Association of migraine and vasomotor symptom severity and duration among naturally menopausal women in Nurses’ Health Study II

Abstract

Objective:

Migraine and vasomotor symptoms (VMS) are prevalent brain conditions linked to female sex hormones and negatively impact the quality of life for middle-aged women. We aimed to quantify the association between migraine phenotype and VMS severity and duration across the final menstrual period.

Methods:

We analyzed data from 21,468 participants in Nurses’ Health Study II, an ongoing prospective cohort study of female registered nurses aged 25-42 at baseline in 1989. We included naturally menopausal individuals who reported a final menstrual period from 2007 through 2015. We analyzed the cross-sectional association between self-reported migraine headaches (with and without self-reported clinician diagnosis) and migraine phenotype (with and without aura) and VMS in midlife using logistic regression, adjusting for demographic, lifestyle, and reproductive health factors.

Results:

Overall, 64% of the cohort reported VMS in the past 4 weeks. Of those experiencing VMS, 7% reported severe VMS and 31% reported VMS lasting ≥ 5 years. Individuals with recent (past 2 y) migraine in midlife (29%) had 30% (21%-40%) greater odds of reporting VMS than those without migraine (68% of women vs 63%). Among women reporting VMS, recent midlife migraine (vs no recent migraine) was also associated with severe (8% vs 6%, OR = 1.53, 95% CI = 1.30-1.81) and prolonged (35% vs 30% ≥ 5 years; OR = 1.24, 95% CI = 1.12-1.36) VMS. Although migraine phenotype was unassociated with VMS frequency or duration, migraine with aura and self-reported clinician-diagnosed migraine were more strongly associated with severe VMS than were migraine without aura or self-reported migraine.

Conclusion:

While the nature of the association remains unclear, our findings highlight the importance of risk assessment and screening for VMS among women with migraine.

Migraine, a disorder of central nervous system dysregulation, is 3 times more common among women than men, possibly due to the role of estrogen cycling in its pathophysiology.¹ Migraine is a major source of pain and disability among middle-aged women, and reaches peak prevalence of 16%-34% in the 40s and 50s.^2-4 While previous research shows an increase in migraine symptoms and headache frequency during the perimenopausal period, followed by a postmenopausal decline, epidemiologic studies of migraine during the menopause transition are few, methodologically inconsistent, and have produced conflicting results.^5-10 Migraine may be associated with vasomotor symptoms (VMS; hot flashes, night sweats) due to vascular dysregulation, shared genetic factors, endothelial dysfunction, and hypothalamic pathways.^6,11,12

Much of the literature on this topic has focused on the association between menopause and headache frequency.^7-10,13 The complexities of the dual burden of migraine and VMS in midlife remained understudied. While one study found that a history of migraine predicted VMS, particularly during late-stage perimenopause, the authors did not have data on migraine frequency or type (migraine with and without aura).¹⁴ Studies of the differential impacts of ovarian hormones on migraine type are inconclusive, although migraine with aura is associated with greater comorbidities, suggesting differences in etiology, which are not well understood.^15-17

This study will be the first to explore the association between migraine phenotype and VMS severity and duration. Given the common pathophysiological pathways and noted differences in migraine phenotype etiology and comorbidities, we hypothesized that women who experience migraines in midlife will also experience more severe VMS, and that this association will be greater among those experiencing MA.

METHODS

Study population

We analyzed data from the Nurses’ Health Study II (NHS II), a large prospective cohort study that began in 1989. A detailed account of NHS II methodology is published elsewhere.^18,19 In brief, NHS II includes 116,408 eligible female registered nurses who were aged 25-42 at enrollment. Participants in NHS II complete follow-up questionnaires every 2 years. For the present analysis, we excluded 54,104 participants who reported surgical menopause and 7,705 participants with uncertain migraine status (single report of lifetime migraine headache) for a cohort of 54,599 participants (Fig. 1). To capture women transitioning across menopause, and therefore, most susceptible to VMS, we included women within (before or after) 2 years of reported final menstrual period (FMP) when data on VMS presence and severity data were collected in 2009 and 2013 (ie, we included women with an FMP 2007-2015), for a final analytic cohort of 21,468 participants. The study protocol was approved by the Institutional Review Board of the Brigham and Women’s Hospital and the Harvard T.H. Chan School of Public Health, and return of the questionnaires was considered implied consent.

FIG. 1.

Study population. NHS II, Nurses' Health Study II.

