Authors: Zongqing He, Qichao Zhang, Yi Sun, Jing Yang, Min Zhao
Categories: Review, Migraine, Mortality risk, Cardiovascular disease, Suicide mortality, Meta-Analysis, All-Cause mortality
Source: The Journal of Headache and Pain
Authors: Zongqing He, Qichao Zhang, Yi Sun, Jing Yang, Min Zhao
Migraine is a common neurological disorder associated with various comorbidities, however its impact on mortality remains controversial. This study aimed to systematically evaluate the associations between migraine and the risk of all-cause, cardiovascular disease (CVD), and suicide mortality.
A comprehensive literature search of PubMed, Embase, and the Cochrane Library was conducted up to July 1, 2025, to identify eligible cohort studies assessing mortality outcomes in individuals with migraine. Random- or fixed-effects models were used to calculate pooled hazard ratios (HRs) with 95% confidence intervals (CIs). Subgroup and meta-regression analyses were performed to explore the sources of heterogeneity. Study quality was assessed using the Newcastle-Ottawa Scale.
Eighteen cohort studies, including 7,928,722 participants and 49,105 all-cause deaths, were included. No significant association was found between any migraine (AM) and all-cause mortality (HR = 0.93, 95% CI: 0.80–1.09, I² = 94.7%), CVD mortality (HR = 0.97, 95% CI: 0.80–1.18, I² = 80.6%), or suicide mortality (HR = 1.11, 95% CI: 0.98–1.26, I² = 15.5%). Considerable heterogeneity was observed across the studies for all-cause and CVD mortality. Subgroup analyses revealed a reduced risk of all-cause mortality in studies based on physician-diagnosed migraines and in those with a follow-up duration of less than 12 years. Meta-regression analyses identified the migraine diagnosis method and mean follow-up duration as potential sources of heterogeneity for all-cause mortality. However, the sex-stratified subgroup analysis demonstrated a higher CVD mortality risk with low heterogeneity in women with migraine (HR = 1.21, 95% CI: 1.08–1.34, I² = 0.0%), particularly in those with migraine with aura (MA) (HR = 1.28, 95% CI: 1.03–1.60, I² = 23.8%). Meta-regression analyses did not identify the sources of heterogeneity in the CVD mortality studies.
This systematic review and meta-analysis found no evidence linking AM with all-cause or cause-specific mortality. However, significant heterogeneity across studies and indications of an elevated CVD mortality risk in women with AM, particularly MA, render these findings inconclusive. Further research is required to elucidate this relationship and its underlying mechanisms.
The online version contains supplementary material available at 10.1186/s10194-025-02164-3.
Migraine is one of the most prevalent forms of headaches and is characterized by recurrent moderate-to-severe unilateral or bilateral pulsating pain, often accompanied by nausea, vomiting, and symptoms such as photophobia and phonophobia. Migraine is classified into two subtypes based on the presence visual disturbances or other temporary central nervous system symptoms before an with aura (MA) and without aura (MO) [1]. As a major global health concern, migraine ranks as the second leading cause of disability worldwide, with a prevalence of 14.1%. Notably, its prevalence in women is nearly three-fold higher than that in men [2]. Beyond the intense pain associated with recurrent attacks, migraine significantly diminishes the quality of life, causes substantial disability, and impairs work and social functioning, thereby contributing to a considerable economic burden [3].
In addition to the primary symptoms, individuals with migraine, particularly those with MA, are more prone to developing comorbid conditions such as depression, anxiety disorders, and cardiovascular disease (CVD) [4]. These comorbidities further increase the risk of all-cause mortality in patients with migraines. Although a previous meta-analysis [5] found no significant relationship between migraine and all-cause and CVD mortality, the limited number of studies included in that analysis hindered the broader applicability of its conclusions. Given the increasing body of research, integrating larger cohort studies, particularly those with more recent and diverse data, is necessary. A 2020 meta-analysis [6] highlighted the higher risk of suicidal behavior and suicide attempts in individuals with migraine; however, the relationship with suicide mortality remains underexplored. More recently, a 2022 study [7] identified an elevated risk of CVD mortality among patients with MA, although other mortality types were not analyzed.
