Frailty and Bleeding Events in Heart Failure: Insights from the JROADHF-NEXT Authors: Taisuke Nakade, Yudai Fujimoto, Tomomi Ide, Keisuke Kida, Shouji Matsushima, Hidetaka Kaku, Nobuyuki Enzan, Masataka Ikeda, Takeshi Kitai, Tatsunori Taniguchi, Takahiro Okumura, Takeshi Tohyama, Hiroyuki Tsutsui, Yuya Matsue Published: 2026 ABSTRACT BackgroundFrail patients are considered to be at higher risk of bleeding than non-frail patients; however, the association between frailty and bleeding risk in patients with heart failure (HF) remains unclear. Therefore, this study aims to evaluate the prevalence of bleeding events, examine the relationship between frailty severity and bleeding risk, and assess the association between bleeding events and subsequent mortality in patients with HF receiving antiplatelet or anticoagulant therapy.MethodsThis post hoc analysis used data from the Japanese Registry of Acute Decompensated Heart Failure–Next, a prospective, multicentre, nationwide registry. Frailty was evaluated using the Clinical Frailty Scale (CFS). Bleeding events were defined as major or clinically relevant non-major bleeding, based on the International Society on Thrombosis and Haemostasis Bleeding Scale.ResultsAmong 4016 patients, 3029 (75.4%) were discharged on antiplatelet or anticoagulant medication. After excluding 89 with missing CFS data, 2940 patients were analyzed (mean age: 75 ± 12 years; 64.0% male). Patients were grouped by CFS scores: 1-3 (n = 1033); 4-5 (n = 1364); and 6-9 (n = 543). At 2 years, 149 (5.0%) experienced bleeding. Higher CFS scores were associated with increased bleeding risk (CFS score 4-5: hazard ratio [HR]: 1.53, 95% confidence interval [CI]: 1.00-2.34; CFS score 6-9: HR: 1.75; 95% CI: 1.04-2.93). In addition, bleeding events were associated with a higher risk of mortality (HR: 2.80; 95% CI: 1.88-4.17, P < 0.001).ConclusionsFrailty is associated with bleeding risk in patients with HF receiving antithrombotic therapy. Bleeding events were associated with a higher risk of subsequent death. Heart failure (HF) is a condition commonly associated with aging and multiple comorbidities. Its prevalence is increasing, and it remains linked to persistently high mortality rates.1, 2, 3 Frailty is a common and clinically important comorbidity in patients with HF.4, 5, 6 A previous systematic review and meta-analysis reported a frailty prevalence of 44.5% in this population, underscoring its substantial clinical burden.7 Moreover, frailty has been associated consistently with worse clinical outcomes in patients with HF.8 Atrial fibrillation and coronary artery disease frequently coexist in patients with HF, resulting in frequent use of antithrombotic therapy in routine clinical practice.9, 10, 11 In a previous cohort of patients hospitalized for acute decompensated HF, approximately half received antiplatelet therapy, and 70% received anticoagulant therapy.12 However, previous studies have shown that older, frail patients face a heightened risk of bleeding. For instance, among patients aged ≥ 65 years with atrial fibrillation, the rate of major bleeding is approximately 1.2 per 100 patient-years, with evidence suggesting that frailty may further increase bleeding risk.13 Nevertheless, whether frailty is associated with bleeding events in patients with HF receiving antiplatelet and/or anticoagulant therapy remains unclear. Moreover, the association between bleeding events and subsequent mortality in this specific population has not been well established. Therefore, to address these gaps in knowledge, we aimed to investigate the association between frailty and bleeding risk in patients with HF and evaluate the association between bleeding events and subsequent mortality. Materials and Methods Study design and patient population This was a post hoc subanalysis of the Japanese Registry of Acute Decompensated Heart Failure–Next (JROADHF-NEXT), a prospective, multicentre registry of patients hospitalized for HF. JROADHF-NEXT enrolled hospitalized patients with HF at 87 cardiovascular care facilities across Japan between April 1, 2019 and April 30, 2021. The primary results of this study have been reported elsewhere.14 Notably, all consecutive patients aged ≥ 20 years who were hospitalized for decompensated HF were eligible. HF decompensation was diagnosed following the Framingham criteria,15 along with the additional requirement of either a B-type natriuretic peptide (BNP) level ≥ 100 pg/mL or an N-terminal-proBNP level ≥ 300 pg/mL on admission. Exclusion criteria were as follows: (i) having a history of heart transplantation; (ii) having acute coronary syndrome requiring urgent revascularization; and (iii) being deemed unsuitable for study participation by the investigators. Only patients who provided written informed consent were enrolled. Relevant medical information and diagnostic information were collected from the Japanese Registry of All Cardiac and Vascular Diseases–Diagnosis Procedure Combination database. All data were anonymized before analysis. The study protocol complied with the principles of the Declaration of Helsinki and was approved by the Kyushu University Ethical Committee (No. 2019-283). Written informed consent was