Received: 2 January 2019 Accepted: 17 July 2019 DOI: 10.1002/gps.5180 RESEARCH ARTICLE Migraine and the risk of all‐cause dementia, Alzheimer's disease, and vascular dementia: A prospective cohort study in community‐dwelling older adults Rebecca E. Morton1 Philip D. St. John2 Suzanne L. Tyas1 1 School of Public Health and Health Systems, University of Waterloo, Waterloo, Ontario, Canada 2 Objectives: Dementia is the most common neurological disease in older adults; head- aches, including migraines, are the most common neurological disorder across all ages. Department of Medicine and Centre on Aging, Section of Geriatric Medicine, Max Rady College of Medicine, University of Manitoba, Winnipeg, Manitoba, Canada The objective of this study was to explore the relationship between migraines and Correspondence S. L. Tyas, PhD, Associate Professor, School of Public Health and Health Systems, University of Waterloo, Waterloo, Canada N2L 3G1. Email: styas@uwaterloo.ca from the Manitoba Study of Health and Aging, a population‐based, prospective Funding information National Health Research and Development Program of Health Canada, Grant/Award Number: 6606‐3954‐MC[S]; Manitoba Health sure (lifetime history of migraines), confounding (age, gender, education, and depres- dementia, including Alzheimer's disease (AD) and vascular dementia (VaD). Methods: Analyses were based on 679 community‐dwelling participants 65+ years cohort study. Participants screened as cognitively intact at baseline had complete data on migraine history and all covariates at baseline and were assessed for cognitive outcomes (all‐cause dementia, AD, and VaD) 5 years later. The association of exposion), and intervening variables (hypertension, myocardial infarction, other heart conditions, stroke, and diabetes) with all‐cause dementia and dementia subtypes (AD and VaD) was assessed using multiple logistic regression models. Results: A history of migraines was significantly associated with both all‐cause dementia (odds ratio [OR]=2.97; 95% confidence interval [CI]=1.25‐6.61) and AD (OR=4.22; 95% CI=1.59‐10.42), even after adjustment for confounding and intervening variables. Migraines were not significantly associated with VaD either before (OR=1.83; 95% CI=0.39‐8.52) or after (OR=1.52; 95% CI=0.20‐7.23) such adjustment. Conclusions: Migraines were a significant risk factor for AD and all‐cause dementia. Despite the vascular mechanisms involved in migraine physiology, migraines were not significantly associated with VaD in this study. Recognition of the long‐term detrimental consequences of migraines for AD and dementia has implications for migraine management, as well as for our understanding of AD etiology. K E Y W OR D S Alzheimer disease, Canada, cohort studies, dementia, headache, logistic models, migraine disorders, odds ratio, risk factors, vascular dementia 1 I N T RO D U CT I O N the most common neurological disorder across all ages, affecting almost half of the global population of adults.2 Migraines are the most As the global population ages, the prevalence of age‐related condi- debilitating form of headaches, affecting 20% of women and 8% of tions, such as dementia, will continue to rise.1 Although dementia is men.3 Both neurological disorders—dementia and migraines—cause the most common neurological disease in older adults, headaches are significant impacts on individuals and their families, as well as on Int J Geriatr Psychiatry. 2019;1–10. wileyonlinelibrary.com/journal/gps © 2019 John Wiley & Sons, Ltd. 1 2 MORTON ET AL. society more broadly. The relationship between migraines and dementia, including Alzheimer's disease (AD) and vascular dementia (VaD), Key points has not yet been clearly established and has the potential to inform • Evidence of an association between migraines and prevention and treatment as well as further understanding of the dementia is unclear and may vary by dementia subtype, etiology of these disorders. headache/migraine measure, gender, comorbidities, and Cognitive consequences of migraines have been more thoroughly clinical vs. community‐based sample. studied for cognition in general than for cognitive states such as dementia. Although one systematic review concluded that migraines • A history of migraines was a significant risk factor for were associated with mild cognitive changes across several domains,4 Alzheimer's disease and for all‐cause dementia, even particularly in patient populations, other reviews5,6 focusing on after adjustment for confounding and intervening stronger cohort study designs have concluded that current evidence variables in a community‐dwelling population of older does not support an association between migraines and cognitive adults. decline. However, evidence is less clear on the impact of migraines • A history of migraines was not significantly associated on dementia. A recent meta‐analysis7 of the limited number of cohort with vascular dementia. studies available noted substantial heterogeneity in these studies. Overall, it concluded that migraines did not increase the risk of dementia, although it did find an association with the broader category 75‐84, and >85 years), and the two oldest age groups were of headache disorders. An earlier meta‐analysis of four case‐control oversampled to ensure sufficient numbers of older participants. At studies found a marginally significant inverse relationship between baseline (1991/1992), 2890 people were contacted to be interviewed; 8 migraines and AD. In