Epilepsy and incident cardiovascular disease: A prospective UK Biobank cohort study Authors: Jie Ma, Yang Guo, Qi Zhao, Xuewen Zhong, Yu Ding Published: 2026 ABSTRACT AbstractBackgroundAs a neurological disorder, epileptic seizures have been documented to correlate with structural and functional cardiac alterations in prior studies. However, the long-term risk association between epilepsy and cardiovascular diseases (CVDs) remains incompletely elucidated.MethodsThis prospective population-based cohort study utilized individual-level data from the UK Biobank. The primary composite endpoint comprised incident CVDs, including ischemic heart disease (IHD), heart failure (HF), atrial fibrillation (AF), stroke, and cardiovascular death. Cox proportional hazards regression models with covariate adjustments, complemented by Fine-Gray competing risk analyses, were employed to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for assessing epilepsy-CVD associations.ResultsA total of 448,487 participants, including 7,220 patients with epilepsy, were included in the analyses. Over a median follow-up of 13.4 years (IQR: 12.5–14.1; range: 3.0–16.2 years), 2,097 (29.0%) and 64,341 (14.6%) participants experienced CVDs in the epilepsy and non-epilepsy cohorts, respectively. Epilepsy was associated with higher risks of overall CVDs (HR: 1.97, 95% CI: 1.88–2.06) and specific cardiovascular outcomes including IHD (HR: 1.63, 95% CI: 1.53–1.74), HF (HR: 1.89, 95% CI: 1.72–2.09), AF (HR: 1.78, 95% CI: 1.65–1.91), stroke (HR: 4.32, 95% CI: 4.01–4.64), and cardiovascular mortality (HR: 2.55, 95% CI: 2.26–2.88), after adjustment for available common variant–based genetic predisposition scores. Furthermore, patients with status epilepticus exhibited a significantly higher CVDs risk compared to those with epilepsy alone (HR: 1.70, 95% CI: 1.43–2.02).ConclusionEpilepsy was associated with higher risks of incident CVDs, including IHD, HF, AF, stroke, and cardiovascular mortality. These findings suggest that patients with epilepsy may represent a population at elevated cardiovascular risk, although causal inference cannot be established.Graphical Abstract Association of epilepsy and risk of cardiovascular events. Epilepsy was associated with higher risks of incident cardiovascular events, including ischemic heart disease, heart failure, atrial fibrillation, stroke, and cardiovascular death. CVD: cardiovascular disease Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-026-28082-w.Keywords: Epilepsy, Status epilepticus, cardiovascular disease, UK Biobank, Prospective cohort study Introduction Epilepsy, one of the most prevalent neurological disorders worldwide, affects over 70 million individuals globally and imposes a significant disease burden [1]. Emerging evidence suggests this chronic brain condition may be intrinsically linked to cardiovascular pathophysiology [2, 3]. Population-based analyses indicate that compared to the general population, individuals with epilepsy face a six-fold increased risk of sudden cardiac death/cardiac arrest [4, 5], five-fold elevated myocardial infarction risk [6], four-fold greater likelihood of developing ischemic heart disease [7], and nearly doubled probability of cardiac arrhythmias [8]. These findings have catalyzed the development of novel conceptual frameworks like “The Epileptic Heart” hypothesis [9] and “Epilepsy-Heart Syndrome” [10], which reflect the strong correlation of epilepsy and cardiovascular diseases (CVDs). Despite these advances, critical knowledge gaps persist regarding the longitudinal cardiovascular trajectories in epilepsy populations. Current evidence predominantly derives from cross-sectional and retrospective studies, limiting definitive assessment of temporal sequencing and causal inference in the epilepsy-CVD relationship. Furthermore, the potential confounding effects of antiseizure medications (ASMs) cardiotoxicity and shared vascular risk factors remain insufficiently characterized in large-scale epidemiological settings [11]. A comprehensive understanding of the association between epilepsy and CVDs is crucial for implementing appropriate medical care in patients with epilepsy. Building on previous UK Biobank evidence linking epilepsy with cardiovascular disease burden and mortality, the present study focused on incident cardiovascular outcomes among participants without baseline cardiovascular disease. We further examined major CVD subtypes, assessed whether epilepsy duration and status epilepticus modified these associations, and explored the roles of baseline ASM use and available common variant–based