Increased risk of all-cause, Alzheimer’s, and vascular dementia in adults with migraine in Korea: a population-based cohort study

Migraine is a common primary headache disorder characterized by episodic disabling headaches and is often accompanied by focal neurological symptoms called aura [1]. Migraine affects approximately 15% of the general population worldwide and ranks second among the top causes of disability [2, 3]. While migraine is most prevalent in young and middle-aged adults, dementia, another leading cause of disability, primarily affects the geriatric population [4].

Previous studies investigating the association between migraine and the risk of dementia have shown inconsistent results. Several cohort studies have reported an increased risk of all-cause dementia, Alzheimer’s dementia (AD), or vascular dementia (VaD) in patients with migraine compared to non-migraine individuals, while others did not find such relationships [5‐8]. Moreover, in two studies, an association between migraine and the risk of dementia was observed only in women [9, 10]. In addition, studies on the relationship between migraine and cognitive function have shown inconsistent findings [11‐13]. The heterogeneity of the results among studies may be attributed to differences in their designs. Specifically, the identification of patients with migraine (incident cases or retrospectively evaluated at baseline), duration of follow-up (5–24 years), the age distribution of the study population, and outcomes (all-cause dementia, AD, or VaD) varied among the studies.

Considering the long preclinical stage of AD and the gap between the prevalent age of migraine and dementia, a sufficient follow-up period should be evaluated to ensure the longitudinal relationship between the two disorders [14]. Additionally, since each cause of dementia (e.g., AD and VaD) has a different pathophysiology, separate analyses investigating the association between migraine and each type of dementia should be performed. A sufficient sample size is required to secure the statistical power to perform separate analyses for each type of dementia. Another research interest is migraine aura, a known risk factor for various conditions, including ischemic stroke and myocardial infarction [15‐17]. Islamoska et al. reported that migraine with aura (MA) was associated with an increased risk of dementia in a Danish cohort study [7]. However, there is still insufficient evidence for the association between MA, migraine without aura (MO), and dementia among the Asian population.

Therefore, this study aimed to investigate the risk of developing all-cause dementia, AD, VaD, and other specified and unspecified dementias after migraine diagnosis over a 16-year follow-up period using Korean population-based data. To assess the temporal context of the association between migraine and dementia, additional analyses were performed according to the age at migraine diagnosis and follow-up duration. Finally, based on the findings of previous studies, we evaluated whether the presence of aura or sex affects the association between migraine and dementia.

Between January 1, 2002, and December 31, 2019, 54,870 individuals met the inclusion criteria for the migraine cohort. Of these, 10,675 individuals were excluded because they had medical records for migraine (n = 6,234) or dementia during the washout period (n = 42), developed dementia before migraine diagnosis (n = 1,472), were included as controls for other patients with migraine (n = 2,745), or showed insufficient matching results (n = 182). The final study sample comprised 44,195 patients with migraine and 44,195 matched controls (Fig. 1). The mean follow-up time was 7.84 years (maximum, 16 years), and 692,867 person-years were generated. There were 8,557 patients with new-onset dementia during the 16-year follow-up period. In the baseline year of 2004, 66.1% of the study subjects were female, and the mean age was 55.3 (± 9.4) years. The standardized differences for all covariates were < 0.1 (Table 1).

The cumulative incidence of dementia during the entire study period showed a statistically significant difference between the migraine and control cohorts (p < 0.001, stratified log-rank test; Fig. 2).

Among the 44,195 patients with migraine, 4,800 developed dementia over 343,871 person-years (IR, 139.6 per 10,000 person-years; 95% CI, 135.7–143.5), while 3,757 developed dementia over 348,995 person-years (IR, 107.7 per 10,000 person-years; 95% CI, 104.3–111.1) among the 44,195 control subjects. Patients with migraine were 1.30 times more likely to develop dementia than their matched controls (HR, 1.30; 95% CI, 1.25–1.35). Patients with migraine showed 1.42 times (HR, 1.42; 95% CI, 1.34–1.51), 1.26 times (HR, 1.26; 95% CI, 1.14–1.40), and 1.16 times (HR, 1.16; 95% CI, 1.05–1.28) higher risk of developing dementia than their matched controls at 0–5, 6–10, and > 10 years after migraine diagnosis, respectively. However, the magnitudes of association between migraine and risk of dementia were similar between sexes, between groups by age at migraine diagnosis, and among groups by the presence of aura (Table 2).

Patients with migraine had 1.29 times (HR, 1.29; 95% CI, 1.23–1.35), 1.35 times (HR, 1.35; 95% CI, 1.19–1.54), 1.36 times (HR, 1.36; 95% CI, 1.00–1.83), and 1.30 times (HR, 1.30; 95% CI, 1.17–1.45) higher risk of developing AD, VaD, mixed or other specified dementias, and unspecified dementia than their matched controls, respectively (Table 3).

