Migraine,weight gain and the risk of becoming overweight and obese: A prospective cohort study

Abstract

Background:

Some cross-sectional studies have suggested an association between migraine and increased body weight. However, prospective data on the association are lacking.

Methods:

We conducted a prospective cohort study among 19,162 participants in the Women’s Health Study who had a body mass index (BMI) of 18.5– <25 kg/m² at baseline. Migraine was self-reported by standardized questionnaires. Main outcome measures were incident overweight (BMI ≥ 25 kg/m²), incident obesity (BMI ≥ 30 kg/m²) and mean weight change. Age- and multivariable-adjusted hazard ratios (HRs) were calculated for the association between migraine and incident overweight and obesity. Differences in weight change were evaluated by analysis of covariance (ANCOVA).

Results:

A total of 3,483 (18.2%) women reported any migraine history. After 12.9 years of follow-up, 7916 incident overweight and 730 incident obesity cases occurred. Migraineurs had multivariable-adjusted HRs (95% confidence interval) of 1.11 (1.05–1.17) for becoming overweight and 1.00 (0.83–1.19) for becoming obese. These associations remained stable after censoring for chronic diseases and were similar according to migraine aura status. Multivariable-adjusted mean weight change from baseline to the end of study was +4.7 kg for migraineurs and +4.4 kg for women without migraine (p = 0.02).

Conclusion:

Results of this large prospective study of middle-aged women do not indicate a consistent association between migraine and incident overweight, obesity or relevant weight gain.

Keywords

Migraine body mass index overweight obesity prospective study

Introduction

Migraine and obesity are both highly prevalent conditions that cause substantial individual and societal burden (1,2). Several studies have evaluated the potential role of obesity on episodic and chronic migraine, migraine attack frequency and migraine features (3). Plausible shared pathophysiological mechanisms have also been proposed to link these two entities (4,5).

Small clinical studies reported a high prevalence of episodic migraine among morbidly obese patients (6,7). However, evidence from population-based cross-sectional studies are inconclusive. Some studies report a positive association between body mass index (BMI) and migraine prevalence (8 –12), while others could not confirm theses results (13 –17). In addition, results of two large population-based studies suggest that obesity is associated with migraine attack frequency and migraine features including photophobia and phonophobia (13,16). Furthermore, obesity is considered a risk factor for disease progression (18,19).

In most previous studies, obesity has been hypothesized as a risk factor of migraine (10,11,13,16,18,20). However, the direction of any association remains unclear in cross-sectional studies. Migraine is a chronic condition that may lead to certain lifestyle modifications. Although some clinical studies report a beneficial effect of physical exercise on migraine (21), many migraineurs with frequent and/or disabling attacks may not be able to adhere to a regular exercise routine. Furthermore, migraine is associated with a variety of comorbid conditions and some migraine-related comorbidities, such as depression, are associated with weight gain and obesity (22). Considering that migraineurs may have differences in lifestyle and a greater risk of certain obesity-related comorbidities, one can envision migraine as a risk factor for becoming obese. Prospective data that can address the temporal issue in the association between migraine, weight gain and incident obesity are lacking.

We therefore conducted a prospective cohort study among female health professionals in the Women’s Health Study (WHS) to examine whether having migraine is a risk factor for becoming overweight or obese.

Methods

Study population

The WHS was a randomized, placebo-controlled trial designed to test the benefits and risks of low-dose aspirin and vitamin E in the primary prevention of cardiovascular disease and cancer among apparently healthy women. The design, methods and results have been described in detail previously (23,24). Briefly, a total of 39,876 US female health care professionals aged 45 years or older at study entry (1992–1995) and without a history of cardiovascular disease, cancer or other major illnesses were randomly assigned to receive aspirin (100 mg on alternate days), active vitamin E (600 IU on alternate days), both active agents or both placebos. All participants provided written informed consent, and the institutional review board of the Brigham and Women’s Hospital, Boston, MA, approved the WHS.

Baseline information was self-reported and collected by mailed questionnaires that asked about many cardiovascular risk factors and lifestyle variables. Every six months in the first year and yearly thereafter, participants were sent follow-up questionnaires asking about study outcomes and other information during the study period. For this analysis, we included follow-up information from the time of randomization through March 2004 (12.9 years of follow-up) of women with a baseline BMI between 18.5 and <25 kg/m².

