Abstract
Aims
Atrial fibrillation (AF) is a major cause of ischaemic stroke (IS), and its risk is assessed using existing risk stratification scores. However, the impact of psychiatric comorbidities such as depression on this risk has not been sufficiently considered. Depression is commonly observed in individuals with AF and is increasingly recognized as an independent risk factor for IS. This study aimed to examine the association between depression and the risk of IS in AF.
Methods and results
A cohort study of individuals with AF was conducted using a large Japanese healthcare database. Hazard ratios (HR) for IS in patients with depression were estimated using a multivariable Cox proportional hazards model, adjusting for established cardiovascular risk factors and medication use (e.g. anticoagulant therapy). Additional analyses included age- and sex-stratified analyses, time-dependent models, and sensitivity analyses to assess reverse causality bias. A total of 22 939 individuals with AF were included in the analysis. Among them, 1774 patients (7.7%) had a diagnosis of depression at baseline. During follow-up, IS occurred at a higher rate in individuals with depression. Multivariate analysis demonstrated that depression was an independent and significant predictor for IS (HR 1.39, 95% confidence interval 1.16–1.66, P < 0.001). Stratified analysis by age revealed this association was particularly pronounced in younger individuals.
Conclusion
This study demonstrates that depression in individuals with AF is a predictor of IS, independent of established cardiovascular risk factors. These findings highlight the potential importance of screening for depression as part of the holistic risk assessment for patients with AF.
Keywords: Atrial fibrillation, Depression, Stroke, Risk factor, Cohort study
Introduction
Atrial fibrillation (AF) is a major cause of ischaemic stroke (IS) and is known to increase the risk of IS approximately five-fold.1 To assess this risk, established clinical risk stratification tools, such as the CHADS2 score and CHA₂DS₂-VASc score, are widely used.2 The CHA₂DS₂-VASc score was developed to more accurately stratify stroke risk and guide the appropriate use of antithrombotic therapy by adding points based on factors such as congestive heart failure, hypertension, older age, diabetes mellitus, prior stroke, vascular disease, and female sex.3 While international guidelines, such as the ‘C’ pillar (Comorbidities) of the Atrial fibrillation Better Care (ABC) pathway,4 advocate for an integrated approach including the management of psychological morbidity,5 these traditional risk scores do not adequately account for psychiatric comorbidities (e.g. depression), which have recently received considerable attention due to their association with cardiovascular disease.
Depression is a highly prevalent comorbidity in individuals with cardiovascular diseases, including AF.6 Further, depression is also widely recognized as an independent predictor for IS in a general population.7–9 Particularly, a close, bidirectional relationship exists between depression and AF. The symptoms of AF, such as palpitations, and the associated decline in quality of life (QOL) can pose a significant psychological burden, potentially triggering depression.10 Conversely, the chronic elevation of stress-induced catecholamines in patients with depression is thought to be involved in the pathophysiology of AF, highlighting a mutually influential relationship.11 Thus, although depression is an independent predictor for IS and is closely associated with AF, the direct relationship of comorbid depression on the incidence of IS specifically in individuals with AF remains limited, and previous studies have reported inconsistent findings.12–14
Therefore, we aimed to perform a cohort study to investigate whether depression in individuals with AF is independently associated with the risk of IS, using a large-scale epidemiological database.
Methods
Data source
The DeSC database is available from DeSC Healthcare Inc. (DeSC Healthcare Inc., Tokyo, Japan; https://desc-hc.co.jp).
Study population
This retrospective observational study utilized the DeSC database (DeSC Healthcare Inc., Tokyo, Japan) spanning from April 2014 to August 2023. The DeSC database collects information from multiple Japanese health insurance systems, including the health insurance for employees of large companies, the National Health Insurance for nonemployees, and the Advanced Elderly Medical Service System for people aged ≥ 75 years. This database includes young, middle-aged, and older individuals, and contains anonymized information derived from claims data and medical checkup data. All medical diagnoses are documented using the International Classification of Diseases, 10th Revision (ICD-10). In addition, the DeSC database compiles data from annual health checkups, including physical measurements (e.g. height, weight, and blood pressure), laboratory tests, and self-administered lifestyle questionnaires. A total of 39 744 individuals with AF (ICD-10 codes: I480, I481, I482, I483, I484, and I489) were identified between April 2014 and August 2023. After excluding individuals with a history of IS (n = 8419), missing smoking status (n = 7164), and those with a follow-up period ≤ 2 months (n = 1222), 22 939 individuals were included in the primary analysis (Figure 1).
Figure 1.
Study flowchart.
Ethics
This study complied with the principles of the Declaration of Helsinki and followed the ethical guidelines of the University of Tokyo (approval number: 2021010NI). The requirement for informed consent was waived because all data were anonymized before analysis. All data handling complied with the International Conference on Harmonization guidelines.
Definition of depression
Depression was defined as a previous diagnosis documented in claims data (ICD-10 codes: F32.0–32.9, F33.0–33.3, F33.8, F33.9, F34.1, and F41.2).
Other variables
Hypertension was defined as systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, or the use of antihypertensive medication. Diabetes mellitus was defined as haemoglobin A1c ≥ 6.5% or the use of glucose-lowering medications. Dyslipidaemia was defined as low-density lipoprotein cholesterol ≥ 140 mg/dL, high-density lipoprotein cholesterol < 40 mg/dL, triglycerides ≥ 150 mg/dL, or the use of lipid-lowering medications. Smoking status, alcohol consumption, and physical activity were obtained using self-administered questionnaires at health checkups. Heart failure, thromboembolism, myocardial infarction, peripheral artery disease, and aortic plaque were identified using ICD-10 codes from claims data. Medication use, including antiplatelet therapy and oral anticoagulants (OAC) within 2 months after the diagnosis of AF, was also assessed. The CHADS₂ and CHA₂DS₂-VASc scores were calculated for each participant.
