Abstract
This study investigates the associations between obesity and migraine. A neurologist clinically assessed 684 women aged 40–74 years attending a population-based mammography screening programme. Body height and weight were measured and body mass index (BMI) calculated. Obesity was defined as a BMI ≥ 30 kg/m2. The proportion of obesity did not differ between women with active migraine, women with inactive migraine or women who had never experienced migraine (P = 0.96). The distribution of frequency, intensity, duration or severity of attacks did not differ between obese and non-obese women with migraine. In this study there were no significant associations between migraine or migraine characteristics on the one hand and obesity on the other.
Keywords
Introduction
Migraine and obesity are prevalent disorders in the general population, but there is little knowledge of their association. Schéle et al. (1) studied the physical characteristics of male conscripts. The authors found that conscripts with migraine were on average 1 cm shorter than control conscripts. The body mass did not differ between the groups. Obesity was reported to be associated with new onset chronic daily headache (2). In a recent study (3), obesity was associated with the frequency and severity of migraine attacks, but not with the prevalence of headaches.
The aim of this study was to further study the associations between obesity and migraine in a population-based cohort of middle-aged and older women.
Study population and methods
All women aged 40–74 years living in a small community in Sweden, Östhammar, are invited every second year to a mammography screening programme (MSP). The women are identified in the official population registry of Statistics Sweden. Most women attend the local mammography unit (MUT) in Östhammar. A minority of women choose to attend the regional MUT or do not respond. The names of women invited to attend the MSP on a particular working day are available at the local MUT.
The source of the study population was women aged 40–74 years who lived in Östhammar during the study period from November 1997 to October 1998 and attended the local MUT, where they were asked to participate in our study. Six hundred and eighty-four women agreed to participate (study population). The study design has been described in greater detail previously (4).
Interview, measures and definitions
Participants were interviewed at the MUT by a board-certified neurologist. Physical and neurological examinations were not performed. If the time for interview was insufficient, the interview was later completed over the telephone (n = 92).
The interview structure was based on the headache criteria proposed by the International Headache Society (5). This study focused on migraine headache. Migraine headache is referred to as migraine. Women had migraine if their headaches (i) fulfilled the criteria for migraine without aura or (ii) fulfilled the criteria for migraine with aura and the headaches in conjunction with the aura fulfilled the criteria for migraine without aura. Active migraine headache was defined as one or more migraine attacks in the previous year. Inactive migraine headache was defined as migraine attacks experienced in the past, but not in the previous year.
All women with migraine experienced inhibition (moderate disability) or prohibition (severe disability) of daily activities during migraine attacks. The pain intensity of untreated or unsuccessfully treated attacks was measured on a verbal analogue scale ranging from ‘no migraine pain’ (= 0) to ‘migraine pain as intense as you can imagine’ (= 10). Frequency was measured in 10 predetermined categories as the number of days in the previous year during which the women had headache (1, 2–3, 4–6, 7–9, 10–12, 13–24, 25–48, 49–96, 97–180, and >180 days). Duration of migraine attacks was measured in 10 predetermined categories as the number of hours during which an attack lasted (<1, 1–3, 4–5, 6–8, 9–12, 13–18, 19–24, 25–48, 49–72, and >72 h).
Educational levels were combined into two classes (upper secondary school or university vs any lower educational level) in all analyses. Age was treated as a continuous variable.
Body height and weight were measured on subjects (n = 684) with light clothing and without shoes. The measurements took place at the MUT. Using WHO guidelines, obesity was defined as BMI ≥ 30 kg/m2.
All women gave their informed consent for participation and the Ethics Committee of the Medical Faculty, University of Uppsala, approved the study (No. 97307).
Data analysis
Differences in proportions were analysed using the Chi-square method. To analyse the association between migraine and obesity, a logistic model was created with migraine (active, inactive and never migraine) as the dependent variable, and age, education and obesity as the independent variables. To analyse the association between migraine and BMI, a logistic model was created with migraine (active, inactive and never migraine) as the dependent variable, and age, education and BMI (continuous) as the independent variables.
The Kruskal–Wallis test was used to assess whether the medians of body height, body weight, age and BMI differed between women with active migraine, inactive migraine and women who never experienced migraine (6).
The relationships between severity of migraine attacks (dependent variable) on the one hand and BMI and obesity (independent variables) on the other were analysed in two separate logistic regression models, taking differences in age and educational level into account. The relationships between frequency, duration and intensity of migraine attacks (dependent variables) and obesity (independent variable) were analysed in three separate quantile regression models, taking differences in age and educational level into account. The relationships between frequency, duration and intensity of migraine attacks (dependent variables) and BMI (independent variable) were analysed in three separate quantile regression models, taking differences in age and educational level into account.
The statistical program packages JMP (version 5.11), Statsdirect (version 2.2.10) and R (version 2.3.1) were used. P-values less than 0.05 were considered significant. All tests were two-sided.
Results
During the study period, 4398 women aged 40–74 years lived in Östhammar. We conducted the study on 32 working days. There were 996 women on the invitation lists on these days. Of the 996 invited women, 838 actually attended the local MUT, where they were asked to participate in our study. Six hundred and eighty-four women agreed to participate (study population), corresponding to a participation rate of 81.6% among attenders.
Official statistics of the MSP said that three of four invited women attend the local MUT, less than 10% choose to attend the regional MUT, and approximately 14% do not respond to the invitation. The attendance rate is known to be lower among older than younger women.