Sensitivity analyses showed that the association between migraine and all-cause dementia was significant when only migraine cases before the age of 60 were included. Patients with migraine also showed higher incidence of each type of dementia in sensitivity analyses. However, statistical significance was observed only for AD in analyses with 2-year washout period whereas VaD and unspecified dementia showed significant relations to migraine in analyses wtih 5-year washout period (Supplementary Table S1, S2).

In this study, we found that patients with migraine exhibited an increased risk of developing all-cause dementia, AD, VaD, and mixed or other specified dementias, and unspecified dementia compared to non-migraine-matched controls. Although the mechanism of this association remains largely unknown, there are several possible explanations for the pathways linking migraine and dementia.

First, migraine is associated with an increased risk of myocardial infarction and ischemic stroke, which are known risk factors for AD or VaD [15, 25]. Second, studies have found that migraine is related to structural and functional brain changes, including alterations in cerebral blood flow, increased white matter hyperintensities, subclinical infarct-like lesions, and brain volume changes [26‐28].

Other possible mechanisms linking migraine and dementia include alterations in the cortisol-hippocampal pathway, inflammation, increased amyloid plaque formation, deficits in nerve growth factors due to comorbid depression, and chronic pain-related changes in the memory network structure of brain [29‐31].

In the subgroup analyses, patients with migraine showed a higher risk of developing dementia than their matched controls, regardless of sex, age at migraine diagnosis, and follow-up duration. However, the magnitude of the association was prominent in patients diagnosed with migraine less than five years ago and attenuated with longer follow-up periods, indicating that a reverse causation (e.g., early diagnosis of dementia after consulting neurologists for migraine treatment) or a shared underlying cause between migraine and dementia may exist. Nevertheless, our findings provide evidence of longitudinal relationship between migraine and dementia development, since there was an increased risk of dementia in patients who had been diagnosed with migraine > 10 years ago or those who were diagnosed with migraine before the age of 60 years (when dementia was less prevalent).

Although the magnitudes of the association were found to be similar between MA and MO, more than half of the patients with migraine did not have information on aura. Moreover, since MA is known to have a higher risk of cardiovascular and cerebrovascular events, or structural abnormalities in the brain than MO, it is difficult to deny the theoretical relevance between MA and dementia with our findings alone [7]. Therefore, the impact of aura on dementia should be re-investigated with an accurate evaluation of migraine aura. Furthermore, future studies need to explore whether the characteristics of migraine attacks, such as severity or frequency, affect the development of dementia.

Contrary to our findings, most previous cohort studies did not find a statistically significant association between migraine and the risk of VaD [5, 32, 33]. The non-significant association may be attributed to the small sample size and lack of statistical power, because those studies showed consistent trends toward an increased risk of VaD in patients with migraine. Using a large sample with a nationwide cohort that provided sufficient statistical power, our study found that migraine was associated with an increased risk of VaD and AD. Moreover, we emulated the effect of prospective studies and reduced the immortal time bias in retrospective claims-based data using the risk-set matching method.

However, certain limitations of this study should be considered when interpreting its findings. First, several variables that could affect the development of dementia, such as educational level or baseline cognitive function, were not available for NHIS-HEALS. However, previous studies reported that educational level was not associated with migraine, or that individuals with higher educational levels were more prevalent with migraine [34, 35]. Additionally, we selected control individuals with the same baseline household income level as patients with migraine, and educational level was strongly correlated with income in South Korea [36]. Thus, in our study, educational level as a confounder had a minimal effect on the relationship between migraine and dementia. Second, the overall age of migraine onset among subjects in this study was greater than the peak prevalent age (35–39 years) of migraine because NHIS-HEALS comprised individuals aged 40–79 years in 2002. Therefore, a portion of the migraine cohort might comprise patients with active period of migraine instead of incident migraine cases. Although sensitivity analyses showed similar results to those of main analysis, findings for each type of dementia were not robust due to a small number of events. Third, because a large proportion of participants did not have information on presence of aura, we cannot draw a robust conclusion about the role of migraine aura on development of PD despite the evidence presented in previous studies and the biologically plausible mechanism. Last, the diagnosis of migraine, dementia, and other comorbidities may be inaccurate, owing to the inherent limitations of claim data. Therefore, to improve diagnostic validity and reduce false-positive cases, we required at least two diagnoses for migraine and at least one hospital admission or two outpatient visits for dementia and other comorbidities. Nevertheless, further studies involving longer follow-up durations (> 30 years), the entire adult population and detailed information regarding migraine are required to enhance the generalizability of our findings.

In conclusion, in the present study, patients with migraine had an increased risk of developing all-cause dementia, AD, VaD, and other dementias compared with their risk-set matched controls. However, further studies are warranted to generalize our findings and elucidate the underlying pathophysiological mechanisms linking migraine and dementia.