Outcomes
The primary outcome was the incidence of IS, defined using ICD-10 codes (I63.0, I63.1–I63.6, I63.8, I63.9 and G45.9).
Statistical analysis
Baseline characteristics were summarized as medians with interquartile ranges (25th–75th percentiles) or as numbers (percentages). Differences between groups with and without depression were compared using the Chi-square test for categorical variables and the Mann–Whitney U test for continuous variables. The incidence of IS was estimated using Kaplan–Meier survival curves and compared using the log-rank test. Cox proportional hazards regression models were applied to evaluate the association between depression and the risk of IS in AF patients. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated. To avoid collinearity between the CHA₂DS₂-VASc score and its individual components, the final multivariate model was adjusted for the CHA₂DS₂-VASc Score (as a continuous variable) and covariates not included in the score: body mass index, dyslipidaemia, smoking status, and medication use (including OAC and antiplatelet therapy). Subgroup analyses were conducted to examine whether the association between depression and IS differed by age (<65 vs. ≥ 65 years) and sex (men vs. women). Interactions between depression and key subgroups [age (<65 vs. ≥ 65 years) and sex] were assessed by adding an interaction term to the primary multivariate model. Statistical significance for interaction was defined as P < 0.1. In addition, sensitivity analyses were performed to assess the robustness of the results. First, time-dependent Cox regression models were applied to account for changes in covariates, including anticoagulant medication status, during the follow-up period. Second, an analysis with a 6-month induction period was conducted, excluding IS events occurring within the first six months after depression diagnosis to reduce potential reverse causality. Third, missing smoking data were imputed using multiple imputations with chained equations and 20 iterations. Finally, we constructed a Fine–Gray competing risk regression model accounting for death as a competing risk, and examined the association between depression and composite stroke events (ischaemic stroke and haemorrhagic stroke) using a Cox proportional hazards regression model.
All statistical analyses were performed using Stata software (version 19; StataCorp LLC, College Station, TX, USA). A two-sided P-value < 0.05 was considered statistically significant.
Results
Study population and baseline characteristics
Study population and selection flow
The study flowchart is shown in Figure 1. This study included 39 744 individuals with AF registered in the DeSC database between April 2014 and August 2023. Several exclusion criteria were considered when defining the study population. Specifically, 8419 patients had a history of IS, 7164 had missing smoking data, and 1222 had a follow-up period of less than 2 months. There were no subjects under 20 years of age. Ultimately, 22 939 individuals were included in the final analysis of this study.
Demographic data and comorbidities of the cohort
Table 1 provides a detailed summary of baseline characteristics for the entire cohort (n = 22 939), stratified by presence or absence of depression. A total of 1774 patients (7.7%) were diagnosed with depression, while 21 165 patients did not have depression. This analysis revealed statistically significant differences between the two groups in multiple characteristics, including cardiovascular risk factors. Specifically, the depression group had a significantly lower proportion of males (49.4% vs. 60.3%, P < 0.001). Additionally, this group had a lower body mass index (BMI) (23.0 kg/m² vs. 23.4 kg/m², P < 0.001), and lower systolic blood pressure (129 mmHg vs. 131 mmHg, P < 0.001) and diastolic blood pressure (74 mmHg vs. 76 mmHg, P < 0.001). Conversely, individuals with depression had significantly higher prevalence rates of dyslipidaemia (63.1% vs. 59.4%, P = 0.002) and heart failure (49.8% vs. 46.0%, P = 0.002). Furthermore, the median CHADS₂ scores (P = 0.002) and CHA₂DS₂-VASc scores (P < 0.001) were numerically higher in the depression group. These results indicate that the depression group had a more complex comorbidity profile.
Table 1.
Baseline characteristics
| Total | Depression (−) | Depression (+) | P value | |
|---|---|---|---|---|
| n = 22 939 | n = 21 165 | n = 1774 | ||
| Age, years | 70 (66–78) | 70 (66–78) | 71 (64–80) | 0.87 |
| Men, n (%) | 13 631 (59.4) | 12 755 (60.3) | 876 (49.4) | <0.001 |
| Body mass index, kg/m2 | 23.3 (21.2–25.7) | 23.4 (21.2–25.7) | 23.0 (20.7–25.5) | <0.001 |
| Systolic blood pressure, mmHg | 131 (120–142) | 131 (121–142) | 129 (118–140) | <0.001 |
| Diastolic blood pressure, mmHg | 76 (68–83) | 76 (68–84) | 74 (67–82) | <0.001 |
| Hypertension, n (%) | 19 471 (84.9) | 17 955 (84.8) | 1516 (85.5) | 0.48 |
| Diabetes mellitus, n (%) | 4103 (17.9) | 3796 (17.9) | 307 (17.3) | 0.51 |
| Dyslipidaemia, n (%) | 13 695 (59.7) | 12 575 (59.4) | 1120 (63.1) | 0.002 |
| Heart failure, n (%) | 10 618 (46.3) | 9734 (46.0) | 884 (49.8) | 0.002 |
| Thromboembolism, n (%) | 817 (3.6) | 740 (3.5) | 77 (4.3) | 0.065 |
| Myocardial infarction, n (%) | 1083 (4.7) | 986 (4.7) | 97 (5.5) | 0.12 |
| Peripheral artery disease, n (%) | 589 (2.6) | 535 (2.5) | 54 (3.0) | 0.19 |
| Aortic plaque, n (%) | 46 (0.2) | 41 (0.2) | 5 (0.3) | 0.43 |
| Cigarette smoking, n (%) | 2996 (13.1) | 2810 (13.3) | 186 (10.5) | <0.001 |
| Haemoglobin A1c, % | 5.7 (5.4–6.0) | 5.7 (5.4–6.0) | 5.6 (5.4–6.0) | 0.03 |
| Low-density lipoprotein cholesterol, mg/dL | 115 (95–135) | 115 (95–135) | 116 (95–137) | 0.1 |
| High-density lipoprotein cholesterol, mg/dL | 60 (50–72) | 60 (50–72) | 59 (49–71) | 0.27 |
| Triglycerides, mg/dL | 97 (70–137) | 97 (70–137) | 98 (71–138) | 0.14 |
| CHADS₂ score, points | 2 (1–3) | 2 (1–3) | 2 (1–3) | 0.002 |
| CHA₂DS₂−VASc score, points | 3 (2–4) | 3 (2–4) | 3 (2–4) | <0.001 |
| Antiplatelet medication, n (%) | 3415 (14.9) | 3151 (14.9) | 264 (14.9) | 0.99 |
| Oral anticoagulant medication within 2 months after the diagnosis of atrial fibrillation, n (%) | 14 364 (62.6) | 13 378 (63.2) | 986 (55.6) | <0.001 |
Values are shown as n (%) or median (interquartile range). P values were calculated using a chi-square test for categorical variables and Mann–Whitney U test for continuous variables.
