Risk of Death From Various Causes According to Prevalent Atrial Fibrillation: A Nationwide Population-Based Study Authors: Young-Kwan Kim, So-Ryoung Lee, Eue-Keun Choi, Hyun Jin Ahn, Nan Young Bae, Kyung-Yeon Lee, JungMin Choi, Hyo-Jeong Ahn, Soonil Kwon, Kyungdo Han, Seil Oh, Gregory Y H Lip Published: 2024 ABSTRACT AbstractBackgroundAtrial fibrillation (AF) is associated with increased risks of adverse events including stroke and all-cause death. Understanding the pattern of causes of death (COD) with the relative risks in patients with AF compared to the non-AF population is essential in planning optimal care for patients with AF. We aimed to analyze the COD and its relative risks in patients with AF, using a nationwide population-based cohort.MethodsUsing the Korean nationwide claims database, people aged 40 or older who received health examinations in 2009 were included if they had no missing values (n = 6,877,929). In total the study included 40,585 people with AF and 6,837,344 without AF. COD was defined by International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnostic codes. Comparison between the AF and non-AF groups was performed with Multivariate Cox proportional regression model.ResultsIn the AF group, cardiovascular diseases were the most common COD, causing 39.8% of all deaths, compared with 19.0% for non-AF subjects. The AF group was associated with a higher risk of death from cardiovascular and cerebrovascular diseases by almost 3-fold than the matched non-AF group (hazard ratios [HR], 3.082; 95% confidence intervals [CIs], 2.963–3.205 for cardiovascular diseases; HR, 2.981; 95% CI, 2.799–3.175 for cerebrovascular diseases, all P < 0.001). Among patients with AF, the risks of all-cause, cardiovascular, and cerebrovascular death were well-stratified by CHA2DS2-VASc scores. The risk of cerebrovascular death was 11 times higher among patients with a CHA2DS2-VASc score ≥ 7.ConclusionCompared to non-AF individuals, patients with AF had a higher risk of death from cardiovascular and cerebrovascular diseases, and the mortality risks were well-stratified by the CHA2DS2-VASc score. Integrated care management of cardiovascular and cerebrovascular diseases for patients with AF might help mitigate mortality.Keywords: Atrial Fibrillation, Cause of Death, CHA2DS2-VASc Score, Population-Based Study, Korea INTRODUCTION Atrial fibrillation (AF) is one of the most common arrhythmias around the world.1 Given the progressive aging of the Asian population,2,3 numerous studies anticipate the prevalence of AF to increase over time.4,5,6 In response to the increasing global burden of AF, contemporary guidelines underscore the significance of an integrated care approach aimed at refining treatment strategies.7 Moreover, this approach accentuates comprehensive patient evaluation, encompassing lifestyle considerations, psychosocial assessment, and concurrent medical conditions.8 Aligned with this paradigm, exploring the causes of mortality among patients with AF, along with assessing their individual risks, has the potential to guide the attention to essential aspects of the disease. Although non-anticoagulated patients with AF are at high risk of thromboembolism and death, the residual risks of death in patients with AF despite taking oral anticoagulation are well recognized, thus emphasizing the need for a holistic or integrated care approach.9 Several previous studies have described the causes of death (COD) in patients with AF.10,11,12 However, the majority of these studies primarily focused on individuals who had already been diagnosed with prevalent AF. This made it difficult to directly compare them with individuals without AF. Moreover, given the known variations in treatment effects and adverse outcomes among patients with AF within the Asian population,13,14 there is an increasing need to conduct research that specifically targets this demographic to investigate the underlying causes of mortality specifically among Asian patients with AF. Furthermore, there is no scoring system specifically designed to predict mortality among patients with AF. Although the CHA2DS2-VASc score is primarily used for predicting stroke risk, many clinicians have also studied its usefulness as a prognostic factor for mortality in patients with AF.15 Conducting extensive longitudinal assessment on a substantial cohort will affirm the predictive efficacy of the CHA2DS2-VASc score. This will provide clinicians with increased confidence when evaluating the overall prognosis of