Outcomes in patients with atrial fibrillation with and without heart failure on oral anticoagulants: a multicenter retrospective cohort study Authors: Xiaomin Dong, Shuyi Wu, Xinhai Huang, Yanxian Lan, Cuifang You, Shuzheng Jiang, Jinhua Zhang Published: 2026 ABSTRACT BackgroundMulticenter real-world data on anticoagulated patients with atrial fibrillation (AF) and heart failure (HF) remain limited, particularly in Chinese populations. This study compared clinical outcomes in AF patients receiving oral anticoagulants (OACs) according to HF status.MethodsWe conducted a multicenter retrospective cohort study across four hospitals in China and included 4,878 AF patients who received OACs after discharge between January 2019 and August 2023. Patients were classified into HF (n = 2,371) and non-HF (n = 2,507) groups and followed for a median of 19 months. a. Logistic regression analysis was used to assess the associations between HF and clinical outcomes. Propensity score matching (PSM) was to matched HF and no-HF cohorts and to compare rivaroxaban with warfarin among patients with HF.ResultsAfter multivariable adjustment, HF was independently associated with higher risk of all-cause death(adjusted OR 1.414, 95% CI 1.118–1.787; P = 0.004), and not significantly different in risks of total bleeding, major bleeding, gastrointestinal bleeding, intracranial hemorrhage, minor bleeding, thrombosis, or ischemic stroke compare with non-HF. After 1:1 PSM, HF remained associated with higher risk of all-cause death. In the matched HF subgroup, rivaroxaban was associated with lower risk of all-cause death than warfarin (OR 0.619, 95% CI 0.422–0.907; P = 0.014).ConclusionAmong AF patients receiving OACs, concomitant HF was associated with increased risk of all-cause death. In the matched HF subgroup, rivaroxaban was associated with lower risk of all-cause death than warfarin. These findings should be interpreted cautiously given the retrospective observational design and the lack of INR and time-in-therapeutic-range data for warfarin-treated patients.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12959-026-00883-4. Introduction Atrial fibrillation (AF) and heart failure (HF) are two common cardiovascular disorders with increasing prevalence worldwide, particularly in aging populations [1–3]. These conditions frequently coexist and are closely interconnected in clinical practice [4]. AF can worsen HF by reducing atrial contribution to ventricular filling, shortening diastolic filling time, promoting rapid and irregular ventricular response, and contributing to tachycardia-induced cardiomyopathy. Conversely, HF promotes the development and maintenance of AF through elevated atrial pressure, chamber dilation, neurohormonal activation, inflammation, and atrial fibrosis [5, 6]. As a result, AF and HF often form a self-reinforcing pathophysiological cycle associated with disease progression and poor prognosis. Previous studies have shown that the coexistence of AF and HF is associated with increased risks of hospitalization, thromboembolic events, and death compared with either condition alone [7–10]. Because oral anticoagulation is central to stroke prevention in AF, understanding the clinical outcomes of anticoagulant therapy in patients with concomitant HF is particularly important. However, the presence of HF may complicate risk assessment because these patients often have a higher burden of comorbidities, impaired renal function, greater frailty, and more frequent polypharmacy, all of which may influence both thrombotic and bleeding outcomes [11–20]. Evidence regarding the safety and effectiveness of oral anticoagulants (OACs) in patients with AF and HF remains incomplete. Most available data are derived from subgroup analyses of randomized trials, registry-based studies, or populations outside Asia. Although direct oral anticoagulants (DOACs) have become widely used in AF, warfarin remains an important option in many real-world settings, and comparative outcome data in Chinese patients with AF and HF are still limited. In addition, differences in patient characteristics, treatment patterns, and anticoagulant selection may affect outcomes in routine clinical practice. Therefore, we conducted a multicenter retrospective cohort study to compare bleeding outcomes, thrombotic events, ischemic stroke, and all-cause death in anticoagulated patients with AF according to HF status. We also performed a propensity score-matched subgroup analysis to compare rivaroxaban with warfarin among patients with AF and HF. Method Study design and study population This multicenter retrospective cohort study was conducted across four hospitals in China between January 2019 and August 2023. Consecutive adult patients with AF treated at the Affiliated Hospital of Guilin Medical University, Minzu Hospital of Guangxi Zhuang Autonomous Region, Traditional Chinese and Western Medicine Hospital of Wuhan, and Ningde