Abstract
Importance
Migraine is a highly prevalent and disabling primary headache disorder that is two to three times more prevalent in young women. Among females, there is a steep increase in incidence from puberty to young adulthood, but the mechanisms for the increase are unknown.
Objective
To determine if age of menarche is a risk factor for developing migraine headache vs. non-migraine headache by young adulthood.
Design
A prospective cohort study, The Growing Up Today Study (GUTS), of adolescents who have been followed since 1996, when they were nine, to 14 years of age. Headache questions were included on the 2007 and 2010 surveys.
Setting
Youth from across the United States who are offspring of women participating in the Nurses’ Health Study II.
Participants
6112 female participants who had provided data on headache symptoms, age at menarche and family history of migraine and were followed through 2007 or 2010 were included in this analysis.
Main outcomes
Migraine or non-migraine headache.
Results
Many females had a history of headaches, with approximately equal numbers reporting symptoms consistent with migraine (29.7%) and non-migraine headaches (25.3%). We found that, independent of age and family history of migraine, each one-year delay in onset of menarche decrease the odds of migraine by 7% (odds ratio (OR) = 0.93, 95% confidence interval (CI) 0.89–0.97), but was not related to non-migraine headaches.
Conclusions and relevance
The findings of this study suggest that early puberty increases the risk of developing migraines by young adulthood. As such, the study emphasizes the need for understanding the pathophysiological links between puberty and developmental changes that occur in the brain during that period and the mechanisms of onset of the migraine disease and its trajectory.
Introduction
Migraine is a common and complex neurological brain disorder that affects about 36 million Americans (12% of the population) and results in approximately $11 billion dollars per year in direct medical care costs (1). Moreover, migraine is the fourth most disabling disease among women worldwide (2,3) and one of the top five most frequent childhood disorders (4,5). Migraine occurs in up to 10.6% of children (five to 15 years of age) (6), and in up to 28% of youth (15 to 19 years of age) (7). While there is no sex disparity in migraine prevalence in pre-pubertal years (it is slightly more common in boys), during and after puberty migraines are 2–3 times more common among girls (8–11) and the attacks frequently become more intense after puberty (9).
There is evidence from basic science, population-based, and clinical-based studies suggesting that ovarian steroids have an important influence on the phenotypic expression of migraine in women. Incidence and frequency of migraine are affected by the phase of the menstrual cycle, pregnancy, peri-menopause, and menopause (12–21). Ovarian steroids levels change substantially during puberty and impact the organization of the brain (11,22,23). The emotional, behavioral, intellectual, and social changes during adolescence, which require remodeling of cortical and limbic circuits, are partially due to regional neuroanatomical changes and/or connectivity changes between regions (11, 24). It is unknown whether developmental changes in the brain during puberty, as well as before or after the onset of puberty, are related to the development of migraines later in life.
Migraine is a heterogeneous disease and its pathophysiology involves complex peripheral and central processes (25) that are not yet fully understood. Despite recent advances in understanding of migraine pathophysiology, the mechanisms of migraine development are still poorly understood. The disease frequently starts in childhood or adolescence, but there are little data available on risk factors for the onset of pediatric migraine. To determine if there is an association between pubertal timing and migraine disease, we investigated whether age at menarche is a risk factor for migraine onset. We hypothesized that earlier age at menarche would be associated with increased risk of migraine onset by young adulthood. To investigate whether the association was specific to migraine or related more generally to headache, we also assessed whether age at menarche was a risk factor for non-migraine headaches.
Methods
Subjects
We assessed the risk of developing headaches among female participants in the Growing Up Today Study (GUTS), an ongoing US longitudinal cohort study of 16,882 children of women participating in the Nurses’ Health Study II (NHS II). The cohort was established in 1996 when the participants were ages 9–-14 years of age, and participants have been followed up annually or biennially since 1996 (26). Parental consent was sought for all the children enrolled in this study before the letters of invitation and baseline questionnaires were sent out to 13,261 girls and 13,504 boys who were enrolled and followed up. For this study, the analysis was limited to females who answered the 2007, which was when migraine headaches were first assessed, or 2010 questionnaire. In 2007 and 2010 the ICHD-II criteria were the current headache criteria and therefore were used to guide the creation of the headache-related questions on the 2007 and 2010 questionnaires. All female participants who had completed the questions on headache symptoms and history, and had reported age of menarche were included in the analysis.