individual patients. In this nationwide population-based study, we analyzed the difference in COD between subjects with AF and those without, using a nationwide population cohort. Furthermore, we investigated whether the CHA2DS2-VASc score could be used to stratify mortality risk in patients with AF. METHODS Database This study relied on information extracted from a comprehensive nationwide population cohort in South Korea. Korean government operates a distinct health insurance framework wherein the government extends healthcare coverage to all citizens, reimbursing healthcare institutions based on services rendered to insured individuals. Consequently, the Korean National Health Insurance System (NHIS) gains access to extensive data encompassing diagnoses, health behaviors, physical assessments, laboratory findings, and patient questionnaires, all furnished by healthcare providers.16,17 Each subject's diagnosis is established through the utilization of revised codes from the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM). Details pertaining to the mortality of each participant derive from a linked dataset connected to the National Health Information Database, overseen by Statistics Korea. This dataset encompasses data regarding the precise and contributory COD, as well as the temporal and geographical particulars associated with individuals whose insurance coverage falls under the purview of the Korean National Health Insurance System (NHIS).16 The validity of the dataset has been previously reported.18,19 Study population The study cohort comprises individuals who participated in a government-conducted national health screening assessment. Mandatory for Korean residents beyond a specified age, this screening encompasses fundamental laboratory tests, physical evaluations, and lifestyle questionnaires. To ensure data with an extended follow-up period, the present investigation is centered on subjects who underwent this screening in 2009 (n = 10,585,843). People aged 40 or older were included in this study (n = 7,240,800). After excluding participants with incomplete data (n = 362,871), the study enrolled a total of 6,877,929 subjects. AF was defined via the ICD-10-CM code I48. Subjects with this diagnostic code necessitating ≥ 2 outpatient visits or ≥1 hospitalization within a year were categorized into the AF group (n = 40,585). All other participants constituted the non-AF group (n = 6,837,344) (Fig. 1). Follow-up extended until December 31st, 2019, or until disqualification due to death or relocation to foreign nations. Flow chart of the study population. AF = atrial fibrillation. Covariate Baseline characteristics and comorbidities (age, sex, hypertension [HTN], diabetes mellitus [DM], dyslipidemia [DL], ischemic heart disease [IHD], ischemic stroke, and chronic kidney disease [CKD]) were evaluated by a composite assessment involving ICD-10-CM diagnostic codes, inpatient and outpatient medical records, clinical examination results, and medication prescription records. Detailed information about definitions of each comorbidity and clinical scores such as CHA2DS2-VASc score are described in Supplementary Table 1. Behavioral traits such as smoking status, level of alcohol consumption, and physical activity were rated based on self-reported questionnaire answers obtained during health examinations. Smoking status was divided into three categories: non-smoker, ex-smoker, or current smoker. Alcohol consumption level was stratified into non-drinker, mild to moderate drinker (0 to 30 g/day), or heavy drinker (> 30 g/day). Regular exercise was defined as performing moderate-intensity exercises 5 or more times per week or vigorous-intensity exercises 3 or more times per week. Cause of death The underlying causes of mortality were categorized based on ICD-10-CM diagnostic codes, with a comprehensive list of these codes and their corresponding abbreviations displayed in Supplementary Table 2. Cardiovascular diseases include disorders with the diagnostic code from I00 to I99, which include rheumatic heart diseases, hypertensive diseases, IHDs, pulmonary heart diseases, valvular heart diseases, cardiomyopathy, arrhythmia, heart failure, cerebrovascular diseases, and diseases of arteries and veins. They are labeled as I in our tables and figures. Non-cardiovascular diseases include disorders with diagnostic codes from A00 to H95 and J00 to Z99. Cerebrovascular diseases include disorders with the diagnostic code I60 to I69, which are subarachnoid hemorrhage, intracerebral