. Total CHD is defined as major CHD, a coronary revascularization procedure, or new occurrence of angina. Men reporting angina at baseline in the PHS are excluded from the latter analysis. Adjusted for age and randomized treatment assignment. Adjusted for age, randomized treatment assignment, body mass index, smoking, alcohol use, exercise, history of hypertension, elevated cholesterol, diabetes, and parental history of MI before age 60. Models in women are adjusted for menopausal status and postmenopausal hormone use, and models for major CHD in men are also adjusted for history of angina (an exclusion criterion in the WHS). porting migraine and 331 (170.7 per 100,000 personyears) among women reporting no migraine. After an average of 12 years of follow-up in the PHS, among the 22,011 physicians free of CHD at baseline, 1,057 men experienced confirmed major CHD events, including 826 cases of nonfatal Ml and 306 deaths due to fatal CHD. Of these, 79 (367.5 per 100,000 person-years) occurred among physicians following a report of migraine and 978 (411.2 per 100,000 person-years) among physicians with no prior report of migraine. With regard to total CHD, 2,108 cases were confirmed in this cohort, 152 (745.1 per 100,000 person-years) among those with prior migraine, and 1,956 (867.7 per 100,000 personyears) among those with no prior report of migraine. As shown in Table 3, neither women nor men reporting migraine had an increased age- and treatmentadjusted relative risk of major CHD compared with those reporting no migraine. In the WHS, the relative risk was 0.86 (95% Cl, 0.56 to 1.30), whereas in the PHS it was 1.00 (95% Cl, 0.80 to 1.26). Further adjustment for coronary risk factors did not materially change the results, with RR estimates of 0.83 (95% Cl, 0.53 to 1.29) in the WHS and 1.02 (95% Cl, 0.79 to 1.31) in the PHS. In addition, when migraine in the WHS was treated as a time-varying exposure in an analysis similar to that used in the PHS, results were similar (RR, 0.85; 95% CI, 0.55 to 1.31). When men reporting angina or coronary revascularization at baseline were excluded from the analysis, the results did not change. Similar null results were found for total CHD in both women and men, with RR estimates of 1.01 (95% Cl, 0.76 to 1.34) in women and 0.98 (95% Cl, 0.82 to 1.18) in men after multivariable adjustment. When considering the separate components of the combined endpoints (Table 4), neither women nor men reporting migraine had an increased risk of nonfatal MI after multivariate adjustment, with RR estimates of 0.91 (95% Cl, 0.58 to 1.43) in the WHS and 1.18 (95% Cl, 0.90 to 1.55) in the PHS. With respect to fatal CHD, only 12 cases of fatal CHD were confirmed among

9 722 September 2002 women, all within the group without migraine. Among men, a total of 306 cases of fatal CHD were confirmed, with 14 among those reporting migraine. No increase in fatal CHD with report of migraine was found in the PHS (RR, 0.58; 95% CI, 0.32 to 1.07). For the 255 reported cases of incident angina in women and 1,482 in men, no differences by migraine status were found, with RR estimates of 1.09 (95% Cl, 0.78 to 1.53) in the WHS and 0.98 (95% Cl, 0.79 to 1.21) in the PHS. When reports of nonmigraine headaches were treated as a separate exposure category in the WHS and PHS, the association of migraine with major CHD when compared with those with neither headache nor migraine was largely unchanged. In the WHS, the RR was 0.87 (95% CI, 0.55 to 1.37), and in the PHS it was 1.01 (95% CI, 0.78 to 1.31). To investigate the validity of self reports of migraine, we examined data from the 5,174 women who reported having had 1 or more migraine attacks in the past year and who answered detailed questions about their migraine characteristics. We defined full agreement with the IHS diagnosis of migraine as fulfilling all of the modified IHS criteria for migraine (IHS code 1.1), and likely agreement as fulfilling all but 1 of the modified IHS criteria (IHS code 1.7, migrainous disorder). 34 We found full agreement between self-reported migraine and the modified IHS criteria among 2,406 (46.5%) women and likely agreement among an additional 1,418 (27.4%), for a total of 3,824 women (73.9%). In addition, 983 (40.9%) of the 2,406 women who fully fulfilled the modified IHS criteria also reported the presence of aura with their migraine attacks. We then performed an additional analysis restricting the exposed group to the 3,824 women who fulfilled the modified IHS criteria for migraine or likely migraine. Twelve cases of major CHD oc- Table 4. RR and 95% CI for Individual CHD Endpoints According to Migraine Status among Women in the WHS and Men in the PHS* Women Men Migraine No Migraine Migraine No Migraine Nonfatal MI Cases Age-adjusted RR % CI (ref) (ref) MV-adjusted RR % CI (ref) (ref) Fatal CHD Cases Age-adjusted RR % CI (ref) MV-adjusted RR % CI (ref) Angina Cases ,373 Age-adjusted RR % CI (ref) (ref) MV-adjusted RR % CI (ref) (ref) *MV, multivariable; ref, reference group. Men reporting angina at baseline in the PHS are excluded from the latter analysis. Adjusted for age and randomized treatment assignment. Adjusted for age, randomized treatment assignment, body mass index, smoking, alcohol use, exercise, history of hypertension, elevated cholesterol, diabetes, and parental history of MI before age 60. Models in women are adjusted for menopausal status and postmenopausal hormone use, and models for major CHD in men are also adjusted for history of angina (an exclusion criterion in the WHS).