contrast, results from individual studies have 1751 people were eligible and agreed to participate (see Figure 1). suggested that migraines significantly increase the risk of dementia, The Modified Mini‐Mental State Exam (3MS) was used to screen for but that the effect may be restricted to specific subgroups (eg, cognitive impairment; participants scoring less than 78 on the 3MS women9) or subtypes of dementia (eg, VaD10), or that headache disor- were invited for an in‐depth clinical examination. Intact cognition 11 12,13 may be risk factors was defined as screening above the cutpoint on the 3MS or below rather than migraines specifically. Evidence of a significant association the cutpoint but subsequently assessed without dementia or any between migraines and dementia has been primarily based on patient lesser impairment (cognitive impairment no dementia or CIND)17 at populations, which are susceptible to selection biases. clinical examination. ders broadly or nonmigrainous headaches The purpose of this study was to determine if migraines are a risk At baseline, 1355 participants were cognitively intact and consti- factor for dementia and its subtypes, AD and VaD, in a population‐ tuted the incidence cohort. Age, gender, and education were deter- based prospective cohort study of community‐dwelling older adults. mined during this baseline interview; a questionnaire on lifetime The impact of possible intervening variables—hypertension, myocar- medical history, including history of migraines, was left with partici- dial infarction, other heart conditions, stroke, and diabetes—was also pants to be returned by mail (n=1039; 76.7%). The sample for this investigated. study was reduced from 1039 to 961 because of missing data on key MSHA‐1 variables (education n=19; lifetime histories of migraine n=23, depression n=23, myocardial infarction n=25, other heart condi- 2 METHODS tions n=37, stroke n=20, diabetes n=20, and hypertension n=26). Missing data for each variable were not mutually exclusive. In addition, 2.1 Sample 37 participants with CIND at follow‐up were excluded as they neither met criteria for dementia nor for the cognitively intact comparison The Canadian Study of Health and Aging (CSHA) is a population‐ group. The analytic sample was thus based on 679 individuals who based, prospective study of aging and dementia in Canada.14 Analyses screened cognitively intact at baseline, provided data on migraine were based on data from the Manitoba Study of Health and Aging history and potential confounders, and completed the cognitive (MSHA), a parallel study to the CSHA that used similar methods for assessment at follow‐up 5 years later (Figure 1). At follow‐up, demen- data collection and diagnoses, but expanded sampling in the province tia diagnoses were determined based on clinical examination, with of Manitoba, Canada.15,16 The community sample of the MSHA was all‐cause dementia diagnosed according to DSM‐IV criteria,18 AD derived by random sampling from the provincial health insurance according to NINCDS‐ADRDA criteria,19 and VaD according to list of adults 65 years and older. This list is one of the most complete NINDS‐AIREN.20,21 sampling frames available because the provincial health insurance plan provides almost universal coverage. Those excluded had federal health coverage (members of the Royal Canadian Mounted Police, 2.2 Statistical analysis the military, and First Nations [North American Indian] living on reserves) or resided in a remote, sparsely inhabited region. The study T tests with Satterthwaite's unequal variance assumption and Pearson population was stratified by health region and age group (65‐74, chi‐square tests, with Fisher's exact tests as appropriate, were used to MORTON 3 ET AL. FIGURE 1 Manitoba Study of Health and Aging study population and derivation of the analytic sample [Colour figure can be viewed at wileyonlinelibrary.com] analyze bivariate associations. The association of exposure (history of the limited number of VaD cases. All analyses were conducted using migraines), confounding (age, gender, education, and depression), and SAS version 9.2 (SAS Institute Inc., Cary, North Carolina). intervening variables (hypertension, myocardial infarction, other heart conditions, stroke, and diabetes) with dementia, AD, and VaD was analyzed using logistic regression models. Confounding and intervening 2.3 Standard protocol approvals, registrations, and patient consents variables assessed in the models were chosen based on supporting literature (eg, depressive symptoms are associated with dementia22 The Faculty of Medicine Committee on the Use of Human Subjects in and major depression increases the risk of migraine, while, in turn, Research at the University of Manitoba approved the original MSHA migraine increases the risk of major depression.23) These variables and MSHA‐2. The present study received ethics approval from the remained in the final models if they met levels of statistical significance Office of Research Ethics at the University of Waterloo. appropriate for regression modeling (alpha ≤.15 for main effects and ≤.05 for interaction terms).24 Standard diagnostics (eg, multicollinearity and residual diagnostics) were conducted to assess model fit. First‐ 3 RESULTS order interactions between migraine history and each of the potential confounding and intervening variables (age, gender, education, depres- The sample was predominantly women (61.9%) with a mean age of sion, diabetes, hypertension, stroke, heart