hemorrhage, cerebral infarction, occlusion of cerebral arteries, and others. They are labeled as I60-69 in our tables and figures. Statistical analysis Descriptive data encompassed baseline characteristics, presented as mean values with standard deviations or medians with interquartile ranges for continuous variables. Conversely, categorical variables were represented as subject counts with corresponding percentages. Incidence rates (IRs) were computed by dividing the death count by the duration of follow-up, subsequently expressed as 1,000 person-years (PY). Comparisons between the AF and non-AF groups were established through hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs), computed utilizing the Multivariate Cox proportional regression model. Adjustments were executed, encompassing baseline characteristics such as age, sex, smoking status, alcohol consumption level, regular physical activities, and comorbidities comprising DM, HTN, DL, CKD, cancer, depression, IHD, and stroke. Given differences in baseline attributes between the two groups, an additional subgroup of non-AF subjects (n = 202,925) was introduced after applying a 5:1 age and sex matching ratio with the AF group. The particulars of deaths, IRs, and baseline characteristics were articulated using the methodology previously outlined. Furthermore, the AF group underwent partitioning based on sex and CHA2DS2-VASc scores. A separate group was formed for patients with a CHA2DS2-VASc score of 7 or more due to a small patient count at such scores. The analysis delved into the aggregate count of deaths and the proportion corresponding to each cause of death. Multivariate Cox proportional regression models adjusting for age, sex, smoking status, alcohol consumption level, and regular physical activities furnished cause-specific HRs and 95% CIs for each CHA2DS2-VASc score group. The cumulative incidence of mortality across these groups was graphically depicted using Kaplan-Meier curves, whereas comparisons were made using log-rank tests. Statistical analyses were undertaken employing SAS software (version 9.4; SAS Institute Inc., Cary, NC, USA). Ethics statement This research was approved by the Institutional Review Board of Seoul National University Hospital (2312-018-1488). Obtaining informed consent was waived because personal identification details were encrypted following strict confidentiality protocols when the cohort was generated. RESULTS Patient characteristics The baseline characteristics of the study population are summarized in Table 1. Patients with AF were older (64.9 ± 10.1 vs. 54.4 ± 10.5 years, P < 0.001) and more likely to be male (61.6% vs. 50.2%, P < 0.001) compared to those without AF. Baseline characteristics of the post-match study population are shown in Supplementary Table 3. Table 1Baseline characteristicsVariablesAF (n = 40,585)Non-AF (n = 6,837,344)P valueAge, yr64.9 ± 10.154.4 ± 10.5< 0.001≥ 6522,405 (55.2)1,300,254 (19.0)< 0.001Sex, men25,000 (61.6)3,433,968 (50.2)< 0.001ComorbidityHypertension29,731 (73.3)2,313,442 (33.8)< 0.001Diabetes mellitus9,429 (23.2)806,526 (11.8)< 0.001Dyslipidemia15,055 (37.1)1,561,650 (22.8)< 0.001Ischemic heart disease15,680 (38.6)386,537 (5.7)< 0.001Ischemic stroke5,950 (14.7)152,462 (2.2)< 0.001Chronic kidney disease9,033 (22.5)562,798 (8.2)< 0.001Cancer1,692 (4.2)144,068 (2.1)< 0.001Physical examination< 0.001Body mass index, kg/m224.3 ± 3.224.0 ± 3.4< 18.51,176 (2.9)157,175 (2.3)18.5 to < 2312,464 (30.7)2,474,524 (36.2)23 to 2510,558 (26.0)1,821,273 (26.6)25 to 3014,628 (36.0)2,158,079 (31.6)≥ 301,759 (4.3)226,293 (3.3)Waist circumference, cm84.4 ± 8.881.2 ± 8.9< 0.001Central obesity, yes13,706 (33.8)1,544,028 (22.6)< 0.001Laboratory findingsFasting glucose, mg/dL105.0 ± 30.0100.1 ± 26.0< 0.001Systolic blood pressure, mmHg125.8 ± 16.3124.3 ± 15.6< 0.001Diastolic blood pressure, mmHg77.3 ± 10.777.2 ± 10.20.025Estimated GFR, mL/min/1.73m276.2 ± 37.184.9 ± 37.5< 0.001Total cholesterol, mg/dL185.6 ± 42.6199.4 ± 42.5< 0.001LDL-cholesterol, mg/dL108.6 ± 93.5119.1 ± 83.4< 0.001HDL-cholesterol, mg/dL54.0 ± 44.356.1 ± 33.9< 0.001Triglyceride, mg/dL118.5 (117.8–119.1)118.0 (118.0–118.1)0.212AST, mg/dL26.2 (26.1–26.3)24.1 (24.1–24.1)< 0.001ALT, mg/dL22.4 (22.2–22.5)21.7 (21.7–21.7)< 0.001γ-GT, mg/dL34.9 (34.6–35.2)27.1 (27.1–27.2)< 0.001Lifestyle questionnairesaSmoking status< 0.001Non-smoker25,301 (62.3)4,340,555 (63.5)Ex-smoker9,589 (23.6)1,061,652 (15.5)Current smoker5,695 (14.0)1,435,137 (21.0)Alcohol consumption< 0.001Non-drinker27,241 (67.1)4,010,563 (58.7)Mild to moderate (> 0 to 30 g/day)10,756 (26.5)2,326,980 (34.0)Heavy (> 30 g/day)2,588 (6.4)499,801 (7.3)Regular exercise8,348 (20.6)1,367,066 (20.0)0.003Continuous variables are presented as mean and standard deviation or median and interquartile range and categorical variables are presented as number and percentage.AF = atrial fibrillation, GFR = glomerular filtration rate, LDL = low density lipoprotein, HDL = high density lipoprotein, ALT = alanine transaminase, AST = aspartate transaminase, γ-GT = γ-glutamyl transferase.aNon-smoker: smoked less than 100 cigarettes in their lifetime; Ex-smoker: smoked more than 100 cigarettes in their lifetime but has not smoked in the last 28 days; Current smoker: smoked more than 100 cigarettes in their lifetime and has smoked in the last 28 days; Regular exercise: performed moderate physical activity for > 30 minutes in ≥ 5 days/week or vigorous physical activity for > 20 minutes in ≥ 3 days/week. Baseline characteristics Continuous variables are presented as mean and standard deviation or median and interquartile range and categorical variables are presented as number and percentage. AF = atrial fibrillation, GFR = glomerular filtration rate, LDL = low density lipoprotein, HDL = high density lipoprotein, ALT = alanine transaminase, AST = aspartate transaminase, γ-GT = γ-glutamyl transferase. aNon-smoker: smoked less than 100 cigarettes in their lifetime; Ex-smoker: smoked more than 100 cigarettes in their lifetime but has not smoked in the last 28 days; Current smoker: smoked more than 100 cigarettes in their lifetime and has smoked in the last 28 days; Regular exercise: performed moderate physical activity for > 30 minutes in ≥ 5 days/week or vigorous physical activity for > 20 minutes in ≥ 3 days/week. Cause-specific mortality of subjects with AF In the total study population, a cumulative count of 490,807 deaths was documented (12,364 in the AF group, and 478,443 in the non-AF group). Notably, cardiovascular death was the major contributor, constituting 39.8% of all death in the AF group, in contrast to 19.0% in the non-AF group (Fig. 2A). Out of the various cardiovascular etiologies, cerebrovascular diseases were the most prevalent. In the AF group, the prevalence of cerebrovascular death was 15.1%, twice of that observed in the non-AF group (7.5%). Among the 37,227 reported deaths in the post-match cohort, 20.3% and 8.0% were attributed to cardiovascular, cerebrovascular causes, respectively. Notably, these percentages were only half of those observed in the AF group (Fig. 2B). Detailed counts and corresponding percentages for each COD are provided in Supplementary Tables 4 and 5. Proportion of various causes of death in AF and non-AF group. (A) Before age, sex matching. (B) After 5:1 age, sex matching. The letters on the right column each stand for International Classification of Diseases, Tenth Revision, Clinical Modification codes which are described on Supplementary Table 2. AF = atrial fibrillation. The AF group showed significantly higher IRs compared to non-AF group in most COD including all-cause mortality (HR, 1.739; 95% CI, 1.708–1.771; P < 0.001, Supplementary Fig. 1 and Supplementary Table 6). Furthermore, patients with AF had a significantly higher risk of death from cardiovascular diseases. The risk of death was nearly three times higher in the AF group compared to the non-AF group (HR, 2.899; 95% CI, 2.814–2.985; P < 0.001). The AF group also had tripled mortality risk attributed to cerebrovascular diseases (HR, 2.911; 95% CI, 2.775–3.054; P < 0.001). Concordantly, comparison with the post-match cohort exhibited a 3-fold higher IR of cardiovascular and cerebrovascular death (HR, 3.082; 95% CI, 2.963–3.205; P < 0.001; HR, 2.981; 95% CI, 2.799–3.175; P < 0.001) (Fig. 3, Supplementary Table 7). Incidence rate and hazard ratio of each cause of death (after 5:1 age, sex matching). AF = atrial fibrillation, CI = confidence interval, IR = incidence rate. aAge, sex, level of smoking, level of alcohol consumption, regular physical activities, diabetes mellitus, dyslipidemia, chronic kidney disease, cancer history, depression, ischemic heart disease, stroke adjusted. Mortality according to CHA2DS2-VASc score and comparisons between men and women For patients with AF, an analysis to determine the cause of death in subgroups with different CHA2DS2-VASc scores was conducted. Our findings indicated that the cumulative incidence of all-cause mortality during the follow-up period exhibited an