10 Headache 723 curred in this group, and the results were essentially unchanged (RR, 0.88; 95% CI, 0.49 to 1.60). When only those completely fulfilling the criteria were included, the risk estimate changed but remained nonsignificant (RR, 1.33; 95% CI, 0.69 to 2.53). In addition, when we restricted the exposed group to those who reported migraine with aura, the RR increased but still remained not statistically significant (RR, 1.50; 95% CI, 0.61 to 3.70). The elevation did not persist when considering total CHD or when women with definite and likely migraine were combined. We examined the association between migraine and major CHD among subgroups of women and men with and without a number of coronary risk factors at baseline (Table 5). No statistically significant effect modification was observed with age, smoking, alcohol consumption, history of hypertension, or history of parental myocardial infarction before age 60, except for an interaction with past smoking in men, resulting in a statistically nonsignificant increased effect of migraine in this group. This finding was not replicated in women, however. Finally, when we classified women at baseline into current (within the past year) and previous migraineurs, no increased risk of major CHD was found in either group compared with those reporting no migraine (current: RR, 0.83; 95% CI, 0.49 to 1.39; previous: RR, 0.90; 95% CI, 0.48 to 1.71). DISCUSSION In this large prospective study conducted among the populations of two randomized trials, self-reported migraine was not associated with increased risk of major or total CHD among women or men without cardiovascular disease at baseline, even after adjust- Table 5. RRs for Major CHD and Migraine Status among Subgroups of Women in the WHS and Men in the PHS, Classified by Baseline Charactaeristics* Women Men Characteristic Cases RR (95% CI) Interaction P value Cases RR (95% CI) Interaction P value Age 60 years ( ) ( ) years ( ) ( ) Cigarette smoking Current ( ) ( ) 0.62 Past ( ) ( ) Never ( ) ( ) Alcohol use 1/week ( ) ( ) /week ( ) ( ) History of hypertension Yes ( ) ( ) 0.95 No ( ) ( ) Parental history of MI age 60 Yes ( ) ( ) 0.50 No ( ) ( ) * ref, reference group. Major CHD is defined as fatal coronary heart disease or nonfatal MI. The number of cases excludes those with missing values on any covariate. Hypertension is defined as self-reported systolic/diastolic blood pressure of 140/90 mmhg or higher, or physician diagnosis of hypertension (women) or current treatment with antihypertensive medication (men). Adjusted for age, randomized treatment assignment, body mass index, smoking, alcohol use, exercise, history of hypertension, elevated cholesterol, diabetes, and parental history of MI before age 60, excluding characteristic of interest. Models in women are adjusted for menopausal status and postmenopausal hormone use, and models for major CHD in men are also adjusted for history of angina (an exclusion criterion in the WHS).