obtained from all participants, who retained the autonomy to withdraw from the study at any time. Relevant study details, including research objectives, inclusion and exclusion criteria, and names of participating institutions, were publicly registered in the University Hospital Information Network (UMIN-CTR: https://www.umin.ac.jp/ctr/index.htm; unique identifier: UMIN000036592) before initiating the initial patient enrollment. Data collection and assessment of frailty Comprehensive physical examinations, prescription records, echocardiographic data, and blood samples were collected under stable conditions before discharge. Notably, only patients on antiplatelet or anticoagulant therapy at discharge were included in this analysis. Frailty was assessed using the Canadian Study of Health and Aging Clinical Frailty Scale (CFS), which evaluates frailty on a 9-point scale based on the patient’s functional autonomy, mobility, and need for assistance with activities of daily living.16 The CFS scoring was performed through direct, face-to-face assessment of patients and, when possible, family members, to determine baseline frailty status before discharge.17 All participating sites used the original published CFS descriptors and pictorial guides, and investigators received standardized written instructions before study initiation to promote consistency across participating hospitals. Although the CFS includes a degree of clinical judgment, it has practical advantages in terms of simplicity and feasibility in routine clinical settings. In addition, previous work has shown that the CFS reflects not only objective physical function but also cognitive function in patients with HF.18 For analysis, CFS scores were categorized into groups (1-3, 4-5, and 6-9) based on the clinical descriptors of the CFS and to ensure sufficient sample sizes in each group, consistent with previous studies using the CFS in cardiovascular populations.17,18 Outcomes Patients were prospectively followed for 2 years after discharge, with a follow-up period of approximately 730 days, and an allowable margin of +90 days. Patients were primarily followed up at their originally enrolled medical facility, with evaluations conducted at 1 and 2 years postdischarge. Prognostic information of those who did not return for in-person follow-up assessment was obtained through telephone interviews. The primary outcome was defined as the onset of a bleeding event, defined before study initiation as either major bleeding or clinically relevant non-major bleeding according to the International Society on Thrombosis and Haemostasis bleeding scale. Major bleeding was defined as fatal bleeding, symptomatic bleeding in a critical area or organ, bleeding causing a fall in hemoglobin of ≥ 2 g/dL, or bleeding leading to transfusion of ≥ 2 units of blood. Clinically relevant non-major bleeding was defined as overt bleeding that required medical attention, hospitalization, or a change in antithrombotic therapy, but did not fulfill the criteria for major bleeding.19 Secondary outcomes included all-cause mortality, cardiovascular death, and noncardiovascular death, to evaluate the relationship between bleeding events and subsequent mortality. Cardiovascular death was defined as death due to myocardial infarction, HF, fatal arrhythmia, stroke, or other cardiovascular causes. Noncardiovascular death was defined as death from any cause not classified as cardiovascular.20 All outcomes were ascertained through review of medical records at each participating site rather than through administrative claims data or diagnosis code–based algorithms. Statistical methods Continuous variables were expressed as medians with interquartile ranges (IQRs), and categorical variables as counts and percentages. When appropriate, variables were log-transformed before analysis. Baseline characteristics were compared using one-way analysis of variance or the Kruskal–Wallis test for continuous variables, and the χ2 test for categorical variables. Cumulative incidence curves for bleeding events from the time of discharge (time 0) were constructed using Fine–Gray competing risk regression analysis, with all-cause mortality unrelated to bleeding treated as a competing risk. Based on previous studies, factors known to influence bleeding risk, including hemoglobin level, estimated glomerular filtration rate, age, hypertension, antiplatelet or anticoagulant therapy, use of nonsteroidal anti-inflammatory drug (topical or oral), and history of bleeding at discharge,21 were included as covariates in multivariable Fine–Gray models. For antiplatelet or anticoagulation therapy as an adjustment variable, we considered the number of drugs used by each patient. Multiple imputation by chained equations, as proposed by van Buuren et al., was performed to account for missing data in the dataset.22 Fifty imputed datasets without missing data were created using the mice package in R Studio (R Foundation for Statistical Computing, Vienna, Austria). In our Fine–Gray analyses, the results were calculated by integrating the 50 datasets using the Rubin rule. Furthermore, to evaluate the association between bleeding events and subsequent mortality, the onset of