Association with incident ischaemic stroke
Figure 2 shows the Kaplan–Meier (KM) curves for the first IS with and without depression. The KM curve for the group with depression rose at a steeper slope than that for the group without depression, resulting in a significant difference between the two groups over time (log rank P < 0.001). Table 2 presents the primary analysis results evaluating the association between IS incidence and various risk factors in the entire cohort using Cox regression models. This analysis included 22 939 individuals, with 1326 IS events occurring during follow-up. The median follow-up period was 2.3 years [interquartile range (IQR): 1.2–3.6 years]. The overall incidence rate was 244.0 events per 10 000 person-years [95% confidence interval (CI): 231.2–257.5]. The incidence rate was significantly higher in the depression group [343.0 (95% CI: 289.4–406.5) per 10 000 person-years] compared to the nondepression group [236.4 (95% CI: 223.4–250.2) per 10 000 person-years]. In the multivariate model, depression was identified as an independent predictor for IS. The hazard ratio (HR) for IS was 1.39 (95% CI: 1.16–1.66). The Schoenfeld residual plot indicated no violation of the proportional hazard assumption (see Supplementary material online, Figure S1).
Figure 2.
Cumulative incidence curves of first-ever IS in patients with and without depression.
Table 2.
Association of risk factors with incident ischaemic stroke in AF patients
| Individuals | P Value | |
|---|---|---|
| Number | 22 939 | — |
| Number of events | 1326 | — |
| Incidence rate (per 10 000 person-years) | 244.0 (231.2–257.5) | — |
| Age | 1.05 (1.04–1.06) | <0.001 |
| Men | 1.20 (1.04–1.39) | 0.013 |
| Body mass index | 0.99 (0.98–1.01) | 0.450 |
| Hypertension | 0.91 (0.75–1.10) | 0.319 |
| Diabetes mellitus | 1.09 (0.92–1.29) | 0.322 |
| Dyslipidaemia | 0.93 (0.83–1.04) | 0.198 |
| Heart failure | 0.86 (0.74–0.99) | 0.036 |
| Cigarette smoking | 1.21 (1.02–1.44) | 0.032 |
| CHA₂DS₂−VASc score | 1.09 (1.00–1.19) | 0.054 |
| Antiplatelet medication | 1.26 (1.10–1.46) | 0.001 |
| Oral anticoagulant medication within 2 months after the diagnosis of atrial fibrillation | 0.90 (0.80–1.00) | 0.060 |
| Depression | 1.39 (1.16–1.66) | <0.001 |
Hazard ratios and 95% confidence intervals were estimated using Cox regression models.
Stratified analysis by age and sex
Stratification by age
Table 3 shows the results of evaluating the association with IS incidence after dividing the cohort into two age groups: 65 years and older, and under 65 years. The ≥65 age group comprised 18 149 individuals with 1197 events, yielding an incidence rate of 295.0 events per 10 000 person-years. In contrast, the <65 age group comprised 4790 individuals with 129 events, resulting in an incidence rate of 93.7 events per 10 000 person-years, which was significantly lower than the older group. In the older group, the HR for depression was 1.31 (P = 0.006), whereas in the group under 65 years old, the HR was 2.11 (P = 0.003), indicating a more pronounced effect. A formal test for interaction confirmed that this age-based difference was statistically significant (P for interaction = 0.0783).
Table 3.
