This is a retrospective observational study and screened 310 patients who were hospitalized due to acute decompensated HF between November 2021 and September 2022. Among them, 116 patients with acute HF who were underwent BIA at the time of admission and at discharge were evaluated. The primary outcome was the composite of in-hospital mortality or rehospitalization for worsening HF within one month of follow-up. The secondary outcomes included length of a hospital stay, dose of diuretics at discharge, and the rate of the guideline-directed medical therapy prescription. HF was diagnosed by cardiologists, internists, or residents, based on HF guidelines.1314 The detailed study flow was illustrated in Fig. 1.

The clinical data of the enrolled patients were retrieved from the electronic medical records. Baseline demographic information, comorbidities, etiology of HF were assessed. The baseline characteristics included sex, age, height, weight, body mass index (BMI), and history of previous HF. We obtained the following clinical information from the medical records: etiology, aggravating factors, comorbidities, vital signs at admission and discharge, discharge medication list, discharge date, readmission date, and follow-up data. Also, we obtained laboratory test results including NT-proBNP or BNP, serum creatinine, estimated glomerular filtration rate (eGFR), hematocrit, echocardiographic examination, and BIA parameters. Chronic kidney disease (CKD) was defined as decreased estimated eGFR to 60 mL/min/1.73 m² or less at least 3 months1516 or who had an already diagnosis of CKD before hospitalization were classified as a CKD patients. We also classified HF with the standard of ejection fraction by HF with preserved ejection fraction (HFpEF) (EF ≥ 50%), HF with mildly reduced ejection fraction (HFmrEF) (EF 41–49%), HF with reduced ejection fraction (HFrEF) (EF ≤ 40%), and the cause of the HF by ischemia, cardiomyopathy, valvular disease, arrhythmia, and the others. Arrhythmia included atrial fibrillation, atrial flutter, and tachycardia-induced arrhythmia. Echocardiographic parameters such as left ventricular ejection fraction (LVEF), left ventricular internal diameter in diastole (LVIDd), left ventricular internal diameter in systole, left atrial volume index, E/e' (The ratio between early mitral inflow velocity and mitral annular early diastolic velocity), and pulmonary artery systolic pressure were considered in this study.

BIA was performed using the portable multi-frequency bio-impedance device (InBody S10; InBody Co. Ltd., Seoul, Korea). The method of BIA with Inbody S10 has been published elsewhere.111718 Briefly, all patients lying posture for 10 minutes before the examination and electrodes were applied to both arm and leg, the foot electrodes should be positioned between patient’s anklebone and heel. Body water composition including ICW, ECW, TBW, and ratio of ECW to TBW (ECW/TBW) were calculated automatically by the BIA device. We converted the ratio of ECW to TBW to percentage to compare quantities effectively. All parameters were measured at the time of admission and discharge. The difference obtained by subtracting the value of the body water compositions (BWCs) measured at the time of hospitalization from the value of body water compositions measured at the time of discharge was defined as a changes in volume status measurements (ΔBWC = BWC _{at predischarge} – BWC _{at admission}). If these differences were high and positive value, it means that the control of volume status in patients with HF is not appropriate.

Categorical variables were presented as absolute counts and percentages, with comparisons by the Chi-squared test, while continuous variables were expressed as mean ± SD or median value (interquartile range [IQR]) with comparisons with the Student’s t-test or Mann-Whitney test. The linear correlation between clinical and laboratory values and BIA parameters was analyzed using Pearson’s bivariate correlation analysis. We determined the cut-off value for the BIA parameters for predicting primary outcome using a receiver operating characteristic (ROC) curve analysis. Furthermore, multiple logistic regression analysis was performed to identify the risk factors associated with primary outcome. The result of primary outcome is reported as the odds ratio (OR) of each variable with a 95% confidence interval (CI). OR adjusted with age, sex, BMI, hypertension, diabetes mellitus (DM), chronic kidney disease, and history of revascularization. Two-tailed p values less than 0.05 were considered to be statistically significant. Statistical analyses were performed using the IBM SPSS Statistics software Version 25.0 (IBM Corp., Armonk, NY, USA) and MedCalc Version 20 were used.

The protocols of this observational study were reviewed and approved by the Institutional Review Board (IRB) of Seoul National University Hospital (IRB No. 2212-023-1383). All the study protocols complied with the ethical guidelines of the 2013 Declaration of Helsinki. The requirement for informed consent from the IRB was waived.