bleeding events was treated as a time-dependent variable. Cumulative incidence curves for all-cause mortality were generated using the time of the bleeding event as the time origin. To evaluate whether bleeding events were associated with subsequent mortality risk, a multivariable Cox proportional hazards model, including bleeding events as a time-dependent covariate, was adjusted for the Meta-Analysis Global Group in Chronic Heart Failure (MAGGIC) risk score and log-transformed BNP levels. The MAGGIC risk score is a validated predictor of mortality in patients with HF.23,24 Adding BNP to the MAGGIC risk score has been shown to improve prognostic accuracy in Japanese patients.24 Statistical significance was defined as a 2-tailed P < 0.05. All statistical analyses were conducted using R Studio (version 4.2.1). Results In total, 4016 patients were enrolled in the JROADHF NEXT study. Of these, 3029 (75.4 %) were prescribed antiplatelet or anticoagulant therapy at discharge. Baseline characteristics of patients with and without antiplatelet and/or anticoagulant therapy are provided in Supplemental Table S1. After excluding 89 patients with missing CFS data, 2940 patients (mean age: 75 ± 12 years; 64.0% male) were included in the analysis cohort. Among them, 821 (27.9%) received antiplatelet therapy alone, 1614 (54.9%) received anticoagulants alone, and 505 (17.2%) received both therapies. Among those on anticoagulants, 1497 (70.6%) were treated with direct oral anticoagulants, and 622 (29.4%) received warfarin. Supplemental Figure S1 displays the distribution of CFS scores; the median score was 4 (IQR: 3-5). Based on CFS scores, patients were stratified into 3 groups— CFS score 1-3 (n = 1033), CFS score 4-5 (n = 1364), and CFS score 6-9 (n = 543). Baseline characteristics across the CFS groups are summarized in Table 1.Table 1Patient baseline clinical characteristicsVariablesCFS 1–3CFS 4–5CFS 6–9Pn = 1033n = 1364n = 543Age, Y69 ± 1376 ± 1182 ± 9< 0.001Male sex787 (76.2)832 (61.0)262 (48.3)< 0.001BMI, kg/m222.6 ± 3.921.7 ± 3.920.6 ± 4.0< 0.001NYHA class III/IV54 (5.2)169 (12.4)124 (22.8)< 0.001LVEF, %42 ± 1545 ± 1647 ± 16< 0.001History of heart failure< 0.001De novo592 (57.3)581 (42.6)196 (36.1)< 1.5 Y129 (12.5)203 (14.9)81 (14.9)≥ 1.5 Y312 (30.2)580 (42.5)266 (49.0)Comorbidities%Atrial fibrillation513 (49.7)735 (53.9)308 (56.7)0.018Ischemic heart disease388 (37.6)569 (41.7)221 (40.7)0.114COPD63 (6.1)114 (8.4)40 (7.4)0.111Diabetes366 (35.4)547 (40.1)210 (38.7)0.064Hypertension713 (69.0)962 (70.5)378 (69.6)0.723Laboratory data at discharge%Hemoglobin, g/dL13.0 ± 2.412.0 ± 2.211.0 ± 1.9< 0.001Albumin, g/dL3.7 ± 0.53.5 ± 0.53.3 ± 0.5< 0.001eGFR, mL/min per 1.73 m248 [35–62]42 [30–54]37 [24–49]< 0.001BNP, pg/mL219 [111–409]277 [135–521]336 [172–608]< 0.001Serum sodium, mmol/L139 ± 3138 ± 4138 ± 40.001Serum potassium, mmol/L4.4 ± 0.54.3 ± 0.54.3 ± 0.60.010HbA1c, %6.4 ± 1.16.3 ± 1.06.3 ± 0.90.039Prescription at discharge%ACE-I/ARB780 (75.5)968 (71.0)329 (60.6)< 0.001ARNI30 (2.9)39 (2.9)12 (2.2)0.884Beta-blocker887 (85.9)1113 (81.6)390 (71.8)< 0.001MRA671 (65.0)835 (61.2)252 (46.4)< 0.001Loop diuretics848 (82.1)1164 (85.3)452 (83.2)0.094Warfarin196 (19.0)296 (21.7)130 (23.9)0.057DOAC552 (53.4)691 (50.7)254 (46.8)0.041Antiplatelet443 (42.9)641 (47.0)242 (44.6)0.130NSAIDs (topical or oral)25 (2.4)39 (2.9)28 (5.2)< 0.001Values are presented as the mean ± standard deviation, median (interquartile range), or n (%), unless otherwise indicated.ACE-I, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; ARNI, angiotensin receptor-neprilysin inhibitor; BMI, body mass index; BNP, brain natriuretic peptide; CFS, clinical frailty score; COPD, chronic obstructive pulmonary disease; DOAC, direct oral anticoagulant; eGFR, estimated glomerular filtration rate; HbA1c, hemoglobin A1c; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; NSAIDs, nonsteroidal anti-inflammatory drug; NYHA, New York Heart Association. Table 1Patient baseline clinical characteristicsVariablesCFS 1–3CFS 4–5CFS 6–9Pn = 1033n = 1364n = 543Age, Y69 ± 1376 ± 1182 ± 9< 0.001Male sex787 (76.2)832 (61.0)262 (48.3)< 0.001BMI, kg/m222.6 ± 3.921.7 ± 3.920.6 ± 4.0< 0.001NYHA class III/IV54 (5.2)169 (12.4)124 (22.8)< 0.001LVEF, %42 ± 1545 ± 1647 ± 16< 0.001History of heart failure< 0.001De novo592 (57.3)581 (42.6)196 (36.1)< 1.5 Y129 (12.5)203 (14.9)81 (14.9)≥ 1.5 Y312 (30.2)580 (42.5)266 (49.0)Comorbidities%Atrial fibrillation513 (49.7)735 (53.9)308 (56.7)0.018Ischemic heart disease388 (37.6)569 (41.7)221 (40.7)0.114COPD63 (6.1)114 (8.4)40 (7.4)0.111Diabetes366 (35.4)547 (40.1)210 (38.7)0.064Hypertension713 (69.0)962 (70.5)378 (69.6)0.723Laboratory data at discharge%Hemoglobin, g/dL13.0 ± 2.412.0 ± 2.211.0 ± 1.9< 0.001Albumin, g/dL3.7 ± 0.53.5 ± 0.53.3 ± 0.5< 0.001eGFR, mL/min per 1.73 m248 [35–62]42 [30–54]37 [24–49]< 0.001BNP, pg/mL219 [111–409]277 [135–521]336 [172–608]< 0.001Serum sodium, mmol/L139 ± 3138 ± 4138 ± 40.001Serum potassium, mmol/L4.4 ± 0.54.3 ± 0.54.3 ± 0.60.010HbA1c, %6.4 ± 1.16.3 ± 1.06.3 ± 0.90.039Prescription at discharge%ACE-I/ARB780 (75.5)968 (71.0)329 (60.6)< 0.001ARNI30 (2.9)39 (2.9)12 (2.2)0.884Beta-blocker887 (85.9)1113 (81.6)390 (71.8)< 0.001MRA671 (65.0)835 (61.2)252 (46.4)< 