Mortality risk stratification in acute decompensated heart failure: ICARUS score for four-pillar era

Acute decompensated heart failure (ADHF) remains a leading cause of hospitalization and short-term death despite quadruple guideline-directed therapy. Early, pragmatic risk stratification at admission is essential.

To identify independent predictors of in-hospital mortality in ADHF within the four-pillar era and to derive a simple bedside score (ICARUS).

A retrospective cohort study was conducted of hospitalized adults, consecutively selected, with acute decompensated heart failure (de novo or chronic). Pre-specified candidate predictors were categorized with clinically informed cutoff and entered into multivariable logistic regression (backward selection). Model fit and discrimination were assessed using the Hosmer-Lemeshow test and the area under the receiver operating characteristic curve (area under the curve [AUC]), respectively. Coefficients were scaled to create a points-based score.

Among 1588 patients (median age 68 years; 69.3% male; median left ventricular ejection fraction [LVEF] 30%), in-hospital mortality was 4.28% (68/1,588; 95% confidence interval [CI] 3.34-5.39). Independent risk factors were diabetes (odds ratio [OR] 1.81), the New York Heart Association III-IV (OR: 1.97), the Minnesota Living with Heart Failure Questionnaire ≥ 41 (OR: 2.09), atrial fibrillation (OR: 2.29), hyperkalemia K+ > 5.5 mmol/L (OR: 2.63), and systolic blood pressure ≤ 110 mmHg (OR: 3.14). Protective factors were β-blocker use at admission (OR: 0.38), implantable device (OR: 0.22), and in-hospital sodium-glucose cotransporter 2 inhibitor therapy (OR: 0.27). The model showed good model fit (Hosmer-Lemeshow p = 0.997) and strong discriminative ability (AUC 0.81). The ICARUS points system performed similarly (AUC: 0.792; 95% CI 0.738-0.845), with stepwise increases in mortality across strata.

Nine readily obtainable variables accurately stratified in-hospital mortality risk in ADHF. The ICARUS score is a pragmatic bedside tool, integrating clinical status and modern therapy, with a good goodness-of-fit and discrimination.