Derivation and Validation of Clinical Prediction Models for Rapid Risk Stratification for Time-Sensitive Management for Acute Heart Failure

Early and rapid risk stratification of patients with acute heart failure (AHF) is crucial for appropriate patient triage and outcome improvements. We aimed to develop an easy-to-use, in-hospital mortality risk prediction tool based on data collected from AHF patients at their initial presentation. Consecutive patients’ data pertaining to 2006–2017 were extracted from the West Tokyo Heart Failure (WET-HF) and National Cerebral and Cardiovascular Center Acute Decompensated Heart Failure (NaDEF) registries (n = 4351). Risk model development involved stepwise logistic regression analysis and prospective validation using data pertaining to 2014–2015 in the Registry Focused on Very Early Presentation and Treatment in Emergency Department of Acute Heart Failure Syndrome (REALITY-AHF) (n = 1682). The final model included data describing six in-hospital mortality risk predictors, namely, age, systolic blood pressure, blood urea nitrogen, serum sodium, albumin, and natriuretic peptide (SOB-ASAP score), available at the time of initial triage. The model showed excellent discrimination (c-statistic = 0.82) and good agreement between predicted and observed mortality rates. The model enabled the stratification of the mortality rates across sixths (from 14.5% to <1%). When assigned a point for each associated factor, the integer score’s discrimination was similar (c-statistic = 0.82) with good calibration across the patients with various risk profiles. The models’ performance was retained in the independent validation dataset. Promptly determining in-hospital mortality risks is achievable in the first few hours of presentation; they correlate strongly with mortality among AHF patients, potentially facilitating clinical decision-making.