Search

Abstract Aims In recent years, there has been remarkable development in machine learning (ML) models, showing a trend towards high prediction performance. ML models with high prediction performance often become structurally complex and are frequently perceived as black boxes, hindering intuitive interpretation of the prediction results. We aimed to develop ML models with high prediction performance, interpretability, and superior risk stratification to predict in‐hospital mortality and worsening heart failure (WHF) in patients with acute heart failure (AHF). Methods and results Based on the Kyoto Congestive Heart Failure registry, which enrolled 4056 patients with AHF, we developed prediction models for in‐hospital mortality and WHF using information obtained on the first day of admission (demographics, physical examination, blood test results, etc.). After excluding 16 patients who died on the first or second day of admission, the original dataset (n = 4040) was split 4:1 into training (n = 3232) and test datasets (n = 808). Based on the training dataset, we developed three types of prediction models: (i) the classification and regression trees (CART) model; (ii) the random forest (RF) model; and (iii) the extreme gradient boosting (XGBoost) model. The performance of each model was evaluated using the test dataset, based on metrics including sensitivity, specificity, area under the receiver operating characteristic curve (AUC), Brier score, and calibration slope. For the complex structure of the XGBoost model, we performed SHapley Additive exPlanations (SHAP) analysis, classifying patients into interpretable clusters. In the original dataset, the proportion of females was 44.8% (1809/4040), and the average age was 77.9 ± 12.0. The in‐hospital mortality rate was 6.3% (255/4040) and the WHF rate was 22.3% (900/4040) in the total study population. In the in‐hospital mortality prediction, the AUC for the XGBoost model was 0.816 [95% confidence interval (CI): 0.815–0.818], surpassing the AUC values for the CART model (0.683, 95% CI: 0.680–0.685) and the RF model (0.755, 95% CI: 0.753–0.757). Similarly, in the WHF prediction, the AUC for the XGBoost model was 0.766 (95% CI: 0.765–0.768), outperforming the AUC values for the CART model (0.688, 95% CI: 0.686–0.689) and the RF model (0.713, 95% CI: 0.711–0.714). In the XGBoost model, interpretable clusters were formed, and the rates of in‐hospital mortality and WHF were similar among each cluster in both the training and test datasets. Conclusions The XGBoost models with SHAP analysis provide high prediction performance, interpretability, and reproducible risk stratification for in‐hospital mortality and WHF for patients with AHF.

Abstract Aims Several studies demonstrated that tricuspid regurgitation (TR) is associated with poor clinical outcomes. However, data on patients with TR who experienced acute heart failure (AHF) remains scarce. The purpose of this study is to evaluate the association between TR and clinical outcomes in patients admitted with AHF, using a large‐scale Japanese AHF registry. Methods and results The current study population consisted of 3735 hospitalized patients due to AHF in the Kyoto Congestive Heart Failure (KCHF) registry. TR grades were assessed according to the routine clinical practice at each participating centre. We compared the baseline characteristics and outcomes according to the severity of TR. The primary outcome was all‐cause death. The secondary outcome was hospitalization for heart failure (HF). The median age of the entire study population was 80 (interquartile range: 72–86) years. One thousand two hundred five patients (32.3%) had no TR, while mild, moderate, and severe TR was found in 1537 patients (41.2%), 776 patients (20.8%), and 217 patients (5.8%), respectively. Pulmonary hypertension, significant mitral regurgitation, and atrial fibrillation/flutter were strongly associated with the development of moderate/severe of TR, while left ventricular ejection fraction

Abstract Aims Little is known about the characteristics and outcomes of patients who undergo coronary angiography during heart failure (HF) hospitalization, as well as those with coronary stenosis, and those who underwent coronary revascularization. Methods and results We analysed 2163 patients who were hospitalized for HF without acute coronary syndrome or prior HF hospitalization. We compared patient characteristics and 1 year clinical outcomes according to (i) patients with versus without coronary angiography, (ii) patients with versus without coronary stenosis, and (iii) patients with versus without coronary revascularization. The primary outcome measure was the composite of all‐cause death or HF hospitalization. Coronary angiography was performed in 37.0% of patients. In the multivariable logistic regression analysis, factors independently associated with coronary angiography were age

Abstract Aims The association between sex and long‐term outcome in patients hospitalized for acute decompensated heart failure (ADHF) has not been fully studied yet in Japanese population. The aim of this study was to determine differences in baseline characteristics and management of patients with ADHF between women and men and to compare 1‐year outcomes between the sexes in a large‐scale database representing the current real‐world clinical practice in Japan. Methods and results Kyoto Congestive Heart Failure registry is a prospective cohort study enrolling consecutive patients hospitalized for ADHF in Japan among 19 centres. Baseline characteristics, clinical presentation, management, and 1‐year outcomes were compared between men and women. A total of 3728 patients who were alive at discharge constituted the current study population. There were 1671 women (44.8%) and 2057 men. Women were older than men [median (IQR): 83 (76–88) years vs. 77 (68–84) years, P

