Migraine,Cognitive Decline,and Dementia in Older Adults: A Population-Based Study

Study design and participants

This is a population-based cohort study. Participants were derived from the Swedish National study on Aging and Care in Kungsholmen (SNAC-K), an ongoing prospective follow-up study of participants who were aged≥60 years and living in the Kungsholmen district, an area of central Stockholm, Sweden, as previously described [21, 22]. In brief, the SNAC-K sample consists of 11 age-stratified cohorts that include four young-old cohorts with a 6-year interval (60, 66, 72, and 78 years) and seven older cohorts with a 3-year interval (81, 84, 87, 90, 93, 96, and 99+ years). Baseline data were collected from 2001 to 2004. Follow-up assessments were performed every 6 years for the young-old cohorts and every 3 years for the older cohorts. This follow-up procedure was undertaken due to more rapid changes in health conditions and higher attrition rates among older cohorts compared with young-old cohorts. Data for the current analysis were derived from baseline examination (2001–2004) and four follow-up assessments in 2004–2007, 2007–2010, 2010–2013, and 2013–2016.

Out of the 4,590 individuals who were alive and eligible to participate in SNAC-K, 3,363 (73.3%) completed the baseline examination [22, 23]. Of these, 3,069 participants with sufficient information on headache were included in the cross-sectional analysis. Participants with prevalent dementia at baseline (n = 175) and no follow-up dementia assessments (n = 711) were excluded, leaving 2,183 participants for the longitudinal analysis of baseline migraine status in association with incident dementia. When analyzing the association of headache and migraine with global cognitive function measured with the Mini-Mental State Examination (MMSE): Of 3,069 participants, we excluded 5 participants with missing data on baseline MMSE and 259 persons with a baseline MMSE score≤24, leaving 2,805 participants for the cross-sectional analysis; Of these 2,805 participants, we further excluded 665 persons without any follow-up MMSE data, leaving 2,140 participants for the longitudinal analysis.

All parts of the SNAC-K population study, including linkage with data from the Swedish Patient Register and Cause of Death Register, were approved by the Ethics Committee at Karolinska Institutet or by the Regional Ethical Review Board in Stockholm, Sweden. Written informed consent was collected from the participants or, in the case of cognitively impaired persons, from a proxy (usually next of kin or guardians).

Data collection

In SNAC-K, data were collected at baseline and follow-ups by the trained nurses and physicians through face-to-face interviews, clinical examinations, and laboratory tests, as previously described [21–23]. We collected data on demographic features (e.g., age, sex, and education), lifestyles (e.g., smoking, alcohol consumption, and physical activity), health history (e.g., hypertension, diabetes, and heart disease), and blood biomarkers. Arterial blood pressure was measured twice at a 5-min interval in a sitting position on the left arm with a sphygmomanometer, and the mean of the two readings was used in the analysis. Peripheral blood samples were taken, and APOE genotype, total cholesterol, and glycated hemoglobin were measured at the university’s laboratory. In addition, health conditions and vital status were ascertained through linkage to the Swedish Patient Register and Cause of Death Register that were linked with SNAC-K database through individual identification number. Diseases in the Patient Register were classified and coded according to the criteria of the International Classification of Diseases, tenth revision (ICD-10). Information on death date and the cause of death for persons who died during the follow-up period was extracted from the Cause of Death Register.

Assessments of migraine status

At baseline, participants were asked a series of questions about current and history of headache (e.g., are you having or have had recurrent headaches?), including headache duration (e.g., how long does/did each headache attack last?), characteristics (e.g., do/did headaches appear one-sided, have pulsating quality, or get worse with physical activity?), related symptoms (e.g., are/were there any symptoms during headaches such as nausea and/or vomiting or sensitivity to sound and/or light?), and presence or absence of aura. A lifetime history of migraine was defined based on the International Classification of Headache Disorders, second edition (ICHD-II) [24]. Briefly, participants with headache attacks lasting 4–72 hours with at least 2 characteristics (unilateral location, pulsating quality, and aggravation by or causing avoidance of physical activity) and at least one related symptom (nausea and/or vomiting, photophobia, and phonophobia) were regarded as having migraine either with aura (MA) or without aura (MO). Participants with a history of headache not fulfilling the criteria for migraine were considered as having non-migraine headache.

Assessments of cognitive function and diagnosis of dementia

At both baseline and follow-up examinations, global cognitive function was assessed using the MMSE. Dementia was diagnosed according to the criteria outlined in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) [25], following a 3-step diagnostic procedure [22], i.e., the examining physician made a preliminary diagnosis of dementia; then a reviewing physician independently made a second preliminary diagnosis; a third opinion from a neurologist was sought in case of disagreement between the two preliminary diagnoses. For participants who died during the follow-up periods, physicians made a diagnosis of dementia by carefully reviewing the clinical charts and medical records as well as the discharge diagnoses from the Swedish Patient Register and death certificates from the Cause of Death Register.