genetic predisposition scores. Methods Study design, setting, and population This prospective cohort study was conducted using individual-level data from the UK Biobank, a large population-based cohort of community-dwelling adults recruited between 2006 and 2010. The UK Biobank enrolled more than 500,000 participants aged 40–69 years at baseline from assessment centers across England, Scotland, and Wales. At baseline, participants completed touchscreen questionnaires and structured nurse-led interviews covering demographic characteristics, lifestyle factors, medical history, and medication use. Standardized physical measurements, including anthropometrics, blood pressure, and 12-lead electrocardiograms, were also performed, and biological samples were collected. Longitudinal health outcomes were ascertained through linkage to routinely collected health records, including hospital inpatient records and national death registry data. Detailed descriptions of the UK Biobank design and assessment procedures have been reported elsewhere. All participants provided written informed consent, and the UK Biobank study protocol was approved by the North West Multi-centre Research Ethics Committee. Access to UK Biobank data for the present study was granted under application 394,593. This study was conducted in accordance with the Declaration of Helsinki. To maintain data integrity, all participants from the UK Biobank cohort were initially included in the analysis. Systematic exclusion criteria were then applied as follows: (1) Participants with pre-existing cardiovascular conditions at baseline, including rheumatic heart disease, hypertensive heart disease, ischemic heart disease (IHD), pulmonary heart disease, pericardial disease, valvular heart disease, myocarditis, cardiomyopathy, arrhythmias, heart failure (HF) and stroke; (2) Missing essential variables required for analysis; (3) Participants without subsequent health outcome data. All disease diagnoses were systematically validated using International Classification of Diseases, Tenth Revision (ICD-10) codes, as detailed in Supplementary Table 1. Definition of epilepsy Epilepsy was defined using ICD-10 diagnostic codes. Participants were classified as having epilepsy if they had a recorded diagnosis of epilepsy or status epilepticus, corresponding to ICD-10 codes G40 and G41, respectively. For analyses by epilepsy status, status epilepticus was identified using ICD-10 code G41, whereas participants with epilepsy but without status epilepticus were classified as the epilepsy-only group. The ICD-10 codes used to define epilepsy and other diseases are provided in Supplementary Table 1. Outcome definitions The endpoints of this study encompassed five cardiovascular outcomes: IHD, HF, atrial fibrillation (AF), stroke, and cardiovascular mortality. Incident nonfatal cardiovascular outcomes, including IHD, HF, AF, and stroke, were ascertained from linked hospital inpatient records using International Classification of Diseases, ICD-10 diagnostic codes. Cardiovascular mortality was identified from the linked national death registry and was defined as death primarily attributed to CVDs based on the ICD-10-coded underlying cause of death. We did not use Read2/CTV3 primary care codes, self-reported disease fields, or UK Biobank pre-defined algorithmic outcome categories as primary outcome definitions. We further evaluated a composite cardiovascular outcome (CVDs), defined as the occurrence of any aforementioned nonfatal cardiovascular event or cardiovascular death during follow-up. The ICD-10 codes used to define each outcome are provided in Supplementary Table 1 [12]. ASM classification and exposure assessment ASMs were identified from baseline medication records in the UK Biobank and classified according to the Anatomical Therapeutic Chemical (ATC) classification system under code N03. All available ASMs recorded in the eligible cohort were summarized in Supplementary Table 2. Individual ASMs with no recorded users or very small exposure counts were not analyzed as separate drug-specific categories to avoid unstable estimates. ASMs were first grouped according to ATC-based pharmacological classes, including barbiturates and derivatives, hydantoin derivatives, succinimide derivatives, benzodiazepine derivatives, carboxamide derivatives, fatty acid derivatives, and other antiepileptics. ASM exposure was defined according to medications recorded at baseline and should therefore be interpreted as baseline recorded ASM use rather than lifetime ASM exposure or medication use throughout follow-up. We further