Abstract Aims The use of tolvaptan is increasing in clinical practice in Japan. However, the characteristics of patients who used tolvaptan and the timing of its use in patients with acute heart failure (AHF) are not fully elucidated. Methods and results Among consecutive 4056 patients in the Kyoto Congestive Heart Failure registry, we analysed 3802 patients after excluding patients on dialysis, prior or unknown tolvaptan use at admission, and unknown timing of tolvaptan use, and we divided them into two groups: tolvaptan use (N = 773) and no tolvaptan use (N = 3029). The prevalence of tolvaptan use varied widely from 48.7% to 0% across the participating centres. Factors independently associated with tolvaptan use were diabetes, poor medical adherence, oedema, pleural effusion, hyponatraemia, estimated glomerular filtration rate

Abstract Aims Little is known about the association between the starting of or dose changes in loop diuretics during acute heart failure (AHF) hospitalization and post‐discharge outcomes. We investigated the clinical impact of starting loop diuretics and changing the loop diuretics dose during hospitalization on post‐discharge outcomes. Methods and results From the Kyoto Congestive Heart Failure registry, 3665 consecutive patients hospitalized for HF and discharged alive were included in this study. We analysed 1906 patients without loop diuretics on admission and were discharged alive and 1759 patients who received loop diuretics on admission and were discharged alive. The primary outcome measure was all‐cause death. Of the 1906 patients without loop diuretics on admission, 1366 (71.7%) patients started loop diuretics during the index AHF hospitalization. Starting loop diuretics was not associated with lower post‐discharge mortality [adjusted hazard ratio (HR) 0.92, 95% confidence interval (CI) 0.68–1.25]. Of the 1759 patients who received loop diuretics on admission, loop diuretic dose was decreased in 23.8%, unchanged in 44.6%, and increased in 31.6% of the patients. Changes in the dose at discharge compared with no change in dose were not associated with lower risk of post‐discharge mortality (decrease relative to no change: adjusted HR 0.98, 95% CI 0.76–1.28; increase relative to no change: adjusted HR 1.00, 95% CI 0.78–1.27). Compared with no loop diuretics at discharge, a loop diuretics dose of ≥80 mg at discharge was associated with higher post‐discharge mortality risk. Conclusions In patients with AHF, we found no association between the starting of loop diuretics and post‐discharge outcomes and between dose changes and post‐discharge outcomes.

Abstract Aims Insomnia is a known risk factor for heart failure (HF) and a predictor of cardiac events in HF patients, but the clinical significance of insomnia in patients with acute HF (AHF) is not adequately evaluated. This study aimed to investigate the association between insomnia and subsequent clinical outcomes in patients with AHF. Methods From the Kyoto Congestive Heart Failure registry, consecutive 3414 patients hospitalized for HF who were discharged alive were divided into the 2 groups at discharge: insomnia group and non‐insomnia group. We compared baseline characteristics and 1 year clinical outcomes according to the presence of insomnia. The primary outcome measure was all‐cause death. Results There were 330 patients (9.7%) and 3084 patients (90.3%) with and without insomnia, respectively. In the multivariable logistic regression analysis, brain‐type natriuretic peptide above median value at discharge (OR = 1.50, 95% CI = 1.08–2.10, P = 0.02) and the presence of oedema at discharge (OR = 4.23, 95% CI = 2.95–6.07, P

Abstract Aims Although the prognostic impact of the high tricuspid regurgitation pressure gradient (TRPG) has been investigated, the association of the decrease in TRPG during follow‐up with clinical outcomes in heart failure (HF) has not been previously studied. The aim of this study was to investigate the association of a decrease in TRPG between hospitalization and 6 month visit with subsequent clinical outcomes in patients with acute decompensated HF (ADHF). Methods and results Among 721 patients with available TRPG data both during hospitalization and a subsequent 6 month visit, the study population was divided into two groups: a decrease in TRPG group (>10 mmHg decrease at 6 month visit) (N = 179) and no decrease in TRPG group (N = 542). The primary outcome measure was a composite of all‐cause death or HF hospitalization. The cumulative 6 month incidence of primary outcome measure was significantly lower in the decrease in TRPG group than in the no decrease in TRPG group (12.2% vs. 18.7%, P = 0.02). After adjusting for confounders, there was a significantly lower risk in decrease in TRPG group than in the no decrease in TRPG group for the measured primary outcome (hazard ratio: 0.56, 95% confidence interval 0.32–0.93, P = 0.02). The lower risk in decrease in TRPG group was not different among the basal TRPG values. Conclusions Heart failure patients with a decrease in TRPG at 6 months after discharge from ADHF hospitalization had lower subsequent risk of all‐cause death and HF hospitalization than those without a decrease in TRPG, regardless of TRPG values.