upward trend corresponding with higher CHA2DS2-VASc scores. Specifically, only 8.5% of subjects with a score of 0 experienced mortality, and displayed a consistent escalation, reaching 72.0% for individuals with a CHA2DS2-VASc score of 7 or higher (Fig. 4, Supplementary Table 8). Cumulative incidence of all-cause death and proportion of cardiovascular, cerebrovascular deaths according to CHA2DS2-VASc score (total population, male, female). A parallel trend was evident concerning the proportion of cardiovascular and cerebrovascular deaths in relation to the CHA2DS2-VASc score. The overall percentages of cardiovascular and cerebrovascular death were recorded at 27.9% and 8.1% in the CHA2DS2-VASc score 0 group, contrasting with 49.2% and 23.2% in the CHA2DS2-VASc score 7 or higher group (Fig. 4, Supplementary Table 9). This increasing trend in the proportion of cardiovascular and cerebrovascular death for higher CHA2DS2-VASc scores was observed in both men and women. Notably, across all CHA2DS2-VASc score groups, the percentage of all-cause death was higher among men compared to that in women. For patients with a score of 7 or more, 79.3% of men died, whereas 69.7% of women died (Fig. 4, Supplementary Table 8). Even considering the sex category score attributed to women, male AF patients with the score of 6 or more still surpass female AF patients with the score of 7 or more (74.8% vs. 69.7%). Unlike the cumulative incidence, the proportion of death due to cardiovascular diseases was higher among women compared to men across all CHA2DS2-VASc score groups. This higher proportion in the women’s group was also evident in cerebrovascular deaths (Fig. 4, Supplementary Tables 10 and 11). With increasing CHA2DS2-VASc scores, the risk of all-cause death increased proportionally (Fig. 5, Supplementary Table 12). The IR of all-cause death for subjects possessing a CHA2DS2-VASc score of 0 was 8.60 per 1,000 PY. This rate increased to 10.75 with a CHA2DS2-VASc score of 1, and further to 117.99 for individuals with a CHA2DS2-VASc score of 7 or higher (HR, 1.060; 95% CI, 0.868–1.295; HR, 3.833; 95% CI, 3.110–4.723; respectively, P < 0.001). This increasing trend in IRs and HRs is congruent across deaths attributed to cardiovascular and cerebrovascular diseases. Commencing from an IR of 3.09 and 0.97 per 1,000 PY, these rates increase to 58.10 and 27.41 per 1,000 PY, respectively (HR, 6.763; 95% CI, 4.625–9.891; P < 0.001; HR, 11.512; 95% CI, 5.775–22.949; P < 0.001). The IRs and HRs of all-cause, cardiovascular, and cerebrovascular death similarly exhibited a discernible stratification across CHA2DS2-VASc scores within men and women (Fig. 5). Detailed numbers are listed in Supplementary Tables 13 and 14. Incidence rate and adjusted hazard ratio of all-cause, cardiovascular, and cerebrovascular deaths according to CHA2DS2-VASc score (Total population, male, female). CI = confidence interval, IR = incidence rate. aAge, sex, level of smoking, level of alcohol consumption, regular physical activities. The Kaplan-Meier analysis confirmed the association between elevated CHA2DS2-VASc scores and progressively higher mortality rates (Supplementary Fig. 2, log-rank P < 0.001). This trend remained consistent for both men and women populations. Similarly, when subjects were stratified based on cause-specific mortality, the outcomes remained stable. Death attributed to cardiovascular and cerebrovascular diseases exhibited an augmented mortality risk among subjects with higher CHA2DS2-VASc scores (Supplementary Figs. 3 and 4, log-rank P < 0.001). DISCUSSION In this nationwide population-based study, we found that patients with AF had an increased risk of all-cause death compared to the non-AF population. Additionally, the risk of cardiovascular and cerebrovascular death was approximately 3-fold higher for patients with AF compared with those without. There are many strengths in our study. To the best of our knowledge, this study is the largest study with a 10-year follow-up period that directly compared AF and non-AF subjects in the Asian population. While several pervious analyses investigated the increased mortality risks of comorbidities in AF patients, most of them included the Caucasian population.20,21 Direct comparison of AF and non-AF subjects through the Korean nationwide claim database allowed a clear analysis of the increased mortality risks of Asian AF patients. Recent population-based studies that reported similar proportions of cause of death underscore the high proportion of