Assessment of migraine

Figure 2 summarizes the timing of migraine and VMS assessment relative to the FMP. In 2009 and 2013, participants were asked, “In the past 2 years, have you had a migraine headache?” If they responded affirmatively, we asked, “Did you sometimes have an aura?” The 2009 migraine assessment was used for individuals with an FMP in 2007-2011, and the 2013 migraine assessment was used for individuals with an FMP in 2012-2015. We further categorized migraine as clinician-diagnosed (participants self-reported a diagnosis of migraine by a health care provider) or self-reported migraine symptoms (participants self-reported experiencing a migraine headache in the past 2 y but had not reported a clinical diagnosis of migraine as of 1995, the last time the provider diagnosis question was asked).

FIG. 2.

Migraine and VMS assessment relative to final menstrual period. FMP, final menstrual period; NHS II, Nurses' Health Study II; VMS, vasomotor symptoms.

Assessment of vasomotor symptoms

Participants also reported whether they experienced hot flashes or night sweats in the past 4 weeks within 2 years of FMP and, if so, whether the symptoms were mild, moderate, or severe. Thus, both migraine and VMS presence and migraine and VMS severity were assessed during overlapping time periods. We ascertained VMS duration from the 2017 questionnaire among individuals who reported VMS and had at least 5 years of follow-up surrounding the FMP. Participants were asked to report whether their symptoms lasted < 5 years, 5-9 years, or 10 years or longer, and we dichotomized responses into long duration (≥ 5 years) or < 5 years.

Assessment of covariates

We assessed race, marital status, and age at menarche from the baseline (1989) NHS II questionnaire. We considered the following potential covariates from the 2009 NHS II questionnaire: parity, current use of oral contraceptives, use of hormone therapy (ever and within 2 years of FMP), alcohol intake, smoking status, and body mass index. We assigned missing indicator levels to missing data, as missingness was low (≤ 3%) for all variables. Potential covariates were selected a priori through literature review. Final covariates were chosen based on subject matter expertise, consideration of exposure-covariate temporality, and model parsimony.

Statistical analysis

We first used summary statistics to describe our analytic cohort and compare the distribution of baseline characteristics among individuals with and without migraine, as well as describe the characteristics of migraine and menopause (Table 1). We modeled the cross-sectional association between migraine and VMS among women within 2 years before and 2 years after the FMP using logistic regression, adjusting for lifestyle, demographic, and reproductive factors. We analyzed the association of migraine in midlife (self-reported clinician-diagnosed/not diagnosed, with/without aura) with VMS presence (yes/no), VMS severity (severe, mild-to-moderate) and VMS duration (< 5 yr, ≥ 5 yr). In final models we adjusted for race (White, Black, Asian, multiracial, other/unknown), age at menarche (≤ 9, 10-12, 13-16, ≥ 17 years, unknown), parity (1, 2, 3, 4 or more births), current contraceptive use (yes vs no), alcohol intake (none, 1, 2, 3, 4 or more days per week), current smoking (yes vs no) and BMI (< 18.5,18.5-25, 25.1-30, > 30 kg/m²).

TABLE 1.

Demographic characteristics of 21,468 NHS II participants by migraine ± 2 years from final menstrual period

Characteristic	Migraine (n = 6,319 29%)	No migraine (n = 15,149 71%)
Age at enrollment (mean, SD)	32 (3.0)	32 (3.1)
Age at menopause (mean, SD)	53 (2.7)	53 (2.7)
Race (%)
White	95	93
Black	< 1	1
Asian	1	2
Multiracial	1	2
Othera/unknown	2	2
Marital status
Never married	16	16
Married	78	77
Divorced	4	5
Separated	1	1
Widowed	< 1	< 1
Unknown	< 1	< 1
Parityb %,
Nulliparous	16	16
1	13	12
2	40	39
3	23	24
4 or more	9	9
Currently using oral contraceptivesb, %	5	4
Ever used hormone therapy, %	26	22
Used hormone therapy within 2 years of FMP	15	13
History of premenstrual syndrome, %	17	12
Age at menarche, years %
9 or younger	1	1
10-12	50	50
13-16	47	47
17+	< 1	1
Unknown	< 1	< 1
Days per week consuming alcoholb, %
None	47	43
1 day	21	19
2 days	10	10
3 days	7	8
4 or more days	19	16
Current smoker, %	4	5
Body mass indexb, kg/m2, %
< 18.5	5	14
18.5-25	43	39
25.1-30	29	25
> 30	23	22
Migraine type, %
Migraine with aura	51	NA
Migraine without aura	49	—
Self-reported clinician-diagnosis of migraine, %	19	NA

FMP, final menstrual period; NA, not applicable; NHS II, Nurses’ Health Study II.