0.001Loop diuretics848 (82.1)1164 (85.3)452 (83.2)0.094Warfarin196 (19.0)296 (21.7)130 (23.9)0.057DOAC552 (53.4)691 (50.7)254 (46.8)0.041Antiplatelet443 (42.9)641 (47.0)242 (44.6)0.130NSAIDs (topical or oral)25 (2.4)39 (2.9)28 (5.2)< 0.001Values are presented as the mean ± standard deviation, median (interquartile range), or n (%), unless otherwise indicated.ACE-I, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; ARNI, angiotensin receptor-neprilysin inhibitor; BMI, body mass index; BNP, brain natriuretic peptide; CFS, clinical frailty score; COPD, chronic obstructive pulmonary disease; DOAC, direct oral anticoagulant; eGFR, estimated glomerular filtration rate; HbA1c, hemoglobin A1c; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; NSAIDs, nonsteroidal anti-inflammatory drug; NYHA, New York Heart Association. Patient baseline clinical characteristics Values are presented as the mean ± standard deviation, median (interquartile range), or n (%), unless otherwise indicated. ACE-I, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; ARNI, angiotensin receptor-neprilysin inhibitor; BMI, body mass index; BNP, brain natriuretic peptide; CFS, clinical frailty score; COPD, chronic obstructive pulmonary disease; DOAC, direct oral anticoagulant; eGFR, estimated glomerular filtration rate; HbA1c, hemoglobin A1c; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; NSAIDs, nonsteroidal anti-inflammatory drug; NYHA, New York Heart Association. Patients with higher CFS scores were generally older, had a lower body mass index (BMI), had a higher left ventricular ejection fraction (LVEF), and had a higher number of previous hospitalizations for HF. Regarding medication use, patients in higher CFS score groups were less likely to receive direct oral anticoagulants; however, no significant differences were observed in the use of warfarin or antiplatelet agents. The median follow-up duration was 727 days (IQR: 582-791), with a 2-year follow-up completion rate of 71.1%. During follow-up, 149 bleeding events were recorded—86 major bleeding events and 63 clinically relevant nonmajor bleeding events, according to the International Society on Thrombosis and Haemostasis definition.19 Figure 1 illustrates the cumulative incidence of bleeding events, accounting for all-cause death unrelated to bleeding as a competing risk. Overall, the cumulative incidence was 2.6% at 1 year and 4.5% at 2 years. The 2-year bleeding event incidence increased with CFS severity—at 1 year and 2 years, respectively: 1.7% and 2.8% in the CFS score 1-3 group; 2.9% and 5.7% in the CFS score 4-5 group; and 3.8% and 7.1% in the CFS score 6-9 group (Fine–Gray model, P < 0.001; Fig. 2).Figure 1Cumulative incidence of bleeding events in patients with heart failure. Cumulative incidence curve for bleeding events, considering all-cause mortality as a competing risk factor, in patients with heart failure taking oral antiplatelet or anticoagulant drugs. The cumulative incidence of bleeding was 2.6% at 1 year and 4.5% at 2 years.Figure 2Cumulative bleeding incidence by Clinical Frailty Scale (CFS) score groups. Patients in the CFS 4-5 and CFS 6-9 score groups exhibited significantly higher bleeding rates than those in the CFS 1-3 score group, as shown using Fine–Gray competing risk analysis (P < 0.001). Cumulative incidence of bleeding events in patients with heart failure. Cumulative incidence curve for bleeding events, considering all-cause mortality as a competing risk factor, in patients with heart failure taking oral antiplatelet or anticoagulant drugs. The cumulative incidence of bleeding was 2.6% at 1 year and 4.5% at 2 years. Cumulative bleeding incidence by Clinical Frailty Scale (CFS) score groups. Patients in the CFS 4-5 and CFS 6-9 score groups exhibited significantly higher bleeding rates than those in the CFS 1-3 score group, as shown using Fine–Gray competing risk analysis (P < 0.001). In a subgroup analysis stratified by LVEF, the stepwise increase in bleeding events according to CFS score category remained significant in patients with LVEF ≥ 50% (P < 0.001). In patients with LVEF < 50%, the cumulative incidence curves showed a similar directional trend, although the association did not reach statistical significance (P = 0.588). However, the interaction between CFS score and LVEF was not statistically significant (P for interaction = 0.120; Supplemental Fig. S2). To examine whether the association between CFS score category and bleeding events differed according to antithrombotic strategy, we further generated cumulative incidence curves separately among patients receiving antiplatelet therapy alone, anticoagulant therapy alone, and combination antithrombotic therapy. In the group receiving anticoagulant therapy alone, higher CFS score categories were significantly associated with a higher cumulative incidence of bleeding events (P = 0.011). In the antiplatelet therapy alone and combination antithrombotic therapy groups, the curves showed a generally similar directional trend, although the differences did not reach statistical significance (P = 0.240 and P = 0.564, respectively; Supplemental Fig. S3). The same analysis also was performed in patients not receiving antiplatelet