Outcomes
Migraine headache
Presence of migraine was determined according to the guidelines set forth by the International Classification of Headache Disorders (ICHD-II) (27). Headaches were classified as migraine if they had at least two of the following symptoms: Unilateral location, pulsating quality, worse when physically active, preventing routine activities, causing difficulty doing normal activities (bed rest necessary), and at least one of the following symptoms: Nausea and/or vomiting, or photophobia and/or phonophobia. The cases that were identified using these criteria were labeled as ‘migraine’ cases. Participants were also asked if they were told by a doctor or healthcare provider that they had migraines.
Non-migraine headache
Presence of non-migraine headaches was also determined according to the guidelines set forth by the International Classification of Headache Disorders (ICHD-II) (27). Headaches were classified as non-migraine headaches if they were present on either one side of the head, or both sides or in the back, front or around the eyes and if they did not have any migraine-like features, i.e. they were not pulsating nor were they worsened by physical activity. In addition, headaches should not have been associated with nausea or vomiting, or produce light and/or sound sensitivity. The cases that were identified using these criteria were labeled as ‘non-migraine headache’ cases. Given the criteria described above, participants who were classified as experiencing migraines could also be experiencing non-migraine headaches, but the participants who were classified as experiencing non-migraine headaches were exclusively experiencing other headaches (presumably mostly tension-type headaches) and not migraines.
Risk factors
Because the risk of migraine increases sharply around adolescence in females, age and age at menarche were used as predicting risk factors in this study. Menarche is the most prominent and reliably measurable pubertal development event in girls that occurs during the later stages of pubertal development. As such, it was used as a marker for the latest stages of pubertal development. Age at menarche was assessed on all GUTS questionnaires between 1996 and 2007. Information from the questionnaires was combined to create a derived age at menarche variable. GUTS surveys between 1996 and 2003 asked the girls in each survey “Have you started having menstrual periods?” and “If yes, age when periods began”. The age at menarche variable was derived by using the information collected on serial questionnaires (i.e., 1996, 1997 … 2003) to create one variable that could be used in a range of analyses. The advantages of this approach are that the variable is not missing for girls who have not yet achieved puberty by the time of completing a particular questionnaire, and there is less missing data than if one had to rely on only using the last questionnaire on which menarche was assessed (2003).
Females with a body mass index (BMI) between 25 and 29.9 kg/m2 were classified as overweight and those with a BMI at or above 30 kg/m2 were classified as obese. Maternal history of migraine was assessed from questions on the 2007 and 2009 NHS II questionnaires that asked, “Have you ever had a migraine headache?”. GUTS participants whose mothers responded on the 2007 or 2009 questionnaires that they had a history of migraine were classified as having a family history of migraine.
Sample for analysis
6457 females who completed the 2007 or 2010 questionnaire and whose mothers provided information on history of migraine on the 2007 or 2009 NHS II questionnaire, thus met the eligibility criteria for this study. Females who did not provide information on headache on either the 2007 or 2010 GUTS questionnaire (n = 238) or age of menarche (n = 107) were excluded, thus reducing the sample to 6112.
Statistical analysis
To account for the correlation between same sex siblings, generalized estimating equations (GEE) were used to examine associations with migraine and non-migraine headache. The GEE is an extension of the generalized linear model that is often used to analyze longitudinal data with correlated response. Since our outcomes were binary, a logit link was used. Separate models were used for each type of headache, while adjusting for age, family history of migraine and weight status. All analyses were performed using SAS version 9.2. (SAS Institute Inc., Cary, North Carolina, USA).
Results
Demographics of the 6112 females in the Growing Up Today Study in 2007.
403 females were missing information on BMI
25 ≤ BMI < 30 kg/m2
BMI ≥ 30 kg/m2
Individuals who met criteria for migraine based on the questionnaires
Individuals diagnosed by a doctor or other healthcare provider
Mean (standard deviation)
As expected, females whose mother reported a history of migraine were more likely to have migraine (odds ratio (OR) = 1.69, 95% confidence interval (CI) 1.52–1.87). However, a maternal history of migraine was protective against other types of headaches (OR = 0.70, 95% CI 0.63–0.77).