11 724 September 2002 ing for coronary risk factors. Nor was migraine associated with risk of developing individual endpoints, including nonfatal MI, fatal CHD, or angina. Previous studies examining the relationship between migraine and CHD have yielded conflicting results. Several cross-sectional studies found significant associations between migraine and CHD. In the Blue Mountains Eye Study, Mitchell et al. 11 found an increased prevalence of vascular disease history, including MI and angina, among those with a lifetime past history of typical migraine. In the Atherosclerosis Risk in Communities Study, Rose et al. 12 found an increased prevalence of angina among men and women with a lifetime history of migraine. In an elderly cohort, Cook et al. 14 found an increased prevalence of history of MI and angina associated with symptoms of migraine. Some investigators have speculated that these associations may be due to Berkson s bias or to the increased surveillance and diagnosis of patients with symptoms of either migraine or CHD. 39 One of the largest assessments of migraine and incident CHD was conducted in a retrospective cohort of 79,588 men and women enrolled in the Kaiser Permanente health maintenance organization in California. 15 Participants underwent comprehensive preventive medical examinations between 1971 and 1973 and were followed until the earliest occurrence of hospitalization for MI, death, termination of enrollment, or termination of the study in December Questionnaires used to determine the frequency of both migraine and chest pain at baseline showed a strong and statistically significant cross-sectional correlation between migraine and chest pain among both men and women. However, after an average follow-up of 15 years, no overall association was observed between migraine and risk of Ml in men or in women. In a subgroup analysis, the investigators did find a statistically significant elevation in risk among women (but not men) with a family history of Ml whose migraine status was determined by self-reported medical diagnosis (RR, 2.4; 95% CI, 1.4 to 4.2). No elevation was observed among a similar group of women whose migraine status was determined by the presence or absence of symptoms. In contrast, in neither the WHS nor the PHS did we find effect modification of the relationship between migraine and CHD by family history of premature MI. In a cohort of 1,310 British women aged 45 years followed for almost 12 years, mortality rates were reduced among women reporting any headache (RR, 0.72; 95% CI, 0.52 to 1.00) and among women reporting headaches with 1 or more migrainous features (RR, 0.78; 95% CI, 0.57 to 1.08). 40 Because this study assessed migraine experienced over the past year rather than lifetime experience, the investigators speculated that disappearance of headache or migraine with age may be related to advancing arterial disease. Although we observed fewer CHD deaths among those reporting migraine in the WHS and PHS, these results are inconclusive because the number of fatal events was small. In addition, when we classified women at baseline as current (within the past year) or previous migraineurs, no effect on CHD risk was found in either group. In a previous study from the PHS, we reported that physicians reporting migraine had significantly increased risks of subsequent total stroke and ischemic stroke compared with those not reporting migraine, with RRs of 1.84 (95% CI, 1.06 to 3.20) for total stroke and 2.00 (95% CI, 1.10 to 3.64) for ischemic stroke after adjusting for age, randomized treatment assignment, and a number of cardiovascular risk factors. 6 A similar significant increase in risk of stroke was observed among individuals with migraine in a 10-year follow-up of more than 12,000 subjects in the first National Health and Nutrition Examination Survey. 7 The fact that we did not find a similar effect in analyses of CHD suggests that mechanisms unrelated to atherosclerosis may be involved. Clinical studies have found positive associations between migraine and MI primarily among patients who had an MI in the presence of normal coronary arteries, 23,24 suggesting that migraine may be related to a vasomotor component in the pathophysiology of MI rather than to atherosclerosis. Other studies relating migraine and Raynaud s phenomenon to vasospastic angina also raise the possibility that migraine may be related to a vasospastic disorder rather than to coronary artery disease itself. Although our data do not allow us to address these interesting hypotheses directly, these clinical data may provide some