Analysis of migraine prevalence in non-participating and non-attending women has been described in greater detail previously (4). To summarize, the prevalence of migraine was significantly lower in attenders of the local MUT who did not participate in our study than in women who attended the MUT and participated in the study. There was no evidence that the prevalence of migraine in women attending the central MUT would differ from the prevalence of migraine in women attending the local MUT (4). For technical reasons this analysis was done in a different sample of non-attenders than that reported in this study. Women who did not respond to the invitation to the MSP were not contacted for ethical reasons. We concluded that response bias would affect estimate of the prevalence of migraine in the source population only to a small extent (4).
We tested if the study population was representative of the corresponding sex and age segment of study area by comparing age-specific participation rates (Table 1). There was a statistically significant difference in age-specific participation rates (Chi-square = 18.5, 6 d.f., P = 0.005). This difference was mainly due to the low participation rate of women aged 70–74 years. After excluding women aged 70–74 years from analysis, no difference emerged (Chi-square = 7.4, 5 d.f, P = 0.19).
Age-specific distribution of female residents, non-participants, and participants
Data from Statistics Sweden.
The prevalence of active and inactive migraine was 19.0% (95% CI, 16.1–21.9) and 13.9 (95% CI, 11.3–16.5), respectively. The prevalence of obesity was 19.3% (95% CI, 16.3–22.2). Characteristics of women with and without migraine are given in Table 2. Active and inactive migraine were not associated with obesity, irrespective of whether we controlled for age and education (LR Chi-square = 0.07, 2 d.f., P = 0.96) or not (LR Chi-square = 0.69, 2 d.f., P = 0.71). The adjusted OR for active migraine was 1.1 (95% CI, 0.6–1.8). The adjusted OR for inactive migraine was 1.0 (95% CI, 0.6–1.7). After adjustments for age and educational levels, BMI was not associated with migraine (LR Chi-square = 0.8, 2 d.f., P = 0.40).
Descriptive statistics of participants according to migraine status
Comparisons between all three groups.
High education was defined as upper secondary school or university.
Obesity was defined as BMI ≥ 30 kg/m2.
The level of education was high in three of 22 (13.6%) obese women with active migraine, and in 49 of 108 non-obese women with active migraine (LR Chi-square = 8.7, 1 d.f., P = 0.003). The median age of obese women with active migraine was 51 years and the median age of non-obese women with active migraine was 47 years (Mann–Whitney U, Z = 1.8, P = 0.08).
Seven of the 22 (31.8%) obese women with active migraine and 41 of the 108 (38.0%) non-obese women with active migraine experienced prohibition of daily activities during attacks. The mean BMI of women who experienced inhibition of daily activities during migraine attacks was 26.6 kg/m2, and mean BMI of women who experienced prohibition of daily activities during migraine attacks was 26.9 kg/m2. Neither BMI (LR Chi-square = 0.00, 1 d.f., P = 0.98) nor obesity (LR Chi-square = 0.93, 1 d.f., P = 0.33) was associated with headache severity after adjustments for differences in age and educational levels. In obese women with active migraine, the median frequency of attacks was 7–9 days during the past year (range 1 to 180–365 days). In non-obese women with active migraine, the median frequency of attacks was 7–9 days during the past year (range 1 to 49–96 days). In obese women with active migraine, the median duration of attacks was 19–24 h (range 4–5 h to 49–72 h). In non-obese women with active migraine, the median duration of attacks was between the predetermined categories 13–18 h and 19–24 h (range 4–5 h to 49–72 h). In obese women with active migraine, the median intensity of attacks was 7.5 (range 4–10). In non-obese women with active migraine, the median intensity of attacks was 8 (range 5–10). There was no statistically significant relationship between frequency (P = 0.26), intensity (P = 1.00) and duration (P = 0.60) of migraine attacks on the one hand and BMI on the other (quantile regression, taking differences in age and educational level into account). There was no statistically significant relationship between frequency (P = 0.11), intensity (P = 0.83) and duration (P = 0.56) of migraine attacks on the one hand and obesity on the other (quantile regression, taking differences in age and educational level into account).
Discussion
In this study of middle-aged and older women we found no association between migraine and obesity. This is in line with the findings of previous studies (1, 3).
The characteristics of migraine attacks were not correlated with BMI and did not differ between obese and non-obese women with active migraine. A recent population study (3) reported that the risk of experiencing 10–15 migraine headache days per month increased from 4.4% in the normal weight group to 13.6% in obese individuals and 20.7% in the morbidly obese subjects. In the present study, 22 of 130 (16.9%) women with active migraine were obese and three women (2.3%) had more than eight migraine days per month. If the findings by Bigal et al. are true, we may have failed to detect a significant correlation between frequency and BMI or obesity due to the limited number of women with frequent migraines in this study.
The main strengths of this study are that the same neurologist interviewed all participants, and that body weight and body height were measured and did not rely on self-reports. In the study by Bigal et al. (3) migraine diagnosis was determined by lay interviews and physical characteristics were based on self-reports. This may have introduced important and unpredictable bias. The population in the present study was representative of the corresponding sex and age segment of the community with the exception of women aged 70–74 years, who were under-represented. This is unlikely to influence the principal conclusions of the study. A potential drawback is the limited number of women with migraine and obesity. This may have limited the possibility of finding true differences between migraine characteristics and obesity.
Conclusion
There was no evidence in this study that obesity affects the prevalence of migraine among middle-aged and older women. If there are true associations between obesity and features of migraine such as frequency, these are likely to be small, and cannot be reliably studied in small or moderately sized community samples.
Acknowledgements
I thank Lille-Mor Jansson, Yvonne Björn and Ingrid Lindquist for excellent research assistance, and statisticians Johan Brink and Felicia Franzén, Statisticon AB, for help with quantile regression analyses.