Association (odds ratios and 95% confidence intervals) of age at menarche with migraine and non-migraine headaches among young adult women in the Growing Up Today Study.
Puberty occurs earlier in obese girls than their leaner peers. Therefore, we examined whether obesity or the interaction between age at menarche and obesity had an impact on the association between earlier age of menarche and development of migraine. We observed that inclusion of these variables had a minimal impact on the results.
Discussion
In this study, we assessed the association of age of menarche with the odds of migraine and non-migraine headaches by early adulthood in a prospective cohort study. We found earlier age of menarche to be associated with a 7% (if age and family history terms in the model were held constant, and 5% if, in addition, the weight status term also remained constant) decrease in odds of migraine by early adulthood, but unrelated to odds for non-migraine headaches. These findings suggest different mechanisms are involved in the pathophysiology of one of the most prevalent primary headache disorders.
There are two potential explanations for our findings on migraine in this study. First, earlier age at menarche may increase the risk of migraine in genetically susceptible individuals (who have a family history of migraine) by exposing the brain to the sex hormone fluctuations from an earlier age and thus precipitating migraine onset by younger adulthood. Reproductive hormonal changes (particularly a sudden decline to low estrogen levels) have been shown by numerous lines of evidence to have significant modulatory influences on triggering migraine attacks. Women with a history of migraine have a faster decline of estradiol prior to menses (late-luteal phase) than controls irrespective of whether they have experienced a headache in that cycle or not (28), which may be an indicator of neuroendocrine vulnerability in female migraineurs. In addition, a decrease in estrogen has been linked to an increased level of symptoms/frequency of migraine attacks (16,17) in some women (29), although not consistently in larger population-based studies (30). In addition, women in their late peri-menopause are at a higher risk of increased migraine attacks compared to pre-menopausal women (31). The mechanisms of such modulatory influences remain largely unknown, and especially there is no data on whether such processes interact. Another possible explanation for the findings is that since menarche occurs during the last stages of pubertal development, earlier age at menarche would also indicate that other phases of pubertal development had started earlier or happened at a faster pace. Thus, the changes that lead to migraine onset could have been happening at earlier phases of pubertal development and hence could be independent of the effect of sex hormones.
These two interpretations may not be mutually exclusive, since sex hormones interact significantly with the structural and functional organization of the brain during the pubertal development period (11,22,23). Risk of migraine, which is a sex-biased neuropathology, may be mediated by the variation in the timing of interactions between the hormones of puberty and the brain development in that period. The actions of hormones on remodeling and structural plasticity involve both direct and indirect genomic actions, as well as hormonal regulation of signaling pathways that directly affect neuronal excitability, metabolism and neuronal survival (32). There may be accumulating changes in the brain starting in childhood that would eventually lead to migraine, and the earlier the age at menarche, the earlier the forming of such changes. Using magnetic resonance imaging we have recently observed structural differences, as well as differences in pain processing, that are sex-specific and only present in adult female migraineurs (33,34). Such changes may be influenced by hormones or may start from or even before the onset of menarche.
Our results also showed increased odds of both non-migraine and migraine headache types by age. Given that stress is a shared trigger for both conditions, the finding of increased odds of these headaches with age may also parallel the more greater stress during adulthood as a result of the potential increase in the social interactions and responsibilities that could lead to increased stress levels from the new and more demanding environments (35). Stress (36, 37) is a shared suspected trigger for most headache types, and so are lack of sleep and tiredness (38).
Our results also showed that a maternal history of migraine was protective against other types of headaches. One explanation for the latter may be that once migraine is recognized in a family, the daughter is more likely to receive appropriate evaluation and diagnosis. Alternatively, the association may be due to genetic risk.