or anticoagulant therapy (n = 987); however, no statistically significant association was found between CFS score and bleeding events (Supplemental Fig. S4). Table 2 presents the univariate and multivariable competing risk analyses. In the multivariable model, both the CFS score 4-5 group (hazard ratio [HR], 1.53; 95% confidence interval [CI], 1.00-2.34) and the CFS score 6-9 group (HR, 1.75; 95% CI, 1.04-2.93) were associated with an increased risk of bleeding after adjustment for established risk factors. In addition, when CFS score was included as a continuous variable in a multivariable competing risk model, the association remained significant (HR per 1-point CFS score increase, 1.19; 95% CI, 1.06-1.33; P = 0.003).Table 2Analysis of bleeding risk using Fine–Gray competing risk regression analysisCategorical valueUnadjusted modelAdjusted model∗HR95% CIPHR95% CIPCFS score 1–31.00(Reference)1.00(Reference)CFS score 4–51.701.13–2.560.0091.531.00–2.340.044CFS score 6–92.151.34–3.440.0011.751.04–2.930.030Continuous valueCFS1.241.12–1.37< 0.0011.191.06–1.330.003CI, confidence interval; CFS, Clinical Frailty Scale; HR, hazard ratio.∗Adjusted for hemoglobin level, estimated glomerular filtration rate, age, hypertension, antiplatelet and/or anticoagulation therapy, history of bleeding at discharge, and use of nonsteroidal anti-inflammatory drug (topical or oral). Table 2Analysis of bleeding risk using Fine–Gray competing risk regression analysisCategorical valueUnadjusted modelAdjusted model∗HR95% CIPHR95% CIPCFS score 1–31.00(Reference)1.00(Reference)CFS score 4–51.701.13–2.560.0091.531.00–2.340.044CFS score 6–92.151.34–3.440.0011.751.04–2.930.030Continuous valueCFS1.241.12–1.37< 0.0011.191.06–1.330.003CI, confidence interval; CFS, Clinical Frailty Scale; HR, hazard ratio.∗Adjusted for hemoglobin level, estimated glomerular filtration rate, age, hypertension, antiplatelet and/or anticoagulation therapy, history of bleeding at discharge, and use of nonsteroidal anti-inflammatory drug (topical or oral). Analysis of bleeding risk using Fine–Gray competing risk regression analysis CI, confidence interval; CFS, Clinical Frailty Scale; HR, hazard ratio. Adjusted for hemoglobin level, estimated glomerular filtration rate, age, hypertension, antiplatelet and/or anticoagulation therapy, history of bleeding at discharge, and use of nonsteroidal anti-inflammatory drug (topical or oral). Also, we evaluated the association between bleeding events and subsequent outcomes. Cumulative incidence curves for all-cause mortality revealed that patients who experienced a bleeding event, treated as a time-dependent variable, had significantly higher mortality rates following the bleeding event than those without bleeding (Fig. 3). Cox proportional hazards analysis showed that bleeding events remained associated with all-cause mortality, even after adjusting for the MAGGIC risk score and log-transformed BNP levels (unadjusted model: HR, 2.79; 95% CI, 1.99-3.90; P < 0.001; adjusted model: HR, 2.80; 95% CI, 1.88-4.17; P < 0.001) (Table 3). In an additional sensitivity analysis that included the CFS score in the model, bleeding events remained significantly associated with all-cause mortality (HR, 2.44, 95% CI, 1.64-3.64, P < 0.001). Bleeding events also were associated with both cardiovascular and noncardiovascular mortality after adjustment. For cardiovascular death, the unadjusted model yielded an HR of 2.71 (95% CI, 1.63-4.52; P < 0.001); the adjusted model yielded an HR of 2.43 (95% CI, 1.31-4.52; P = 0.005). For noncardiovascular death, the unadjusted HR was 2.85 (95% CI, 1.82–4.45; P < 0.001), and the adjusted HR was 3.15 (95% CI, 1.87-5.30; P < 0.001).Figure 3Cumulative incidence curves for all-cause mortality rates. Cumulative incidence curves demonstrated that patients who experienced a bleeding event, treated as a time-dependent variable, had significantly higher all-cause mortality rates than those without a bleeding event.Table 3Cox proportional hazards analysis for the prediction of all-cause mortalityVariableUnadjusted modelAdjusted model∗HR95% CIPHR95% CIPBleeding2.791.99–3.90< 0.0012.801.88–4.17< 0.001CI, confidence interval; HR, hazard ratio.∗Adjusted for the Meta-analysis Global Group in Chronic Heart Failure risk score and log-transformed brain natriuretic peptide level. Cumulative incidence curves for all-cause mortality rates. Cumulative incidence curves demonstrated that patients who experienced a bleeding event, treated as a time-dependent variable, had significantly higher all-cause mortality rates than those without a bleeding event. Table 3Cox proportional hazards analysis for the prediction of all-cause mortalityVariableUnadjusted modelAdjusted model∗HR95% CIPHR95% CIPBleeding2.791.99–3.90< 0.0012.801.88–4.17< 0.001CI, confidence interval; HR, hazard ratio.