attack, or other heart condi- 75.9 years (SD=6.1). At follow‐up 5 years later, 7.5% (n=51) of partic- tion) were assessed for dementia and AD, but not for VaD because of ipants had developed dementia, 5.1% (n=34) had developed AD, and 4 MORTON ET AL. 1.9% (n=12) had developed VaD. No men reporting a history of cognitively intact participants, those with dementia, AD, or VaD were migraines were diagnosed with dementia. In bivariate analyses significantly older, and those with dementia or AD had significantly (Table 1), migraines were significantly associated with AD: a history less education. Stroke was significantly more common in participants of migraines was noted in 23.5% of participants with AD compared with dementia or VaD than in those who were cognitively intact. with 9.9% of cognitively intact participants (p=.01). Compared with The association of other heart conditions with VaD was marginally significant (.05≤p<.10). TABLE 1 Bivariate analyses of the association of cognitive health outcomes with exposure, confounding and intervening variables in the Manitoba Study of Health and Aging Variables AD VaD (n=628) (n=34) (n=12) (%) (%) (%) (%) 19.6 23.5* 16.7 Exposure History of migraines the association between migraines and dementia strengthened (OR=3.28; 95% CI = 1.41‐7.21). Although stroke was not an independent statistically significant predictor of dementia, it met statistical criteria to enter the model and affected the strength of the association between migraines and dementia; it was thus included in the fully 9.9 adjusted model. This full model indicated that individuals with dementia were three times more likely (OR=2.97; 95% CI=1.25‐6.61) than Confounding variables Age, mean (SD) 2.23; 95% confidence interval [CI] = 1.06‐4.66) as those without dementia to have a history of migraines in unadjusted models (Table 2). After adjustment for confounding by age and education, Intact Cognitiona Dementia (n=51) Individuals with dementia were twice as likely (odds ratio [OR] = nondemented individuals to have a history of migraines (Table 2). 75.1 (5.7) 81.5*** (5.3) 81.5*** (5.6) 81.7** (5.5) 65‐74 years 45.5 11.8*** 11.8*** 16.7** adjustment for age and education, those with AD were four times 75‐84 years 48.3 56.9 55.9 50.0 more likely to have a history of migraines (OR=4.22; 95% CI=1.59‐ 6.2 31.4 32.4 33.3 unadjusted models (OR=2.81; 95% CI=1.22‐6.47) (Table 3). After Age 85+ years either dementia or AD models. 61.5 66.7 67.7 66.7 Educational level Did not complete primary school 10.42). None of the putative intervening variables contributed significantly to the model. No significant interactions were observed in Gender Female AD was significantly associated with a history of migraines in VaD was not significantly associated with a history of migraines in either unadjusted models (OR=1.83; 95% CI=0.39‐8.52) or models 6.9 19.6 23.5 16.7 adjusted for significant confounding by age and depression (OR=2.21; 95% CI=0.32‐9.77) (Table 4). When history of stroke was included in the model, however, the association between migraines Completed 47.8 primary school 45.1* 44.1** 58.3 Completed high school 27.4 21.6 14.7 25.0 Completed college/ university 18.0 13.7 17.7 0 8.6 11.8 8.8 25.0 Depressionb stroke was an intervening variable. Stroke was also a significant independent predictor of VaD. 4 DISCUSSION Identifying risk factors for dementia may facilitate early identification Intervening variablesb Hypertension and VaD weakened (OR=1.52; 95% CI=0.20‐7.23), suggesting that 31.7 41.2** 35.3 50.0*** Diabetes 6.4 9.8 11.8 0 Stroke 4.3 13.7 5.9 33.3 Myocardial infarction 8.0 9.8 8.8 8.3 Other heart condition 16.6 23.5 17.7 41.7 of at‐risk individuals and preventive strategies. Migraines were a significant risk factor for the development of all‐cause dementia and AD in this study. The significant association between migraines and dementia may be driven by the strong association between migraines and AD. This interpretation is supported by the weaker association for dementia than for AD, reflecting a dilution of the association with migraines across all types of dementia including VaD, where a significant association was not found. However, the limited number of VaD Abbreviations: AD: Alzheimer's disease; VaD: vascular dementia. a Excludes cognitively impaired, not demented. b Self‐reported lifetime history (presence). cases in our study precludes firm conclusions on the association between migraines and VaD. The significant increased risk of dementia with migraines noted in Comparison to cognitively intact group: this study is consistent with those of other cohort studies9,11,13 and *p<.05. in contrast with other individual studies12 and a meta‐analysis of **p<.01. cohort studies7 that found no association. Additional evidence sup- ***p<.001. ports an association of dementia with headaches overall (including MORTON 5 ET AL. TABLE 2 models Association of migraine history with dementia in the Manitoba Study of Health and Aging: Adjusted and unadjusted logistic regression Dementia Unadjusted Adjusted for Confounding Variables Adjusted for Confounding and Intervening Variables OR (95% CI) OR (95% CI) OR (95% CI) Exposure 2.23 (1.06‐4.66) Migraine history 3.28 (1.41‐7.21) 2.97 (1.25‐6.61) Confounding variables Age 65‐74 years 1.0 (reference) 75‐84 years 85+ years Gender (male vs. female) 1.0 4.36 (1.89‐11.90) 4.21 (1.82‐11.58) 21.94 (8.24‐66.06) 20.60 (7.73‐62.06) ns ns Educational level Did not complete primary school 1.0 1.0 Completed primary