# Prediction model of rehospitalization and mortality in heart failure patients with preserved and mildly reduced ejection fraction: the AD2NNER risk score

## TL;DR

This study developed a risk score called AD2NNER to predict rehospitalization and mortality in heart failure patients with preserved or mildly reduced ejection fraction.

## Contribution

The AD2NNER score is a novel, simple six-variable model for predicting rehospitalization and mortality in HFpEF and HFmrEF patients.

## Key findings

- The AD2NNER score includes age, T2DM, serum sodium, NYHA class, etiology, and rhythm as independent predictors.
- Higher AD2NNER scores were associated with reduced event-free survival and increased mortality.
- The score performed well in HFpEF patients but not in HFmrEF patients alone.

## Abstract

This study aimed to identify predictors of heart failure (HF) rehospitalization and explore their association with mortality in patients with preserved (HFpEF), and mildly reduced (HFmrEF) ejection fraction, leading to the development of a multivariable risk prediction score.

We enrolled 1,022 HFpEF and HFmrEF inpatients discharged between January 2019 and May 2023. Demographic, clinical, biological, and imaging data were collected for analysis.

After a mean follow-up of 3.5 ± 1.4 years, 308 (30.1%) patients experienced HF rehospitalization. Univariable analysis revealed several parameters associated with HF rehospitalization, including age (p < 0.001), male sex (p = 0.015), type 2 diabetes mellitus (T2DM, p = 0.016), arterial hypertension (p = 0.018), smoking (p = 0.029), NYHA class at discharge (p = 0.006), atrial fibrillation (p = 0.003), ischemic or congenital etiology (p = 0.011), serum sodium (p = 0.002), and several echocardiographic measures. Multivariate Cox regression revealed six independent predictors: age (HR = 0.98, p < 0.001), T2DM (HR = 1.31, p = 0.026), NYHA class (HR = 1.39, p = 0.010), ischemic or congenital etiology (HR = 1.33, p = 0.037), atrial fibrillation (HR = 0.65, p = 0.001), and serum sodium level (HR = 0.97, p = 0.005). These formed the AD2NNER (age, T2DM, serum natrium, NYHA class, etiology, rhythm) score, ranging from 0 to 9 points. Kaplan–Meier analysis confirmed reduced event-free survival in patients with scores ≥4 (log-rank p = 0.005). Comparative Kaplan–Meier curves using an unweighted risk count (0–6) showed less distinct stratification. Subgroup analysis revealed robust score performance in HFpEF, but not HFmrEF alone. Higher AD2NNER scores were also associated with all-cause mortality.

The AD2NNER risk score is a simple, six-variable model that effectively predicts rehospitalization, and is also associated with mortality in patients with HFpEF and HFmrEF.