Assessment of migraine

On the baseline questionnaire, participants were asked: “Have you ever had migraine headaches?” and “In the past year, have you had migraine headaches?” Women who experienced migraine headache in the past year were further asked details about their migraine attacks, including attack duration of 4–72 hours; unilateral location and pulsating quality of pain; inhibition of daily activities; aggravation by routine physical activity; nausea or vomiting; and sensitivity to light or sound. Furthermore, participants who reported migraine during the past year were asked whether they had an “aura or any indication a migraine is coming” as well as details about their attack frequency (daily, weekly, monthly, every other month, <6 times/year). Based on this information, we categorized women into “women without migraine history”, “women who reported any migraine history”, “women who reported active migraine” (migraine in the year prior completing the baseline questionnaire), and “women who reported prior migraine”, which includes women who reported ever having had a migraine but none in the year prior to completing the questionnaire. Women with “active migraine” were further classified into “active migraine with aura” and “active migraine without aura”, similar to previous studies (25,26). Our “any migraine” category included women with “active migraine” (migraine with aura, migraine without aura) and “prior migraine”.

In previous studies of the WHS (25,27), we have shown good agreement between our migraine classification and modified 1988 diagnostic criteria of the International Headache Society (IHS) (28). Furthermore, another study using a subsample of the WHS showed excellent agreement between self-reported migraine and migraine classification according to the 2004 IHS diagnostic criteria. Over 87% of women with active migraine could be diagnosed as migraine without aura (71.5%) or probable migraine without aura (16.2%) when applying the 2004 diagnostic criteria (29).

Assessment of covariates

Participants were asked to report age, race, smoking status, exercise, menopausal status, postmenopausal hormone use, history of hypertension, history of hypercholesterolemia and a history of diabetes on the baseline questionnaire. Information on depression has been assessed during follow-up. On the 48-month follow-up questionnaire, women were asked to report whether they had ever been diagnosed with depression.

A total of 39,310 of the 39,876 randomized participants completed a 131-item, self-administered semiquantitative food-frequency questionnaire at baseline. For each food item, a commonly used unit or portion size was specified and participants indicated on a 9-point scale, ranging from “never” to “six or more times per year”, how often they had consumed that specific amount, on average, during the last year. Nutrient scores were calculated by multiplying the intake frequency of each unit of food by the nutrient content of the specific portion size according to food composition tables from the Harvard School of Public Health, Boston, Massachusetts. Intake of each nutrient was further adjusted for total energy intake using the residual method. The semiquantitative food-frequency questionnaire has demonstrated reasonable validity and reliability as a measure of long-term average dietary intake (30). We defined quintiles of food intake (fruit and vegetables, whole grain, refined grains, red meat, low-fat dairy products, high-fat dairy products) and nutrient intake (total fat, saturated fat, monounsaturated fat, polyunsaturated fat, carbohydrates, protein, fiber).

Outcome ascertainment

Participants were asked to report their height and weight on the baseline questionnaire and information on weight was updated on 2-, 3-, 5-, 6-, 9-, 11-. 12- and 13-year follow-up questionnaires. Based upon this information, we calculated BMI values at baseline and at each of the eight follow-up time points. A BMI of 18.5– <25 kg/m² was considered normal, a BMI of 25– <30 kg/m² was defined as overweight and a BMI ≥ 30 kg/m² as obese. Furthermore, we calculated the weight change between baseline and each follow-up time point.

Consistent with a previous study in the WHS (31), we defined women with a normal BMI at baseline who reported a BMI of ≥25 kg/m² at any follow-up point as incident cases of becoming overweight or obese. To estimate the time when the BMI crossed the cutoff point, a regression line was modeled from the last reported BMI of <25 kg/m² to the first reported BMI of ≥25 kg/m² during follow-up. Using the same approach, we defined incident obesity using a BMI cutoff point of 30 kg/m².

Person-time was calculated from the date of randomization until the time when a person became overweight or obese, or the latest date when a BMI < 25 kg/m² (or <30 kg/m² for analyses of incident obesity) was reported, whichever occurred first. Women who developed diabetes during follow-up prior to becoming overweight or obese were censored on the date of diabetes diagnosis, as diabetes management typically includes weight control.

Statistical analysis

Of the 39,876 randomized participants, we excluded 20,313 women with a BMI ≥ 25 kg/m² or <18.5 kg/m² at baseline, 128 women with no updated body weight during the entire follow-up period, 194 women with baseline diabetes, 53 women with missing information on migraine and 26 women who reported cardiovascular disease or cancer events prior to receiving the baseline questionnaire, leaving a total of 19,162 women for this analysis.