Given the high prevalence and considerable burden of migraines, understanding its relationship with mortality is crucial for improving patient outcomes and advancing public health. This study aimed to provide a comprehensive and updated perspective by integrating the latest cohort studies. By examining migraine as an exposure factor, we sought to quantify its impact on all-cause and cause-specific mortality while offering new insights through subgroup and sensitivity analyses.
This study was conducted according to the updated Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [8]. The study protocol was pre-registered in PROSPERO (CRD420251088240) and has been updated as the research progressed.
A comprehensive literature search was performed across three databases (PubMed, Embase, and the Cochrane Library) for cohort studies published from database inception through July 1, 2025. The search strategy combined the use of medical subject headings (MeSH) and keywords. The search terms included “migraine disorder,” “mortality,” “all-cause mortality,” “risk,” and “cohort studies” (Supplementary Tables 1–3).
Two authors (Zongqing, He and Qichao, Zhang) independently screened the studies based on the predefined inclusion and exclusion criteria. Disagreements were resolved by a third author (Yi, Sun).
Inclusion Exposure: diagnosis of migraine based on a clear definition; Control: no history of migraine or headache; Outcome: all-cause mortality or specific-cause mortality, with sufficient data to calculate adjusted hazard ratios (HRs) and 95% confidence intervals (CIs); Study cohort studies.
Conference abstracts, study protocols, duplicate publications, and studies without healthy controls were excluded. In case of multiple studies derived from the same cohort, the study with the longest follow-up period was included.
Relevant data were extracted by two independent authors (Qichao, Zhang and Yi, Sun) and discrepancies were resolved by a third author (Jing, Yang). The extracted information included the first author’s name, year of publication, study location, sex, age, sample size, number of deaths, mean follow-up duration, migraine subtype, specific mortality outcomes, and adjusted confounders.
The quality of the included studies was assessed using the Newcastle-Ottawa Scale (NOS) [9], Where NOS scores range from 0 to 9, with scores >7 indicating high quality; 4–6, moderate quality; and, < 3, low quality.
The adjusted HRs and 95% CIs from each study were used to calculate the effect sizes for migraine disorders and the risks of all-cause and cause-specific mortality. Heterogeneity between studies was evaluated using Cochran Q test and I² statistics [10]. When the I² value exceeded 50%, a random-effects model was applied to combine the results. Otherwise, a fixed-effects model [11] was used, and a sensitivity analysis was performed by sequentially excluding one study at a time to assess the robustness of the overall findings [12]. Publication bias was visually examined using a funnel plot, and Egger’s regression test was conducted to statistically assess the bias [13]. Subgroup analyses were performed according to sex, migraine type, region, data source type, migraine diagnosis, medication adjustment, and mean follow-up duration to detect potential sources of heterogeneity. A meta-regression analysis was used to explore whether the above variables influenced the results. All statistical analyses were performed using the Stata statistical software (version 14.0; Stata Corp, College Station, Texas, USA) and R software (version 4.3.1).
A preliminary search across three major electronic databases yielded 1,088 studies. After removing the duplicates, 893 studies remained. Based on the titles and abstracts, 859 studies were excluded. Following a full-text review, 16 additional studies were 8 lacked full text, 3 did not involve migraine as the exposure, 3 did not report mortality as an outcome, and 2 were from the same cohort. A total of 18 studies were included in the final analysis (Fig. 1).
Fig. 1Study flowchart depicting the methodology used in selecting articles for inclusion in the study
Eighteen studies [14–31] met the final inclusion and exclusion criteria (Table 1). The follow-up period across studies ranged from 1.5 to 25.9 years. These studies included 7,928,722 participants, and 49,105 all-cause deaths were recorded. The populations represented were from 10 American, six European, and two Asian cohorts. Of these, 12 were prospective cohort studies and six were retrospective cohort studies. Among these studies, the data from nine were derived from electronic databases, whereas nine originated from clinical investigations. Regarding migraine diagnosis, 10 studies relied on physician diagnosis, and 8 were based on self-reported migraine diagnosis. All the studies accounted for potential confounders using multivariable adjustments. Regarding the types of mortality, 10 studies evaluated all-cause mortality; 10, CVD mortality; and, 4, suicide-related mortality, with definitions and recording methods detailed in Supplementary Table 4. Twelve studies included participants of both sexes; five, women; and, one, men. Additionally, four studies provided data on MA. The quality scores of the individual studies, based on the NOS, ranged from 6 to 8 (Supplementary Table 5).