classified ASMs according to clinically relevant pharmacological properties. Enzyme-inducing ASMs included phenobarbital, primidone, phenytoin, carbamazepine, and oxcarbazepine, whereas the remaining ASMs were classified as non-enzyme-inducing ASMs. In addition, given prior evidence linking sodium channel blockers to cardiovascular risk, a separate sodium channel blocker category was defined, including phenytoin, carbamazepine, oxcarbazepine, lamotrigine, and topiramate [13]. Statistical analysis Statistical analyses were executed using R software (version 4.3.0), with statistical significance set at a two-sided P-value of less than 0.05. Categorical variables were expressed as frequencies and percentages, whereas continuous variables were summarized as median and interquartile range (IQR) because of their skewed distributions. For continuous variables, the Mann-Whitney U test was used for two-group comparisons, and the Kruskal-Wallis test was used for comparisons involving more than two groups. Categorical variables were compared using the chi-square test. Time-to-event analysis was performed using the Kaplan-Meier method, with cumulative incidence curves constructed to visualize cardiovascular risk across epilepsy status groups. Between-group differences were statistically assessed through log-rank testing. Cox regression models were implemented to quantify the independent relationship between epilepsy status and cardiovascular outcomes, with proportional hazards assumptions verified through Schoenfeld residual testing. To account for the potential bias introduced by death events, which may preclude the observation of cardiovascular outcomes, we implemented Fine and Gray’s subdistribution hazard models. In the analysis of cardiovascular outcomes, all-cause mortality was designated as the competing event. For cardiovascular-specific mortality assessment, non-cardiovascular deaths were explicitly defined as the competing risk. To evaluate the associations between epilepsy and the study outcomes, three sequential models were constructed with progressive covariate adjustment: (1) Model 1: Adjusted for demographic factors (age, sex, ethnicity); (2) Model 2: Additionally incorporated lifestyle variables (body mass index [BMI], alcohol consumption, smoking status); (3) Model 3 (fully adjusted): Further included clinical comorbidities (dyslipidemia, hypertension, hyperthyroidism, diabetes, renal failure). Covariates for the statistical models were selected based on existing literature, biological plausibility, and availability of data as shown in the directed acyclic graph (Supplemental Fig. 1). We further examined whether epilepsy status (chronic epilepsy and status epilepticus) modified the association between epilepsy and CVD risk. To validate the robustness of our primary findings, we implemented several complementary approaches. Firstly, to mitigate potential confounder bias, we applied propensity score matching (PSM) to create a matched cohort. The detailed PSM procedure is provided in the Supplementary Methods, and the baseline characteristics after matching are shown in Supplementary Table 5. Covariate balance before and after matching was assessed using absolute standardized mean differences (SMDs), with an SMD < 0.1 considered acceptable balance. Because P values may be overly sensitive in large samples, SMDs were used as the primary measure of post-matching balance. In the matched cohort, Cox regression models were additionally adjusted for the covariates included in the propensity score model to further reduce potential residual confounding. Secondly, a sensitivity analysis was conducted by excluding participants with certain diseases to assess the robustness of the primary findings. The exclusion criteria were predefined based on clinical relevance and potential confounding effects. Thirdly, stratified analyses were conducted based on key variables such as age, sex, ethnicity, smoking status, alcohol consumption, and BMI to assess potential effect modification. Fourthly, to account for the potential long-term relationship between epilepsy and CVDs, we performed the analysis by sequentially excluding participants with less than 1, 3, or 5 years of follow-up. Fifthly, to evaluate the impact of epilepsy duration on cardiovascular risk, we stratified the study population into four exposure categories: 0–1 years, > 1–5 years, > 5–20 years, and > 20 years since epilepsy diagnosis. For epilepsy duration analyses, duration categories were additionally modeled as an ordinal variable to test for linear trend across increasing duration groups. P values for