Association of risk factors with incident ischaemic stroke in AF patients stratified by age
| Age ≥ 65 years | Age < 65 years | |||
|---|---|---|---|---|
| Individuals | P value | Individuals | P Value | |
| Number | 18 149 | — | 4790 | — |
| Number of events | 1197 | — | 129 | — |
| Incidence rate (per 10 000 person-years) | 295.0 (278.8–312.2) | — | 93.7 (78.9–111.4) | — |
| Age | 1.05 (1.04–1.06) | <0.001 | 1.04 (1.01–1.07) | 0.008 |
| Men | 1.18 (1.01–1.38) | 0.032 | 1.12 (0.64–1.98) | 0.691 |
| Body mass index | 0.99 (0.98–1.01) | 0.434 | 1.00 (0.95–1.04) | 0.9 |
| Hypertension | 0.83 (0.68–1.02) | 0.078 | 1.67 (0.90–3.10) | 0.103 |
| Diabetes mellitus | 1.14 (0.95–1.36) | 0.154 | 0.79 (0.42–1.47) | 0.455 |
| Dyslipidaemia | 0.92 (0.81–1.04) | 0.163 | 1.06 (0.72–1.54) | 0.776 |
| Heart failure | 0.84 (0.72–0.99) | 0.034 | 1.17 (0.68–1.99) | 0.573 |
| Cigarette smoking | 1.21 (1.00–1.46) | 0.055 | 1.23 (0.82–1.84) | 0.314 |
| CHA₂DS₂−VASc score | 1.08 (0.98–1.19) | 0.118 | 1.00 (0.69–1.45) | 0.999 |
| Antiplatelet medication | 1.23 (1.06–1.43) | 0.007 | 1.91 (1.19–3.09) | 0.008 |
| Oral anticoagulant medication within 2 months after the diagnosis of atrial fibrillation | 0.89 (0.79–1.01) | 0.062 | 0.93 (0.65–1.33) | 0.706 |
| Depression | 1.31 (1.08–1.60) | 0.006 | 2.11 (1.29–3.45) | 0.003 |
Hazard ratios and 95% confidence intervals were estimated using Cox regression models.
Stratification by sex
Table 4 presents the results of analyses stratified by sex (men: n = 13 631; women: n = 9308). The incidence rate was 233.8 per 10 000 person-years for men and 259.2 per 10 000 person-years for women. This analysis confirmed that depression was a statistically significant predictor for IS in both men (HR 1.40, P = 0.012) and women (HR 1.39, P = 0.009). The hazard ratios were nearly identical in both groups, strongly supporting that the association between depression and IS risk is a consistent phenomenon independent of sex. This was supported by a nonsignificant test for interaction (P for interaction = 0.9753). Sex-specific differences in effect were observed for several other covariates. The CHA₂DS₂-VASc score showed a significant association only in women (HR 1.16, P = 0.028) and was not significant in men (HR 1.04, P = 0.510). Furthermore, anticoagulant use was associated with a significant risk reduction in men (HR 0.85, P = 0.039), but no significant association was observed in women (HR 0.95, P = 0.547).
Table 4.
Association of risk factors with incident ischaemic stroke in AF patients stratified by sex
| Men | Women | |||
|---|---|---|---|---|
| Individuals | P value | Individuals | P value | |
| Number | 13 631 | — | 9308 | — |
| Number of events | 758 | — | 568 | — |
| Incidence rate (per 10 000 person-years) | 233.8 (217.7–251.0) | — | 259.2 (238.7–281.4) | — |
| Age | 1.06 (1.04–1.07) | <0.001 | 1.05 (1.04–1.06) | <0.001 |
| Body mass index | 0.99 (0.97–1.02) | 0.612 | 0.99 (0.97–1.02) | 0.613 |
| Hypertension | 0.99 (0.77–1.26) | 0.905 | 0.81 (0.61–1.10) | 0.175 |
| Diabetes mellitus | 1.23 (0.99–1.52) | 0.056 | 0.89 (0.67–1.17) | 0.396 |
| Dyslipidaemia | 0.98 (0.84–1.14) | 0.788 | 0.87 (0.73–1.03) | 0.104 |
| Heart failure | 0.92 (0.76–1.12) | 0.399 | 0.78 (0.62–0.97) | 0.027 |
| Cigarette smoking | 1.16 (0.96–1.41) | 0.122 | 1.45 (0.96–2.20) | 0.078 |
| CHA₂DS₂−VASc score | 1.04 (0.92–1.17) | 0.510 | 1.16 (1.02–1.32) | 0.028 |
| Antiplatelet medication | 1.17 (0.97–1.40) | 0.104 | 1.41 (1.13–1.76) | 0.003 |
| Oral anticoagulant medication within 2 months after the diagnosis of atrial fibrillation | 0.85 (0.73–0.99) | 0.039 | 0.95 (0.80–1.12) | 0.547 |
| Depression | 1.40 (1.08–1.82) | 0.012 | 1.39 (1.09–1.78) | 0.009 |
Hazard ratios and 95% confidence intervals were estimated using Cox regression models.
Sensitivity analysis
Time-dependent analysis
Table 5 presents results from a time-dependent Cox regression analysis accounting for anticoagulant medication status changes over the study period. This analysis included 23 703 individuals and 1491 events. In this model, depression remained a statistically significant predictor (HR 1.35, P = 0.001).
Table 5.
Association of risk factors with incident ischaemic stroke in AF patients (time-dependent analysis)
| Individuals | P value | |
|---|---|---|
| Number | 23 703 | — |
| Number of events | 1491 | — |
| Incidence rate (per 10 000 person-years) | 263.5 (250.5–277.3) | — |
| Age | 1.05 (1.04–1.06) | <0.001 |
| Men | 1.16 (1.01–1.33) | 0.031 |
| Body mass index | 0.99 (0.98–1.01) | 0.416 |
| Hypertension | 0.92 (0.77–1.10) | 0.38 |
| Diabetes mellitus | 1.10 (0.94–1.28) | 0.245 |
| Dyslipidaemia | 0.91 (0.82–1.02) | 0.102 |
| Heart failure | 0.85 (0.74–0.97) | 0.018 |
| Cigarette smoking | 1.24 (1.05–1.45) | 0.009 |
| CHA₂DS₂−VASc score | 1.08 (1.00–1.18) | 0.053 |
| Antiplatelet medication | 1.38 (1.21–1.57) | <0.001 |
| Oral anticoagulant medication | 0.95 (0.86–1.06) | 0.367 |
| Depression | 1.35 (1.13–1.60) | 0.001 |
Hazard ratios and 95% confidence intervals were estimated using Cox regression models.