Of the 310 patients screened, 116 patients with acute HF who underwent BIA at the time of admission and at discharge were evaluated. Table 1 shows baseline characteristics. In all subjects, the median age was 77 years [IQR, 67–82 years], the median value of BMI was 23.0 (IQR, 20.6–26.0) kg/m², and 57.8% was the male subjects. Of the patients, 51.7% had hypertension, 49.1% had DM, 18.1% had dyslipidemia, 37.1% had a chronic kidney disease, 42.2% had atrial fibrillation, and 42.2% had a history of previous coronary revascularizations. The mean LVEF was 40.7 ± 14.6%. Regarding HF subtypes based on LVEF, the proportion of HFrEF, HFmrEF, and HFpEF was 55.8%, 11.5%, and 32.7%, respectively. The primary etiology of HF in this study was cardiomyopathy followed by ischemia, arrhythmia, and valvular heart disease. The quantitative values of body water composition (ICW, ECW, and TBW) and segmental edema indices (total, trunk, right leg, and left leg ECW/TBW) were presented in Table 1. Notably, the mean value of ECW/TBW (total, trunk, lower extremities) of patients with HF at the time of admission was more than 40.0% (normal range: 36.0–39.0%). However, the baseline characteristics did not show significant difference between the group with a high and positive ECW to TBW ratio difference (ΔECW/TBW=+0.3% to +5.1%) and the group with low and negative ECW to TBW ratio difference (ΔECW/TBW= −5.3% to +0.2%) except for ECW to TBW ratio and serum sodium concentration.

We evaluated the correlation between BMI, echocardiographic parameters, natriuretic peptides, and BIA parameters (Table 2). Body fluid parameters (ICW, ECW, and TBW) showed a statistically significant positive correlation with BMI. However, the ratio of ECW to TBW, as an edema index showed a negative correlation with BMI (Pearson correlation r = −0.2, P = 0.002). Furthermore, edema index showed a statistically significant positive correlation with BNP level (Pearson correlation r = +0.4, P = 0.010) and NT-proBNP (Pearson correlation r = +0.23, P = 0.020). Regarding echocardiographic parameters, body fluid parameters (ICW, ECW, and TBW) showed statistically significant positive correlation with LVIDd. However, there was no significant correlation between LVEF, E/e’ and any parameters of BIA.

We evaluated the association between the change of BIA parameters during hospitalization and the primary outcome. In analysis of pairwise comparison of volume status-related variables, ROC curves were used to predict primary outcome. The changes of body water composition (ΔICW, ΔECW, and ΔTBW) did not show the significant association with the primary outcome. However, ΔECW/TBW during hospitalization showed a good prognostic accuracy of the primary outcome (Fig. 2 and Supplementary Table 1). The area under the curve (AUC) of ΔECW/TBW was 0.71 (95% CI 0.61 – 0.79, P = 0.006) with sensitivity and specificity of 72.2% and 61.2%. In ROC curve analysis, the predictive accuracy of BNP (measured at the time of admission) was good. The AUC for prediction of the primary outcome (death or rehospitalization) was 0.85 (95% CI, 0.69–0.95; P < 0.001). Further, the areas under the ROC curve of NT-proBNP for prediction of the primary outcome was 0.66 (95% CI, 0.59–0.73; P = 0.040). However, the areas under the ROC curve of E/e’ for prediction of the primary outcome was 0.62 (95% CI, 0.51–0.72; P = 0.129) (Supplementary Table 1).

Univariate and multivariate logistic regression analysis showed that ΔECW/TBW has a significant association with primary outcome (Supplementary Table 2). Parameters considered in univariable analysis were age, gender, diabetic status, BMI, hypertension, prior revascularization, atrial fibrillation, LVEF, chronic kidney disease, Number of GDMT, diuretics prescription, ICW, ECW, ECW/TBW, ΔBody weight, ΔECW/TBW (median), ΔECW, ΔICW, and NT-proBNP. Among those parameters, parameters having a lower P value than 0.05 were considered in multivariate logistic regression analysis. Therefore, chronic kidney disease, diuretics prescription, number of GDMT (guideline-directed medical therapy), NT-proBNP, and ΔECW/TBW were included in multivariate analysis. The result of multivariate analysis about those parameters were as followed; chronic kidney disease (adjusted OR, 4.63; 95% CI, 0.96–22.26; P = 0.056), diuretics prescription (adjusted OR, 0.05; 95% CI, 0.06–0.38; P = 0.004), NT-proBNP (adjusted OR, 1.77; 95% CI, 0.39–8.07; P = 0.463), and ΔECW/TBW (adjusted OR, 12.55; 95% CI, 1.25–97.03; P = 0.018). In multivariate logistic regression analysis, even when NT-proBNP and other covariates were included, the change of the edema index (ΔECW/TBW) appeared as an independent predictor.

Next, the HF patients divided into two groups based on the median value of ΔECW/TBW to evaluate whether the change of the edema index associated with clinical outcomes. Groups with high and positive ECW to TBW ratio difference (Δ ECW/TBW = +0.3% to +5.1%), whose edema indices were increased during hospitalization, showed a significantly higher risk of the primary outcome (23.2% vs. 8.3%; adjusted OR, 4.8; 95% CI, 1.2–19.3; P = 0.029) than those with a low and negative ECW to TBW ratio difference (ΔECW/TBW = −5.3% to +0.2%) (Tables 3 and 4). This result was consistent even when the value of the Δ ECW/TBW was divided by the optimal cut-point value or tertiles (Supplementary Tables 3 and 4). Secondary parameters did not show statistically significant difference between two groups (Table 3).

Subgroup analysis (Fig. 3) was performed to evaluate homogenous results across the subgroup population. Although confidence interval was wide, the odds ratios in the subgroup analysis showed a homogeneity according to age, sex, BMI, diabetic status, LVEF, or diuretic use.