We compared baseline characteristics with respect to migraine status using t-tests for continuous variables and chi-square tests for categorical variables. Cox proportional hazards models were used to evaluate the association between baseline migraine status and the risk of becoming overweight (BMI ≥ 25 kg/m²), as well as the risk of becoming obese (BMI ≥ 30 kg/m²). Age- and multivariable-adjusted hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs) were calculated. To determine the mean change in body weight from baseline to each follow-up point according to migraine status, age- and multivariable-adjusted analysis of covariance (ANCOVA) models were used.

We further used Cox proportional hazards models to evaluate the association between migraine frequency and the risk of becoming overweight or obese among 2407 women with active migraine who provided migraine frequency information. Because of low cell counts in the daily and weekly migraine frequency categories we defined the following attack frequency categories: at least weekly, monthly, every other month and less than six times/year (reference category).

The multivariable Cox proportional hazards models and the ANCOVA models were adjusted for age (continuous), race (white/other), randomized treatment assignments, baseline BMI (continuous), smoking status (never, past, current), exercise (never, < 1 time/week, 1–3 times/week, ≥4 times/week) alcohol consumption (never, 1–3 drinks/month, 1–6 drinks/week, ≥1 drink/day), history of hypertension (yes/no), history of hypercholesterolemia (yes/no), total energy intake (<1600 kcal/day, 1600–2400 kcal/day, >2400 kcal/day), postmenopausal status (premenopausal, postmenopausal, biologically uncertain, unclear), postmenopausal hormone use (never, past, current) and history of depression (yes/no).

Further adjustment for quintiles of food intake (fruit and vegetables, whole grain, refined grains, red meat, low-fat dairy products, high-fat dairy products) and nutrient intake (total fat, saturated fat, monounsaturated fat, polyunsaturated fat, carbohydrates, protein, fiber), oral contraceptive use, family history of myocardial infarction (MI), and caffeine intake did not substantially change the results and are therefore not included in the final models.

Because calorie intake and depression can also be viewed as a consequence of migraine we run separate multivariable models excluding these two covariates to prevent adjustment for potential intermediating factors.

To test the proportional hazards assumption, we included an interaction term for migraine status with the logarithm of time and found no statistically significant violation.

Effect modification was evaluated for baseline age (<55, 55– <65, 65– <75, ≥75 years), smoking status (never, past, current) exercise (never, <1 time/week, 1–3 times/week, ≥4 times/week) and alcohol consumption (never, 1–3 drinks/month, 1–6 drinks/week, ≥ 1 drink/day). We tested statistically significant effect modification by contrasting models with and without an interaction term indicator variable using the likelihood ratio test.

We performed a sensitivity analysis by censoring participants who developed myocardial MI, stroke or cancer during follow-up in the Cox proportional hazards models, as these diseases are potentially associated with subsequent weight change. For each ANCOVA model, we excluded women who developed MI, stroke or cancer during the specific time period analyzed.

A missing value indicator was incorporated into the outcome models for covariates if the number of missing women was greater or equal to 100. We assigned participants with missing values to the covariate reference category if the number of missing information was less than 100.

For all analysis, we used SAS (version 9.1; SAS Institute Inc., Cary, NC, USA). All p values were two-tailed and p < 0.05 was considered to be statistically significant.

Results

Among a total of 19,162 women having a BMI 18.5– <25 kg/m² at baseline, 3483 (18.2%) women reported any history of migraine, of whom 974 reported migraine aura. In Table 1, baseline characteristics of the participants according to migraine status are presented. Migraineurs had a mean age of 53.6 years and were younger compared to women without migraine. They were more likely to rarely/never drink alcohol, to never smoke cigarettes, to exercise less than once per week, to have a total calorie intake of ≥2400 kcal/day, to have a history of hypertension, a history of cholesterol ≥240 mg/dl and a history of depression.

Table 1.

Baseline characteristics according to migraine status (N = 19,162).