Municipal Hospital Affiliated to Ningde Normal University were screened for eligibility. The study was registered at the Chinese Clinical Trial Registry (ChiCTR2300067734). Patients were eligible if they were aged 18 years or older, had a documented diagnosis of AF, and received OACs therapy after discharge. Patients were excluded if they did not receive OAC therapy after discharge, had insufficient data for subsequent analyses, or were lost to follow-up. A total of 4,878 patients were included, comprising 2,371 patients with HF and 2,507 without HF. Data collection and definitions Baseline demographic and clinical data were extracted from hospital information systems. Follow-up information on clinical events was obtained from medical records and follow-up interviews with patients or their relatives. AF was identified by electrocardiography and/or documented medical history during hospitalization. Collected baseline variables included age, sex, height, weight, smoking status, alcohol consumption, hypertension, diabetes mellitus, coronary artery disease, malignancy, peripheral vascular disease, gout, chronic obstructive pulmonary disease (COPD), liver insufficiency, renal insufficiency, and laboratory variables including albumin, platelet count, hemoglobin, D-dimer, and creatinine. HF was defined according to a documented clinical diagnosis in the medical record or discharge diagnosis at the participating center, based on treating physician assessment in routine clinical practice. Because of the retrospective multicenter design, detailed echocardiographic information and consistent classification of HF subtype, including HF with reduced ejection fraction, mildly reduced ejection fraction, or preserved ejection fraction, were not uniformly available and were therefore not analyzed separately. Hemorrhage history and history of thrombosis were recorded as baseline clinical history variables and were analyzed separately from other comorbidities. Anticoagulant exposure and concomitant medications The OACs included warfarin and the following DOACs: apixaban, rivaroxaban, edoxaban, and dabigatran. Concomitant medications, including antiplatelet agents, non-steroidal anti-inflammatory drugs (NSAIDs), proton pump inhibitors (PPIs), statins, amiodarone, angiotensin receptor blockers (ARBs), beta-blockers, calcium channel blockers (CCBs), and digoxin, were also recorded. Clinical risk scores The CHA2 DS2 - VASc score [21] was calculated based on congestive heart failure/left ventricular dysfunction, hypertension, age, diabetes mellitus, prior stroke or transient ischemic attack, vascular disease, age 65–74 years, and female sex. The HAS-BLED score [22] was calculated using hypertension, abnormal liver function, abnormal renal function, stroke history, hemorrhage history, age older than 65 years, drugs, and alcohol use. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Creatinine clearance (CrCl) was estimated using the Cockcroft-Gault formula [23]. Study outcomes The safety outcomes were total bleeding, major bleeding, gastrointestinal bleeding, intracranial hemorrhage (ICH), and minor bleeding. The efficacy outcomes were thrombosis, ischemic stroke, and all-cause death. Total bleeding included all bleeding events recorded during follow-up. Major bleeding was defined according to the International Society on Thrombosis and Haemostasis (ISTH) criteria [24, 25]. Gastrointestinal bleeding and ICH were analyzed separately as site-specific bleeding outcomes. ICH was defined as clinically evident ICH confirmed by brain imaging or autopsy [26]. Gastrointestinal bleeding was defined as bleeding with clinically significant hematemesis, melena, or endoscopic confirmation [24]. Minor bleeding included clinically relevant non-major bleeding or other bleeding events that did not meet ISTH criteria for major bleeding. Because bleeding outcomes were categorized both by severity and by anatomical site, major bleeding should not be interpreted as the simple sum of gastrointestinal bleeding and ICH. Thrombosis was defined as ischemic stroke and/or systemic embolism. Ischemic stroke was defined as an acute focal or global neurological deficit consistent with cerebral ischemia [27]. All-cause death included death from any cause during follow-up. Statistical analysis All statistical analyses were performed using R software version 4.3.1 (R Core Team, Vienna, Austria). All tests were two-sided, and a P value < 0.05 was considered statistically significant. Descriptive analysis The distribution of continuous variables was assessed using the Shapiro-Wilk test. Because most continuous variables were not normally distributed, they are presented as median (interquartile range [IQR]) and were compared using the Mann-Whitney U test or Wilcoxon rank-sum test, as appropriate. Categorical variables are presented as counts (percentages) and were compared using the chi-square test or Fisher’s exact test, as