Table 1Characteristics of the included studiesStudyYearCountryStudydesignSexFollow-up,yearsAge,yearsMigraineassessmentDeaths/total cohortControlsdescriptionMigrainesubtypesMortalitystudiedMultivariableadjustmentKim et al. [14]2025KoreaRetrospective cohort studyBoth9–11 (length)20–39Database physician diagnosis(ICD-10 criteria)13,040/6,539,547Non-migraineAnyAurano auraSuicideSex, age, income, smoking, drinking, exercise, obesity and depression.Elser et al. [15]2025DanishProspective cohort studyBoth8.6 (median)40.1 (median)Database physician diagnosis(ICD-10 criteria)372/452,418Non-headacheAnySuicideSex, age, income, education, cancer, cardiovascular disease, COPD, stroke, head injury, mood disorders, and substance use disorders.Liu et al. [16]2023AmericaProspective cohort studyBoth16.8 (median)46.4 (mean)Database self-reported symptoms3,078/10,644Non-headacheAnyAll-causeCVDAge, sex, marital status, race, education, income, and NHANES cycle, smoking and drinking, BMI, metabolic syndrome, diabetes, and aspirin use.Rohmannet al. [17]2020AmericaProspective cohort studyFemale22.7 (median)53.6 (mean)Questionnaire self-reported symptoms(ICHD-Ⅱ criteria)4,371/27,844Non-headacheAnyAurano auraAll-causeCVDAge, BMI, total cholesterol, smoking, drinking, physical activity, diabetes, hypertension, family history of myocardial infarction, and family history of breast cancer, cervical cancer, endometrial cancer, ovarian cancer, and uterine cancer.Harnod et al. [18]2018ChinaRetrospective cohort studyBoth12 (length)46.0 (mean)Database physician diagnosis(ICD-9 criteria)6,743/99,063Non-migraineAnyAll-causeAge, sex, and comorbidities of depression, mental disorders, insomnia, alcohol-related illness, and anxiety.Rambaratet al. [19]2017AmericaProspective cohort studyFemale9.5 (median)54.0 (mean)Questionnaire self-reported symptoms174/917Non-migraineAnyAll-causeCVDAge, race, BMI, diabetes, hypertension, dyslipidemia, smoking, family history of CAD, CAD severity score and aspirin use.Kruth et al. [20]2016AmericaProspective cohort studyFemale9.0 (mean)34.2 (mean)Questionnaire physician diagnosis (ICD-9 criteria)223/17,531Non-migraineAnyCVDAge, cholesterol, diabetes, hypertension, BMI, smoking, drinking, exercise, postmenopausal hormone use, menopausal status, ever used oral contraceptive, aspirin use, acetaminophen, non-steroidal, anti-inflammatory drug use, and family history of myocardial infarction before age of 60.Asberg et al. [22]2016NorwayProspective cohort studyBoth14.1 (mean)44.0 (mean)Questionnaire self-reported symptoms(ICHD-Ⅰ criteria)9,408/51,853Non- headacheAnyAurano auraAll-causeCVDAge, sex, education, cardiovascular disease, BMI, exercise, smoking, drinking, HADS, metabolic syndrome and systolic BP.Colman et al. [21]2015CanadaProspective cohort studyBoth7.3 (mean)41.0 (mean)Database physician diagnosis55/101,114Non-migraineAnySuicideADG score.Monteith et al. [23]2015AmericaProspective cohort studyBoth11.0 (mean)68.0 (mean)Questionnaire self-reported symptoms(ICHD-ⅠⅠ criteria)178/1,554Non-migraineAnyAurano auraCVDAge, sex, race, smoking, drinking, exercise, BMI, hypertension, hypercholesterolemia, diabetes.Ilgen et al. [24]2013AmericaRetrospective cohort studyBoth3 (length)AnyDatabase physician diagnosis (ICD-9)127/82,807Non-migraineAnySuicideAge, Sex, Charlson Score, and Concomitant Psychiatric Conditions.Gudmundsson et al. [25]2010IcelandProspective cohort studyBoth25.9 (median)50.6 (mean)Questionnaire self-reported symptoms(ICHD-ⅠⅠ criteria)7,477/13,071Non- headacheAnyAurano auraAll-causeCVDAge, BMI, smoking, education, systolic and diastolic blood pressure, and use of