# Limited predictive value of traditional comorbidities for readmission in acute decompensated heart failure

## TL;DR

Common heart failure comorbidities like diabetes and kidney disease are not strong predictors of readmission, suggesting the need for better risk models.

## Contribution

Demonstrates that traditional comorbidities have limited predictive value for readmission in acute decompensated heart failure.

## Key findings

- Comorbidities like CKD and diabetes showed statistical associations but poor predictive power for readmission.
- AUC values for 30-day and 100-day readmission remained low, even when combining multiple comorbidities.
- The study highlights the need for more robust models incorporating dynamic clinical and patient-centered factors.

## Abstract

Common comorbidities in heart failure (HF), including chronic kidney disease (CKD), diabetes mellitus (DM), ischemic heart disease (IHD), and atrial fibrillation, are frequently presumed to predict hospital readmission. However, recent studies have challenged their predictive strength, raising questions about their clinical utility for risk stratification.

We conducted a retrospective cohort study of 7,652 patients admitted with acute decompensated heart failure (ADHF) at a tertiary center between 2007 and 2017. Associations between comorbidities and readmission at 30 and 100 days were assessed using Fine-Gray competing risk models, with death as a competing event. Subdistribution hazard ratios (sHRs) were reported. Model performance was evaluated using receiver operating characteristic (ROC) analysis and area under the curve (AUC) values, assessing individual comorbidities and incremental combinations. All comorbidities were included irrespective of univariable significance, based on clinical relevance.

Several comorbidities were significantly associated with readmission, including CKD (sHR 1.16–1.23), DM (sHR 1.18–1.27), IHD (sHR 1.10–1.15), and anemia (sHR 1.11). However, predictive power was poor. For 30-day readmission, AUC values ranged from 0.516 (COPD) to 0.529 (CKD), with a maximal AUC of 0.555 when combining the four strongest predictors. For 100-day readmission, AUC values ranged from 0.528 (DM) to 0.545 (CKD), with a maximal combined AUC of 0.593.

Despite consistent statistical associations, common comorbidities perform poorly as predictive tools for identifying individual patients at risk of HF readmission. These findings highlight the need for more robust risk models integrating dynamic clinical, laboratory, and patient-centered factors.