	No migraine history	Any migraine history	p value
	N = 15,679	N = 3,483	p value
Demographic information
Mean age, years (SD)	54.7 (7.2)	53.6 (6.5)	<0.01
Race, %
White	14,903 (95.7)	3288 (95.3)	0.26
Lifestyle factors
Mean baseline BMI, kg/m² (SD)	22.4 (1.6)	22.5 (1.6)	<0.01
Alcohol consumption, %			<0.01
Rarely/never	5910 (37.7)	1413 (40.6)
1–3 drinks/month	1965 (12.5)	532 (15.3)
1–6 drinks/week	5658 (36.1)	1151 (33.1)
≥1 drink/day	2141 (13.7)	387 (11.1)
Smoking status, %			0.06
Never	7862 (50.2)	1.823 (52.4)
Past	5612 (35.8)	1195 (34.4)
Current	2195 (14.0)	461 (13.3)
Exercise, %			<0.01
Rarely/never	5126 (32.7)	1144 (32.9)
<1/week	2871 (18.3)	729 (20.9)
1–3 times/week	5476 (34.9)	1157 (33.2)
≥4 times/week	2200 (14.0)	451 (12.7)
Total calorie intake (kcal/day)			0.04
<1600	7054 (46.3)	1516 (45.0)
1600– <2400	6684 (43.8)	1471 (43.7)
≥2400	1504 (9.9)	380 (11.3)
Other covariates, %
Postmenopausal status at baseline			<0.01
Premenopausal	4564 (29.2)	1002 (28.8)
Postmenopausal	8541 (54.6)	1763 (50.7)
Biological uncertain	1947 (12.4)	578 (16.6)
Unclear/subject unsure	604 (3.9)	135 (3.9)
Postmenopausal hormone use, %			<0.01
Never	7332 (46.8)	1462 (42.1)
Past	1396 (8.9)	328 (9.4)
Current	6925 (44.2)	1683 (48.5)
History of hypertension	2477 (15.8)	594 (17.1)	0.07
History of cholesterol ≥240 mg/dl	3849 (24.6)	949 (27.3)	<0.01
History of depression	1310 (8.7)	474 (14.2)	<0.01

Percentages may not add up to 100% because of rounding or missing values. BMI: body mass index; SD: standard deviation.

During 12.9 years of follow-up, 7916 initially normal-weight women became overweight (BMI ≥ 25 kg/m²) and 730 became obese (BMI ≥ 30 kg/m²). In Table 2, age- and multivariable-adjusted HRs for incident overweight are shown. Women who reported any migraine had a multivariable-adjusted HR of 1.11 (1.05–1.17) for becoming overweight. The associations did not differ according to migraine aura status. Compared with women without migraine, those with any history of migraine were not at increased risk of becoming obese (Table 3).

Table 2.

Age- and multivariable-adjusted* HRs (95% CI) for becoming overweight according to migraine status (N = 19,162).

	Total number	Incident overweight (BMI ≥ 25 kg/m²)	Age-adjusted HR (95% CI)	Multivariable-adjusted HR (95% CI)
			HR (95% CI)	HR (95% CI)
Any migraine	3483	1617	1.18 (1.12, 1.25)	1.11 (1.05, 1.17)
No migraine	15,679	6299	1.00	1.00
Migraine subtypes
Migraine with aura	974	459	1.18 (1.07, 1.30)	1.10 (1.00, 1.21)
Migraine without aura	1464	686	1.18 (1.09, 1.28)	1.13 (1.05, 1.23)
Prior migraine	1045	472	1.18 (1.08, 1.30)	1.07 (0.98, 1.18)
No migraine	15,679	6299	1.00	1.00

Multivariable models were adjusted for age, race, randomized treatment assignments, baseline BMI, total calorie intake, alcohol consumption, exercise, smoking status, postmenopausal hormone use, postmenopausal status, history of cholesterol >240 mg/dl, history of hypertension and history of depression.

HR: hazard ratio; BMI: body mass index; CI: confidence interval.

Table 3.

Age- and multivariable-adjusted* HRs (95% CI) for becoming obese according to migraine status (N = 19,162).

	Total number	Incident obesity (BMI ≥ 30)	Age-adjusted HR (95% CI)	Multivariable-adjusted HR (95% CI)
			HR (95% CI)	HR (95% CI)
Any migraine	3483	155	1.16 (0.97, 1.39)	1.00 (0.83, 1.19)
No migraine	15,679	575	1.00	1.00
Migraine subtypes
Migraine with aura	974	46	1.20 (0.89, 1.62)	1.02 (0.76, 1.38)
Migraine without aura	1464	75	1.29 (1.01, 1.64)	1.11 (0.87, 1.41)
Prior migraine	1045	34	0.92 (0.65, 1.31)	0.79 (0.56, 1.12)
No migraine	15,679	575	1.00	1.00

Multivariable models were adjusted for age, race, randomized treatment assignments, baseline BMI, total calorie intake, alcohol consumption, exercise, smoking status, postmenopausal hormone use, postmenopausal status, history of cholesterol ≥240 mg/dl, history of hypertension and history of depression.

HR: hazard ratio; BMI: body mass index; CI: confidence interval.