anti-hypertensive drug.Kruth et al. [27]2007AmericaProspective cohort studyMale15.7 (mean)56.5 (mean)Questionnaire self-reported symptoms866/20,084Non-migraineAnyCVDAge, hypertension, diabetes, smoking, exercise, BMI, drinking, hypercholesterolemia, parental history of MI before age 60 years, and randomized treatment assignments.Liew et al. [26]2007AustraliaProspective cohort studyFemale6.0 (mean)69.1 (mean)Questionnaire physician diagnosis(ICHD-ⅠⅠ criteria)60/2,331Non-migraineAnyAurano auraCVDAge, smoking, diabetes, fasting cholesterol and systolic blood pressure.Becker et al. [28]2007United KingdomRetrospective cohort studyBoth7 (length)AnyDatabase physician diagnosis458/103,376Non-migraineAnyAll-causeAge, sex, smoking, BMI, triptan use, diabetes mellitus, hypertension, dyslipidemiaVelentgaset al. [29]2004AmericaRetrospective cohort studyBoth1.5 (mean)38.2 (mean)Database physician diagnosis544/260,822Non-migraineAnyAll-causeCVDAge, sex, triptan use, comorbidities, oral contraceptive use, and estrogen replacement therapy use.Hall et al. [30]2004United KingdomRetrospective cohort studyBoth3.0 (mean)AnyDatabase physician diagnosis1,744/140,814Non-migraineAnyAll-causeCVDHypertension, diabetes, obesity, hypercholesterolemia, oral contraceptive use, and smoking.Waters et al. [31]1983United KingdomProspective cohort studyFemale12 (length)20–64Questionnaire self-reported symptoms187/2932Non- headacheAnyAll-causeAge, smoking, analgesic use.BMI body mass index, CAD coronary artery disease, HADS hospital anxiety and depression scale, ADG aggregated diagnosis groups, ACE angiotensin-converting enzyme, ICD International Classification of Diseases code, ICHD-I International Classification of Headache Disorders, first edition,* ICHD-II* International Classification of Headache Disorders, second edition
Seven studies [16, 18, 22, 25, 28–30] investigated the association between any migraine (AM) and the risk of all-cause mortality. The pooled analysis using a random-effects model revealed no significant association between AM and all-cause mortality, with substantial heterogeneity (HR = 0.93; 95% CI: 0.83–1.99, I² = 94.7%, P < 0.001) (Fig. 2). The sensitivity analysis confirmed that no individual study notably altered the pooled hazard ratio, thereby supporting the robustness of the results (Supplementary Fig. 1.A). Subgroup analyses based on sex yielded similar findings (HR = 0.90, 95% CI: 0.75–1.07, I² = 90.4%, P < 0.001 for women; HR = 0.97, 95% CI: 0.77–1.23, I² = 88.4%, P < 0.001 for men). However, pooled data from two studies indicated that MA significantly increased the risk of all-cause mortality, with reduced heterogeneity (HR = 1.21; 95% CI: 1.12–1.30, I² = 0.0%, P = 0.840). Stratified analysis according to different diagnostic methods demonstrated that AM was associated with a significantly reduced risk of all-cause mortality in studies based on physician-diagnosed migraine, accompanied by markedly lower heterogeneity (HR = 0.80; 95% CI: 0.76–0.86; I² = 21.4%, P = 0.282), whereas no such association was found for self-reported migraine cases. The subgroup analysis by mean follow-up duration indicated a substantial reduction in HR only in studies with mean follow-up duration ≤ 12years (HR = 0.80; 95% CI: 0.76–0.86; I² = 21.4%, P = 0.282); in contrast, the pooled results of studies with mean follow-up duration >12 years indicated a trend toward an increased risk of death associated with AM (Table 2). Meta-regression analyses identified diagnostic approach for migraine (P = 0.016) and mean follow-up duration (P = 0.016) as potential sources of heterogeneity (Supplementary Table 7).