^a“Other” race includes American Indian and Native Hawaiian/Pacific Islander, as well as individuals who self-identified as none of the above categories.

^bAscertained from 2009 questionnaire.

RESULTS

Descriptive results

Participants were an average of 32 years old at baseline in 1989, and 53 years old at menopause (Table 1). The majority were White, married, parous, and nonsmokers. The overall prevalence of midlife migraine in the cohort was 29%. Nearly 1 in 5 (19%) participants reported having been diagnosed with migraine by a health care provider. An equal proportion of participants with migraine reported experiencing migraine with and without aura. We found similar proportions of hormone therapy use within 2 years of FMP between individuals with and without migraine (15% vs 13%). Individuals with and without migraine were similar in terms of other demographic, lifestyle, and reproductive health factors, except for a history of premenstrual syndrome (PMS). Individuals with migraine were nearly more likely to self-report clinician-diagnosed PMS than individuals without migraine (17% vs 12%).

Overall, 64% of individuals within 2 years of FMP reported the occurrence of any VMS within the past month. Of those with VMS, 7% indicated experiencing severe symptoms. Of individuals, 31% experiencing VMS reported a VMS duration of ≥ 5 years.

Migraine and vasomotor symptoms

Among individuals with midlife migraine, 68% reported experiencing VMS in the past month, compared with 63% of individuals without migraine (Table 2). The adjusted odds ratio (aOR) was 1.30 (95% CI: 1.21-1.40; Fig. 3). Effect estimates were similar across self-reported clinician-diagnosed migraine (aOR = 1.30, 95% CI: 1.17-1.45) and migraine phenotype (migraine with aura aOR = 1.27, 95% CI: 1.16-1.39, migraine without aura aOR = 1.23, 95% CI: 1.12-1.34).

TABLE 2.

Recent migraine and recent VMS, among women ± 2 years surrounding final menstrual period in NHS II

	Past month VMS n (%)	No past month VMS n (%)	Crude odds ratio (95% CI)	Adjusteda odds ratio (95% CI)
Migraine	3,334 (68)	1,573 (32)	1.26 (1.18-1.36)	1.30 (1.21-1.40)
Clinician-diagnosed migraine	1,330 (67)	639 (32)	1.24 (1.12-1.37)	1.30 (1.17-1.45)
Self-reported migraine symptoms	2,004 (68)	934 (32)	1.28(1.17-1.39)	1.30 (1.19-1.43)
Migraine with aura	1,836 (68)	849 (32)	1.28 (1.17-1.40)	1.27 (1.16-1.39)
Migraine without aura	1,800 (67)	867 (32)	1.23 (1.13-1.34)	1.23 (1.12-1.34)
No migraine	7,988 (63)	4,762 (37)	Ref	Ref

BMI, body mass index; NHS II, Nurses’ Health Study II; Ref, reference; VMS, vasomotor symptoms.

^aAdjusted for covariates: race, age at menarche, parity (2009), oral contraceptive use (2009), alcohol intake (2009), smoking (2009), BMI (2009).

FIG. 3.

Migraine within the past 2 years and recent (past 4 wk) VMS, among women ± 2 years from final menstrual period in NHS II. NHS II, Nurses' Health Study; VMS, vasomotor symptoms.

We also examined VMS severity and duration among individuals reporting VMS. Overall, individuals with migraine were more likely to report severe (vs mild or moderate) VMS and long-duration VMS (VMS lasting 5+ y vs VMS lasting < 5 y). The adjusted odds ratio for the relationship between migraine overall and severe VMS was 1.53 (95% CI: 1.30-1.81; Table 3). This association appeared to be stronger among participants with self-reported clinician-diagnosed migraine (aOR = 1.78 95% CI: 1.44-2.21) and those with migraine with aura (aOR = 1.77, 95% CI: 1.45-2.16) than among those with self-reported migraine symptoms (aOR = 1.36, 95% CI: 1.11-1.67) or migraine without aura (aOR = 1.32, 95% CI: 1.06-1.65), respectively (Fig. 4A). The adjusted odds ratio for the relationship between migraine overall and long duration VMS (≥ 5 y) was 1.24 (1.12-1.36; Table 4). Estimates were similar across migraine diagnoses and phenotypes, with adjusted odds ratios for long-duration VMS ranging from 1.23 to 1.30 (Fig. 4B).

TABLE 3.