appropriate. Regression analysis Univariable and multivariable logistic regression analysis were performed to assess the associations between HF and the study outcomes, including total bleeding, major bleeding, ICH, gastrointestinal bleeding, minor bleeding, thrombosis, ischemic stroke, and all-cause death. Multivariable models were outcome-specific and were constructed using covariates selected on the basis of clinical relevance, prior literature, and variables associated with each outcome in univariable analyses. Given the clinical importance of anticoagulant therapy for both bleeding and thrombotic outcomes, OAC exposure was included in the modeling strategy. Propensity score matching To reduce baseline imbalance, propensity score matching (PSM) was performed at a 1:1 ratio using nearest-neighbor matching with a caliper of 0.05 [28]. PSM was used both to construct comparable HF and no-HF cohorts and to compare rivaroxaban with warfarin among patients with HF. Covariate balance after matching was assessed using standardized mean differences (SMDs), with an SMD ≤ 0.10 considered indicative of adequate balance. Because statistical significance testing may be influenced by sample size after matching, SMDs rather than P values were used to assess balance in matched cohorts [29]. Ethics approval The study complied with the Declaration of Helsinki and was approved by the Ethics Committee of Fujian Maternity and Child Health Hospital (registration number: ChiCTR2300067734; registration date: January 19, 2023). The requirement for informed consent was waived because of the retrospective nature of the study. Patient data were deidentified to ensure confidentiality. Results Baseline characteristics A total of 4,878 patients with AF receiving OAC therapy were included in the study, of whom 2,371 had HF and 2,507 did not. The median age of the overall cohort was 71 years (IQR, 63–78), and 58% were male. The cohort had a substantial burden of comorbidities, particularly hypertension, coronary artery disease, and prior stroke (Table 1). The median CHA2 DS2 - VASc score was 4 (IQR, 2–5), and the median HAS-BLED score was 1 (IQR, 1–2) Fig. 1. Table 1Patient characteristics at baselineCharacteristicOverall, N = 4,878No HF, N = 2,507HF, N = 2,371p-valueSex, male, n (%)2,844 (58%)1,493 (60%)1,351 (57%)0.072Age, years, Median (25%,75%)71(63,78)70(63,77)71(63,79)0.183BMI, Median (25%,75%)23.4(21.3,25.9)23.4(21.4,25.8)23.4(21.1,26.0)0.908Smoking, n (%)910 (19%)453 (18%)457 (19%)0.286Alcohol, n (%)844 (17%)419(17%)425 (18%)0.272HAS-BLED0.114Score ≥ 3, n (%)669 (14%)363 (14%)306 (13%)Median (25%,75%)1(1.00,2.00)1(1.00,2.00)1(1.00,2.00)CHA2DS2-VASc< 0.001Score ≥ 3, n (%)3,762 (77%)1,797 (72%)1,965 (83%)Median (25%,75%)4.00(2.00,5.00)4.00(2.00,5.00)4.00(2.00,5.00)Baseline clinical historyHemorrhage history, n (%)232 (4.8%)126 (5.0%)106 (4.5%)0.382History of thrombosis, n (%)1,296 (27%)715 (29%)581 (25%)0.002Comorbidities, n (%)Hypertension2,798 (57%)1,506 (60%)1,292 (54%)< 0.001Diabetes mellitus1,042 (21%)552 (22%)490 (21%)0.263Gout382 (7.8%)162 (6.5%)220 (9.3%)< 0.001Coronary heart disease1,647 (34%)726 (29%)921 (39%)< 0.001COPD178 (3.6%)71 (2.8%)107 (4.5%)0.002Stroke1,548 (32%)917 (37%)631 (27%)< 0.001Liver insufficiency393 (8.1%)170 (6.8%)223 (9.4%)< 0.001Renal insufficiency749 (15%)252 (10%)497 (21%)< 0.001Cancer117 (2.4%)76 (3.0%)41 (1.7%)0.004Peripheral vascular disease1,399 (29%)773 (31%)626 (26%)< 0.001Catheter ablation329 (6.7%)199 (7.9%)130 (5.5%)< 0.001LaboratoriesAlbumin, g/L, Median (25%,75%)38.5(35.3,41.6)39.1(35.8,42.1)38.0(34.8,40.9)< 0.001CrCl, ml/min, Median (25%,75%)57(43,75)61(46,79)53(38,70)< 0.001PLT, 109/L, edian (25%,75%)202(163,250)204(168,249)198(158,250)0.003Hb, g/L, Median (25%,75%)129(115,143)130(117,144)129(113,142)< 0.001D-dimer, mg/L, Median (25%,75%)0.66(0.31,1.56)0.59(0.28,1.39)0.75(0.35,1.74)< 0.001Oral anticoagulants, n (%)< 0.001Apixaban177 (3.6)106 (4.2)71 (3)Rivaroxaban, n (%)2718 (55.7)1455 (58)1263 (53.3)Edoxaban, n (%)249 (5.1)105 (4.2)144 (6.1)Dabigatran, n (%)306 (6.3)199 (7.9)107 (4.5)Warfarin, n (%)1,428 (29%)642 (26%)786 (33%)Combined medication, n (%)Antiplatelet drugs1,372 (28%)747 (30%)625 (26%)0.008NSAIDs824 (17%)424 (17%)400 (17%)0.970PPI2,487 (51%)1,219 (49%)1,268 (53%)< 0.001Statins2,952 (61%)1,621 (65%)1,331 (56%)< 0.001Amiodarone624 (13%)348 (14%)276 (12%)0.021ARB1,553 (32%)562 (22%)991 (42%)< 0.001β-Blockers2,913 (60%)1,329 (53%)1,584 (67%)< 0.001CCB1,127 (23%)726 (29%)401 (17%)< 0.001Digoxin1,109 (23%)259 (10%)850 (36%)< 0.001Note: HF, Heart failure; BMI, Body mass index; COPD, Chronic obstructive pulmonary disease; PLT, Platelet count; Hb Hemoglobin; CrCl, Creatinine clearance; OACs, Oral anticoagulants; DOACs, Direct Oral Anticoagulants; NSAIDs, non-steroidal anti-inflammatory drugs; PPI, Proton pump inhibitor; ARB, Angiotensin receptor blockers; CCB, Calcium calcium entry blockers Table 1Patient characteristics at baselineCharacteristicOverall, N = 