In sensitivity analyses that censored participants with cardiovascular disease and cancer during follow-up, the associations between any migraine and incident overweight and obesity remained stable in multivariable-adjusted models (HR incident overweight: 1.10, 95% CI: 1.04–1.16, HR incident obesity: 0.97, 95% CI: 0.81–1.17).

When we excluded calorie intake and depression from the multivariable model, the association between any migraine and becoming overweight (HR: 1.13; 95% CI: 1.07–1.19) or becoming obese (HR: 1.04; 95% CI: 0.87–1.25) remained very similar.

The associations between migraine and incident overweight and obesity were not modified by age, smoking, alcohol consumption or postmenopausal status (all p values for interaction >0.09). The relationship between migraine and incident overweight appeared to be significantly modified by exercise (p interaction = 0.03), indicating that women with migraine who exercised less than once per week, but not those who exercised more, may have an increased risk of becoming overweight (multivariable-adjusted HR of 1.17, 95% CI: 1.09–1.26).

In Table 4, age- and multivariable-adjusted mean weight change according to migraine status is presented. During follow-up, both migraineurs and women without migraine constantly gained weight. After 12.9 years of follow-up, women who reported migraine at baseline had a multivariable-adjusted mean weight change of 4.65 kg, and women without migraine had a mean weight change of 4.36 kg (p = 0.02). In sensitivity analyses, the differences in mean weight change between migraineurs and women without migraine attenuated and lost statistical significance in multivariable-adjusted models for most follow-up time points (data not presented).

Table 4.

Age- and multivariable-adjusted* mean weight change (kg) according to migraine status (N = 19,162).

Follow-up duration	No migraine (N = 15,679)	Any migraine (N = 3,483)	p value
2 years (N = 18,942)	N = 15,504	N = 3438
Age-adjusted	1.34	1.44	0.15
Multivariable-adjusted	1.70	1.72	0.74
3 years (N = 18,045)	N = 14,755	N = 3290
Age-adjusted	1.76	1.95	0.01
Multivariable-adjusted	1.99	2.09	0.23
5 years (N = 18,012)	N = 14,753	N = 3259
Age-adjusted	2.44	2.80	<0.01
Multivariable-adjusted	3.13	3.35	0.01
6 years (N = 17,724)	N = 14,521	N = 3203
Age-adjusted	2.73	3.12	<0.01
Multivariable-adjusted	3.19	3.43	0.01
9 years (N = 15,297)	N = 12,526	N = 2771
Age-adjusted	3.16	3.56	<0.01
Multivariable-adjusted	4.13	4.34	0.06
11 years (N = 16,274)	N = 13,319	N = 2955
Age-adjusted	3.17	3.62	<0.01
Multivariable-adjusted	4.05	4.30	0.04
12 years (N = 15,945)	N = 13,031	N = 2914
Age-adjusted	3.24	3.75	<0.01
Multivariable-adjusted	4.04	4.36	0.01
13 years (N = 15,590)	N = 12,737	N = 2853
Age-adjusted	3.26	3.74	<0.01
Multivariable-adjusted	4.36	4.65	0.02

In Table 5, the associations between migraine frequency and incident overweight and obesity are presented. Compared to women with a migraine attack frequency of less than six times/year, migraineurs who reported an attack frequency of every other month, monthly or at least weekly had multivariable-adjusted HRs of 1.31 (0.74–2.31), 0.95 (0.59–1.53) and 0.61 (0.24–1.53), respectively, for becoming obese.

Table 5.

Age- and multivariable-adjusted HRs (95% CI) for becoming overweight or obese according to migraine frequency (N = 2,407).

Migraine frequency	Total number	Incident overweight (BMI ≥ 25)	Age-adjusted	Multivariable-adjusted
Daily + weekly	131	60	0.95 (0.73, 1.23)	0.97 (0.75, 1.27)
Monthly	496	224	0.93 (0.80, 1.08)	1.00 (0.86, 1.17)
Every other month	249	127	1.10 (0.91, 1.33)	1.34 (1.11, 1.63)
<6 times/year	1531	720	1.00	1.00
Migraine frequency	Total number	Incident obesity (BMI ≥ 30)	Age-adjusted	Multivariable-adjusted
Daily + weekly	131	5	0.80 (0.32, 1.98)	0.61 (0.24, 1.53)
Monthly	496	23	0.90 (0.57, 1.44)	0.95 (0.59, 1.53)
Every other month	249	15	1.19 (0.68, 2.07)	1.31 (0.74, 2.31)
<6 times/year	1531	75	1.00	1.00

Multivariable models were adjusted for age, race, randomized treatment assignments, baseline BMI, total calorie intake, alcohol consumption, exercise, smoking status, postmenopausal hormone use, postmenopausal status, history of cholestero l >240 mg/dl, history of hypertension and history of depression.