Recent migraine and VMS severity, among women ± 2 years surrounding final menstrual period with VMS in NHS II

	Past month severe VMS n (%)	Past month mild or moderate VMS n (%)	Crude odds ratio (95% CI)	Adjusteda odds ratio (95% CI)
Migraine	277 (8)	3,057 (92)	1.48 (1.27-1.73)	1.53 (1.30-1.81)
Clinician diagnosed migraine	130 (10)	1,200 (90)	1.77 (1.44-2.17)	1.78 (1.44-2.21)
Self-reported migraine symptoms	147 (7)	1,857 (93)	1.29 (1.07-1.57)	1.36 (1.11-1.67)
Migraine with aura	161 (10)	1,625 (90)	1.72 (1.43-2.08)	1.77 (1.45-2.16)
Migraine without aura	116 (7)	1,508 (93)	1.26 (1.02-1.56)	1.32 (1.06-1.65)
No migraine	461 (6)	7,527 (94)	Ref	Ref

BMI, body mass index; NHS II, Nurses’ Health Study II; Ref, reference; VMS, vasomotor symptoms.

^aAdjusted for covariates: race, age at menarche, parity (2009), oral contraceptive use (2009), alcohol intake (2009), smoking (2009), BMI (2009).

FIG. 4.

Recent migraine and VMS severity (A) and duration (B) among women ± 2 years from final menstrual period in NHS II with VMS. NHS II, Nurses' Health Study II; VMS, vasomotor symptoms.

TABLE 4.

Recent migraine and VMS duration, among women ± 2 years surrounding final menstrual period with VMS in NHS II

	VMS lasting ≥ 5 years n (%)	VMS lasting < 5 years n (%)	Crude odds ratio (95% CI)	Adjusteda odds ratio (95% CI)
Migraine	1,044 (34)	2,030 (66)	1.18 (1.08-1.29)	1.24 (1.12-1.36)
Clinician diagnosed migraine	436 (35)	808 (65)	1.24 (1.09-1.41)	1.30 (1.13-1.48)
Self-reported migraine symptoms	608 (33)	1,222 (67)	1.14 (1.02-1.28)	1.20 (1.06-1.34)
Migraine with aura	533 (34)	1,023 (66)	1.23 (1.09-1.38)	1.25 (1.10-1.41)
Migraine without aura	506 (34)	990 (66)	1.20 (1.07-1.35)	1.23 (1.09-1.39)
No migraine	2,152 (30)	5,064 (70)	ref	ref

BMI, body mass index; NHS II, Nurses’ Health Study II; VMS, vasomotor symptoms.

^aAdjusted for covariates: race, age at menarche, parity (2009), oral contraceptive use (2009), alcohol intake (2009), smoking (2009), BMI (2009).

DISCUSSION

Migraine occurring across the final menstrual period was associated with VMS presence, greater severity, and longer duration during overlapping time periods in midlife women. We found that individuals with migraine within 2 years surrounding the FMP had ~30% greater odds of experiencing VMS (68% of women with migraine vs 63% of women without migraine), and 24% greater odds of experiencing VMS lasting 5 years or more (35% vs 30%). Among women with VMS, individuals with migraine overall had 53% increased odds of severe VMS (8% vs 6%). This association appeared to be stronger among individuals with self-reported clinician-diagnosed migraine (78% greater odds) and migraine with aura (77% greater odds). The cross-sectional nature of this study limits causal inference. However, VMS and migraine are common, potentially debilitating conditions among midlife women with high public health relevance. This study highlights women with migraine as an at-risk group for severe and long-duration VMS.

Several previous studies have shown that the prevalence of migraine in premenopause and perimenopause is nearly double the prevalence of postmenopausal migraine, and that the frequency and severity of migraine improve after menopause.^5,7-10 Fewer studies have examined the association between migraine and VMS, the most common symptom of the menopause transition.²⁰ Only one small study (n = 728 women between 40 and 74 yr old) found a reduced prevalence of past-year migraine with aura among women with VMS, as compared with women without VMS (OR = 0.71, 95% CI: 0.26-1.66), and an increased prevalence of past-year migraine without aura among women with VMS (OR = 1.12, 95% CI: 0.71-1.72).³ While this study points to associations that may differ depending on the presence of aura, estimates were imprecise and consistent with negative, null, or positive associations.