Fig. 2Forest plot of all-cause mortality in any migraine
Table 2Subgroup analysis of the all-cause mortality in migraineSubgroupNo.studyTest of associationTest of heterogeneityHR95% CIP valueP valueI^2^Sex Female60.900.75–1.070.216< 0.00190.4 Male30.970.77–1.230.817< 0.00188.4Migraine subtype MO20.960.86–1.070.4360.14453.2 MA21.211.12–1.30< 0.0010.8400.0Region America21.070.81–1.430.6200.00488.2 Europe40.910.74–1.120.354< 0.00194.8 Asia10.810.76–0.87< 0.001//Data source type Database50.890.73–1.090.260< 0.00194.4 Clinic trial21.040.86–1.270.685< 0.00192.6Migraine diagnosis Self-reported31.100.95–1.260.206< 0.00190.4 Physician diagnosis40.800.76–0.86< 0.0010.28221.4Medication adjustment Yes50.960.78–1.180.702< 0.00194.8 No20.870.75–1.010.0620.00786.1Mean follow-up duration (year) ﹥1231.100.95–1.280.206< 0.00190.4 ≤ 1240.800.76–0.86< 0.0010.28221.4AM any migraine, MO migraine without aura, MA migraine with aura
Six studies [16, 22, 23, 25, 29, 30] reported the association of CVD mortality with migraine. The pooled analysis showed that AM was associated with a slight decrease in CVD mortality (HR = 0.97; 95% CI: 0.798–1.182, I² = 80.6%, P < 0.001); however, this difference was not statistically significant (Fig. 3). The sensitivity analysis confirmed the robustness of these results (Supplementary Fig. 1.B), and no evidence indicated the existence of publication bias in the included studies using Egger’s test (Supplementary Tables 6 and Supplementary Fig. 2.A). However, sex-specific subgroup analyses revealed a positive association between AM and elevated CVD mortality risk, with no statistical heterogeneity (HR = 1.21; 95% CI: 1.08–1.34, I² = 0.0%, P = 0.703 for women; HR = 1.26; 95% CI: 1.10–1.45, I² = 9.9%, P = 0.330 for men). Three studies [22, 23, 25] explored the association between MO and CVD mortality and found no significant association (HR = 0.90; 95% CI: 0.71–1.15, I² = 69.0%, P = 0.040). Sex-stratified analyses yielded similar results. Three studies [22, 23, 25] examined the relationship between MA and CVD mortality. The pooled analysis indicated that MA significantly increased the risk of CVD mortality, with no evidence of statistical heterogeneity (HR = 1.25; 95% CI: 1.12–1.40, I² = 0.0%, P = 0.519), particularly among women (HR = 1.28; 95% CI: 1.03–1.60, I² = 23.8%, P = 0.269). Subgroup analyses based on region, data source, migraine diagnosis, medication adjustment, and mean follow-up duration revealed no significant association between AM and CVD mortality (Table 3). The meta-regression analysis did not identify any source of heterogeneity (Supplementary Table 8).
