Predicting survival in patients with severe heart failure: Risk score validation in the HELP-HF cohort

Aims: Accurate selection of patients with severe heart failure (HF) who might benefit from advanced therapies is crucial. The present study investigates the performance of the available risk scores aimed at predicting the risk of mortality in patients with severe HF. Methods and results: The risk of 1-year mortality was estimated in patients with severe HF enrolled in the HELP-HF cohort according to the MAGGIC, 3-CHF, ADHF/NT-proBNP, and GWTG-HF risk scores, the number of criteria of the 2018 HFA-ESC definition of advanced HF, I NEED HELP markers, domains fulfilled of the 2019 HFA-ESC definition of frailty, the frailty index, and the INTERMACS profile. In addition, we tested the performance of different machine learning (ML)-based models to predict 1-year mortality. At 1-year follow-up, 265 patients (23.1%) died. The prognostic accuracy, tested in the subgroup of patients with completeness of all data regarding the variables included in the scores (497/1149 patients), resulted moderate for MAGGIC, GWTG-HF, and ADHF/NT-proBNP scores (area under the curve [AUC] ≥0.70) and only poor for the other tools. All the scores lost accuracy in estimating the rate of 1-year mortality in patients at the highest risk. Support vector machine-based model had the best AUC among ML-based models, slightly outperforming most of the tested risk scores. Conclusion: Most of the scores used to predict the risk of mortality in HF performed poorly in real-world patients with severe HF and provided inaccurate estimate of the risk of 1-year mortality in patients at the highest risk. ML-based models did not significantly outperform the currently available risk scores and their use must be validated in large cohort of patients.