∗Adjusted for the Meta-analysis Global Group in Chronic Heart Failure risk score and log-transformed brain natriuretic peptide level. Cox proportional hazards analysis for the prediction of all-cause mortality CI, confidence interval; HR, hazard ratio. Adjusted for the Meta-analysis Global Group in Chronic Heart Failure risk score and log-transformed brain natriuretic peptide level. To further validate these findings, a sensitivity analysis was conducted, excluding patients with cancer, as malignancy could confound the association between bleeding events and clinical outcomes. At baseline, 251 patients (8.3%) had a history of cancer, and during the 2-year follow-up period, 153 additional patients (6.5%) were diagnosed with cancer. Cox proportional hazards analysis was repeated in this cancer-free cohort, yielding similar results (unadjusted model: HR, 2.80; 95% CI, 1.61-4.84; P < 0.001; adjusted model: HR, 2.82; 95% CI, 1.46-5.44; P = 0.002). These findings reinforce the robustness of the association between bleeding events and mortality, irrespective of cancer status. Discussion In this study of patients hospitalized for HF and receiving antiplatelet or anticoagulant therapy, we investigated the association between frailty and bleeding risk, as well as the association between bleeding events and subsequent mortality. Our findings revealed the following: (i) 75.4% of patients with HF were prescribed antiplatelet or anticoagulant medication at discharge; (ii) frailty severity was associated with bleeding risk in a stepwise manner; and (iii) the onset of bleeding events was associated with increased subsequent mortality in patients with HF (Central Illustration).Central IllustrationFrailty, bleeding events, and subsequent mortality in patients with heart failure receiving antithrombotic therapy. Among patients hospitalized for heart failure receiving antiplatelet or anticoagulant therapy at discharge, higher Clinical Frailty Scale score categories were associated with a stepwise increase in bleeding events. Bleeding events were subsequently associated with a higher risk of all-cause mortality. These findings suggest that frailty assessment may help identify patients who require closer follow-up and careful management of modifiable bleeding risk factors during antithrombotic therapy. AHF, acute heart failure; JROADHF-NEXT, Japanese Registry of Acute Decompensated Heart Failure–Next. Frailty, bleeding events, and subsequent mortality in patients with heart failure receiving antithrombotic therapy. Among patients hospitalized for heart failure receiving antiplatelet or anticoagulant therapy at discharge, higher Clinical Frailty Scale score categories were associated with a stepwise increase in bleeding events. Bleeding events were subsequently associated with a higher risk of all-cause mortality. These findings suggest that frailty assessment may help identify patients who require closer follow-up and careful management of modifiable bleeding risk factors during antithrombotic therapy. AHF, acute heart failure; JROADHF-NEXT, Japanese Registry of Acute Decompensated Heart Failure–Next. Frailty and bleeding events Notably, no prior study has directly evaluated the association between frailty and bleeding risk in patients with HF; however, several studies have reported findings consistent with ours. Chung et al. evaluated 72 patients with HF who required ventricular assist device implantation and found that low handgrip strength was associated with post-procedural bleeding events.25 A post-hoc analysis of the Warfarin versus Aspirin in Reduced Cardiac Ejection Fraction (WARCEF) trial reported that 97 of 2305 patients (4.2%) with HF and reduced ejection fraction experienced major bleeding events over a mean follow-up period of 3.5 years; older age and lower BMI were associated with significant bleeding risk.26 These findings support our observation of an association between frailty and bleeding risk in patients with HF. Moreover, the association between frailty and bleeding risk in patients receiving antiplatelet or anticoagulant therapy has been well documented in non-HF populations. For instance, Dodson et al. demonstrated that, among 129,330 patients aged ≥ 65 years with acute myocardial infarction who underwent cardiac catheterization, the incidence of bleeding increased significantly with frailty severity, as defined by impairments in walking, cognition, and activities of daily living.27 In addition, the association between frailty and bleeding risk has been documented in patients with atrial fibrillation.28,29 Therefore, the findings of our study suggest that frailty severity is associated with an increased risk of bleeding in patients with HF receiving antiplatelet or anticoagulant therapy, which is consistent with previous research. Our study was not designed to elucidate the mechanisms underlying this association; however, frailty could increase the risk of bleeding through multiple mechanisms, including systemic inflammation, changes in haemostasis (such as alterations in coagulation factors and platelet reactivity), variable pharmacokinetics, inappropriate medication dosing due to low muscle mass, and increased vascular fragility.30, 31, 32 This finding suggests that frailty alone may not inherently elevate bleeding risk, but rather that the risk is synergistically elevated when frail patients with HF are treated with antiplatelets or anticoagulants. Therefore, further research is warranted to investigate these underlying mechanisms. Association between bleeding events and subsequent mortality Previous studies involving patients