Fig. 3Forest plot of CVD mortality in any migraine
Table 3Subgroup analysis of the CVD mortality in migraineSubgroupNo.studyTest of associationTest of heterogeneityHR95% CIP valueP valueI^2^Sex Female61.211.08–1.340.0010.7030.0 Male31.261.10–1.440.0010.3309.9Migraine subtype MO30.900.71–1.150.3950.04069.0 MO in female41.000.82–1.230.9850.24527.8 MO in male20.900.57–1.410.6340.02779.6 MA31.251.12–1.40< 0.0010.5190.0 MA in female41.281.03–1.600.0280.26923.8 MA in male21.290.89–1.880.1810.22033.4Region America30.830.48–1.470.5290.00581.4 Europe31.010.80–1.270.9570.00186.7Data source type Database30.980.70–1.370.8900.01576.1 Clinic trial30.940.70–1.280.705< 0.00188.5Migraine diagnosis Self-reported41.030.81–1.310.809< 0.00185.0 Physician diagnosis20.820.56–1.210.3180.17246.3Medication adjustment Yes41.070.87–1.320.1200.23429.5 No20.820.65–1.050.5050.01073.5Mean follow-up duration (year) ﹥1231.110.88–1.410.388< 0.00186.9 ≤ 1230.790.59–1.050.1030.19239.5*CVD *cardiovascular disease, AM any migraine, MO migraine without aura, MA migraine with aura
Given the low heterogeneity (I² = 15.5%, P = 0.314), a fixed-effects model was used to pool data from the four studies [14, 15, 21, 24] on suicide mortality (Fig. 4). The analysis revealed a slight but non-significant increase in the risk of suicide mortality in individuals with AM (HR = 1.11; 95% CI: 0.98–1.26). The sensitivity analysis further supported the robustness of the results (Supplementary Fig. 1.C). The Egger’s test indicated no evidence of publication bias in the included studies (Supplementary Tables 6 and Supplementary Fig. 2.B). The sex-stratified subgroup analysis indicated that AM only slightly increased the risk of suicide mortality in women (HR = 1.18; 95% CI: 1.00–1.38, I² = 0.0%, P = 0.573) (Table 4).
Fig. 4Forest plot of suicide mortality in any migraine
Table 4Subgroup analysis of the suicide mortality in migraineSubgroupNo.studyTest of associationTest of heterogeneityHR95% CIP valueP valueI^2^SexFemale21.181.01–1.380.0430.5730.0Male20.980.84–1.150.8130.6270.0AM any migraine
Visual inspection of the funnel plot revealed no evidence of significant publication bias regarding the association between migraine and the risk of all-cause mortality (Fig. 5). The Egger’s regression test (P = 0.683) confirmed the absence of publication bias in our meta-analysis.
Fig. 5Publication bias of the risk of all-cause mortality by any migraine
This meta-analysis did not reveal a significant association between AM and all-cause, cardiovascular, or suicide mortality, although significant heterogeneity was observed across studies. According to the sex-specific analysis, AM was associated with a 1.21-fold higher risk of CVD mortality in women, with no statistical heterogeneity. There are some indications that MA increases the risk CVD mortality by 1.25-fold, particularly in women; however, these results were pooled from a small number of studies. Thus, uncertainty persists regarding the true association between AM, including MA and MO, and all-cause mortality as well as specific-cause mortality.
Our findings are consistent with those of previous studies [5, 7], which also reported that in the general population with AM, there was no significant association of AM with all-cause or cardiovascular mortality, although significant heterogeneity across studies was noted. However, this meta-analysis adds to the existing literature by including more recent studies and conducting detailed subgroup analyses based on the sex, migraine type, region, data source type, migraine diagnosis, medication adjustment, and mean follow-up duration to detect heterogeneity. It was found that the diagnostic method and mean follow-up duration may be sources of heterogeneity in studies on all-cause mortality, which provides new insights compared to previous research. Notably, we observed an association between male patients with AM and a higher risk of cardiovascular mortality, which was characterized by lower heterogeneity. This finding extends previous reports that such a significant association was only observed in female populations [6]. Additionally, this is the first large-scale cohort study to examine the relationship between AM and suicide mortality. Although the number of studies on suicide mortality is limited, our results suggest that AM, particularly in women, is associated with a higher risk of suicide mortality.