HR: hazard ratio; BMI: body mass index; CI: confidence interval.

Discussion

In this large prospective study of female health professionals aged 45 years or older at study entry, we did not find meaningful associations between any history of migraine and an increased risk for becoming overweight or obese. The associations did not differ according to migraine aura status. Both women with and without migraine gained weight, but the magnitude of weight change did not substantially differ between both groups after 12.9 years of follow-up. The differences in weight change were attenuated after excluding women who were diagnosed with chronic diseases during follow-up.

The potential relationship between episodic migraine and obesity has been studied frequently, with inconsistent results. Several cross-sectional studies are in-line with our finding of no association between migraine and obesity (13 –17,32). In a cross-sectional study among 31,865 Danish twin individuals, a BMI ≥ 30 kg/m² was not associated with migraine (OR: 0.98, 95% CI: 0.87–1.10), migraine with aura (OR: 0.92, 95% CI: 0.77–1.10) or migraine without aura (OR: 1.02, 95% CI: 0.88–1.17) after multivariable adjustment. One potential limitation of that large population-based study is that migraine diagnosis and BMI calculations were based upon self-reported information and misclassification is possible (14). Another cross-sectional study using measured BMI indices as well as a migraine diagnosis based on neurologist interviews found no relationship between BMI and migraine prevalence among 684 women aged 40–74 years. Migraine was divided into active (migraine during the past year) and inactive migraine (migraine in the past but not in the year preceding the interview), and BMI was not significantly associated with either active (OR: 1.1, 95% CI: 0.6–1.8) or inactive migraine (OR: 1.0, 95% CI: 0.6–1.7) (15).

Contrary to our results, several studies found modest associations between BMI and migraine (8,11,12). Windsvold et al. evaluated the association between migraine and cardiovascular risk factors among 48,713 participants of the HUNT study. Migraine was defined according to modified IHS criteria, and BMI was calculated using measured height and weight. Compared to a BMI < 25 kg/m², a BMI ≥ 30 kg/m² was associated with a multivariable-adjusted OR of 1.17 (1.04–1.32) for migraine without aura and an OR of 1.32 (1.04–1.32) for migraine with aura after adjusting for age, gender and education level (12). A cross-sectional study evaluating data from the National Health and Nutrition Survey 1999–2002 found an association between BMI and severe headache or migraine. Among 7601 participants aged ≥20 years, a BMI of ≥30 kg/m² was associated with an increased multivariable-adjusted OR of 1.37 (1.09–1.72). Information on measured BMI indices was available in this study, but migraine was not defined according to IHS criteria and was combined with severe headaches (8).

Only few studies have specifically evaluated the relationship between migraine and weight gain. Recently, Vo et al. evaluated the association between BMI, weight gain and migraine among 3733 participants of the Omega study, who were interviewed during early pregnancy. BMI was estimated using pregravid weight. A pregravid BMI ≥ 30 kg/m² was associated with a multivariable-adjusted OR of 1.28 (1.12, 1.96) for adult migraine. Women with pediatric migraine had an OR of 1.67 (1.13–2.47) for adult weight gain of ≥10 kg in this study. In our study, migraine diagnosis was associated with a slightly increased mean weight change, but we could not observe an association between migraine and obesity. The study population and design of the Omega study differ from our study; study participants were younger and of reproductive age, compared to our study population (9).

Age, gender and postmenopausal status are important factors that may contribute to inconsistent results across studies. In a cross-sectional study of 21,783 participants of the National Health and Nutrition Examination Survey 1999–2004, Peterlin et al. reported an association between severe headache or migraine and total body obesity for those aged ≤55, but not >55 years (11). In an elderly Brazilian population, Bensenor et al. evaluated the association between cardiovascular risk factors and migraine, and found that mean BMI values did not differ according to migraine status (32). The lack of association between migraine and incident obesity we have observed in our population could be due to the fact that participants in our study were 45 years and older at study entry and a high percentage of participants was already postmenopausal. However, prospective data in younger populations evaluating the association between migraine and incident obesity are lacking.