trend were reported in Supplementary Table 8. To account for available common variant–based genetic predisposition to cardiovascular outcomes, we incorporated CVD-specific polygenic risk scores (PRS) derived from the UK Biobank Standard PRS Catalog (Category 301) into exploratory analyses. The PRS were calculated using genome-wide association study (GWAS) summary statistics and were included as additional covariates in multivariable models to examine whether the associations between epilepsy and cardiovascular outcomes persisted after adjustment for measured genetic susceptibility. We further performed analyses stratified by PRS quartiles and tested multiplicative interaction terms between epilepsy status and PRS for selected cardiovascular outcomes, including IHD, AF, and stroke. No formal mediation analysis was performed. Detailed information regarding genotyping, imputation, and quality control procedures in the UK Biobank has been reported previously [14]. No formal adjustment for multiple comparisons was applied because the primary cardiovascular outcomes were prespecified and clinically correlated; therefore, Bonferroni-type correction was considered overly conservative. The interpretation of findings was based primarily on effect estimates and 95% confidence intervals rather than P-values alone. Subgroup analyses, ASM-related analyses, PRS-related analyses, and sensitivity analyses were considered secondary or exploratory. To assess the impact of ASM use on CVD risk, participants were stratified into four groups: (1) non-epileptic ASM non-users, (2) non-epileptic ASM users, (3) epileptic ASM non-users, and (4) epileptic ASM users. Associations were evaluated using multivariable Cox regression models adjusted for relevant covariates. Participants with missing values for variables required in the main analyses, including epilepsy status, cardiovascular outcomes, follow-up time, and covariates included in the fully adjusted models, were excluded. Thus, the primary analyses were conducted using a complete-case approach, and multiple imputation was not performed. Missing values in variables not used in the present analyses did not affect participant inclusion. Results Baseline characteristics A total of 448,487 participants were included in the final analysis, of whom 7,220 (1.6%) had epilepsy (Supplementary Fig. 2). The median age of the study population was 57 years (IQR: 49–63), and 43.9% were male. Compared with participants without epilepsy, those with epilepsy had a higher prevalence of dyslipidemia, hypertension, diabetes, and renal failure. Baseline characteristics are shown in Table 1. Among participants with epilepsy, 1,434 (19.9%) had recorded ASM use at baseline. The most frequently recorded ASMs were carbamazepine, lamotrigine, and phenytoin; detailed baseline ASM use is provided in Supplementary Table 2. Table 1Baseline CharacteristicsVariablesTotal cohort(n = 448,487)With epilepsy(n = 7,220)Without epilepsy(n = 441,267)P valueAge57 (49, 63)58 (50, 63)57 (49, 63)< 0.001Male197,405 (43.9%)3,502 (48.5%)193,543 (43.9%)< 0.001British396,480 (88.4%)6,491 (89.9%)389,989 (88.4%)< 0.001Smoking status< 0.001 Current47,664 (10.6%)1,062 (14.7%)46,602 (10.6%) Never250,439 (55.8%)3,749 (51.9%)246,690 (55.9%) Previous150,384 (33.5%)2,409 (33.4%)147,975 (33.5%)Drinking status< 0.001 Current414,328 (92.4%)6,232 (86.3%)408,096 (92.5%) Never19,284 (4.3%)459 (6.4%)18,825 (4.3%) Previous14,875 (3.3%)529 (7.3%)14,346 (3.3%) BMI26.6 (24.1, 29.6)26.8 (24.3, 30.1)26.6 (24.1, 29.6)< 0.001 Hypertension106,832 (23.8%)1,974 (27.3%)104,858 (23.8%)< 0.001 Diabetes11,253 (2.5%)270 (3.7%)10,983 (2.5%)< 0.001 Dyslipidemia53,508 (11.9%)1,075 (14.9%)52,433 (11.9%)< 0.001 Hyperthyroidism4,636 (1.0%)77 (1.1%)4,559 (1.0%)0.781 Renal failure4,157 (0.9%)91 (1.3%)4,066 (0.9%)0.003BMI Body mass index Baseline Characteristics BMI Body mass index Epilepsy and risk of cardiovascular outcomes Over a median follow-up of 13.4 years (IQR: 12.5–14.1; range: 3.0–16.2 years), 2,097 (29.0%) and 64,341 (14.6%) participants experienced CVDs in the epilepsy and non-epilepsy cohorts, respectively. Kaplan-Meier curves showed consistently higher cumulative incidence of overall CVD, IHD, HF, AF, stroke, and cardiovascular mortality among participants with epilepsy (Fig. 1). Cumulative incidence of cardiovascular disease among groups with and without epilepsy In fully adjusted Cox regression models, epilepsy was associated with higher risks of overall CVD (HR: 1.97, 95% CI: 1.88–2.06), IHD (HR: 1.63, 95% CI: 1.53–1.74), HF (HR: 1.89, 95% CI: 1.72–2.09), AF (HR: 1.78, 95% CI: 1.65–1.91), stroke (HR: 4.32, 95% CI: 4.01–4.64), and cardiovascular mortality (HR: 2.55, 95% CI: 2.26–2.88) (Supplementary Table 4). Competing risk analyses yielded consistent results (Fig. 2). Participants with status epilepticus had higher cardiovascular risks than those with epilepsy alone, particularly for stroke and cardiovascular mortality (Table 2 and Supplementary Fig. 4). Association of epilepsy and new-onset cardiovascular diseases. Model 1 was adjusted for age, sex, and ethnicity; Model 2 was further adjusted for body mass index, alcohol consumption, smoking status; Model 3 was further adjusted for dyslipidemia, hypertension, hyperthyroidism, diabetes, and renal failure. HR, hazard ratio; CI, confidence interval Table 2Association of epilepsy status and cardiovascular outcomesOutcomesWith epilepsy only (n = 6,912)†With status epilepticus (n = 308)†Status epilepticus vs. epilepsy onlyIncidence rate*HR (95% CI)P valueIncidence rate*HR (95% CI)P valueHR (95% CI)P valueAll cardiovascular events235.5 (225.3, 246.2)1.84 (1.75, 1.93)< 0.001404.7 (343.2, 477.4)3.13 (2.65, 3.69)< 0.0011.70 (1.43, 2.02)< 0.001Ischemic heart disease112.4 (105.6, 119.6)1.52 (1.43, 1.63)< 0.001150.9 (116.7, 195.2)1.99 (1.54, 2.58)< 0.0011.31 (1.01, 1.71)0.047Heart failure42.9 (38.8, 47.3)1.66 (1.49, 1.84)< 0.00181.9 (58.2, 115.2)3.15 (2.24, 4.44)< 0.0011.90 (1.33, 2.71)< 0.001Atrial fibrillation87.5 (81.6, 93.9)1.73 (1.64, 1.85)< 0.001145.4 (118.9, 188.9)2.75 (2.11, 3.57)< 0.0011.59 (1.21, 2.08)0.001Stroke78.3 (72.7, 84.4)4.09 (3.77, 4.43)< 0.001210.1 (168.2, 262.2)10.7 (8.6, 13.4)< 0.0012.62 (2.08, 3.31)< 0.001Cardiovascular death24.1 (21.3, 27.3)2.29 (2.00, 2.61)< 0.00168.6 (48.3, 97.6)6.30 (4.42, 8.98)< 0.0012.75 (1.90, 4.00)< 0.001The above analyses were based on the fully adjusted model: age, sex, ethnicity, body mass index, alcohol consumption, smoking status, dyslipidemia, hypertension, diabetes, hyperthyroidism, and kidney disease. HR, hazard ratio; CI, confidence interval*Per 10 000 person years (95% confidence interval)†With participants without epilepsy as the reference group Association of epilepsy status and cardiovascular outcomes The above analyses were based on the fully adjusted model: age, sex, ethnicity, body mass index, alcohol consumption, smoking status, dyslipidemia, hypertension, diabetes, hyperthyroidism, and kidney disease. HR, hazard ratio; CI, confidence interval *Per 10 000 person years (95% confidence interval) †With participants without epilepsy as the reference group Sensitivity analyses After propensity score matching, baseline covariates were generally balanced between participants with and without epilepsy, as assessed by absolute SMDs reported in Supplementary Table 5. In the matched cohort, the associations between epilepsy and cardiovascular outcomes remained consistent after additional multivariable adjustment (Supplementary Tables 5 and Supplementary Fig. 3). Similar findings were observed after excluding participants with major comorbidities and after excluding events occurring within the first 1, 3, and 5 years of follow-up, reducing the likelihood of reverse causation (Supplementary Table 6.1 and 6.2). Stratified analyses suggested that the epilepsy-CVD association varied by age, sex, and BMI for selected outcomes (Supplementary Table 7.1 and 7.2). In addition, longer epilepsy duration was associated with progressively higher cardiovascular risk, with the highest risks observed among participants with epilepsy duration exceeding 20 years; trend tests across duration categories were added to Supplementary Table 8. Genetic susceptibility analysis Higher CVD-related PRS was associated with increased risks of cardiovascular outcomes (Supplementary Tables 9 and Supplementary Fig. 5). After adjustment for PRS, epilepsy remained associated with increased risks of overall CVD, IHD, AF, and stroke across PRS strata (Supplementary Table 10). No significant multiplicative interaction was observed between epilepsy status and PRS, suggesting that the observed associations were not fully explained by measured genetic susceptibility. Analysis of ASM effects In analyses jointly considering epilepsy status and ASM use, participants with epilepsy had elevated cardiovascular risks regardless of ASM use (Supplementary Fig. 6 and Supplementary Table 11). Analyses by ASM pharmacological properties showed heterogeneous associations across cardiovascular outcomes, supporting the exploratory nature of the ASM-related analyses. Detailed estimates are provided in Supplementary Table 11. Discussion A recent UK Biobank study by Shah et al. reported a higher