12 Headache 725 basis for the differing effect of migraine on stroke and CHD. The strengths of our study are its size, its prospective design, and the relatively homogeneous nature of the cohorts that minimize confounding by several variables, including early symptom awareness, access to medical care, educational attainment, and socioeconomic status. At the same time, it has several potential limitations. The exposure data were based on self-reports, which can lead to misclassification. However, studies of health professionals have found self-reporting to be reliable for cardiovascular risk factors. 45 In addition, women health professionals and men physicians are likely to accurately report migraine, which is supported by our finding that of the 5,174 WHS participants who reported having had 1 or more migraine attacks in the previous year, 74% satisfied all or all but 1 of the modified IHS criteria for a diagnosis of migraine. We cannot exclude, however, the possibility of under-diagnosis of migraine in our study populations. In addition, these participants in randomized trials may be generally healthier than the general population, suggesting that the prevalence of migraine in these studies may not be generalizable. Another possible limitation is that we had no information on use of antimigraine drugs such as betablockers and some calcium-channel blockers. The effect of these medications on angina and blood pressure could partially explain the lack of association we found between migraine and CHD. However, because most migraineurs would not be on preventative medications, the concurrent medication use would probably not have a significant impact on these results. In addition, our study populations were limited to those aged 40 years and over in men and 45 and over in women. Some data suggest that migraine is a potent risk factor for ischemic stroke in women younger than 45 years of age but not those in older age groups In conclusion, although several cross-sectional studies have shown a positive association between migraine and history of angina or CHD, these results are not supported by data from two large, prospective cohorts. Our results indicate that migraine is not associated with increased risk of subsequent major or total CHD in women or in men. Acknowledgments: The authors would like to acknowledge the crucial contributions of the entire staff of the WHS, under the leadership of David Gordon, and of the PHS, under the leadership of Charlene Belanger. We would also like to acknowledge the dedicated participation and commitment of the 39,876 participants of the WHS and the 22,071 participants of the PHS. This work was supported by grants HL26490, HL34595, HL43851, CA34944, CA40360, CA47988, and NS34108 from the National Institutes of Health (Bethesda, MD). I.M.B. ( ) and P.A.L. ( ) are recipients of fellowships from Fundação de Amparo a Pesquisa do Estado de São Paulo (São Paulo, Brazil). REFERENCES 1. Stewart WF, Lipton RB, Celentano DD, Reed ML. Prevalence of migraine headache in the United States. Relation to age, income, race, and other sociodemographic factors. JAMA. 1992;267: Lipton RB, Stewart WF. Migraine in the United States: a review of epidemiology and health care use. Neurology. 1993;43(6 suppl 3):S6-S Rozen TD, Swanson JW, Stang PE, McDonnell SK, Rocca WA. Increasing incidence of medically recognized migraine headache in a United States population. Neurology. 1999;53: Janeway TC. A clinical study of hypertensive cardiovascular disease. Arch Intern Med. 1913;12: Broderick JP, Swanson JW. Migraine-related strokes. Clinical profile and prognosis in 20 patients. Arch Neurol. 1987;44: Buring JE, Hebert P, Romero J, et al. Migraine and subsequent risk of stroke in the Physicians Health Study. Arch Neurol. 1995;52: Merikangas KR, Fenton BT, Cheng SH, Stolar MJ, Risch N. Association between migraine and stroke in a large-scale epidemiological study of the United States. Arch Neurol. 1997;54: Lafitte C, Even C, Henry-Lebras F, de Toffol B, Autret A. Migraine and angina pectoris by coronary artery spasm. Headache. 1996;36: Wayne VS. A possible relationship between migraine and coronary artery spasm. Aust N Z J Med. 1986;16: Fournier JA, Fernandez-Cortacero JA, Granado C, Gascon D. Familial migraine and coronary artery spasm in two siblings. Clin Cardiol. 1986;9: Mitchell P, Wang JJ, Currie J, Cumming RG, Smith W.