Several studies evaluated the association between BMI and migraine attack frequency (13,15,16,20). Two cross-sectional studies suggested a J-shaped association between BMI and migraine frequency (16,20). A BMI of ≥35 kg/m² was associated with the highest effect estimates for a daily migraine frequency in a previous study using WHS data (OR 3.11, 95% CI: 1.12–8.67) and very frequent headache in the American Migraine Prevalence and Prevention study (OR: 1.74, 95% CI: 1.41–1.93) (16,20). In contrast, a cross-sectional study from Sweden found no association between BMI and migraine frequency (15). But these findings do not easily compare with our WHS data, because women with a BMI of 18.5– < 25 kg/m² at baseline were included in our study and we evaluated whether women with a high attack frequency are at increased risk for becoming overweight and/or obese. Our data suggest that among the lean, a high migraine attack frequency is associated with a decreased risk for becoming obese.

Our study has several strengths. We followed a large number of women with long-term follow-up of 12.9 years. In addition, migraine cases were defined according to diagnostic criteria of the IHS and we were able to distinguish migraine subtypes. In addition, comprehensive information on potential confounders was available for our analyses. Furthermore, we were able to evaluate the association between migraine and incident overweight and obesity in a population free of prevalent cardiovascular disease, cancer and other chronic conditions at baseline, which may be associated with weight change. The following limitations should also be considered when interpreting our results. First, information on migraine was self-reported and misclassification is possible. However, we have shown excellent agreement between self-reported migraine and the 2004 diagnostic criteria in a previously published study using WHS data (29). Second, we cannot rule out misclassification of BMI due to the self-reported height and weight information. However, our cohort consists of health professionals known to accurately report health information (33). Specifically, self-reported and measured weight information was highly correlated in (r = 0.97) in a comparable cohort of female health professionals (34). Third, no information on migraine-specific treatment known to be associated with weight gain or weight loss, such as topiramate and calcium-channel blockers, was available. However, because both drugs are used for preventive treatment of migraine, the effect of these drugs would probably not have a large impact on our results. Finally, our study population consists of female health professionals aged 45 years and older at study entry, characteristics which may limit the generalizability of our results to other populations.

In conclusion, in this large prospective study, we did not observe a meaningful association between migraine and becoming overweight or obese, nor did migraineurs have a relevant weight gain.

Footnotes

Funding

The Women’s Health Study is supported by grants from the National Heart, Lung, and Blood Institute (HL-043851 and HL-080467); and the National Cancer Institute (CA-47988). The sponsors of the study played no role in design and conduct of the study, including collection, management, analysis and interpretation of the data; and preparation, review or approval of the manuscript.

Acknowledgements

We are indebted to the participants in the Women’s Health Study for their outstanding commitment and cooperation, and to the entire Women’s Health Study staff for their expert and unfailing assistance.

Conflicts of interest statement

We report a full disclosure for the last two years for each author:

Dr Winter has received an international postdoctoral research fellowship of the American Association of University Women and a research fellowship of the German Research Foundation (DFG).

Dr Wang has nothing to disclose.

Dr Buring has received investigator-initiated research funding and support from the National Institutes of Health and Dow Corning Corporation; research support for pills and/or packaging from Bayer Health Care and the Natural Source Vitamin E Association.

Dr Sesso has received investigator-initiated research funding as principal investigator from not-for-profit entities including the National Institutes of Health, the California Strawberry Commission and the Tomato Products Wellness Council. Dr Sesso has also received investigator-initiated research funding as principal investigator from for-profit entities, including Cambridge Theranostics, Ltd. Dr Sesso has also served as a consultant to Iovate Health Sciences USA, Inc.

Dr Kurth has received investigator-initiated research funding from the French National Research Agency, the US National Institutes of Health, Merck, the Migraine Research Foundation and the Parkinson’s Disease Foundation. He is a consultant to World Health Information Science Consultants, LLC, and has received honoraria from Allergan, the American Academy of Neurology and Merck for educational lectures, and from MAP Pharmaceutical for contributing to a scientific advisory panel, and from the BMJ for editorial services.

References

Blumenfeld

Varon

Wilcox

. Disability, HRQoL and resource use among chronic and episodic migraineurs: results from the International Burden of Migraine Study (IBMS). Cephalalgia 2011; 31: 301–315.

Withrow

Alter

. The economic burden of obesity worldwide: a systematic review of the direct costs of obesity. Obes Rev 2011; 12: 131–141.

Evans

Williams

Rapoport

Peterlin

. The association of obesity with episodic and chronic migraine. Headache 2012; 52(4): 663–6671.

Bigal

Lipton

Holland

Goadsby

. Obesity, migraine, and chronic migraine: possible mechanisms of interaction. Neurology 2007; 68: 1851–1861.

