Abstract
The objective was to study the cross-sectional association between body mass index (BMI) and the prevalence of severe headaches or migraines in a national sample of US adults. We used data from 7601 men and women aged ≥ 20 years who participated in the National Health and Nutrition Examination Survey 1999-2002. The age-adjusted prevalence of severe headaches or migraines during the previous 3 months was 34.0, 18.9, 20.7 and 25.9± among participants with a BMI< 18.5, 18.5 to< 25, 25 to< 30 and ≥ 30 kg/m2, respectively. After adjusting for a variety of covariates in a logistic regression model, those with a BMI< 18.5 kg/m2 [odds ratio (OR) 2.01; 95± confidence interval (CI) 1.34, 3.02] or ≥ 30 kg/m2 (OR 1.37; 95± CI 1.09, 1.72) had a significantly elevated OR for having a headache compared with participants with a BMI of 18.5-< 25 kg/m2. BMI is associated with the prevalence of severe headaches or migraines in a non-linear manner.
Introduction
Headache is a common source of morbidity in the US population, and its burden is substantial. The prevalence of various types of headaches such as migraine ranges from 7 to 16% across countries (1). In the USA, its prevalence ranges from 6% in men to 18.2% in women (2). In 1998, the direct costs of migraine were estimated to be about $1 billion, and the indirect economic costs for people aged 20–64 years were estimated to be about $12 billion (3). In 2002, the cost of lost productive time due to headaches was estimated at almost $20 billion (4). In addition, headaches are a major determinant of impaired quality of life (5).
Because of the considerable burden imposed by headaches on populations in the USA and elsewhere, an understanding of potentially modifiable risk factors for this condition is key to helping to reduce the morbidity associated with it. In recent years, researchers have examined possible associations between obesity and headaches (6–10). Given the rapid increase in obesity in the USA, a possible link between obesity and headaches is of considerable interest and would suggest new avenues for reducing the public health burden of headaches. Therefore, the objective of this study was to examine the association between body mass index (BMI) and headaches in a national sample of US adults.
Materials and methods
In the National Health and Nutrition Examination Survey 1999–2002, a national sample was recruited using a multistage, stratified sampling design. The survey was designed to produce results representative of the civilian, non-institutionalized US population. The participants were interviewed at home and invited to attend a mobile examination centre, where they were asked to complete additional questionnaires, undergo various examinations and provide a blood sample. The study received human subjects' approval, and participants were asked to sign an informed consent form. Details about the survey may be found elsewhere (11).
Participants who responded affirmatively to the question ‘During the past 3 months, did you have severe headaches or migraines?’ were considered to have headaches. No additional information about the location of the headaches, their severity or possible accompanying symptoms was requested.
BMI (kg/m2) was calculated from measured height and weight and categorized into the following categories: < 18.5, 18.5 to < 25, 25 to < 30 and ≥ 30 kg/m2. Covariates included in the analyses were age, sex, race or ethnicity (White, African-American, Mexican-American, other), education (< high school, high school diploma or general equivalency diploma, > high school), marital status (married, widowed, divorced, separated, never married, living with partner), smoking status, alcohol consumption, physical activity, systolic blood pressure, concentrations of total cholesterol, concentrations of C-reactive protein, and diagnosed diabetes (yes, no). Current smokers were defined as participants who had smoked ≥ 100 cigarettes during their lifetime and were still smoking. Former smokers were defined as participants who had smoked ≥ 100 cigarettes during their lifetime but had stopped. Participants who had smoked < 100 cigarettes during their lifetime were classified as never having smoked. Alcohol consumption (g/day) was determined from a single 24-h dietary recall. Participants were asked questions about their participation in moderate and vigorous activities. The weekly time spent in each activity was multiplied by the metabolic equivalent (MET) value for that activity, yielding a MET-h index for each activity, and the product for each activity was summed over all activities. One MET corresponds to a metabolic rate of 3.5 ml of oxygen per kg body weight per minute. Up to four blood pressure measurements were obtained from each participant. We used the average of the last two measurements for participants who had three or four measurements, the second one for participants with only two measurements, and the only one for participants who had one measurement. Serum cholesterol was measured enzymatically on a Hitachi 717 Analyzer (Boehringer Mannheim Diagnostics, Indianapolis, IN, USA) using commercial reagents. Concentrations of C-reactive protein were measured by using latex-enhanced nephelometry (N High Sensitivity CRP assay) on a BN II nephelometer (Dade Behring Inc., Deerfield, IL, USA). Participants who responded affirmatively to the question ‘Have you ever been told by a doctor or health professional you have diabetes or sugar diabetes?’ were considered to have diagnosed diabetes. Those who answered that they had not been so told or that they had borderline diabetes were not considered to have diagnosed diabetes.