4,878No HF, N = 2,507HF, N = 2,371p-valueSex, male, n (%)2,844 (58%)1,493 (60%)1,351 (57%)0.072Age, years, Median (25%,75%)71(63,78)70(63,77)71(63,79)0.183BMI, Median (25%,75%)23.4(21.3,25.9)23.4(21.4,25.8)23.4(21.1,26.0)0.908Smoking, n (%)910 (19%)453 (18%)457 (19%)0.286Alcohol, n (%)844 (17%)419(17%)425 (18%)0.272HAS-BLED0.114Score ≥ 3, n (%)669 (14%)363 (14%)306 (13%)Median (25%,75%)1(1.00,2.00)1(1.00,2.00)1(1.00,2.00)CHA2DS2-VASc< 0.001Score ≥ 3, n (%)3,762 (77%)1,797 (72%)1,965 (83%)Median (25%,75%)4.00(2.00,5.00)4.00(2.00,5.00)4.00(2.00,5.00)Baseline clinical historyHemorrhage history, n (%)232 (4.8%)126 (5.0%)106 (4.5%)0.382History of thrombosis, n (%)1,296 (27%)715 (29%)581 (25%)0.002Comorbidities, n (%)Hypertension2,798 (57%)1,506 (60%)1,292 (54%)< 0.001Diabetes mellitus1,042 (21%)552 (22%)490 (21%)0.263Gout382 (7.8%)162 (6.5%)220 (9.3%)< 0.001Coronary heart disease1,647 (34%)726 (29%)921 (39%)< 0.001COPD178 (3.6%)71 (2.8%)107 (4.5%)0.002Stroke1,548 (32%)917 (37%)631 (27%)< 0.001Liver insufficiency393 (8.1%)170 (6.8%)223 (9.4%)< 0.001Renal insufficiency749 (15%)252 (10%)497 (21%)< 0.001Cancer117 (2.4%)76 (3.0%)41 (1.7%)0.004Peripheral vascular disease1,399 (29%)773 (31%)626 (26%)< 0.001Catheter ablation329 (6.7%)199 (7.9%)130 (5.5%)< 0.001LaboratoriesAlbumin, g/L, Median (25%,75%)38.5(35.3,41.6)39.1(35.8,42.1)38.0(34.8,40.9)< 0.001CrCl, ml/min, Median (25%,75%)57(43,75)61(46,79)53(38,70)< 0.001PLT, 109/L, edian (25%,75%)202(163,250)204(168,249)198(158,250)0.003Hb, g/L, Median (25%,75%)129(115,143)130(117,144)129(113,142)< 0.001D-dimer, mg/L, Median (25%,75%)0.66(0.31,1.56)0.59(0.28,1.39)0.75(0.35,1.74)< 0.001Oral anticoagulants, n (%)< 0.001Apixaban177 (3.6)106 (4.2)71 (3)Rivaroxaban, n (%)2718 (55.7)1455 (58)1263 (53.3)Edoxaban, n (%)249 (5.1)105 (4.2)144 (6.1)Dabigatran, n (%)306 (6.3)199 (7.9)107 (4.5)Warfarin, n (%)1,428 (29%)642 (26%)786 (33%)Combined medication, n (%)Antiplatelet drugs1,372 (28%)747 (30%)625 (26%)0.008NSAIDs824 (17%)424 (17%)400 (17%)0.970PPI2,487 (51%)1,219 (49%)1,268 (53%)< 0.001Statins2,952 (61%)1,621 (65%)1,331 (56%)< 0.001Amiodarone624 (13%)348 (14%)276 (12%)0.021ARB1,553 (32%)562 (22%)991 (42%)< 0.001β-Blockers2,913 (60%)1,329 (53%)1,584 (67%)< 0.001CCB1,127 (23%)726 (29%)401 (17%)< 0.001Digoxin1,109 (23%)259 (10%)850 (36%)< 0.001Note: HF, Heart failure; BMI, Body mass index; COPD, Chronic obstructive pulmonary disease; PLT, Platelet count; Hb Hemoglobin; CrCl, Creatinine clearance; OACs, Oral anticoagulants; DOACs, Direct Oral Anticoagulants; NSAIDs, non-steroidal anti-inflammatory drugs; PPI, Proton pump inhibitor; ARB, Angiotensin receptor blockers; CCB, Calcium calcium entry blockers Patient characteristics at baseline Note: HF, Heart failure; BMI, Body mass index; COPD, Chronic obstructive pulmonary disease; PLT, Platelet count; Hb Hemoglobin; CrCl, Creatinine clearance; OACs, Oral anticoagulants; DOACs, Direct Oral Anticoagulants; NSAIDs, non-steroidal anti-inflammatory drugs; PPI, Proton pump inhibitor; ARB, Angiotensin receptor blockers; CCB, Calcium calcium entry blockers Compared with patients without HF, those with HF had lower prevalences of history of thrombosis, hypertension, prior stroke, peripheral vascular disease, cancer, and catheter ablation, but higher prevalences of gout, coronary artery disease, COPD, liver insufficiency, and renal insufficiency. Patients with HF also had lower albumin, CrCl, platelet count, and hemoglobin levels, as well as higher D-dimer levels. In addition, a greater proportion of patients in the HF group had a CHA2 DS2 - VASc score ≥ 3 than those in the no-HF group (83% vs. 72%). All patients received OAC therapy. Warfarin was used in 1,428 patients (29%), and DOACs were used in 3,450 patients, including apixaban in 177, rivaroxaban in 2,718, edoxaban in 249, and dabigatran in 306 patients. Compared with the no-HF group, the HF group had a higher proportion of warfarin use and a lower proportion of DOAC use. Clinical outcomes During a median follow-up of 19 months, 508 patients (10.4%) experienced total bleeding, 206 (4.2%) experienced major bleeding, 178 (3.6%) experienced gastrointestinal bleeding, 29 (0.6%) experienced ICH, and 313 (6.4%) experienced minor bleeding. In addition, 275 patients (5.6%) experienced thrombosis, including 153 ischemic strokes (3.1%), and 419 patients (8.6%) died during follow-up (Table 2). Table 2Clinical outcomes of HF vs. no HF in AF patients taking OACsOutcomeNo HF(n = 2,507)HF(n = 2,371)Total events, n = 4878)OR (95% CI) P Adjusted OR (95%CI) P Total bleeding, n (%)283 (11.3%)225 (9.5%)508 (10.41%)0.824(0.685–0.991)0.0440.827(0.671–1.019) a0.075Major bleeding, n (%)108(4.3%)98(4.1%)206 (4.22%)0.958 (0.724–1.266)0.7620.882 (0.656–1.186) b0.406Intracranial haemorrhage, n (%)21 (0.8%)8(0.3%)29 (0.59%)0.401 (0.177–0.907)0.0280.492 (0.216–1.122) b0.092Gastrointestinal bleeding, n (%)88 (3.5%)90(3.8%)178 (3.65%)1.085 (0.804–1.463)0.5950.976 (0.71–1.34) b0.879Minor