school 0.41 (0.18‐1.01) 0.41 (0.17‐1.01) Completed high school 0.32 (0.12‐0.87) 0.32 (0.12‐0.86) Completed college/university 0.30 (0.10‐0.88) 0.32 (0.10‐0.94) Depression a ns ns Intervening variablesa Hypertension ns Diabetes ns Stroke 2.52 (0.90‐6.42) Heart attack ns Other heart condition ns Note. n=679. Bolded values indicate statistically significant results. Abbreviations: CI: confidence interval; ns: not significant for inclusion in model (α>.15 for main effects); OR: odds ratio. a Self‐reported lifetime history (presence vs. absence). migraines)7,11 and nonmigrainous headaches.12,13 However, the asso- gender may occur in these dementia subtypes as noted for all‐cause ciation between migraines and dementia was significant only among dementia. An early meta‐analysis of case‐control studies8 found an 9 women in models stratified by gender : this same gender effect was overall inverse relationship between AD risk and migraines, with the observed for all headaches25 and nonmigrainous headaches.13 The association reaching significance only among women in three pooled present study reflected this strong association of migraines with studies. In contrast, although Tyas et al26 also noted a significant and dementia in women; it was unable to assess potential gender differ- stronger association between migraines and AD in women, the risk ences because no male participants had dementia as well as a history of AD was higher rather than lower in women. However, gender dif- of migraines. ferences with AD were not seen in a population‐based linkage Relatively few studies have focused on the association between study.10 It also did not find gender differences for VaD but did for migraines and subtypes of dementia, with reports of no association mixed dementia, with headaches a significant risk factor in men but 10,12 of AD with migraines 13 contrasting not in women. While significant associations between various defini- with the increased risk of AD with migraines seen in the current study. tions of headaches/migraines and dementia subtypes other than The current study builds on a previous overview of multiple risk fac- VaD and AD were found across a series of studies based on a health tors for AD26 that reported an association between migraines and insurance database in Taiwan,11,13,25 interpretation of this subtype is AD in the same data source by focusing in this study on migraine, unclear. In one study,25 this other dementia category excluded AD expanding the analysis to assess additional confounding variables and VaD yet contributed almost all of the dementia cases. Other stud- and investigate possible intervening variables and interactions, and ies using the same database11,13 defined their other dementia subtype extending the outcomes examined to include VaD and all‐cause based simply on exclusion of VaD and did not classify it further by AD. dementia. For VaD, our lack of an association with migraines was con- However, based on results from Yin et al,25 this dementia category 11,12 sistent with some or nonmigrainous headaches 10 but not all reports. Effect modification by would similarly be expected to be predominantly non‐AD dementia, 6 MORTON ET AL. TABLE 3 Association of migraine history with Alzheimer's disease in the Manitoba Study of Health and Aging: Adjusted and unadjusted logistic regression models Alzheimer's Disease Unadjusted Adjusted for Confounding Variables Adjusted for Confounding and Intervening Variables OR (95% CI) OR (95% CI) OR (95% CI) Exposure 2.81 (1.22‐6.47) Migraine history 4.22 (1.59‐10.42) 4.22 (1.59‐10.42) Confounding variables Age 65‐74 years 1.0 (reference) 75‐84 years 4.16 (1.51‐14.67) 1.0 4.16 (1.51‐14.67) 23.03 (7.12‐90.12) 85+ years Gender (male vs. female) ns 23.03 (7.12‐90.12) ns Educational level Did not complete primary school 1.0 Completed primary school 1.0 0.35 (0.14‐0.99) 0.35 (0.14‐0.99) Completed high school 0.18 (0.05‐0.61) 0.18 (0.05‐0.61) Completed college/university 0.32 (0.09‐1.05) 0.32 (0.09‐1.05) Depressiona Intervening variables ns ns a Hypertension ns Diabetes ns Stroke ns Heart attack ns Other heart condition ns Note. n=662. Bolded values indicate statistically significant results. Abbreviations: CI: confidence interval; ns: not significant for inclusion in model (α>.15 for main effects); OR: odds ratio. a Self‐reported lifetime history (presence vs. absence). raising issues of generalizability to other populations where AD would for the effects of stroke, consistent with the weakening of the associ- be a more common contributor to the prevalence of dementia. ation between migraines and VaD in our study after adjustment Diagnostic criteria for exposures and outcomes may also affect for stroke. In addition, as the authors note, the significant association whether a significant relationship between migraines and dementia is with VaD may reflect bias, because those with headaches or found. The studies used in the case‐control meta‐analysis for AD8 migraines may be more likely to seek medical care, and thus also more used a wide variety of self‐report migraine criteria, ranging from likely to be diagnosed with dementia.10 Since their study linked severe headaches to migraines, with one study validating self‐reports headache/migraine reports with a dementia registry, this may have with medical records. The current study studied migraines rather than created an association between headaches and dementia. headaches and relied on self‐report for this condition. Both headaches Thus, in