Peterlin

Rapoport

Kurth

. Migraine and obesity: epidemiology, mechanisms, and implications. Headache 2009; 50: 631–648.

Horev

Wirguin

Lantsberg

Ifergane

. A high incidence of migraine with aura among morbidly obese women. Headache 2005; 45: 936–938.

Peres

Lerario

Garrido

Zukerman

. Primary headaches in obese patients. Arq Neuropsiquiatr 2005; 63: 931–933.

Ford

Pearson

. Body mass index and headaches: findings from a national sample of US adults. Cephalalgia 2008; 28: 1270–1276.

Ainalem

Qiu

. Body mass index and adult weight gain among reproductive age women with migraine. Headache 2011; 51: 559–569.

10.

Keith

Wang

Fontaine

. BMI and headache among women: results from 11 epidemiologic datasets. Obesity (Silver Spring) 2008; 16: 377–383.

11.

Peterlin

Rosso

Rapoport

Scher

. Obesity and migraine: the effect of age, gender and adipose tissue distribution. Headache 2009; 50: 52–62.

12.

Winsvold

Hagen

Aamodt

. Headache, migraine and cardiovascular risk factors: the HUNT study. Eur J Neurol 2010; 18: 504–511.

13.

Bigal

Liberman

Lipton

. Obesity and migraine: a population study. Neurology 2006; 66: 545–550.

14.

Tfelt-Hansen

Skytthe

. Association between migraine, lifestyle and socioeconomic factors: a population-based cross-sectional study. J Headache Pain 2011; 12: 157–172.

15.

Mattsson

. Migraine headache and obesity in women aged 40–74 years: a population-based study. Cephalalgia 2007; 27: 877–880.

16.

Winter

Berger

Buring

Kurth

. Body mass index, migraine, migraine frequency and migraine features in women. Cephalalgia 2009; 29: 269–278.

17.

Winter

Hoffmann

Meisinger

. Association between lifestyle factors and headache. J Headache Pain 2011; 12: 147–155.

18.

Bigal

Lipton

. Obesity is a risk factor for transformed migraine but not chronic tension-type headache. Neurology 2006; 67: 252–257.

19.

Scher

Stewart

Ricci

Lipton

. Factors associated with the onset and remission of chronic daily headache in a population-based study. Pain 2003; 106: 81–89.

20.

Bigal

Tsang

Loder

. Body mass index and episodic headaches: a population-based study. Arch Intern Med 2007; 167: 1964–1970.

21.

Busch

Gaul

. Exercise in migraine therapy—is there any evidence for efficacy? A critical review. Headache 2008; 48: 890–899.

22.

Pan

Sun

Czernichow

. Bidirectional association between depression and obesity in middle-aged and older women. Int J Obes (Lond) 2012; 36(4): 595–602.

23.

Rexrode

Lee

Cook

. Baseline characteristics of participants in the Women's Health Study. J Womens Health Gend Based Med 2000; 9: 19–27.

24.

Ridker

Cook

Lee

. A randomized trial of low-dose aspirin in the primary prevention of cardiovascular disease in women. N Engl J Med 2005; 352: 1293–1304.

25.

Kurth

Gaziano

Cook

. Migraine and risk of cardiovascular disease in women. JAMA 2006; 296: 283–291.

26.

Rist

Buring

Kase

. Migraine and functional outcome from ischemic cerebral events in women. Circulation 2010; 122: 2551–2557.

27.

Bensenor

Cook

Lee

. Low-dose aspirin for migraine prophylaxis in women. Cephalalgia 2001; 21: 175–183.

28.

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): 1–96.

29.

Schurks

Buring

Kurth

. Agreement of self-reported migraine with ICHD-II criteria in the Women's Health Study. Cephalalgia 2009; 29: 1086–1090.

30.

Willett

Sampson

Stampfer

. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 1985; 122: 51–65.

31.

Wang

Lee

Manson

. Alcohol consumption, weight gain, and risk of becoming overweight in middle-aged and older women. Arch Intern Med 2010; 170: 453–461.

32.

Bensenor

Goulart

Lotufo

. Cardiovascular risk factors associated with migraine among the elderly with a low income: the Sao Paulo Ageing & Health Study (SPAH). Cephalalgia 2010; 31: 331–337.

33.

Colditz

Martin

Stampfer

. Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women. Am J Epidemiol 1986; 123: 894–900.

34.

Rimm

Stampfer

Colditz

. Validity of self-reported waist and hip circumferences in men and women. Epidemiology 1990; 1: 466–473.