with conditions other than HF have evaluated the prognostic impact of bleeding events. For instance, a post hoc analysis of the multinational Frailty Aortic Valve Replacement cohort study, which prospectively enrolled 1195 patients aged ≥ 70 years who underwent transcatheter aortic valve replacement or surgical aortic valve replacement, found that a significant proportion received antiplatelet or anticoagulant therapy as part of perioperative management.33 In this cohort, frailty assessed using the Essential Frailty Toolset was an independent predictor of major bleeding, and the incidence of major bleeding was significantly associated with a threefold increase in 1-year mortality following transcatheter or surgical aortic valve replacement.33 Eikelboom et al. examined the relationship between bleeding events and mortality in 34,146 patients with acute coronary syndrome. They reported that major bleeding was associated with a 5fold higher incidence of death within the first 30 days in these patients than in those without bleeding (12.8% vs 2.5%; P < 0.0001).34 In contrast, few studies have specifically investigated the association between bleeding events and all-cause mortality in patients with HF. A previous study included 7160 patients undergoing percutaneous coronary intervention and assessed the relationship between bleeding events within 30 days of this procedure and all-cause mortality over 3 years in patients with and without HF.35 However, only 384 patients with HF were included, with just 14 experiencing bleeding events. This limited sample size rendered the study underpowered to draw reliable conclusions, and the study did not consider bleeding events as a time-dependent variable. In contrast, with a larger sample size, our study accounted for bleeding events as a time-dependent variable and showed an association between bleeding events and subsequent mortality in patients with HF. Clinical significance Using a sufficiently large patient cohort with prospectively assessed frailty status and bleeding events, this study showed that frailty was associated with bleeding events after adjustment for established clinical predictors in hospitalized patients with HF receiving antithrombotic therapy for conditions such as atrial fibrillation or coronary artery disease. Our findings further suggest that bleeding events are associated with subsequent all-cause mortality; therefore, frailty assessment may be useful not to justify withdrawal of indicated antithrombotic therapy, but to identify patients who may benefit from closer follow-up and more aggressive management of modifiable bleeding risk factors. This approach is particularly crucial given that a considerable proportion of patients with HF are on such medications, and that patients with HF and frailty are more likely to experience noncardiovascular rather than cardiovascular death,36 which may not be mitigated by the increase in HF therapy. However, this hypothesis warrants further evaluation in future clinical trials. Limitations This study has some limitations. First, it was a post hoc analysis of a prospective study, with inherent limitations. Second, frailty was assessed only once, at baseline before discharge, and changes in frailty status during the follow-up period were not captured. Therefore, we could not evaluate whether temporal changes in frailty influenced bleeding risk or subsequent outcomes. In addition, frailty was assessed only by the CFS. Although we have previously reported that the CFS score correlates well with objective physical and cognitive function measures,18 the CFS still relies partly on clinical judgement and is therefore inherently subjective. In addition, we did not assess frailty using other established models, such as the Fried phenotype or a frailty index. Third, although we assessed whether patients were receiving antiplatelet, anticoagulant, or both therapies, we did not consider the dosages of these medications. The dosages used were standard dosages covered by medical insurance in Japan; however, the potential impact of varying drug dosages cannot be completely excluded. In addition, we did not account for time in therapeutic range among those on warfarin. Fourth, the study focused solely on all-cause mortality and did not account for other clinically important outcomes, such as all-cause or bleeding-related hospitalization. Because hospitalizations other than HF-related admissions were not systematically collected in this registry, we were unable to evaluate these outcomes. Future studies should examine hospitalization outcomes, including all-cause hospitalization and bleeding-related readmission, in addition to mortality. Fifth, particularly in patients with severe frailty, goals of care may not necessarily prioritize survival and may instead place greater emphasis on symptom relief, quality of life, or avoidance of hospitalization. Because these data were not systematically collected in this registry, we were unable to account for their potential influence on the observed association between bleeding events and mortality. Sixth, detailed information on changes in antithrombotic therapy after bleeding events, such as switching agents, dose reduction, or