Analysis with a 6-month induction period
Supplementary material online, Table S1 presents the results of another sensitivity analysis, which excluded IS events occurring within the first 6 months after diagnosis. The purpose of this analysis was to mitigate the possibility that the latent stroke event caused the depression diagnosis, or that both conditions share the same underlying pathology (reverse causality). This model included 17 941 subjects and 804 events. In this analysis, depression remained a statistically significant predictor (HR 1.42, P = 0.003).
Association of risk factors with incident ischaemic stroke in AF patients after performing multiple imputations
In Supplementary material online, Table S2, sensitivity analyses were performed after imputing missing data, including cigarette smoking, using multiple imputations. Association of predictors with incident ischaemic stroke in AF patients after performing multiple imputations was similar.
Competing risk analysis
Supplementary material online, Table S3 shows the results of the Fine–Gray competing risk regression model, which was constructed to account for death as a competing risk for ischaemic stroke. Even after adjusting for the competing risk of death, depression remained a statistically significant predictor of ischaemic stroke.
Association with composite stroke events
We also examined the association between depression and composite stroke events, defined as either ischaemic stroke or haemorrhagic stroke. As presented in Supplementary material online, Table S4, the Cox proportional hazards regression model demonstrated that depression was significantly associated with an increased risk of composite stroke events.
Discussion
This study consistently demonstrates an independent association between depression and the elevated risk of IS in individuals with AF, based on a large-scale retrospective cohort study using an epidemiological database of more than 20 000 patients with AF. European guidelines already recommend an integrated approach to AF care, including the ABC (Atrial fibrillation Better Care) pathway, in which ‘C’ emphasizes the management of Comorbidities and cardiovascular risk factors.5 Our results strongly suggest that depression should be actively screened for and managed as a key component of this ‘C’ pillar, as depression is strongly associated with poor outcomes.15 Given the importance of holistic care,16 collaboration with mental health professionals and provision of psychotherapy could become integral components of routine AF management.
Our main analyses revealed that depression was an independent predictor for IS, with a hazard ratio of 1.39. Importantly, there is a growing body of evidence indicating plausible biological and behavioural pathways through which depression increases stroke risk. Specifically, depression is associated with endothelial dysfunction, platelet hyperreactivity, systemic inflammation, and autonomic dysregulation, each of which can promote a prothrombotic milieu.17–19 In addition, behavioural factors such as physical inactivity, poor adherence to prescribed medications, and other adverse health behaviours commonly observed in depressed patients may further augment stroke susceptibility.20 In contrast, a large nationwide cohort study from Finland reported that mental health conditions, including depression, were not independently associated with the risk of incident IS in patients with AF after adjustment for comorbidities and oral anticoagulant use.21 Several factors related to population characteristics and study design may explain this discrepancy. First, the Finnish FinACAF study included more than 200 000 patients with incident AF identified from nationwide hospital and primary care registers, with a mean age of approximately 72 years and nearly 50% women. By comparison, our study analysed a Japanese insurance-based cohort of 22 939 individuals who tended to be slightly younger and had a higher prevalence of cardiovascular risk factors such as hypertension and dyslipidaemia (Table 1). These demographic and clinical differences may have influenced the observed associations. Second, the classification of mental health conditions differed between studies. Teppo et al. evaluated several psychiatric disorders and defined ‘any mental health condition’ partly based on psychotropic medication use, whereas our study focused specifically on clinically diagnosed depression. This narrower definition may have reduced heterogeneity and increased the likelihood of detecting a robust association with IS risk. Our results are also comparable to those of a nationwide registry-based cohort study in Denmark,13 which suggested that depression may contribute to IS incidence in AF patients. However, unlike their findings—where the crude association between severe depression and stroke was attenuated and became nonsignificant after adjustment for stroke risk factors, comorbidities, and oral anticoagulant use—our study demonstrated that depression itself remained a significant and independent contributor to IS risk. The discrepancies with both the Finnish and Danish studies may also reflect differences in external validity. Our study, based on a Japanese (Asian) cohort, may be influenced by ethnic-specific factors or differences in healthcare systems (e.g. diagnostic practices for depression) compared to these European populations. It is also important to consider the potential for underdiagnosis. The 7.7% prevalence of depression in our study was based on ICD-10 codes, representing only diagnosed cases. It is widely recognized, particularly in cardiovascular populations, that a significant proportion of depression goes unnoticed or is misattributed to somatic symptoms.15 Therefore, the true prevalence of depression in AF patients is likely much higher. If this unmeasured depression is also associated with IS, our results may underestimate the true magnitude of the association between depression and IS risk in this population.
The robustness of our key findings was further reinforced by sensitivity analyses. First, depression remained a significant predictor even in time-dependent models that accounted for changes in covariates over the study period, strengthening confidence in the stability of our results. Second, when excluding the first 6 months after depression diagnosis to reduce the possibility that the latent stroke event could have triggered depression diagnoses (reverse causality), depression still emerged as significantly associated with the outcome. This is an important consideration, as cardiovascular events themselves can induce psychological distress leading to post-traumatic stress disorder (PTSD)22 or depression.23,24 Together, these analyses strongly support the conclusion that depression is not merely a marker of preceding conditions but an association that influences stroke risk over time. They emphasize the need for novel clinical approaches to identify risks not fully captured by the CHA₂DS₂-VASc score, reinforcing calls from major clinical guidelines to systematically consider psychosocial factors in cardiovascular disease prevention.15
Subgroup analyses revealed age-related differences. While depression was significantly associated with IS across all age groups, its impact was particularly pronounced among younger patients (<65 years). This suggests that, in older patients, conventional comorbidities such as hypertension and diabetes may dominate IS risk, whereas in younger patients, depression emerges more clearly as an independent associated factor. Clinically, this underscores the need to carefully evaluate depression as a key factor in IS risk assessment for younger individuals with AF. Finally, although prior studies have shown that the psychosocial burden of AF differs by sex with women experiencing greater anxiety, depression, and impaired quality of life compared to men,23,25,26 our study did not observe significant sex differences in the pathophysiological association between depression and IS risk.