The relationship between migraine and all-cause mortality remains unclear. This meta-analysis found no significant association between AM and all-cause mortality, with substantial statistical heterogeneity. This may be attributed to the primary association between migraine and CVD, psychiatric disorders, and neurological conditions [32–34]. Consequently, migraine may specifically increase the risk of CVD or suicide-related mortality; however, its impact on other causes of death requires further investigation. To address this high heterogeneity, we conducted meta-regression analyses to explore sex, migraine type, region, data source type, migraine diagnosis, medication adjustment, and mean follow-up duration. Diagnostic approach and mean follow-up duration have been identified as potential sources of heterogeneity. Further subgroup analyses revealed that MA was associated with a higher risk of all-cause mortality; however, this finding was unreliable owing to the inclusion of only two studies. Conversely, a significantly lower all-cause mortality risk and lower heterogeneity were observed in studies using physician diagnoses and in those with a follow-up duration shorter than 12 years. There are several possible explanations for these findings. First, self-reported migraine diagnoses depend on patients’ subjective recall, which increases the likelihood of misclassification and compromises exposure accuracy. By contrast, physician diagnoses based on standardized criteria are generally more accurate and consistent, resulting in reduced heterogeneity in studies employing this method. Second, patients receiving formal physician diagnoses typically have access to more comprehensive clinical management, including systematic evaluation, appropriate medication regimens, and regular follow-up care. Such structured medical interventions may effectively mitigate the burden of migraine itself as well as associated comorbidities, thereby contributing to the observed reduction in mortality risk. The observation that the mortality rate of the AM population is lower than that of the general population in short-term studies stems primarily from two types of bias. First, insufficient accumulation of death Mortality is a low-incidence outcome. During short-term follow-up (< 12 years), the small number of death events makes the results vulnerable to random variation, which exaggerates the spurious between-group differences. In contrast, longer follow-up duration leads to more accumulated death events, diluting random errors and yielding more reliable results. Regarding outcomes such as mortality and disease progression, the driving factors are more concentrated for short-term outcomes, while long-term outcomes are influenced by the superposition of multiple etiologies, which expands the differences in results across studies.
In this meta-analysis, we found no evidence linking AM to an increased risk of CVD mortality. However, this result should be interpreted with caution because of the high heterogeneity observed in the pooled estimates. Although the meta-regression analyses failed to identify statistically significant sources of heterogeneity, sources of clinically relevant heterogeneity remained. Interestingly, sex-specific analyses found that women with AM, particularly those with MA, exhibited a significantly higher risk of CVD mortality, with low statistical heterogeneity. The mechanisms underlying the association between migraine and CVD are complex and multifactorial. Potential contributing factors include genetic predisposition, endothelial dysfunction, anatomical abnormalities, and sex hormone dysregulation, particularly in women [35, 36]. Additionally, our meta-analysis revealed that women with AM, particularly those with MA, exhibited a significantly elevated risk of CVD mortality. Female patients are more likely to use oral contraceptives and experience estrogen fluctuations, both of which may exacerbate oxidative stress in endothelial cells, leading to endothelial activation and increased cardiovascular risk [37]. Additionally, men with AM had a higher risk of CVD mortality. This may be associated with the higher prevalence of CVD risk factors, such as hypertension and dyslipidemia, in men [38]. Subgroup analyses revealed that MA was associated with a higher risk of CVD mortality, with reduced heterogeneity compared with AM. This association may be driven by specific factors in patients with MA, such as cortical spreading depression, which can trigger localized inflammation and oxidative stress, impair vascular endothelial function, and contribute to increased cardiovascular events [39]. Furthermore, up to 60% of patients with MA have a patent foramen ovale, which may increase the risk of myocardial infarction and CVD mortality by allowing thrombi to enter the coronary arteries [40]. Although the association between MA and CVD mortality risk in female populations can be explained mechanistically and clinically, the number of studies included in our analysis was limited, typically three to four, and the reliability of these results still requires validation.
Our study suggests a potentially higher suicide mortality risk among patients with migraines, although this association was not statistically significant. Chronic pain, particularly in patients with migraines, may contribute to depressive symptoms and hopelessness compounded by impairments in occupational and social functioning [41]. These factors can precipitate suicidal ideation and behaviors potentially influenced by dysregulated serotonin (5-HT) levels [42, 43]. The inclusion of four cohort studies [14, 15, 21, 24] in this meta-analysis indicated that depression attenuates the relative risk of suicide in patients with migraine compared to non-depressed individuals, suggesting that comorbid depression is a significant driver of suicide risk in patients with migraine. However, the residual risk associated with migraines, independent of depression, warrants further investigation. Furthermore, we observed a significantly higher suicide risk among female patients with migraine; however, the limited number of available studies requires caution when interpreting this result.