bleeding, n (%)178 (7.1%)135 (5.7%)313 (6.42%)0.79 (0.627–0.995)0.0460.787 (0.617–1.003) c0.053Thrombosis, n (%)135 (5.4%)140 (5.9%)275 (5.64%)1.103 (0.864–1.406)0.4321.236 (0.949–1.61) d0.117Ischemic stroke, n (%)87 (3.5%)66 (2.8%)153 (3.14%)0.796 (0.575–1.102)0.1700.99 (0.693–1.414) d0.954All-cause death, n (%)172 (6.9%)247 (10%)419 (8.59%)1.579(1.288–1.935)< 0.0011.414(1.118–1.787) a0.004AF, atrial fbrillation; HF, heart failure; OR, Odds ratio; 95%CI, 95% Confdence interval;a: Adjust for heart failure, sex, age, body mass index, smoking, alcoho, hemorrhage history, history of thrombosis, hypertension, diabetes, gout, coronary artery disease, chronic obstructive pulmonary disease, stroke, renal insufficiency, hypohepatia, cancer, peripheral vascular disease, catheter ablation, albumin, creatinine clearance, platelet count, hemoglobin, D-dimer, OACs, antiplatelet agents, non-steroidal anti-inflammatory drugs, proton pump inhibitors, statins, amiodarone, angiotensin receptor blockers, beta-blockers, digoxin, calcium blockers;b: Adjust for heart failure, hemorrhage history, stroke, OACs;c: Adjust for heart failure, age, smoking, alcohol, hemorrhage history, history of thrombosis, gout, chronic obstructive pulmonary disease, renal insufficiency, creatinine clearance, hemoglobin, non-steroidal anti-inflammatory drugs, proton pump inhibitors, amiodarone, angiotensin receptor blockers, OACs;d: Adjust for heart failure, sex, age, body mass index, history of thrombosis, hypertension, stroke, hypohepatia, peripheral vascular disease, creatinine clearance, non-steroidal anti-inflammatory drugs, statins, digoxin, calcium blockers, OACsAdjusted models were outcome-specific and were constructed based on clinically relevant covariates and variables associated with each outcome in univariable analyses (P < 0.20) Table 2Clinical outcomes of HF vs. no HF in AF patients taking OACsOutcomeNo HF(n = 2,507)HF(n = 2,371)Total events, n = 4878)OR (95% CI) P Adjusted OR (95%CI) P Total bleeding, n (%)283 (11.3%)225 (9.5%)508 (10.41%)0.824(0.685–0.991)0.0440.827(0.671–1.019) a0.075Major bleeding, n (%)108(4.3%)98(4.1%)206 (4.22%)0.958 (0.724–1.266)0.7620.882 (0.656–1.186) b0.406Intracranial haemorrhage, n (%)21 (0.8%)8(0.3%)29 (0.59%)0.401 (0.177–0.907)0.0280.492 (0.216–1.122) b0.092Gastrointestinal bleeding, n (%)88 (3.5%)90(3.8%)178 (3.65%)1.085 (0.804–1.463)0.5950.976 (0.71–1.34) b0.879Minor bleeding, n (%)178 (7.1%)135 (5.7%)313 (6.42%)0.79 (0.627–0.995)0.0460.787 (0.617–1.003) c0.053Thrombosis, n (%)135 (5.4%)140 (5.9%)275 (5.64%)1.103 (0.864–1.406)0.4321.236 (0.949–1.61) d0.117Ischemic stroke, n (%)87 (3.5%)66 (2.8%)153 (3.14%)0.796 (0.575–1.102)0.1700.99 (0.693–1.414) d0.954All-cause death, n (%)172 (6.9%)247 (10%)419 (8.59%)1.579(1.288–1.935)< 0.0011.414(1.118–1.787) a0.004AF, atrial fbrillation; HF, heart failure; OR, Odds ratio; 95%CI, 95% Confdence interval;a: Adjust for heart failure, sex, age, body mass index, smoking, alcoho, hemorrhage history, history of thrombosis, hypertension, diabetes, gout, coronary artery disease, chronic obstructive pulmonary disease, stroke, renal insufficiency, hypohepatia, cancer, peripheral vascular disease, catheter ablation, albumin, creatinine clearance, platelet count, hemoglobin, D-dimer, OACs, antiplatelet agents, non-steroidal anti-inflammatory drugs, proton pump inhibitors, statins, amiodarone, angiotensin receptor blockers, beta-blockers, digoxin, calcium blockers;b: Adjust for heart failure, hemorrhage history, stroke, OACs;c: Adjust for heart failure, age, smoking, alcohol, hemorrhage history, history of thrombosis, gout, chronic obstructive pulmonary disease, renal insufficiency, creatinine clearance, hemoglobin, non-steroidal anti-inflammatory drugs, proton pump inhibitors, amiodarone, angiotensin receptor blockers, OACs;d: Adjust for heart failure, sex, age, body mass index, history of thrombosis, hypertension, stroke, hypohepatia, peripheral vascular disease, creatinine clearance, non-steroidal anti-inflammatory drugs, statins, digoxin, calcium blockers, OACsAdjusted models were outcome-specific and were constructed based on clinically relevant covariates and variables associated with each outcome in univariable analyses (P < 0.20) Clinical outcomes of HF vs. no HF in AF patients taking OACs AF, atrial fbrillation; HF, heart failure; OR, Odds ratio; 95%CI, 95% Confdence interval; a: Adjust for heart failure, sex, age, body mass index, smoking, alcoho, hemorrhage history, history of thrombosis, hypertension, diabetes, gout, coronary artery disease, chronic obstructive pulmonary disease, stroke, renal insufficiency, hypohepatia, cancer, peripheral vascular disease, catheter ablation, albumin, creatinine clearance, platelet count, hemoglobin, D-dimer, OACs, antiplatelet agents, non-steroidal anti-inflammatory drugs, proton pump inhibitors, statins, amiodarone, angiotensin receptor blockers, beta-blockers, digoxin, calcium blockers; b: Adjust for