addition to the influences of gender, dementia subtype, and migraines were investigated in Hagen et al.10 They found no and headache/migraine measure, differences in results may reflect significant association with AD. As reported elsewhere,12 the compar- the different sources used to recruit subjects. Unlike the current 10 was problematic in that participants did not study, other studies have included patient populations, either complete a cognitive assessment to determine if they were free of directly8,10,12 or through health insurance databases.9,11,13,25 Patient dementia or other cognitive impairment. Their AD cases would thus populations would disproportionately include those with more severe be more similar to their comparison group, and this may have contrib- migraines, as they would be more likely to seek medical attention for uted to the lack of significant association between migraines and AD management of their migraines. More severe migraines would also in their study. However, this would not explain why they found a be more likely to be associated with dementia than milder migraines, significant association with VaD while we did not. Hagen et al10 had assuming that there is a causal association between migraines and a larger sample size for VaD than our study and thus greater statistical dementia. Our study was not vulnerable to this bias because partici- power to detect an association. Another explanation may be that the pants were assessed for dementia as part of the study and not through association found in their study was affected by not fully accounting their interaction with the health care system. ison group in this study MORTON 7 ET AL. TABLE 4 Association of migraine history with vascular dementia in the Manitoba Study of Health and Aging: Adjusted and unadjusted logistic regression models Vascular Dementia Unadjusted Adjusted for Confounding Variables Adjusted for Confounding and Intervening Variables OR (95% CI) OR (95% CI) OR (95% CI) Exposure Migraine history 1.83 (0.39‐8.52) 2.21 (0.32‐9.77) 1.52 (0.20‐7.23) Confounding variables Age group 65‐74 years 1.0 (reference) 75‐84 years 3.04 (0.69‐21.00) 1.0 2.87 (0.64‐20.09) 17.38 (3.16‐133.44) 85+ years Gender (male vs. female) 14.28 (2.46‐112.64) ns ns Did not complete primary school ns ns Completed primary school ns ns Educational level Completed high school ns ns Completed college/university ns ns Depressiona 3.51 (0.73‐13.03) Intervening variables 2.95 (0.57‐11.61) a Hypertension ns Diabetes ns 7.90 (1.82‐29.71) Stroke Heart attack ns Other heart condition ns Note. n=640. Bolded values indicate statistically significant results. Abbreviations: CI: confidence interval; ns: not significant for inclusion in model (α>.15 for main effects); OR: odds ratio. a Self‐reported lifetime history (presence vs. absence). The overlap between the underlying biological mechanisms of migraines and dementia suggests that migraines may increase the risk vascular dysfunction in the brain, as vascular variables did not influence this association. of dementia. Vascular risk factors and events, such as diabetes, hyper- An alternative explanation for the relationship between migraines tension, heart attacks, and stroke, are related to the development of and AD is that long‐term neurological damage from migraines and sub- dementias, including AD.20,27-29 A relationship between vascular risk sequent development of late‐life adverse cognitive outcomes may be factors and migraines has also been observed.30-32 Migraines, in turn, specific to cognitive domains affected by migraine neurophysiology. are also a risk factor for various diseases and disorders, such as cardio- Clarification of the cognitive domains affected by migraines may aid 31,33 Many of the mechanisms involved in migraine in determining vulnerability to specific types of cognitive impairment neurophysiology, such as inflammation and reduced cerebral blood in late life. A greater understanding of the effect of migraine neuro- flow, are also underlying causes of dementia.34-36 Repeated activation physiology on cognitive function and decline has important clinical of these pathways in chronic migraineurs has been shown to cause implications including identifying, screening, and providing preventive vascular disease. 35 permanent neurological and vascular damage. The extent of this treatments for at‐risk individuals. damage may be related to the severity and frequency of migraine Examining the vascular aspects of long‐term migraine neurophysi- attacks; the varying levels of neurological damage may be similarly ology may clarify associations of migraine neurophysiology with stroke correlated with severity of dementia.37,38 Migraines have been linked and VaD. Individuals with cardiovascular risk factors are at high risk 39 although they may not modify the for stroke and VaD. Stroke was a significant risk factor for VaD, association between migraines and cognition.40 Thus, vascular factors consistent with previous reports and diagnostic criteria for VaD.29,41 were hypothesized in our study to be intervening variables in the Furthermore, migraines have been reported to be a significant risk association between a history of migraines and dementia. However, factor for stroke,31,32,36 which was also found in our study (data not while an association of migraines with all‐cause dementia and AD shown). A history of migraines was not a significant risk factor for was observed in this study, it did not appear to be explained by VaD in