Limitation
This study is based on large-scale real-world data, but several important limitations must be acknowledged. First, as an observational study, it cannot establish causality between depression and IS. Second, the observed association between antiplatelet therapy and an increased IS event may indicate that individuals who should have been prescribed OAC were instead treated with antiplatelet agents, reflecting inadequate stroke prevention. Furthermore, the low overall prevalence of OAC use in both groups is a major limitation of this real-world cohort and a significant potential confounder, even though OAC use was included as a covariate in the model. Third, due to database constraints, many individuals had missing data for several important variables (such as smoking status), and excluding these individuals may have introduced selection bias. However, we performed multiple imputations for missing information on cigarette smoking, and the results of sensitivity analyses remained consistent. Fourth, reliance on the ICD-10 diagnostic codes in the database resulted in a lack of detailed information regarding the severity, duration, or specific therapy of depression. Fifth, we relied on ICD-10 codes for case identification. The precise sensitivity and specificity of these codes for AF and depression diagnoses within this specific database are unknown, which could lead to misclassification bias. Sixth, we assessed OAC use at baseline but could not assess medication adherence (e.g. persistence or proportion of days covered) during follow-up, which is a key factor in stroke prevention. Finally, our models could not adjust for several potential confounders, including the type of OAC (direct oral anticoagulants vs. vitamin K antagonists), the specific AF treatment strategy (rate vs. rhythm control, or use of ablation), and the severity of depression. These factors could all influence both the depression diagnosis and the stroke outcome. Regarding renal dysfunction, which is also a critical predictor, we performed an additional sensitivity analysis adjusting for eGFR (see Supplementary material online, Table S5). Even after adjusting for eGFR, our main findings were unchanged, supporting the robustness of our results.
Conclusion
Our analysis using a large-scale epidemiological database suggests that depression is independently associated with an increased risk of IS in individuals living with AF. Given the observational nature of this study and the potential for unmeasured confounders, a causal relationship cannot be established. However, these findings highlight the need for greater clinical awareness and systematic screening for depression in AF patients, as it may identify risks not captured by the CHA₂DS₂-VASc score. Our findings suggest that systematic screening for depression and integration of mental health care should be considered as part of a holistic, integrated care approach (e.g. the ABC pathway) for individuals with AF, particularly among younger individuals.
Supplementary Material
Contributor Information
Keitaro Senoo, Department of Cardiac Arrhythmia Research and Innovation, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kyoto 602-8566, Japan; Department of Cardiovascular Medicine, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kyoto 602-8566, Japan.
Hidehiro Kaneko, Department of Cardiovascular Medicine, The University of Tokyo, Tokyo 113-8655, Japan; Department of Advanced Cardiology, The University of Tokyo, Tokyo 113-8655, Japan.
Yuta Suzuki, Department of Cardiovascular Medicine, The University of Tokyo, Tokyo 113-8655, Japan; Center for Outcomes Research and Economic Evaluation for Health, National Institute of Public Health, Saitama 351-0197, Japan.
Akira Okada, Department of Prevention of Diabetes and Lifestyle-Related Diseases, Graduate School of Medicine, The University of Tokyo, Tokyo 113-8655, Japan.
Toshiyuki Ko, Department of Cardiovascular Medicine, The University of Tokyo, Tokyo 113-8655, Japan; Department of Frontier Cardiovascular Science, Graduate School of Medicine, The University of Tokyo, Tokyo 113-8655, Japan.
Katsuhito Fujiu, Department of Cardiovascular Medicine, The University of Tokyo, Tokyo 113-8655, Japan; Department of Advanced Cardiology, The University of Tokyo, Tokyo 113-8655, Japan.
Norifumi Takeda, Department of Cardiovascular Medicine, The University of Tokyo, Tokyo 113-8655, Japan.
Hiroyuki Morita, Department of Cardiovascular Medicine, The University of Tokyo, Tokyo 113-8655, Japan.
Koichi Node, Department of Cardiovascular Medicine, Saga University, Saga 849-8501, Japan.
Norihiko Takeda, Department of Cardiovascular Medicine, The University of Tokyo, Tokyo 113-8655, Japan.
Hideo Yasunaga, Department of Clinical Epidemiology and Health Economics, School of Public Health, The University of Tokyo, Tokyo 113-8655, Japan.
Satoaki Matoba, Department of Cardiac Arrhythmia Research and Innovation, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kyoto 602-8566, Japan; Department of Cardiovascular Medicine, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kyoto 602-8566, Japan.
Supplementary material
Supplementary material is available at European Heart Journal Open online.
Ethics
This study was conducted in accordance with the ethical guidelines of our institution (approval by the Ethical Committee of the University of Tokyo: 2021010NI) and the principles of the Declaration of Helsinki.
Funding
This work was supported by grants from the Ministry of Health, Labour and Welfare, Japan (23AA2003).
Data availability
This dataset is available for use by anyone who is a purchaser or contract holder.