heart failure, hemorrhage history, stroke, OACs; c: Adjust for heart failure, age, smoking, alcohol, hemorrhage history, history of thrombosis, gout, chronic obstructive pulmonary disease, renal insufficiency, creatinine clearance, hemoglobin, non-steroidal anti-inflammatory drugs, proton pump inhibitors, amiodarone, angiotensin receptor blockers, OACs; d: Adjust for heart failure, sex, age, body mass index, history of thrombosis, hypertension, stroke, hypohepatia, peripheral vascular disease, creatinine clearance, non-steroidal anti-inflammatory drugs, statins, digoxin, calcium blockers, OACs Adjusted models were outcome-specific and were constructed based on clinically relevant covariates and variables associated with each outcome in univariable analyses (P < 0.20) Bleeding outcomes The incidence of total bleeding was 9.5% in the HF group and 11.3% in the no-HF group. Major bleeding occurred in 4.1% of patients with HF and 4.3% of patients without HF. Gastrointestinal bleeding occurred in 3.8% and 3.5% of patients, ICH in 0.3% and 0.8%, and minor bleeding in 5.7% and 7.1% of patients in the HF and no-HF groups, respectively. In univariable analysis, HF was associated with lower odds of total bleeding (OR 0.824, 95% CI 0.685–0.991; P = 0.044), ICH (OR 0.401, 95% CI 0.177–0.907; P = 0.028), and minor bleeding (OR 0.790, 95% CI 0.627–0.995; P = 0.046). However, after multivariable adjustment, HF was not significantly associated with total bleeding (adjusted OR 0.827, 95% CI 0.671–1.019; P = 0.075), major bleeding (adjusted OR 0.822, 95% CI 0.656–1.186; P = 0.406), ICH (adjusted OR 0.492, 95% CI 0.216–1.122; P = 0.092), gastrointestinal bleeding (adjusted OR 0.976, 95% CI 0.710–1.340; P = 0.879), or minor bleeding (adjusted OR 0.787, 95% CI 0.617–1.003; P = 0.053). Multivariable analyses of risk factors for bleeding outcomes are presented in the Supplementary Tables 1–5 and Supplementary Figure 1–5. In adjusted models, history of thrombosis, gout, peripheral vascular disease, catheter ablation, NSAIDs use, amiodarone use, and beta-blocker use were associated with total bleeding, whereas male sex was associated with lower odds of total bleeding. History of thrombosis was associated with gastrointestinal bleeding, and amiodarone use was associated with minor bleeding. Thrombotic outcomes The incidence of thrombosis was 5.9% in the HF group and 5.4% in the no-HF group. Ischemic stroke occurred in 2.8% of patients with HF and 3.5% of patients without HF. In univariable analysis, HF was not significantly associated with thrombosis (OR 1.103, 95% CI 0.864–1.406; P = 0.432) or ischemic stroke (OR 0.796, 95% CI 0.575–1.102; P = 0.170). These associations remained non-significant after multivariable adjustment for thrombosis (adjusted OR 1.236, 95% CI 0.949–1.610; P = 0.117) and ischemic stroke (adjusted OR 0.990, 95% CI 0.693–1.414; P = 0.954). As shown in the Supplementary Tables 6–7 and Supplementary Figure 6–7, history of thrombosis and prior stroke were independently associated with thrombosis and ischemic stroke, whereas male sex was associated with lower odds of these outcomes. All-cause death All-cause death occurred in 247 patients (10.4%) in the HF group and 172 patients (6.9%) in the no-HF group. HF was associated with significantly higher odds of all-cause death in both univariable analysis (OR 1.579, 95% CI 1.288–1.935; P < 0.001) and multivariable analysis (adjusted OR 1.414, 95% CI 1.118–1.787; P = 0.004). In the multivariable model for all-cause death, older age, history of thrombosis, cancer, liver insufficiency, higher platelet count, higher D-dimer level, and NSAIDs use were associated with higher odds of death, whereas higher hemoglobin, higher albumin, OAC exposure, and ARB use were associated with lower odds of death (Supplementary Table 8, Supplementary Figure 8). Propensity score-matched analysis After 1:1 PSM, 1,413 patients with HF were matched to 1,413 patients without HF (Supplementary Table 9). Baseline characteristics were well balanced after matching, with all standardized mean differences ≤ 0.10. In the matched cohort, HF remained associated with a higher risk of all-cause death (OR 1.394, 95% CI 1.110–1.751; P < 0.001). Subgroup analysis of rivaroxaban versus warfarin in patients with HF We further compared rivaroxaban and warfarin among patients with AF and HF. Baseline characteristics before and after matching are presented in the Supplementary Material. After 1:1 PSM, 615 patients remained in each treatment group, and baseline covariates were generally well balanced (Supplementary Table 10). In the matched HF cohort, rivaroxaban was associated with a lower risk of all-cause death than warfarin (OR 0.619, 95% CI 0.422–0.907; P = 0.014) (Table 3). Table 3Association of HF and anticoagulant type with all-cause death after propensity score matchingAnalysisOR (95% CI)P-valueHF vs. no HF in matched cohort a1.394 (1.110–1.751)P<0.001Rivaroxaban vs. Warfarin