The analyses were limited to men and women aged ≥ 20 years. To age-adjust statistics, direct adjustment to the US population aged ≥ 20 years in the year 2000 was performed (12). Logistic regression analyses were conducted to assess the independent association between headaches (dependent variable) and BMI (independent variable). SUDAAN (Software for the Statistical Analysis of Correlated Data; RTI Inc., Research Triangle Park, NC, USA) was used for the analyses to account for the complex sampling design.
Results
A total of 10 291 participants aged ≥ 20 years were interviewed, and 9471 attended an examination. After excluding participants with missing data for one or more variables, 7601 were included in the analyses. The prevalence of headache was 21.7% (S.E. 0.8) (men 15.2%, women 27.9%). About 2.0% of participants had a BMI < 18.5 kg/m2, 33.0% had a BMI 18.5 to < 25 kg/m2, 34.8% a BMI 25 to < 30 kg/m2, and 30.3% a BMI ≥ 30 kg/m2.
The prevalence of headache was inversely related to age (p Wald χ2 < 0.001) (Table 1). The age-adjusted prevalence was higher in women than men (p Wald χ2 < 0.001), higher in participants who were African-American or of another race or ethnicity than in participants who were White (p Wald χ2 = 0.001), higher among participants with lower educational status than those with higher educational attainment (p Wald χ2 < 0.001), higher among those who were living with a partner than among other groups (p Wald χ2 < 0.001), higher among current smokers than never smokers (p Wald χ2 = 0.014) and lower among users of alcohol than those not using alcohol (p Wald χ2 < 0.001) (Table 1). The age-adjusted prevalence of headache was related to BMI in a non-linear fashion (p Wald χ2 < 0.001). The prevalence was higher among those with a BMI < 18.5 and among those with a BMI ≥ 30 kg/m2 than among those with a normal BMI.
Age-specific and age-adjusted prevalence of severe headaches or migraines during the previous 3 months by selected characteristics, National Health and Nutrition Examination Survey 1999–2002
After adjustment for age, sex, ethnicity, educational status, marital status, smoking status, alcohol use, systolic blood pressure, concentrations of C-reactive protein and total cholesterol, physical activity and self-reported diabetes in a logistic regression model, BMI was significantly associated with the prevalence of headache. Compared with participants with a BMI of 18.5 to < 25 kg/m2, those with a BMI < 18.5 kg/m2[odds ratio (OR) 2.01; 95% confidence interval (CI) 1.34, 3.02] or ≥ 30 kg/m2 (OR 1.37; 95% CI 1.09, 1.72) had a significantly elevated OR for having a headache (Table 2). The association was similar for men and women (p Wald χ2 for interaction between sex and BMI class < 0.394). Significant associations with having severe headaches or migraines were found for age (inverse), sex (women greater than men), educational status (lower educational attainment greater than higher), marital status (living with a partner higher than other categories), smoking status (current smokers higher than never smokers), alcohol use (inverse), and, among men only, systolic blood pressure (positive). No significant associations were found for race or ethnicity, concentrations of C-reactive protein and total cholesterol, or physical activity.
Adjusted∗ odds ratios and 95% confidence intervals for the associations between body mass index and other characteristics and the prevalence of severe headaches or migraines during the previous 3 months from multiple logistic regression, National Health and Nutrition Examination Survey 1999–2002
Odds ratios for each variable are adjusted for all other variables in table.
Using responses to the question ‘Are you limited in any way in any activity because of a physical, mental or emotional problem?’, we also examined whether obesity was related to the prevalence of physical, mental or emotional problems among participants with headaches. Because the sample size for underweight participants was small, the analyses were limited to participants with a BMI ≥ 18.5 kg/m2. The unadjusted prevalence was 5.0% among normal weight participants, 9.2% among overweight participants, and 13.8% among obese participants. After adjustment for age, sex, ethnicity, educational status, marital status, smoking status, alcohol use, systolic blood pressure, concentrations of C-reactive protein and total cholesterol, physical activity, and self-reported diabetes, BMI was significantly associated with the occurrence of physical, mental or emotional problems among participants who reported having headaches. Compared with participants who had a BMI of 18.5–25 kg/m2, elevated ORs were noted for participants who were overweight (OR 2.05; 95% CI 1.01, 4.16) and obese (OR 3.73; 95% CI 2.11, 6.60). We also examined whether headache status was an effect modifier of the association between BMI and the occurrence of physical, mental or emotional problems by adding an interaction term between BMI and headache status to the logistic regression model. However, no effect modification by headache status existed (P = 0.193), suggesting that the associations between categories of BMI and the occurrence of physical, mental or emotional problems were similar for participants with and without headaches.
