Early prognostic indicators in acute heart failure
K Ntalekou, N Katsillis, G P Barakos, V Gardikioti, A E Kalesi, P Skantzikas, T Pallas, S Kapsokolis, A Archontikis, E Bousoula, N Papakonstantinou, N Kasinos, A Theodosis Georgilas, N PatsourakosAbstract
Background/Introduction
Acute heart failure (AHF) is associated with high mortality. Clinical presentation, underlying etiology, and comorbidities play a pivotal role in patient outcome (1,2). Early identification of high-risk patients is essential; however, no validated risk stratification model has yet been established.
Purpose
This study aimed to collect epidemiological data and identify predictors of in-hospital mortality in patients with AHF to develop a risk stratification model.
Methods
This retrospective observational study included patients with acute heart failure (AHF) hospitalized between January and June 2025 at our Cardiology Department. Data was obtained from the hospital database, and the primary endpoint was in-hospital mortality. To identify statistically significant predictive parameters, univariate logistic regression was performed. Principal Component Analysis (PCA) and Factor Analysis of Mixed Data (FAMD) were used as an exploratory tool to assess model development. Variables with p-value < 0.05 or established clinical relevance were entered into a multivariable logistic regression model using Firth’s penalized method to identify independent prognostic factors. The R programming language version 4.5.2 was used.
Results
A total of 119 patients admitted for acute heart failure were included in our study (62% male, median age 75 years, interquartile range [IQR] 68–84 years), and 6 in-hospital deaths were recorded (5%). Descriptive statistics are presented in Table 1. Regarding FAMD, the first component (8.9% variance) highlighted coronary artery disease and ischemia as key discriminating variables, whereas in the second component, variability was driven by aminotransferase levels (ALT/AST) and diastolic blood pressure (DBP) (Figure 1). From PCA, principal component 1 (PC1) was characterized by elevated AST/ALT and B-Natriuretic Peptide (BNP) levels, with negative associations with estimated Glomerular Filtration Rate (eGFR), systolic blood pressure (SBP), and DBP- accounting for 14.2 % of variance. From PC2, age and echocardiographic parameters were the main contributors to variability- 9.7% (Figure 2). Univariate logistic regression identified SBP, DBP, oxygen saturation (SpO₂), cardiogenic shock, pericardial effusion, eGFR, ALT, AST, and Ejection Fraction of 41–49% as statistically significant predictors (Table 1). In our Firth’s logistic regression model DBP and SpO2 were independently associated with mortality.
Conclusions
The present analysis provides a preliminary qualitative evaluation of key variables—including vital signs, echocardiographic parameters and biomarkers—that appear to be associated with early mortality in AHF. Although the limited sample size precludes the development of a robust regression-based predictive model, our study aims to support ongoing prospective data collection, enabling these observations to be reinforced as the cohort expands.TABLE 1For image description, please refer to the figure legend and surrounding text.picture 2For image description, please refer to the figure legend and surrounding text.