In contrast, we found a consistently positive association between migraine (with or without aura) and VMS. Discrepancies in these findings may be due to the low prevalence of migraine with aura in the previous study (3%), and differing exposure windows for “active” or “recent” migraine relative to menopause (1 y in the prior study; 2 y in the current study). A second study¹⁰ found an OR of 1.5 (95% CI: 1.0-2.2) and 2.1 (95% CI: 1.3-3.4) for migraine prevalence among women with hot flashes and night sweats, respectively, compared with women without these symptoms. While our findings were directionally consistent, our study contributes importantly to the literature through further understanding of associations by migraine phenotype.

Our findings are consistent with a smaller recent study showing that individuals with migraine are more likely to experience more severe VMS.²¹ A 2023 cross-sectional study of 5,708 women aged 45-60 reported an odds ratio of 1.34 (95% CI: 1.08-1.66) for the association between migraine and the presence of severe to very severe VMS, as compared with no VMS. The present study’s estimates of migraine and severe vs. no VMS ranged from 1.68 (95% CI: 1.34-2.11) for migraine without aura to 2.20 (95% CI: 1.76-2.76) for self-reported clinician-diagnosed migraine (Supplemental Table 1, Supplemental Digital Content 1, http://links.lww.com/MENO/B396). Our study, in a much larger sample, suggests a stronger association between migraine and VMS severity and adds additional detail on how this association may vary across migraine characteristics.

Our findings build on the existing literature by examining VMS presence, severity, and duration among individuals self-reporting migraine surrounding FMP, those self-reporting a clinician diagnosis of migraine, and those reporting migraine with and without aura. While we conducted a cross-sectional analysis, this analysis was nested within a large prospective cohort study in which participants reported migraine and VMS within 2 years of the FMP. While a previous validation study among women in the general population found high correlation between self-reported migraine and migraine meeting International Classification of Headache Disorder criteria,²² our study population of professional nurses may more accurately differentiate between migraine and other headaches and characterize their migraine phenotype.

While restricting to individuals who were within 2 years of their FMP at the time of migraine and VMS assessment ensured that we focused on women undergoing the menopause transition, we were unable to establish temporality between midlife migraine and VMS in this analysis. We did not have sufficient data to examine incident migraine after VMS presentation or incident VMS after midlife migraine presentation. Therefore, we cannot determine whether migraine reflects shared underlying pathophysiology that may predispose an individual to VMS, if VMS triggers migraine, or if migraine triggers VMS.

Although overall misclassification of migraine and VMS is unlikely, there is likely some misclassification of migraine diagnosis and VMS duration due to questionnaire timing and content. Participants were asked if they received a migraine diagnosis from a health care provider until 1995. After 1995, migraine assessment focused on recent migraine symptoms. While participants receiving a new migraine diagnosis after 1995 (average age ~40 years) would be correctly categorized as having migraine overall, they would not be captured as having clinician-diagnosed migraine. In a large, representative study of migraine in the United States, the average age at migraine diagnosis much younger (24 yr, SD: 12 yr).²³ By collecting data on lifetime clinician diagnosis of migraine when participants were ~40 years old, we would expect to have captured most migraine diagnoses, limiting misclassification by this mechanism.

While our study was 4-30 times larger than previous studies and collected robust data on migraine phenotype, we did not have detailed data to examine this association across menopause stages. Although we collected data on an array of potential confounders, we did not collect data on underlying factors such as genetics or other menopause symptoms, which may influence the risk of migraine and VMS. We also did not account for hormone therapy use in our final models, as we found similar proportions of hormone therapy use within 2 years of FMP between individuals with and without migraine (15% vs 13%) and lacked enough detailed data to determine the exact temporality of migraine onset, hormone therapy use, and menopause symptoms. If women were prescribed or denied hormone therapy because of their migraine histories, adjusting for hormone therapy would lead to a biased estimate. Finally, our cohort was relatively homogenous, comprised of mostly White, well-educated women in the United States who reached menopause in the 2000s and early 2010s. Therefore, our findings may not be generalizable to other populations who differ from the NHS II cohort on traits that may influence migraine and VMS.

CONCLUSIONS

Overall, this study provides a detailed cross-sectional analysis nested within a large prospective cohort study that examines the association between robustly characterized migraine and VMS characteristics. Migraine is a potentially debilitating condition among midlife women, and severe and long-duration VMS were more frequently reported among women who experienced midlife migraine. These findings highlight women with midlife migraine as an at-risk group and underscore the relevance of risk assessment and screening for migraine and VMS burdens as women enter the menopause transition.

Supplementary Material

Supplemental Digital Content is available for this article. Direct URL citations are provided in the HTML and PDF versions of this article on the journal’s website, www.menopause.org.