cardiovascular disease burden, mortality, and sudden death risk among people with epilepsy. Building on this work, our study focused on incident cardiovascular outcomes after excluding participants with baseline cardiovascular disease, evaluated major CVD subtypes, and explored whether the associations differed by epilepsy duration, status epilepticus, baseline ASM use, and available common variant–based genetic predisposition scores. Overall, epilepsy was associated with higher risks of incident CVD, including IHD, HF, AF, stroke, and cardiovascular mortality. These associations were consistent across several sensitivity analyses and persisted after adjustment for available CVD-related PRS, although this does not exclude the potential contribution of genetic factors not captured by standard PRS. The link between epilepsy and cardiovascular dysfunction has long been recognized, from early reports of cardiac arrest during seizures to more recent concepts such as the “epileptic heart” and “epilepsy–heart syndrome” [9, 10, 15]. However, previous studies have reported inconsistent findings for ischemic cardiovascular events and HF. Wilson et al. reported an increased risk of myocardial infarction after epilepsy onset [16], whereas a larger nationwide cohort found a lower risk of myocardial infarction and HF when accounting for competing risks [17]. These discrepancies may reflect differences in study design, population characteristics, baseline cardiovascular disease status, outcome definitions, covariate adjustment, and handling of competing mortality. Because patients with epilepsy may have higher risks of both cardiovascular and non-cardiovascular death, death before the diagnosis of nonfatal outcomes such as myocardial infarction or HF may influence risk estimates. In the present study, we excluded participants with pre-existing cardiovascular disease and performed Fine-Gray competing risk analyses; the associations with IHD and HF remained consistent, suggesting that these findings were not fully explained by competing death. Our findings also support previous evidence linking epilepsy with arrhythmias and stroke. Wang et al. reported an association between epilepsy and incident AF in the UK Biobank [8], and we further observed that longer epilepsy duration was associated with progressively higher risks of AF and other cardiovascular outcomes. This duration-response pattern may be consistent with cumulative cardiovascular effects related to recurrent seizures, autonomic dysfunction, hypoxemia, catecholamine excess, or shared vascular risk pathways [18–21]. Stroke deserves particular attention because it is both a common cause of late-onset epilepsy and a potential outcome after epilepsy. By excluding participants with baseline stroke and other cardiovascular diseases, our analysis provides prospective evidence that epilepsy is associated with subsequent stroke risk, although residual confounding and undetected preclinical cerebrovascular disease cannot be fully excluded. Status epilepticus was associated with higher cardiovascular risks than epilepsy without status epilepticus, particularly for stroke and cardiovascular mortality. This finding is clinically plausible given the severe systemic stress, autonomic disturbance, hypoxemia, and neuronal injury associated with status epilepticus [22, 23]. Nevertheless, status epilepticus may also be a marker of more severe epilepsy or greater comorbidity burden. Therefore, this result should be interpreted as evidence of risk stratification rather than proof of a causal effect. The ASM-related findings require cautious interpretation. EIASMs may influence cardiovascular risk through hepatic enzyme induction, adverse lipid changes, increased homocysteine levels, systemic inflammation, and drug-drug interactions with cardiovascular medications, including anticoagulants and antiarrhythmic agents [24–30]. These mechanisms may affect atherosclerotic, thromboembolic, and arrhythmic pathways differently, which could partly explain the heterogeneous associations across cardiovascular outcomes. In within-epilepsy comparisons, participants with recorded ASM use appeared to have lower risks of several cardiovascular outcomes than those without recorded ASM use. This should not be interpreted as a protective causal effect of ASMs. Potential explanations include differences in epilepsy phenotype, seizure control, treatment selection, healthcare contact, medication adherence, baseline-only medication ascertainment, and residual confounding. Therefore, ASM-related analyses should be considered exploratory. From a clinical