Discussion
In this large nationally representative sample of US adults, BMI was associated with headache in a non-linear manner. Participants with a BMI < 18.5 kg/m2 or ≥ 30 kg/m2 had a higher likelihood of having a headache than did participants whose BMI was in the normal range. Furthermore, the prevalence of physical, mental or emotional problems among participants with headaches was higher among those who were overweight or obese compared with those with a normal BMI.
The relationships between the various sociodemographic characteristics and the prevalence of severe headaches or migraines were consistent with the results of previous studies (1, 2, 13–15). Thus, age was inversely associated with headaches, women had a higher prevalence than men, race or ethnicity was not significantly associated with headaches, and those with lower levels of educational attainment had a higher prevalence than those with more education.
Previous cross-sectional and prospective studies have produced mixed findings regarding a possible link between obesity and headache. In a Canadian cross-sectional study of 50 347 men and women aged 20–64 years, migraine was not significantly associated a BMI < 18.5 kg/m2 or ≥ 30 kg/m2 (6). In a US study of 1932 men and women conducted from 1997 to 1999, BMI was significantly associated with prevalent (OR 1.34; 95% CI 1.0, 1.8) and incident (OR 5.28; 95% CI 1.3, 21.1) headache (7). In addition, conversion to chronic daily headache occurred far more frequently among obese patients with episodic headaches than among normal weighted patients. Data were collected through telephone interviews, and weight and height were self-reported. In a Brazilian study, primary headaches were more common among 74 obese participants (76%) than among 70 controls (43%) (8). In a cross-sectional analysis of 30 215 patients from a US study conducted by telephone from 1997 to 2000, BMI was not associated with the prevalence of migraine among men or women (9). However, obesity was associated with the frequency and severity of headache attacks. Recent findings based on 30 703 participants who had episodic headaches from the American Migraine Prevalence and Prevention study have shown that obesity was associated with an increased frequency of headaches and increased disability among patients with migraines, but not severe tension-type headaches (16). Finally, a Swedish cross-sectional study of 684 women aged 45–74 years has reported no significant association between obesity and prevalent migraine or the frequency, intensity and duration of migraine attacks (10).
Several mechanisms support a possible aetiological link between obesity and headaches. Concentrations of calcitonin gene-related peptide, which are increased in obesity (17), are higher among people with migraine and other headaches than among those without headaches (18, 19). That obesity is a low-grade inflammation is shown by increases in the concentrations of circulating cytokines such as C-reactive protein, tumour necrosis factor-alpha, and interleukins such as IL-6 (20). Inflammation has been proposed as a possible contributing mechanism for migraines (21–23). However, several studies have not found any association between inflammation and headaches (24–27). Likewise, in our study, we found no significant independent association between concentrations of C-reactive protein and headaches. In addition to inflammation, high concentrations of circulating lipids and free fatty acids, which are elevated in obese people, may be important in the pathogenesis of migraines (28). Furthermore, obesity is associated with sleep apnoea (29). Although some reports have shown that headaches occur more frequently among those with than those without sleep apnoea (30–36), not all studies have corroborated these findings (37–42).
Why participants who were underweight had an elevated risk for having headaches is less clear. Underlying health conditions could possibly have contributed to the increased prevalence of headaches in this group. Because the prevalence of underweight was only about 2%, the population impact of being underweight on the prevalence of headaches is limited. Not many data on this topic appear to have been reported. In a cross-sectional analysis of Canadian data, underweight was not significantly associated with migraine (6). Thus, our findings in this regard require confirmation. Of interest is that the prevalence of moderately severe and severe depression was elevated among underweight participants with headaches in the American Migraine Prevalence and Prevention Study (43).
In contrast to several other studies that relied on self-reported weight and height to estimate BMI, we were able to use weight and heights from actual measurements. An additional strength of this study is that we were able to adjust for a variety of physiological and biochemical factors. A weakness of our study, however, was the non-specific definition of headache. Thus, we were unable to separate out the various primary headaches. Nevertheless, the sociodemographic correlates of headache in our analyses are consistent with what is known about the epidemiology of headaches. We were also unable to examine the possible effects of BMI on other parameters of headaches such as frequency, severity, or duration. The cross-sectional nature of our study also limits the interpretation of the directionality of the association between obesity and headaches.
In conclusion, we found that the prevalence of severe headaches or migraines was higher among both underweight and obese participants than among those with a BMI in the normal range. If obesity increases the risk for headaches, weight management might be a useful approach in headache management (44). In light of the large increase in the prevalence of obesity in the USA during the last several decades, establishing whether obesity is causally related to the development of headaches is worthy of additional research.