References
- 1. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke 1991;22:983–988. [DOI] [PubMed] [Google Scholar]
- 2. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest 2010;137:263–272. [DOI] [PubMed] [Google Scholar]
- 3. Olesen JB, Lip GY, Hansen ML, Hansen PR, Tolstrup JS, Lindhardsen J, Selmer C, Ahlehoff O, Olsen A-MS, Gislason GH, Torp-Pedersen C. Validation of risk stratification schemes for predicting stroke and thromboembolism in patients with atrial fibrillation: nationwide cohort study. BMJ 2011;342:d124–d124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Lip GYH. The ABC pathway: an integrated approach to improve AF management. Nat Rev Cardiol 2017;14:627–628. [DOI] [PubMed] [Google Scholar]
- 5. Van Gelder IC, Rienstra M, Bunting KV, Casado-Arroyo R, Caso V, Crijns H, De Potter TJR, Dwight J, Guasti L, Hanke T, Jaarsma T, Lettino M, Løchen M-L, Lumbers RT, Maesen B, Mølgaard I, Rosano GMC, Sanders P, Schnabel RB, Suwalski P, Svennberg E, Tamargo J, Tica O, Traykov V, Tzeis S, Kotecha D, Dagres N, Rocca B, Ahsan S, Ameri P, Arbelo E, Bauer A, Borger MA, Buccheri S, Casadei B, Chioncel O, Dobrev D, Fauchier L, Gigante B, Glikson M, Hijazi Z, Hindricks G, Husser D, Ibanez B, James S, Kaab S, Kirchhof P, Køber L, Koskinas KC, Kumler T, Lip GYH, Mandrola J, Marx N, Mcevoy JW, Mihaylova B, Mindham R, Muraru D, Neubeck L, Nielsen JC, Oldgren J, Paciaroni M, Pasquet AA, Prescott E, Rega F, Rossello FJ, Rucinski M, Salzberg SP, Schulman S, Sommer P, Svendsen JH, ten Berg JM, Ten Cate H, Vaartjes I, Vrints CJ, Witkowski A, Zeppenfeld K, Simoni L, Kichou B, Sisakian HS, Scherr D, Cools F, Smajić E, Shalganov T, Manola S, Avraamides P, Taborsky M, Brandes A, El-Damaty AM, Kampus P, Raatikainen P, Garcia R, Etsadashvili K, Eckardt L, Kallergis E, Gellér L, Guðmundsson K, Lyne J, Marai I, Colivicchi F, Abdrakhmanov AS, Bytyci I, Kerimkulova A, Kupics K, Refaat M, Bheleel OA, Barysienė J, Leitz P, Sammut MA, Grosu A, Pavlovic N, Moustaghfir A, Yap S-C, Taleski J, Fink T, Kazmierczak J, Sanfins VM, Cozma D, Zavatta M, Kovačević DV, Hlivak P, Zupan I, Calvo D, Björkenheim A, Kühne M, Ouali S, Demircan S, Sychov OS, Ng A, Kuchkarov H. 2024 ESC guidelines for the management of atrial fibrillation developed in collaboration with the European association for cardio-thoracic surgery (EACTS). Eur Heart J 2024;45:3314–3414. [DOI] [PubMed] [Google Scholar]
- 6. Senoo K, Yukawa A, Ohkura T, Iwakoshi H, Nishimura T, Teramukai S, Narumoto J, Matoba S. Depression and quality of life in older adults with atrial fibrillation: a cross-sectional community-based study. Geriatr Gerontol Int 2022;22:505–510. [DOI] [PubMed] [Google Scholar]
- 7. Pan A, Sun Q, Okereke OI, Rexrode KM, Hu FB. Depression and risk of stroke morbidity and mortality: a meta-analysis and systematic review. JAMA 2011;306:1241–1249. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Yuan M, Xiao ZL, Zhou HY, Rao W, Huang G, Nie HB, Cao W-F, Xu R-S. Bipolar disorder and the risk for stroke incidence and mortality: a meta-analysis. Neurol Sci 2022;43:467–476. [DOI] [PubMed] [Google Scholar]
- 9. Pérez-Piñar M, Ayerbe L, González E, Mathur R, Foguet-Boreu Q, Ayis S. Anxiety disorders and risk of stroke: a systematic review and meta-analysis. Eur Psychiatry 2017;41:102–108. [DOI] [PubMed] [Google Scholar]
- 10. Garnock-Jones KP. Spotlight on dabigatran etexilate in the prevention of stroke and systemic embolism in patients with atrial fibrillation. Drugs Aging 2011;28:415–419. [DOI] [PubMed] [Google Scholar]
- 11. Aviles RJ, Martin DO, Apperson-Hansen C, Houghtaling PL, Rautaharju P, Kronmal RA, Tracy RP, Van Wagoner DR, Psaty BM, Lauer MS, Chung MK. Inflammation as a risk factor for atrial fibrillation. Circulation 2003;108:3006–3010. [DOI] [PubMed] [Google Scholar]
- 12. Teppo K, Jaakkola J, Lehto M, Biancari F, Airaksinen KEJ. The impact of mental health conditions on oral anticoagulation therapy and outcomes in patients with atrial fibrillation: a systematic review and meta-analysis. Am J Prev Cardiol 2021;7:100221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Søgaard M, Skjøth F, Kjældgaard JN, Larsen TB, Hjortshøj SP, Riahi S. Atrial fibrillation in patients with severe mental disorders and the risk of stroke, fatal thromboembolic events and bleeding: a nationwide cohort study. BMJ Open 2017;7:e018209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Schauer DP, Moomaw CJ, Wess M, Webb T, Eckman MH. Psychosocial risk factors for adverse outcomes in patients with nonvalvular atrial fibrillation receiving warfarin. J Gen Intern Med 