in matched HF cohort e0.619 (0.422–0.907)0.014a: Adjust for heart failure, sex, age, body mass index, smoking, alcohol, hemorrhage history, history of thrombosis, hypertension, diabetes, gout, coronary artery disease, chronic obstructive pulmonary disease, stroke, renal insufficiency, hypohepatia, cancer, peripheral vascular disease, catheter ablation, albumin, creatinine clearance, platelet count, hemoglobin, D-dimer, OACs, antiplatelet agents, non-steroidal anti-inflammatory drugs, proton pump inhibitors, statins, amiodarone, angiotensin receptor blockers, beta-blockers, digoxin, calcium blockers;e: Adjust for heart failure, sex, age, body mass index, smoking, alcohol, hemorrhage history, history of thrombosis, hypertension, diabetes, gout, coronary artery disease, chronic obstructive pulmonary disease, stroke, renal insufficiency, hypohepatia, peripheral vascular disease, antiplatelet agents, non-steroidal anti-inflammatory drugs, proton pump inhibitors, statins Table 3Association of HF and anticoagulant type with all-cause death after propensity score matchingAnalysisOR (95% CI)P-valueHF vs. no HF in matched cohort a1.394 (1.110–1.751)P<0.001Rivaroxaban vs. Warfarin in matched HF cohort e0.619 (0.422–0.907)0.014a: Adjust for heart failure, sex, age, body mass index, smoking, alcohol, hemorrhage history, history of thrombosis, hypertension, diabetes, gout, coronary artery disease, chronic obstructive pulmonary disease, stroke, renal insufficiency, hypohepatia, cancer, peripheral vascular disease, catheter ablation, albumin, creatinine clearance, platelet count, hemoglobin, D-dimer, OACs, antiplatelet agents, non-steroidal anti-inflammatory drugs, proton pump inhibitors, statins, amiodarone, angiotensin receptor blockers, beta-blockers, digoxin, calcium blockers;e: Adjust for heart failure, sex, age, body mass index, smoking, alcohol, hemorrhage history, history of thrombosis, hypertension, diabetes, gout, coronary artery disease, chronic obstructive pulmonary disease, stroke, renal insufficiency, hypohepatia, peripheral vascular disease, antiplatelet agents, non-steroidal anti-inflammatory drugs, proton pump inhibitors, statins Association of HF and anticoagulant type with all-cause death after propensity score matching a: Adjust for heart failure, sex, age, body mass index, smoking, alcohol, hemorrhage history, history of thrombosis, hypertension, diabetes, gout, coronary artery disease, chronic obstructive pulmonary disease, stroke, renal insufficiency, hypohepatia, cancer, peripheral vascular disease, catheter ablation, albumin, creatinine clearance, platelet count, hemoglobin, D-dimer, OACs, antiplatelet agents, non-steroidal anti-inflammatory drugs, proton pump inhibitors, statins, amiodarone, angiotensin receptor blockers, beta-blockers, digoxin, calcium blockers; e: Adjust for heart failure, sex, age, body mass index, smoking, alcohol, hemorrhage history, history of thrombosis, hypertension, diabetes, gout, coronary artery disease, chronic obstructive pulmonary disease, stroke, renal insufficiency, hypohepatia, peripheral vascular disease, antiplatelet agents, non-steroidal anti-inflammatory drugs, proton pump inhibitors, statins Because this was an observational analysis and data on warfarin anticoagulation quality, including INR control and time in therapeutic range, were unavailable, this finding should be interpreted cautiously. Discussion In this multicenter retrospective cohort study of anticoagulated patients with AF, concomitant HF was independently associated with a higher risk of all-cause death, whereas no statistically significant associations were observed between HF and adjusted risks of bleeding outcomes, thrombosis, or ischemic stroke. In a propensity score-matched HF subgroup, rivaroxaban was associated with lower all-cause mortality than warfarin. The coexistence of AF and HF is clinically important because these two disorders interact through multiple hemodynamic, structural, and neurohormonal pathways [5, 6]. HF contributes to atrial stretch, fibrosis, inflammation, and electrical remodeling, thereby facilitating AF. Conversely, AF may further worsen HF by impairing ventricular filling, reducing cardiac output, and promoting tachycardia-related myocardial dysfunction. This bidirectional relationship likely contributes to the greater clinical vulnerability of patients with both conditions and may help explain the increased mortality observed in our HF cohort. Our finding that HF was associated with higher all-cause mortality is consistent with prior literature showing that AF-HF coexistence identifies a high-risk population. Importantly, this association persisted after multivariable adjustment and after propensity score matching, suggesting that HF provides prognostic information beyond measured baseline characteristics alone. At the same time, the attenuation of the effect estimate after adjustment indicates that the excess risk is also influenced