this study, despite the vascular mechanisms involved in to white matter abnormalities, 8 MORTON ET AL. migraine biology. This is consistent with our results that suggest An additional limitation was that the sample was restricted to par- vascular factors are not involved in the association between migraine ticipants with complete data. Participants who died during the follow‐ and AD. The lack of a significant association between VaD and up period (including those with dementia) were not included in the migraines may reflect the lack of a direct link between migraines and analyses because we could not diagnose cognitive status in those VaD and indicate that stroke is an intervening variable. We did see who did not survive until the follow‐up assessment. This introduces stronger intervening effects of vascular factors for VaD than for AD a potential selection bias that may affect the generalizability of the or overall dementia. However, although the association slightly weak- results. This is a standard limitation of longitudinal studies: the disad- ened after inclusion of stroke, there is no clear evidence that stroke vantage of attrition during the follow‐up period, however, is generally mediated the association between migraine and VaD and the results considered to be outweighed by the advantage of establishing tempo- for VaD should be interpreted with caution. The sample size for VaD rality (ie, migraine exposure data were collected before development was small, and we noted a nonsignificant trend towards a higher risk of dementia as we included only incident dementia cases) compared of VaD in those with migraines vs. those without. Our analyses thus with cross‐sectional study designs. This remains a limitation, however, may be underpowered to determine an effect. as loss to follow‐up, in particular attrition due to mortality, can influ- The association between migraines and AD may also be influenced by genetic factors. Individuals with familial AD due to presenilin‐1 ence incidence rates for dementia, as has been demonstrated in this same overall study population.46 mutations are more likely to suffer from migraines or recurrent head- Although the established higher prevalence of both migraines and aches.42 Research has implicated chromosomes 1 and 19 in both dementia in women compared with men is logically reflected in fewer migraines and AD.43,44 Although neither APOE nor MTHFR genotype male migraineurs with dementia, the absence in our study of any male modified the association between migraine and cognitive decline,40 participants with migraines who developed dementia meant that it further studies of these or other genotypes may help to explain the was not feasible to assess effect modification by gender. Clarifying association between migraines and AD and to identify high‐risk these gender effects on dementia overall as well as subtypes of individuals. dementia is an important area for further study. The relatively small Limitations of this study include self‐reported data at baseline, number of VaD cases was also a limitation in this study as it decreased such as the single migraine measure. Migraine data did not include the statistical power of the VaD analytic models and precluded the medical records with diagnoses based on standardized migraine assessment of interactions. Although no effect modification by APOE criteria (ie, International Headache Society ‐ The International Clas- or MTHFR genotype has been reported for migraines and cognition,40 sification of Headache Disorders II [IHS: ICHD‐II]). However, self‐ the lack of genetic data to explore potential effect modification for reported migraine has been shown to have excellent agreement dementia outcomes is also a limitation. with the IHS: ICHD‐II criteria for migraine diagnoses in a large Strengths of this study include data from a large, prospective population‐based study.45 The MSHA migraine measure did not distin- cohort study. The study design allowed for identification of incident guish between migraines with and without aura. Although there are cases of AD, VaD, and all‐cause dementia, reducing methodological some reports of more severe cognitive consequences in migraines concerns, such as survival bias, common to studies using prevalent with aura,4 the effect of aura on the association between migraine cases. Our population‐based sample of older adults living in the com- and cognitive impairment appears to be inconsistent.6 This study munity reduced the selection biases common to patient samples found a significant association between migraines and AD despite restricted to those seeking medical care for migraines or dementia. the inclusion of migraine without aura in the general migraine mea- Whereas previous evidence has primarily been based on these patient sure. The inclusion of migraine without aura would dilute this effect populations, we found a significant association of migraine history if migraine with aura causes greater neurological damage, and thus, with increased risk of dementia and AD in a community‐dwelling, our findings that individuals with AD were four times more likely to population‐based sample. Clinical assessment and diagnosis were con- have a history of migraines may be a conservative estimate. However, ducted using established criteria. Analyses included consideration of it is also possible that migraine with aura may have more of an impact confounding and intervening variables. on the association between migraine and VaD due to