Assessments of covariates

Demographics (e.g., age, sex, and education), lifestyles (e.g., smoking, physical activity, and alcohol consumption), health conditions (e.g., hypertension, diabetes, and cardiovascular diseases), use of medication, and APOE genotype were considered as covariates. Current use of medications was classified following the Anatomical Therapeutic Chemical (ATC) classification system. Education was measured as the maximum years of formal schooling, and was divided into elementary school, middle or high school, and university. Smoking status was dichotomized into current versus non-current smoking. Alcohol consumption as no or occasional, light to moderate, and heavy drinking. Physical inactivity was defined as participating less than 3 times per month in light-to-moderately intense exercise. We defined hypertension as arterial blood pressure≥140/90 mmHg or current use of antihypertensive agents (ATC codes C02, C03, and C07-C09). Diabetes was ascertained according to self-reported history, records of diabetes in the inpatient register (ICD-10 codes E10-E14), use of oral hypoglycemic agents or insulin injection (ATC code A10), or glycated hemoglobin≥6.5%. Total cholesterol levels were classified as normal (< 5.17 mmol/L), borderline high (5.17–6.21 mmol/L), and high cholesterol (≥6.22 mmol/L). Heart disease was ascertained from a self-reported history and through the Swedish Patient Register, including coronary heart disease (ICD-10 codes I20-I25), atrial fibrillation (ICD-10 code I48), and heart failure (ICD-10 code I50). Similarly, cerebrovascular disease was defined by a self-report history of stroke or records in the patient register (ICD-10 codes I60-I69). Depression was identified as current and previous diagnosis of depression based on the self-administered questionnaire. APOE genotype was dichotomized into any ɛ4 allele versus no ɛ4 allele. Antimigraine medications included specific antimigraine medications that covered ergot alkaloids (ATC code N02CA), selective serotonin agonists (ATC code N02CC), and other antimigraine drugs (ATC code N02CX), and non-specific antimigraine medications that included anti-inflammatory and anti-rheumatic products, and non-steroids (ATC code M01A), salicylic acid and derivatives (ATC code N02BA), and paracetamol (ATC code N02BE01). People who took at least one specific or non-specific antimigraine medication were considered as users of specific or non-specific antimigraine medications.

Statistical analysis

Baseline characteristics of participants by migraine status were compared using analysis of covariance for continuous variables and chi-square test for categorical variables. Logistic and Cox proportional-hazard models were used to assess the cross-sectional and longitudinal associations of migraine status with prevalent and incident dementia, respectively. Linear mixed-effects models were used to examine both the cross-sectional and longitudinal associations between migraine status and MMSE score, taking into account the effect of random intercept and random slope. The linear-mixed effects models were fitted by including migraine status, follow-up time, and an interaction term between migraine status and follow-up time together with covariates. The β coefficient for migraine status reflects the mean differences in MMSE score across groups at baseline, and the β coefficient for the interaction term represents the differences in the average rate of MMSE score change associated with baseline migraine status. We tested the statistical interaction of migraine status with age (< 78 versus≥78 years), sex, cerebrovascular disease, use of specific and non-specific antimigraine medications, or APOE ɛ4 allele on cognitive outcomes. When a statistical interaction was detected, we further performed stratifying analysis to verify the direction and magnitude of the interaction. Finally, we repeated the analyses by classifying persons with non-migraine headache into the reference group. We reported the main results from two models: model 1 was adjusted for age, sex, and education; and model 2 was additionally adjusted for smoking, alcohol consumption, physical activity, hypertension, diabetes, total cholesterol, heart disease, cerebrovascular disease, depression, APOE ɛ4 allele, and use of antimigraine medications. In addition, we performed a sensitivity analysis by excluding the dementia cases diagnosed only based on clinical charts and death certificates among deceased participants during the follow-up periods.

We used IBM SPSS Statistics Version 23 for Windows (IBM Corp. Released 2015. Armonk, NY: IBM Corp) for all the analyses. We considered the 2-tailed test p < 0.05 as statistical significance.

Conclusion

This population-based cohort study of Swedish older adults showed no evidence supporting the association of migraine and its two main subtypes with cognitive decline and dementia. Given the high prevalence of migraine in the general population, the lack of epidemiological evidence for the migraine-cognition association may provide reassuring evidence for migraineurs, with regard to the cognitive consequences of migraine. However, further research is warranted to clarify why observational epidemiological studies do not support the apparently plausible hypothesis of migraine in association with cognitive consequences in old age.