deprescribing, was not systematically collected. Therefore, we could not assess how post-bleeding prescribing patterns may have influenced subsequent bleeding risk or mortality. Finally, the generalizability of our findings to non-Japanese populations, particularly North American and European populations, should be interpreted with caution. Japanese patients generally have lower body mass index and different body composition compared with Western populations, and the present cohort also showed relatively low BMI values across CFS score categories. Differences in body size and body composition may influence bleeding susceptibility, comorbidity profiles, and overall risk stratification. In addition, antithrombotic prescribing patterns and management strategies differ across regions. In the present study, nearly 30% of patients receiving anticoagulant therapy were treated with warfarin, whereas direct oral anticoagulants predominate in many contemporary North American cohorts. Additionally, antithrombotic management in Japan differs from Western practice in several respects, including lower international normalized ratio (INR) targets for older patients receiving warfarin and lower approved doses of rivaroxaban.37,38 These differences may reflect distinct bleeding and thrombotic risk profiles in East Asian populations and may influence both the absolute bleeding event rates and the observed association between frailty and bleeding events. Therefore, further studies in other racial, ethnic, and regional populations are needed to confirm the external validity of our findings. Conclusions Frailty is a significant risk factor for bleeding events in patients with HF receiving oral antiplatelet or anticoagulant therapy. In addition, the occurrence of bleeding events is strongly associated with an increased risk of subsequent mortality. Clinically, these findings suggest that greater attention to frailty may be warranted when patients with HF are treated with antithrombotic therapy, and that careful follow-up for bleeding may be particularly important in frail individuals. Future studies should examine whether frailty-informed management and monitoring approaches can reduce bleeding events and improve outcomes. Ethics Statement The study protocol complied with the principles of the Declaration of Helsinki and was approved by the Kyushu University Ethical Committee (No. 2019-283). Patient Consent Written informed consent for participation and publication was obtained from all participants (or their legally authorized representatives) in accordance with the Declaration of Helsinki. Funding Sources This work was supported by a Japan Agency for Medical Research and Development grant (19ek0210080h0003, to H.T.) and Health Sciences Research Grants from the Japanese Ministry of Health, Labour and Welfare (Comprehensive Research on Cardiovascular Diseases; 20FC1051, to H.T.). The funders had no role in the design or conduct of the study or in the collection, management, analysis or interpretation of the data, or the preparation, review, or submission of the manuscript for publication. Disclosures N.E. reports being supported by an American Heart Association postdoctoral fellowship. T.T. reports being affiliated with the collaborative research division of Mochida Pharmaceuticals (FY2024; April 1, 2023 – March 31, 2024) and being supported by the Cardiovascular Innovative Conference grant and the Uehara Memorial Foundation Fellowship. M.I. reports receiving research funding from MEDINET; and being affiliated with the collaborative research division of MEDINET (FY2023; April 1, 2022 – March 31, 2023). S.M. reports receiving grants from Astellas pharma and Tsumura Corporation, outside the submitted work. Y.M. reports receiving honoraria from Otsuka Pharmaceutical, EN Otsuka Pharmaceutical, Novartis Pharma K.K., Bayer, Pfizer Japan, Nippon Boehringer Ingelheim, and AstraZeneca; and receiving a collaborative research grant from Pfizer Japan, Otsuka Pharmaceutical, EN Otsuka Pharmaceutical, Roch Diagnostics K.K., and Nippon Boehringer Ingelheim. T.K. reports receiving honoraria from AstraZeneca, Alnylam Japan, Bayer, Ono Pharmaceutical, Otsuka Pharmaceutical, EN Otsuka Pharmaceutical, Novartis Pharma K.K., Nippon Boehringer Ingelheim, and Pfizer Japan. T.T. reports being the founder and a shareholder of Remohab. K.K. reports receiving honoraria from AstraZeneca K.K., Ono Pharmaceutical, Nippon Boehringer Ingelheim, Bayer Yakuhin, Otsuka Pharmaceutical, and Novartis Pharmaceuticals. T.O. reports being affiliated with a department that is sponsored by Medtronic Japan; receiving honoraria from AstraZeneca, Pfizer, Boehringer Ingelheim, Novartis, Otsuka, and Ono Yakuhin; and receiving research grants from Pfizer, Alnylam, Alexion, and Ono Yakuhin. H.T. reports receiving consultancy fees from Ono Pharmaceutical, Bayer Yakuhin, Nippon Boehringer Ingelheim, and Novo Nordisk Pharma; remuneration from AstraZeneca KK, Ono Pharmaceutical, Otsuka Pharmaceutical, Daiichi Sankyo, Tsumura & Co, Terumo, Nippon Boehringer Ingelheim, Novartis Pharma K.K., Novo Nordisk Pharma, Bayer Yakuhin, Pfizer Japan, and Viatris; and receiving manuscript fees from IGAKU-SHOIN, and MEDICAL VIEW.