Eur J Intern Med. 2026 May 8:106860. doi: 10.1016/j.ejim.2026.106860. Online ahead of print.
ABSTRACT
BACKGROUND: Real-world uptake of guideline-directed medical therapy (GDMT) at hospital discharge and clinical predictors of complete decongestion in acute heart failure (AHF) populations remain insufficiently described.
METHODS: The BRING-UP 3 HF study is an observational, prospective, nationwide investigation involving 179 Italian cardiology sites. This report summarizes baseline data from the first enrollment phase hospitalized cohort and assesses the predictors of decongestion via a machine learning model.
RESULTS: Among 1373 patients (mean age 71 years; 30% females; 43% de-novo HF), HF with reduced ejection fraction (HFrEF) predominated (70%). Hypertension, atrial fibrillation, diabetes mellitus, and chronic kidney disease were reported in 75%, 43%, 35%, and 33% of patients, respectively. In HFrEF, discharge prescriptions rose markedly with respect to admission, with 57% of patients receiving all four pillars of GDMT. Successful decongestion was achieved in 469/681 evaluable patients (69%). A random-forest model identified higher estimated glomerular filtration rate, younger age, lower urea/creatinine ratio, lower C-reactive protein, and smaller left-atrial volumes as the strongest predictors of a successful decongestion, with good discrimination (AUC 0.80).
CONCLUSIONS: Contemporary Italian cardiology practice shows high adherence to discharge GDMT across the spectrum of EF in AHF. Nevertheless, nearly one-third of patients leaves the hospital with residual congestion. The identified machine learning model predictors may provide an objective framework for risk stratification. These variables may help clinicians identify a high-risk patient profile that requires intensified in-hospital decongestive strategies and more aggressive post-discharge transitional care to reduce the risk of early rehospitalization.
CLINICALTRIAL: GOV: NCT06279988.
PMID:42218046 | DOI:10.1016/j.ejim.2026.106860