13 726 September 2002 Prevalence and vascular associations with migraine in older Australians. Aust N Z J Med. 1998;28: Rose KM, Paton CC, Brown A, et al. The association of migraine headaches with angina in the Atherosclerosis Risk in Communities Study [abstract]. 1998;97: Franceschi M, Colombo B, Rossi P, Canal N. Headache in a population-based elderly cohort. An ancillary study to the Italian Longitudinal Study of Aging (ILSA). Headache. 1997;37: Cook NR, Evans DA, Funkenstein HH, et al. Correlates of headache in a population-based cohort of elderly. Arch Neurol. 1989;46: Sternfeld B, Stang P, Sidney S. Relationship of migraine headaches to experience of chest pain and subsequent risk for myocardial infarction. Neurology. 1995;45: Fiermonte G, Annulli A, Pierelli F. Transcranial Doppler evaluation of cerebral hemodynamics in migraineurs during prophylactic treatment with flunarizine. Cephalalgia. 1999;19: Fiermonte G, Pierelli F, Pauri F, Cosentino FI, Soccorsi R, Giacomini P. Cerebrovascular CO 2 reactivity in migraine with aura and without aura. A transcranial Doppler study. Acta Neurol Scand. 1995;92: Sakai F, Meyer JS. Abnormal cerebrovascular reactivity in patients with migraine and cluster headache. Headache. 1979;19: Hanington E, Jones RJ, Amess JA, Wachowicz B. Migraine: a platelet disorder. Lancet. 1981;2: Tousoulis D, Davies G, McFadden E, Clarke J, Kaski JC, Maseri A. Coronary vasomotor effects of serotonin in patients with angina. Relation to coronary stenosis morphology. Circulation.1993;88: Wright L, Homans DC, Laxson DD, Dai XZ, Bache RJ. Effect of serotonin and thromboxane A 2 on blood flow through moderately well developed coronary collateral vessels. J Am Coll Cardiol. 1992;19: Vikenes K, Farstad M, Nordrehaug JE. Serotonin is associated with coronary artery disease and cardiac events. Circulation. 1999;100: Ammann P, Marschall S, Kraus M, et al. Characteristics and prognosis of myocardial infarction in patients with normal coronary arteries. Chest. 2000; 117: Ciraulo DA, Bresnahan GF, Frankel PS, Isely PE, Zimmerman WR, Chesne RB. Transmural myocardial infarction with normal coronary angiograms and with single vessel coronary obstruction. Clinicalangiographic features and five-year follow-up. Chest. 1983;83: Maassen Van Den Brink A, Reekers M, Bax WA, Ferrari MD, Saxena PR. Coronary side-effect potential of current and prospective antimigraine drugs. Circulation. 1998;98: Kelly KM. Cardiac arrest following use of sumatriptan. Neurology. 1995;45: Ottervanger JP, Paalman HJ, Boxma GL, Stricker BH. Transmural myocardial infarction with sumatriptan. Lancet. 1993;341: Mueller L, Gallagher RM, Ciervo CA. Vasospasminduced myocardial infarction with sumatriptan. Headache. 1996;36: O Connor P, Gladstone P. Oral sumatriptan-associated transmural myocardial infarction. Neurology. 1995;45: Buring JE, Hennekens CH for the Women s Health Study Research Group. The Women s Health Study: Rationale, background and summary of the study design. J Myocardial Ischemia. 1992;4:27-29 and Rexrode KM, Lee IM, Cook NR, Hennekens CH, Buring JE. Baseline characteristics of participants in the Women s Health Study. J Womens Health Gend Based Med. 2000;9: Manson JE, Buring JE, Satterfield S, Hennekens CH. Baseline characteristics of participants in the Physicians Health Study: a randomized trial of aspirin and beta-carotene in U.S. physicians. Am J Prev Med. 1991;7: Findings from the aspirin component of the ongoing Physicians Health Study. N Engl J Med. 1988;318: Headache Classification Committee of the International Headache Society. Classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain. Cephalalgia. 1988;8(suppl 7): Rose GA, Blackburn H, Gillum RR, Prineas RJ. Cardiovascular Survey Methods. Monograph series no. 56. Geneva, Switzerland: World Health Organization; Cox D. Regression models and life tables. J Roy Stat Soc. 1972;34: Chen TC, Leviton A, Edelstein S, Ellenberg JH. Migraine and other diseases in women of reproductive age. The influence of smoking on observed associations. Arch Neurol. 1987;44:

14 Headache Volans GN, Castleden CM. The relationship between smoking and migraine. Postgrad Med J. 1976; 52: Peroutka SJ, Price SC, Jones KW. The comorbid association of migraine with osteoarthritis and hypertension: complement C3F and Berkson s bias. Cephalalgia. 1997;17: Waters WE, Campbell MJ, Elwood PC. Migraine, headache, and survival in women. Br Med J. 1983; 287: Nakamura Y, Shinozaki N, Hirasawa M, et al. Prevalence of migraine and Raynaud s phenomenon in Japanese patients with vasospastic angina. Jpn Circ J. 2000; 64: Miller D, Waters DD, Warnica W, Szlachcic J, Kreeft J, Theroux P. Is variant angina the coronary manifestation of a generalized vasospastic disorder? N Engl J Med. 1981;304: Heupler FA, Jr. Syndrome of symptomatic coronary arterial spasm with nearly normal coronary arteriograms. Am J Cardiol. 1980;45: Robertson D, Oates JA. Variant angina and Raynaud s phenomenon. Lancet. 1978;1: Colditz GA, Martin P, Stampfer MJ, et al. Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women. Am J Epidemiol. 1986;123: Tzourio C, Iglesias S, Hubert JB, et al. Migraine and risk of ischaemic stroke: a case-control study. BMJ. 1993;307: Tzourio C, Tehindrazanarivelo A, Iglesias S, et al. Case-control study of migraine and risk of ischaemic stroke in young women. BMJ. 1995;310: Chang CL, Donaghy M, Poulter N. Migraine and stroke in young women: case-control study. The World Health Organisation Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception. BMJ. 1999;318:13-18.