This meta-analysis had several limitations that merit careful consideration. First, the use of all-cause and CVD mortality as outcomes may have been overly broad, contributing to substantial heterogeneity in the pooled results. Only two studies clearly defined all-cause mortality, with notable variability in definitions, and only four studies specified CVD mortality, with similar inconsistencies. This variability is likely to drive the observed heterogeneity. Additionally, the incomplete reporting of migraine diagnoses hindered our ability to fully explore the sources of heterogeneity. Although meta-regression analyses identified diagnostic methods and mean follow-up duration as potential contributors, the limited number of studies warrants cautious interpretation owing to reduced statistical power. Second, the key factors influencing survival, such as age, were not included in the subgroup or meta-regression analyses. This was due to the overlapping age ranges across the studies, compounded by the incomplete reporting of these data, which may have influenced the results. Third, the limited representation of countries and regions, particularly in Asia, restricts the generalizability of our findings. Fourth, although we performed meta-and subgroup analyses based on diagnostic methods and data sources, missing data on the diagnostic criteria precluded further exploration, limiting the depth of our analysis. Fifth, the included studies did not stratify migraine severity or attack frequency, thus preventing an analysis based on these clinically relevant variables. Sixth, although no publication bias was detected, the small number of studies, particularly on suicide-related mortality, reduced the statistical power of Egger’s test, limiting the reliability of this assessment. Finally, although definitive evidence linking migraine medications to mortality is lacking [44], studies, including those in this meta-analysis, suggest that commonly used treatments in patients with migraines, such as triptans [30, 45], calcitonin gene-related peptide inhibitors (CGRP) [46], ergotamines [29], antiepileptic drugs (AEDs) [47], and oral contraceptives [48], may increase the risk of cardiovascular events. Conversely, other frequently used medications such as aspirin [49] and beta-blockers [50] may confer vascular protective effects, potentially influencing mortality outcomes. Furthermore, migraine is frequently comorbid with psychiatric disorders such as depression and anxiety, which may lead to increased antidepressant use and heightened suicide risk [51]. These factors may have compromised the robustness of the findings. Of the studies we included, 7 controlled for drug factors related to migraine prevention and treatment, while 11 did not adjust for such factors—this may have a potential impact on the results.
Our meta-analysis revealed no association between AM and increased risks of all-cause, CVD, and suicide mortality, with high heterogeneity. However, AM appeared to elevate the risk of suicide mortality, with subgroup analyses suggesting that specific migraine populations, particularly women, may also have a higher risk of CVD mortality. Considering that the Systematic Coronary Risk Evaluation (SCORE) does not yet include migraine as a risk factor, coupled with the elevated suicide risk associated with psychiatric disorders [52, 53], there is an urgent need for interdisciplinary collaboration among headache, stroke, cardiology, and psychiatry centers to conduct long-term follow-up cohort studies, which are essential to investigate whether migraine prevention can reduce specific mortality risks or endpoint events. The American Registry for Migraine Research (ARMR) has taken a significant step forward by integrating multi-institutional data [54]. Future efforts should aim to integrate international migraine cohort data to enhance the generalizability of the research results and further conduct stratified analyses according to age, sex, ethnicity, migraine subtype, and severity to accurately identify high-risk populations. Furthermore, exploring potential mechanisms through which migraine may influence mortality, such as vascular endothelial dysfunction, immune dysregulation, and neurotransmitter imbalances, alongside clinical studies to develop targeted preventive strategies, has profound public health implications.
This study found no significant association between AM and all-cause, CVD, or suicide-related mortality. However, the substantial heterogeneity among studies introduces uncertainty into the findings. Sex-specific analyses indicated that AM may increase CVD mortality in women, particularly those with MA, although the limited study numbers necessitate cautious interpretation of this result. Given the high prevalence of migraine, further research is needed to elucidate its relationship with all-cause and cause-specific mortality and explore the underlying mechanisms.
Supplementary Material 1.