by comorbidities, frailty, organ dysfunction, and treatment-related factors. In contrast, HF was not independently associated with significantly higher adjusted risks of bleeding, thrombosis, or ischemic stroke in our study. This differs from some previous reports and may reflect differences in study population, anticoagulant exposure, baseline comorbidity profiles, and routine care patterns. For example, compared with patients without HF, those with HF in our cohort had lower prevalences of prior stroke and history of thrombosis, which may have influenced thrombotic risk estimates [11–14, 16–19]. In addition, real-world prescribing patterns and heterogeneity in HF severity may have contributed to the observed results. Overall, our findings suggest that in anticoagulated patients with AF, the excess burden associated with HF may be reflected more strongly in mortality than in anticoagulation-related bleeding or thrombotic events. The comparison between rivaroxaban and warfarin in the matched HF subgroup is clinically relevant but should be interpreted cautiously. One possible explanation for the lower mortality observed with rivaroxaban is more stable anticoagulation in routine practice, given that warfarin effectiveness and safety are strongly affected by diet, drug interactions, monitoring quality, and maintenance of therapeutic INR [30]. It is also possible that factor Xa inhibition may have broader vascular or pleiotropic effects, as suggested in some prior studies, although such mechanisms cannot be established in our dataset [31]. Importantly, the present study was observational, and residual confounding by indication or treatment selection remains possible. We also acknowledge that other DOACs, including apixaban, dabigatran, and edoxaban, are established treatment options for stroke prevention in AF. Because rivaroxaban accounted for the majority of DOAC exposure in our cohort, meaningful comparisons among individual DOAC agents were not feasible. Therefore, our findings regarding rivaroxaban should not be generalized to all DOACs without caution. Our study also revealed that female patients had a higher risk of both thrombosis and bleeding, which is consistent with findings from other studies on anticoagulant therapies [32, 33]. This may be due to sex-related factors such as hormonal influences, inflammation, and platelet aggregation [34–37]. These findings highlight the need to consider gender differences when designing treatment strategies, as they allow for more comprehensive risk assessments and the development of individualized treatment plans. Clinicians should be aware of these differences to optimize therapy and minimize adverse outcomes. In adjusted analyses, male sex was associated with lower odds of some bleeding and thrombotic outcomes; however, the present study was not specifically designed to evaluate sex-based differences, and no formal sex-stratified analyses were performed. These observations should therefore be considered exploratory rather than definitive. This study has several limitations. First, as a retrospective observational study, it is subject to selection bias, information bias, and residual confounding. Second, HF was identified from routine clinical documentation, and detailed information on HF subtype, ejection fraction, severity, and HF-specific treatment was not uniformly available. Third, we did not have data on medication adherence during follow-up. Fourth, INR values and time in therapeutic range were unavailable for warfarin-treated patients, which is a major limitation when interpreting the comparison between warfarin and rivaroxaban. Fifth, because rivaroxaban was the predominant DOACs in this cohort, comparisons among individual DOACs were limited. Finally, some secondary outcomes, particularly ICH, had relatively few events, which may have reduced statistical power. Overall, our findings support the need for comprehensive and individualized management in anticoagulated patients with AF and HF. In addition to anticoagulant selection, optimization of comorbidity management and closer risk stratification may be important for improving long-term outcomes in this clinically complex population. Conclusion Among patients with AF receiving oral anticoagulant therapy, concomitant HF was associated with increased all-cause mortality but was not independently associated with significantly higher adjusted risks of bleeding or thrombotic outcomes. In the propensity score-matched HF subgroup, rivaroxaban was associated with lower all-cause mortality than warfarin. Given the retrospective design, potential residual confounding, and the absence of INR and time-in-therapeutic-range data for warfarin-treated patients, these findings should be interpreted cautiously and require confirmation in prospective studies. Fig. 1Selection of the study population Selection of the study population Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1