We used ICHD-II guidelines for classification of the headache types (27) for to designing the questions in the surveys and assessing migraine and tension tension-type headache. Since patients who experience migraine could also experience less severe non-migraine headaches (such as tension-type headache) in addition to their migraines, for the purpose of classification and analysis we did not distinguish between patients who exclusively experienced migraine attacks and those who experienced a combination of both non-migraine and migraine headaches. For non-migraine headache, however, patients had to have no migraineous features associated with their headaches and their headaches had to be exclusively non-migraineous. Tension-type headache is the most prevalent and common form of non-migraine headache, and it is likely that the majority of non-migraine headaches were tension-type headache. The estimated prevalence for of non-migraine headache in our study is, however, slightly lower compared to the figures reported in the literature for tension-type headache. This may be due to the added exclusivity criterion, and that migraine patients who may also have been experiencing non-migraine headaches such as tension tension-type headaches were excluded. It should be noted that in tension-type headache, headache pain has a different presentation (no throbbing pain, usually tightening) and it is also milder compared to migraine and much less disabling. However there are further differences between these two common types of headaches. For instance, there is a prodrome, or premonitory phase, associated with migraine. During the premonitory phase of migraine, which could last for hours or days before the headache pain itself, changes in the brain systems occur that present with symptoms such as mood change, irritability, and increased sensitivity to light and noise (39–41). Such differences may further suggest different underlying pathological mechanisms for the two conditions. While more studies are needed to determine the underlying neurological basis for migraine, we believe that, given the differences in the presentation, trigger profile and symptoms associated with migraine, it is very likely that there is a distinct pathophysiological basis for migraine onset among young females.
Our study was a large prospective investigation of pubertal timing and the development of headache that included a large number of girls with migraine, with a number of strengths. First, information on exposures was collected prospectively and before the onset of migraine. Another strength is the size of the cohort and the duration of follow-up. The Growing Up Today Study is one of the largest ongoing cohort studies in the United States. The participants live throughout the United States, thus the results should be generalizable to adolescents and young adults residing across the United States. We are unaware of any other prospective longitudinal epidemiological studies in the United States studying pediatric migraine. It is important, however, to acknowledge that there are some limitations to the study. First, we used questionnaires to classify participants. Because they live throughout the United States, bringing in all members for an evaluation by a neurologist is logistically impossible. However, the use of questionnaires may have introduced some misclassification. The fact the results differed by headache type provides some face validity to the self-reports. Another limitation is that the cohort is more than 90% Caucasian, thus it is unclear if the results are generalizable to young adults of color. Another limitation relates to the potential underestimation of the family history of migraine, because a non-trivial number of migraineurs will not be diagnosed.
It should also be noted that while GUTS is an ongoing study, the data that was used in this investigation were limited to the surveys that were returned up to and including the 2010 survey, when the majority of the subjects were in their mid-20s. It is likely that some subjects developed migraine after 2010 or that their migraines were not yet correctly diagnosed by 2010. Thus a limitation of our findings is that they do not address whether later onset of menarche leads to later migraine onset, has a protective effect and reduces the risk of early and late onset migraine, or is unrelated to migraine risk.
Conclusion
In conclusion, our study suggests that early puberty increases the risk of developing migraine. The next step is to investigate the pathophysiological links between puberty and developmental changes that occur in the brain during that period and the mechanisms driving the onset of migraine disease. Migraines can cause severe impairment, and the individuals who suffer from the disease have to live and cope with it for many years of their lives. Therefore it is of major public health importance to understand the cause of developing migraine in young adults so that prevention programs can be developed. Developmental changes in the brain that occur during puberty (11,23,24, 42–46) may create individual variation in the disease evolution trajectory in susceptible individuals. If the mechanisms of such changes are better understood, we may be able to modify such trajectories and prevent migraine development.
Key findings
Early puberty increases the risk of developing migraines by young adulthood, but not the risk of other types of headaches. Independent of age and family history of migraine, each one-year delay in onset of menarche decreases the odds of migraine by 7%. As expected, headaches in general are highly common (95% of the sample), and females whose mothers reported a history of migraine are more likely to have migraine, but not other types of headaches.
Footnotes
Acknowledgements
Data collection was supported by research grants from the National Institutes of Health (MH087786, DK59570, DK46200, HL68041, and HD049889). The authors would like to thank the thousands of participants in the Growing Up Today Study and their mothers. Dr. Maleki was supported by Boston Children’s Hospital Office of Faculty Development Fellowship.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Drs Maleki and Field have indicated they have no financial relationships relevant to this article to disclose. Dr Kurth is a research methodology consultant for a project at the Massachusetts General Hospital, Boston, MA, for which the hospital has received a research grant from Merck Sharp & Dohme Corp. He further has received honoraria from the BMJ and Cephalalgia for editorial services and travel support from the International Headache Society.