the vascular components of migraine with aura. Nevertheless, significant associations of migraines with AD and all‐cause dementia were noted in our 5 CO NC LUSIO NS study, even with a single, simple migraine measure. Additional data on migraine characteristics, such as use of migraine medications, Migraines were a significant risk factor for AD and all‐cause dementia frequency of migraine attacks, and severity or intensity of migraine in a population‐based prospective cohort study of community‐ attacks, would aid in determining the effects of migraines on cognitive dwelling older adults. Evidence did not support an association outcomes.4-6 Differences across migraineurs in factors unrelated to between migraines and VaD. Identifying predictors of dementia is crit- migraine (eg, comorbidities) may also help to explain inconsistent ical, given the current increases and expected further growth in the associations between migraines and cognition.4,6 Although our data proportion of older adults in the population. Identifying a midlife risk on migraine characteristics were limited, analyses did incorporate data factor for dementia, such as migraines, enables earlier detection of on comorbidities, such as depression and cardiovascular disease. at‐risk individuals in addition to contributing to our understanding of MORTON 9 ET AL. AD etiology. It also provides a rationale for the development of new preventive strategies for AD and treatments targeting migraines and associated intervening variables. Implications for clinical practice include earlier screening for cognitive decline in migraine sufferers, as well as more aggressive treatment of potential intervening variables to delay dementia, improve quality of life, and increase the likelihood of healthy aging. Acknowledgements The Manitoba Study of Health and Aging (MSHA‐1) was funded primarily by Manitoba Health, with additional funding provided through the Canadian Study of Health and Aging by the Seniors Independence Research Program of the National Health Research and Development Program of Health Canada (project no. 6606‐3954‐MC[S]). The Mani- 6. Suhr JA, Seng EK. Neuropsychological functioning in migraine: clinical and research implications. Cephalalgia. 2012;32(1):39‐54. 7. Wang J, Xu W, Sun S, Yu S, Fan L. Headache disorder and the risk of dementia: a systematic review and meta‐analysis of cohort studies. J Headache Pain. 2018;19(1):95. 8. Breteler MM, van Duijn CM, Chandra V, et al. Medical history and the risk of Alzheimer's disease: a collaborative re‐analysis of case‐control studies. EURODEM Risk Factors Research Group. Int J Epidemiol. 1991;20(2):S36‐S42. 9. Lee SY, Lim JS, Oh DJ, Kong IG, Choi HG. Increased risk of neurodegenerative dementia in women with migraines: a nested case– control study using a national sample cohort. Medicine. 2019;98(7): e14467. 10. Hagen K, Stordal E, Linde M, Steiner TJ, Zwart J‐A, Stovner LJ. Headache as a risk factor for dementia: a prospective population‐based study. Cephalalgia. 2014;34(5):327‐335. toba Study of Health and Aging follow‐up study (MSHA‐2) was funded primarily by Manitoba Health's Healthy Communities Development 11. Tzeng N‐S, Chung C‐H, Lin F‐H, et al. Headaches and risk of dementia. Am J Med Sci. 2017;353(3):197‐206. Fund with additional funding provided through the Canadian Study of Health and Aging by the Seniors Independence Research Program of 12. Stræte Røttereng AK, Bosnes O, Stordal E, et al. Headache as a predictor for dementia: the HUNT Study. J Headache Pain. 2015;16(1):89. the National Health Research and Development Program of Health on Aging or Manitoba Health is intended or should be inferred. 13. Yang F‐C, Lin T‐Y, Chen H‐J, Lee J‐T, Lin C‐C, Kao C‐H. Increased risk of dementia in patients with tension‐type headache: a nationwide retrospective population‐based cohort study. PLoS ONE. 2016;11(6): e0156097. Conflicts o f int erest 14. McDowell I, Hill G, Lindsay J. An overview of the Canadian Study of Health and Aging. Int Psychogeriatr. 2001;13(1):7‐18. Canada (project no. 6606‐3954‐MC[S]). The results and conclusions are those of the authors and no official endorsement by the Centre 15. MSHA Research Group. Manitoba Study of Health and Aging: technical section. Winnipeg, MB: Centre on Aging, University of Manitoba; 1995. None declared. D a t a a v a il a b i l i t y s t at e m e n t The data that support the findings of this study are available upon request from the data access committee of the Manitoba Study of Health and Aging. Restrictions apply to the availability of these data, which were used under license for this study. Data are available upon request from the authors with the permission of the data access committee of the Manitoba Study of Health and Aging. ORCID Suzanne L. Tyas https://orcid.org/0000-0003-3007-3940 RE FE R ENC E S 1. Prince M, Bryce R, Albanese E, Wimo A, Ribeiro W, Ferri CP. The global prevalence of dementia: a systematic review and metaanalysis. Alzheimers Dement. 2013;9(1):63‐75. 2. Stovner LJ, Hagen K, Jensen R, et al. The global burden of headache: a documentation of headache prevalence and disability worldwide. Cephalalgia. 2007;27(3):193‐210. 3. Diener HC, Beck CA. Migraine and risk of cardiovascular disease in women: learning about relative and absolute risk. Neurology. 2009;73(8):576‐577. 4. de Araújo CM, Barbosa IG, Lemos SMA, Domingues RB, Teixeira AL. Cognitive impairment in migraine: a systematic review. Dement Neuropsychol. 2012;6(2):74‐79. 