perspective, these findings suggest that patients with epilepsy may represent a population at elevated cardiovascular risk. Cardiovascular risk assessment and prevention may be particularly relevant for patients with long-standing epilepsy or status epilepticus. The results also support closer collaboration between neurology and cardiology in the long-term care of patients with epilepsy. However, because this was an observational study, causal inference cannot be established. Strengths and limitations This study has several strengths. It was based on a large prospective cohort with long-term follow-up and sufficient power to examine multiple cardiovascular outcomes. We excluded participants with baseline cardiovascular disease, assessed major CVD subtypes, performed competing risk and sensitivity analyses, and explored the roles of epilepsy duration, status epilepticus, baseline ASM use, and available CVD-related PRS. Several limitations should also be acknowledged. First, detailed epilepsy-related information, including epilepsy subtype, severity, seizure frequency, seizure type, and family history of cardiovascular disease, was unavailable. Information on other relevant illnesses or medications and ASM mono- or polytherapy was also incomplete, leaving potential residual confounding. Second, cardiovascular outcomes were mainly identified using ICD-10-coded hospital inpatient records and death registry data; therefore, asymptomatic or less severe events not requiring hospitalization may have been missed. Third, ASM exposure was based on baseline medication records and did not capture lifetime ASM use, treatment adherence, dose, duration, treatment changes during follow-up, or prescription indication. Baseline ASM use may therefore have been under-ascertained. Fourth, PRS analyses were based on UK Biobank Standard PRS; potential sample overlap between PRS discovery datasets and UK Biobank participants cannot be fully excluded, and PRS captures only part of inherited risk. Finally, although multiple covariates were adjusted for, residual confounding from unmeasured factors, such as physical activity, remains possible. Conclusion This large-scale prospective cohort study showed that epilepsy was associated with higher risks of incident CVDs, including IHD, HF, AF, stroke, and cardiovascular mortality. These findings suggest that patients with epilepsy may represent a population at elevated cardiovascular risk. However, given the observational design, causal inference cannot be established, and further studies are needed to clarify the underlying mechanisms. Supplementary Information Supplementary Material 1. Acknowledgements This research has been conducted using data from UK Biobank under application 394593. We thank the UK Biobank participants and the UK Biobank team for access to this resource. Declaration of AI-assisted language editing During manuscript revision, AI-assisted tools were used solely to improve language clarity and readability. The authors carefully reviewed and edited all AI-assisted text and take full responsibility for the content of the manuscript. No AI tool was used for data analysis, result generation, interpretation of findings, or clinical decision-making. Authors' contributions Jie Ma and Prof. Yu Ding conceived the study. Jie Ma performed the statistical analyses and drafted the manuscript. Yang Guo, Qi Zhao, and Xuewen Zhong contributed to data interpretation and critically revised the manuscript. Prof. Yu Ding provided supervision and final approval. All authors reviewed and approved the final manuscript. Funding The authors declare that no funding was received for this study. Data availability This study was conducted using individual-level data from the UK Biobank under Application 394593. The UK Biobank data are available to bona fide researchers via the UK Biobank Access Management System upon reasonable application and approval (www.ukbiobank.ac.uk/). Declarations Ethics approval and consent to participate All participants provided written informed consent. The UK Biobank study protocol was approved by the North West Multi-centre Research Ethics Committee (11/NW/0382). This research was conducted in accordance with the Declaration of Helsinki and relevant institutional and national guidelines. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement This study was conducted using individual-level data from the UK Biobank under Application 394593. The UK Biobank data are available to bona fide researchers via the UK Biobank Access Management System upon reasonable application and approval (www.ukbiobank.ac.uk/).