non-cardiovascular etiologies.10,11,22 In the Fushimi AF registry based on the Japanese population (n = 4,045), mortality due to cardiovascular causes were approximately 25.5%, with non-cardiovascular causes comprising 54.0% of deaths observed in patients with AF.11 Moreover, the GARFIELD-AF registry (n = 17,162) also reported that the percentage of death from cardiovascular diseases was 40.5%, with the other 59.5% being non-cardiovascular causes.22 A recent prospective multicenter nationwide AF registry (COOL-AF, n = 3,405) from Thailand presented consistent findings with the percentage of cardiovascular deaths being 31.8%, whereas non-cardiovascular death consisted 49.7% of total number of deaths.10 Our study also shows cohesive data that non-cardiovascular deaths contribute to approximately 60.0% of all-cause deaths amongst patients with AF. In addition, death risks due to non-cardiovascular diseases were higher for AF subjects compared with non-AF subjects. This association between AF and non-cardiovascular deaths is not clearly explained. While some studies showed the association between AF and cancer,23,24 the mechanism of increased cancer incidence was not clearly defined. A study using data from the SAKURA AF registry found that being male, underweight, and having anemia were significant factors associated with nonvascular death.25 Including these factors in the adjustment variables might show a different result. Systemic inflammation linked to AF may be the key factor underlying this elevated risk.26,27 Moreover, our findings may indicate that AF represents a manifestation of severe systemic disease, thereby serving as a risk marker.28 Nevertheless, the relative risks of non-cardiovascular deaths for patients with AF were low. While the relative risks for death due to non-cardiovascular diseases were between 1.2 and 1.7, the relative risks for cardiovascular and cerebrovascular death were 2.9 and 2.9, respectively (95% CI, 2.814–2.985, 2.775–3.054). This data provides background to previous articles that underscored the importance of multi-dimensional management for patients with AF including cardiovascular and cerebrovascular diseases.29,30 Several studies have also emphasized comorbid heart failure of patients with AF as the main COD. A Chinese cohort comprised of nonvalvular patients with AF with anticoagulation (n = 2,248) showed that the most common cause of death was cardiovascular disease (55%, 120/218) with heart failure contributing to most (18.3%, 40/218) of them.31 A systematic review of data from randomized controlled trials of non-vitamin K antagonist oral anticoagulants also reported that 15% of deaths were due to heart failure.32 Even though the data about death due to heart failure is not available in this study, considering the results of recent studies investigating the association between heart failure and AF,33,34 further comparison of heart failure-specific risk of death between AF patients and non-AF subjects might help evaluate and emphasize the excess mortality of heart failure due to AF. Previous studies have proposed markers that were associated with increased mortality in patients with AF.35,36 However, they were not able to deduce a simple scoring system to independently predict long-term AF mortality. The newly introduced GARFIELD-AF risk assessment is noteworthy, though it requires considerable time to compute. In contrast, the CHA2DS2-VASc score remains the most widely used and the simplest scoring system to use directly at the patient's side. Although several studies that proposed the predictive value of CHA2DS2-VASc score relative to mortality risk (notwithstanding that this score was proposed for stroke risk, not mortality), the follow-up duration has been relatively short. Our study shows that CHA2DS2-VASc score can be used as a tool for clinicians to roughly estimate the long-term mortality risk each patient has for the next 10 years. The sex differences have been indicated as an important factor influencing mortality risk in patients with AF.37,38 Previous studies emphasized the higher risk of cerebrovascular disease among women than men.12,39,40 According to the Kaplan-Meier analyses of present study, the cumulative risk of death due to cerebrovascular diseases seems to be higher for women with AF with CHA2DS2-VASc score of 7 or more. Compared with all-cause death, the gap between the groups with a score of 6 versus the group with a score of 7 or higher was wider for cerebrovascular deaths in women (Supplementary