2005;20:1114–1119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Bueno H, Deaton C, Farrero M, Forsyth F, Braunschweig F, Buccheri S, Dragan S, Gevaert S, Held C, Kurpas D, Ladwig K-H, Lionis CD, Maas AHEM, Merșa CO, Mindham R, Pedersen SS, Rojnic Kuzman M, Szmit S, Taylor RS, Uchmanowicz I, Vilchinsky N, Koskinas KC, Vranckx P, Adamo M, Arbelo E, Asselbergs FW, Boriani G, Brida M, Burke N, Byrne RA, de Laval I, Einsfeld L, Frigoli E, Galiuto L, Gandjbakhch E, Grace SL, Grand J, Heidecker B, Heldner M, Helmark C, Gaber MH, Hill L, Hynes L, Ibanez B, James S, Kharlamov A, Kindermans H, Køber L, Landmesser U, Lip GYH, Magavern EF, McEvoy JW, McGreavy P, Mihaylova B, Moelgaard I, Abreu M, Nielsen JC, Pasquet AA, Prescott E, Rakisheva A, Rocca B, Rossello X, Salzwedel A, Sannino A, Shantsila E, Ski C, Steptoe A, Tanner FC, Telles D, Vaartjes I, von Kemp B, Vrints C, Zeppenfeld K, Bouzid MEA, Karimli E, De Pauw M, Ferhatbegović L, Karamfiloff K, Rickovic V, Moutiris J, Zwisler AD, Elghalid A, Irs A, Monsuez JJ, Agladze R, Herrmann-Lingen C, Dritsas A, Zima E, Karlsson H, Agmon IN, Mattioli AV, Tundybayeva M, Elezi S, Mirrakhimov E, Mintale I, Lapinskas T, Banu C-A, Revenco V, Asfalou I, Kuijpers P, Kostov J, Rice E, Mierzynska A, Abreu M, Mitu F, Bertelli L, Tosevski DL, Katreniakova Z, Fras Z, Campuzano R, Hlebowicz J, Hermann M, Gungor B, Sirenko Y, Makhkamova N. 2025 ESC clinical consensus statement on mental health and cardiovascular disease: developed under the auspices of the ESC clinical practice guidelines committee. Eur Heart J 2025;46:4156–4225. [DOI] [PubMed] [Google Scholar]
- 16. Krittayaphong R, Treewaree S, Lip GYH. Components of the atrial fibrillation better care pathway for holistic care of patients with atrial fibrillation: a win ratio analysis from the COOL-AF registry. Europace 2024;26:euae237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Pasic J, Levy WC, Sullivan MD. Cytokines in depression and heart failure. Psychosom Med 2003;65:181–193. [DOI] [PubMed] [Google Scholar]
- 18. Carney RM, Freedland KE, Veith RC. Depression, the autonomic nervous system, and coronary heart disease. Psychosom Med 2005;67:S29–S33. [DOI] [PubMed] [Google Scholar]
- 19. Murck H, Held K, Ziegenbein M, Künzel H, Koch K, Steiger A. The renin-angiotensin-aldosterone system in patients with depression compared to controls–a sleep endocrine study. BMC Psychiatry 2003;3:15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Pająk A, Jankowski P, Kotseva K, Heidrich J, de Smedt D, De Bacquer D. Depression, anxiety, and risk factor control in patients after hospitalization for coronary heart disease: the EUROASPIRE III Study. Eur J Prev Cardiol 2013;20:331–340. [DOI] [PubMed] [Google Scholar]
- 21. Teppo K, Jaakkola J, Biancari F, Halminen O, Putaala J, Mustonen P, Haukka J, Linna M, Kinnunen J, Tiili P, Kouki E, Penttilä T, Hartikainen J, Aro AL, Airaksinen KEJ, Lehto M. Mental health conditions and risk of first-ever ischaemic stroke and death in patients with incident atrial fibrillation: a nationwide cohort study. Eur J Clin Invest 2022;52:e13801. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Vilchinsky N, Ginzburg K, Fait K, Foa EB. Cardiac-disease-induced PTSD (CDI-PTSD): a systematic review. Clin Psychol Rev 2017;55:92–106. [DOI] [PubMed] [Google Scholar]
- 23. Rafiei S, Raoofi S, Baghaei A, Masoumi M, Doustmehraban M, Nejatifar Z, Sanaei M, Bagheribayati F, Vaziri Shahrebabak Es, Shayestehbonyan M, Javan Biparva A, Raoofi N, Mir M, Momeni Z, Yousefy S, Hashemi Khoonigh Z, Rezaei F, Pashazadeh Kan F, Ghashghaee A. Depression prevalence in cardiovascular disease: global systematic review and meta-analysis. BMJ Support Palliat Care 2023;13:281–289. [DOI] [PubMed] [Google Scholar]
- 24. Karami N, Kazeminia M, Karami A, Salimi Y, Ziapour A, Janjani P. Global prevalence of depression, anxiety, and stress in cardiac patients: a systematic review and meta-analysis. J Affect Disord 2023;324:175–189. [DOI] [PubMed] [Google Scholar]
- 25. Senoo K, Kaneko H, Ueno K, Suzuki Y, Okada A, Fujiu K, Jo T, Takeda N, Morita H, Kamiya K, Ako J, Node K, Yasunaga H, Komuro I. Sex differences in the association between depression and incident cardiovascular disease. JACC Asia 2024;4:279–288. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Moussavi S, Chatterji S, Verdes E, Tandon A, Patel V, Ustun B. Depression, chronic diseases, and decrements in health: results from the world health surveys. Lancet 2007;370:851–858. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
This dataset is available for use by anyone who is a purchaser or contract holder.