5. Rist PM, Kurth T. Migraine and cognitive decline: a topical review. Headache. 2013;53(4):589‐598. 16. MSHA‐2 Research Group. Follow‐up to the Manitoba Study of Health and Aging (MSHA‐2): methodology. Winnipeg, MB: Centre on Aging, University of Manitoba; 1998. 17. Tuokko HA, Frerichs RJ, Kristjansson B. Cognitive impairment, no dementia: concepts and issues. Int Psychogeriatr. 2001;13(1):183‐202. 18. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994. 19. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer's disease: report of the NINCDS‐ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease. Neurology. 1984;34(7): 939‐944. 20. Hebert R, Lindsay J, Verreault R, Rockwood K, Hill G, Dubois MF. Vascular dementia: incidence and risk factors in the Canadian Study of Health and Aging. Stroke. 2000;31(7):1487‐1493. 21. Roman GC, Tatemichi TK, Erkinjuntti T, et al. Vascular dementia: diagnostic criteria for research studies. Report of the NINDS‐AIREN International Workshop. Neurology. 1993;43(2):250‐260. 22. Diniz BS, Butters MA, Albert SM, Dew MA, Reynolds CF. Late‐life depression and risk of vascular dementia and Alzheimer's disease: systematic review and meta‐analysis of community‐based cohort studies. Br J Psychiatry. 2013;202(5):329‐355. 23. Breslau N, Lipton RB, Stewart WF, Schultz LR, Welch KM. Comorbidity of migraine and depression: investigating potential etiology and prognosis. Neurology. 2003;60(8):1308‐1312. 24. Tyas SL, Koval JJ, Pederson LL. Does an interaction between smoking and drinking influence the risk of Alzheimer's disease? Results from three Canadian data sets. Stat Med. 2000;19(11‐12):1685‐1696. 10 25. Yin J‐H, Tsai C‐L, Lee P‐J, et al. Age‐specific and gender‐dependent impact of primary headache disorders on dementia risk: population‐ based longitudinal study. Medicine. 2018;97(52):e13789. 26. Tyas SL, Manfreda J, Strain LA, Montgomery PR. Risk factors for Alzheimer's disease: a population‐based, longitudinal study in Manitoba, Canada. Int J Epidemiol. 2001;30(3):590‐597. 27. de la Torre JC. Cardiovascular risk factors promote brain hypoperfusion leading to cognitive decline and dementia. Cardiovasc Psychiatry Neurol. 2012;2012:1‐15. 28. Duron E, Hanon O. Vascular risk factors, cognitive decline, and dementia. Vasc Health Risk Manag. 2008;4(2):363‐381. 29. Patterson C, Feightner J, Garcia A, MacKnight C. General risk factors for dementia: a systematic evidence review. Alzheimers Dement. 2007;3(4):341‐347. 30. Bigal ME, Kurth T, Santanello N, et al. Migraine and cardiovascular disease: a population‐based study. Neurology. 2010;74(8):628‐635. 31. KurthT, Schurks M, Logroscino G, Buring JE. Migraine frequency and risk of cardiovascular disease in women. Neurology. 2009;73(8):581‐588. 32. Scher AI, Terwindt GM, Picavet HS, Verschuren WM, Ferrari MD, Launer LJ. Cardiovascular risk factors and migraine: the GEM population‐based study. Neurology. 2005;64(4):614‐620. 33. Adelborg K, Szépligeti SK, Holland‐Bill L, et al. Migraine and risk of cardiovascular diseases: Danish population based matched cohort study. BMJ. 2018;360:k96. 34. Bernecker C, Ragginer C, Fauler G, et al. Oxidative stress is associated with migraine and migraine‐related metabolic risk in females. Eur J Neurol. 2011;18(10):1233‐1239. 35. Hamed SA, Hamed EA, Ezz Eldin AM, Mahmoud NM. Vascular risk factors, endothelial function, and carotid thickness in patients with migraine: relationship to atherosclerosis. J Stroke Cerebrovasc Dis. 2010;19(2):92‐103. 36. Scher AI, Gudmundsson LS, Sigurdsson S, et al. Migraine headache in middle age and late‐life brain infarcts. JAMA. 2009;301(24):2563‐2570. 37. O'Bryant SE, Marcus DA, Rains JC, Penzien DB. The neuropsychology of recurrent headache. Headache. 2006;46(9):1364‐1376. MORTON ET AL. 38. Calandre EP, Bembibre J, Arnedo ML, Becerra D. Cognitive disturbances and regional cerebral blood flow abnormalities in migraine patients: their relationship with the clinical manifestations of the illness. Cephalalgia. 2002;22(4):291‐302. 39. Bashir A, Lipton RB, Ashina S, Ashina M. Migraine and structural changes in the brain: a systematic review and meta‐analysis. Neurology. 2013;81(14):1260‐1268. 40. Rist PM, Dufouil C, Glymour MM, Tzourio C, Kurth T. Migraine and cognitive decline in the population‐based EVA study. Cephalalgia. 2011;31(12):1291‐1300. 41. Hachinski VC, Iliff LD, Zilhka E, et al. Cerebral blood flow in dementia. Arch Neurol. 1975;32(9):632‐637. 42. Ringman JM, Romano JD, Medina LD, et al. Increased prevalence of significant recurrent headache in preclinical familial Alzheimer's disease mutation carriers. Dement Geriatr Cogn Disord. 2008;25(4):380‐384. 43. Breslau N, Rasmussen BK. The impact of migraine: epidemiology, risk factors, and co‐morbidities. Neurology. 2001;56(6 Suppl 1):S4‐S12. 44. Burns A, Iliffe S. Alzheimer's disease. BMJ. 2009;338:b158. https://doi. org/10.1136/bmj.b158 45. Schurks M, Buring JE, Kurth T. Agreement of self‐reported migraine with ICHD‐II criteria in the Women's Health Study. Cephalalgia. 2009;29(10):1086‐1090. 46. Tyas SL, Tate RB, Wooldrage K, Manfreda J, Strain LA. Estimating the incidence of dementia: the impact of adjusting for subject attrition using health care utilization data. Ann Epidemiol. 2006;16(6): 477‐484. How to cite this article: Morton RE, St. John PD, Tyas SL. Migraine and the risk of all‐cause dementia, Alzheimer's disease, and vascular dementia: A prospective cohort study in community‐dwelling older adults. Int J Geriatr Psychiatry. 2019;1–10. https://doi.org/10.1002/gps.5180