Fig. 4C). The wider gap is also prominent when compared with men. Further studies targeting patients with high initial CHA2DS2-VASc score to strictly manage cerebrovascular diseases might help diminish mortality, especially for women. The limitations of our study are as follows. First, since the analysis was based on an observational cohort, only the relationship, not the causality, can be proven. Second, the data we used were based on ICD-10-CM codes. The potential inaccuracy regarding underestimation and overestimation of the number of patients with specific cause of death cannot be overlooked. Third, the prescription rates and type of anticoagulation each patient received were not available in the database. If we had compared the patients based on the type of anticoagulation they were taking, we could have conducted a more comprehensive investigation and formulated improved recommendations for anticoagulation strategies. Fourth, factors of AF that might influence mortality risks, such as AF type and duration, were not available in the database. Additional analysis based on the type and duration of AF may offer more insight into our findings. Lastly, the changes in baseline characteristics (i.e., comorbidities, CHA2DS2-VASc score) during follow-up period were not considered in our study. The changes in baseline characteristics might have affected our results. Patients with AF showed higher risks of death from cardiovascular and cerebrovascular diseases than patients without AF, and the risks of mortality were clearly stratified according to CHA2DS2-VASc scores. Integrated care management of cardiovascular and cerebrovascular diseases for patients with AF might help mitigate mortality. Footnotes Funding: This research was supported by a grant from the Patient-Centered Clinical Research Coordinating Center (PACEN) funded by the Ministry of Health & Welfare, Republic of Korea (grant number: RS-2021-KH119931), and supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (IRIS RS-2024-00340590). Disclosure: Young-Kwan Kim, So-Ryoung Lee, Hyun Jin Ahn, Nan Young Bae, Kyung-Yeon Lee, JungMin Choi, Hyo-Jeong Ahn, Soonil Kwon, Kyungdo Han, Seil Oh: nothing to disclose. Eue-Keun Choi: Research grants or speaking fees from Abbott, Bayer, BMS/Pfizer, Biosense Webster, Chong Kun Dang, Daewoong Pharmaceutical Co., Daiichi-Sankyo, DeepQure, Dreamtech Co., Ltd., Jeil Pharmaceutical Co. Ltd, Medtronic, Samjinpharm, Seers Technology, and Skylabs, but he made no influence on this work in relation with the company or its products. Gregory Y. H. Lip: Consultant and speaker for BMS/Pfizer, Boehringer Ingelheim and Daiichi-Sankyo. No fees are received personally. He is a National Institute for Health and Care Research (NIHR) Senior Investigator and co-principal investigator of the AFFIRMO project on multimorbidity in AF, which has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 899871, but he made no influence on this work in relation with the company or its products. Author Contributions: Conceptualization: Lee SR, Choi EK, Han K.Data curation: Han K, Choi EK.Formal analysis: Han K, Kim YK, Lee SR.Funding acquisition: Choi EK.Investigation: Kim YK, Lee SR, Choi EK, Han K.Methodology: Lee SR, Choi EK, Han K.Software: Kim YK, Lee SR, Choi EK, Han K.Supervision: Ahn HJ1, Bae NY, Lee KY, Choi JM, Ahn HJ2, Kwon SI, Oh S, Lip GYH.Validation: Ahn HJ1, Bae NY, Lee KY, Choi JM, Ahn HJ2, Kwon SI, Oh S, Lip GYH.Visualization: Kim YK, Lee SR, Choi EK.Writing - original draft: Kim YK, Lee SR.Writing - review & editing: Lee SR, Choi EK, Lip GYH. Conceptualization: Lee SR, Choi EK, Han K. Conceptualization: Lee SR, Choi EK, Han K. Data curation: Han K, Choi EK. Data curation: Han K, Choi EK. Formal analysis: Han K, Kim YK, Lee SR. Formal analysis: Han K, Kim YK, Lee SR. Funding acquisition: Choi EK. Funding acquisition: Choi EK. Investigation: Kim YK, Lee SR, Choi EK, Han K. Investigation: Kim YK, Lee SR, Choi EK, Han K. Methodology: Lee SR, Choi EK, Han K. Methodology: Lee SR, Choi EK, Han K. Software: Kim YK, Lee SR, Choi EK, Han K. Software: Kim YK, Lee SR, Choi EK, Han K. Supervision: Ahn HJ1, Bae NY, Lee KY, Choi JM, Ahn HJ2, Kwon SI, Oh S, Lip GYH. Supervision: Ahn HJ1, Bae NY, Lee KY, Choi JM, Ahn HJ2, Kwon SI, Oh S, Lip GYH. Validation: Ahn HJ1, Bae NY, Lee KY, Choi JM, Ahn HJ2, Kwon SI, Oh S, Lip GYH. Validation: Ahn HJ1, Bae NY, Lee KY, Choi JM, Ahn HJ2, Kwon SI, Oh S, Lip GYH. Visualization: Kim YK, Lee SR, Choi EK. Visualization: Kim YK, Lee SR, Choi EK. Writing - original draft: Kim YK, Lee SR. Writing - original draft: Kim YK, Lee SR. Writing - review & editing: Lee SR, Choi EK, Lip GYH. Writing - review & editing: Lee SR, Choi EK, Lip GYH. Ahn HJ1, Hyun Jin Ahn; Ahn HJ2, Hyo-Jeong Ahn. SUPPLEMENTARY MATERIALS Supplementary Table 1 Definitions of comorbidities and clinical scores Supplementary Table 2 ICD-10-CM codes of cause of mortality Supplementary Table 3 Baseline characteristics (after 5:1 age, sex matching) Supplementary Table 4 The comparison of cause-specific mortality percentage between atrial fibrillation and non-atrial fibrillation population according to ICD-10-CM code Supplementary Table 5 The comparison of cause-specific mortality percentage between atrial fibrillation (after 5:1 age, sex matching) and non-atrial fibrillation population according to ICD-10-CM code Supplementary Table 6 The incidence rate of cause-specific mortality of subjects with and without atrial fibrillation Supplementary Table 7 The incidence rate of cause-specific mortality of subjects with and without atrial fibrillation (after 5:1 age, sex matching) Supplementary Table 8 Percentage of mortality according to CHA2DS2-VASc score (total population, male, female) Supplementary Table 9 The comparison of cause-specific mortality percentage according to CHA2DS2-VASc score (total population) Supplementary Table 10 The comparison of cause-specific mortality percentage according to CHA2DS2-VASc score (male) Supplementary Table 11 The comparison of cause-specific mortality percentage according to CHA2DS2-VASc score (female) Supplementary Table 12 The incidence rate and hazard ratio difference between CHA2DS2-VASc score (total population) Supplementary Table 13 The incidence rate and hazard ratio difference between CHA2DS2-VASc score (male) Supplementary Table 14 The incidence rate and hazard ratio difference between CHA2DS2-VASc score (female) Supplementary Fig. 1 Incidence rate and HR of each cause of death. Supplementary Fig. 2 Kaplan-Meier curve of all-cause death according to CHA2DS2-VASc score. (A) Total population, (B) male, (C) female. Supplementary Fig. 3 Kaplan-Meier curve of cardiovascular death according to CHA2DS2-VASc score. (A) Total population, (B) male, (C) female. Supplementary Fig. 4 Kaplan-Meier curve of cerebrovascular death according to CHA2DS2-VASc score. (A) Total population, (B) male, (C) female. References Associated Data Supplementary Materials Supplementary Table 1 Definitions of comorbidities and clinical scores Supplementary Table 2 ICD-10-CM codes of cause of mortality Supplementary Table 3 Baseline characteristics (after 5:1 age, sex matching) Supplementary Table 4 The comparison of cause-specific mortality percentage between atrial fibrillation and non-atrial fibrillation population according to ICD-10-CM code Supplementary Table 5 The comparison of cause-specific mortality percentage between atrial fibrillation (after 5:1 age, sex matching) and non-atrial fibrillation population according to ICD-10-CM code Supplementary Table 6 The incidence rate of cause-specific mortality of subjects with and without atrial fibrillation Supplementary Table 7 The incidence rate of cause-specific mortality of subjects with and without atrial fibrillation (after 5:1 age, sex matching) Supplementary Table 8 Percentage of mortality according to CHA2DS2-VASc score (total population, male, female) Supplementary Table 9 The comparison of cause-specific mortality percentage according to CHA2DS2-VASc score (total population) Supplementary Table 10 The comparison of cause-specific mortality percentage according to CHA2DS2-VASc score (male) Supplementary Table 11 The comparison of cause-specific mortality percentage according to CHA2DS2-VASc score (female) Supplementary Table 12 The incidence rate and hazard ratio difference between CHA2DS2-VASc score (total population) Supplementary Table 13 The incidence rate and hazard ratio difference between CHA2DS2-VASc score (male) Supplementary Table 14 The incidence rate and hazard ratio difference between CHA2DS2-VASc score (female) Supplementary Fig. 1 Incidence rate and HR of each cause of death. Supplementary Fig. 2 Kaplan-Meier curve of all-cause death according to CHA2DS2-VASc score. (A) Total population, (B) male, (C) female. Supplementary Fig. 3 Kaplan-Meier curve of cardiovascular death according to CHA2DS2-VASc score. (A) Total population, (B) male, (C) female. Supplementary Fig. 4 Kaplan-Meier curve of cerebrovascular death according to CHA2DS2-VASc score. (A) Total population, (B) male, (C) female.