# Identification of Frailty Clusters Using Cross-Sectional Frailty and Frailty Trajectory: Cohort of Heart Failure Veterans

## TL;DR

This study identifies four frailty clusters in heart failure patients using cross-sectional and longitudinal measures, showing that the high-high cluster has the worst outcomes.

## Contribution

The study introduces a novel approach combining cross-sectional frailty indices with frailty trajectories to improve risk stratification in heart failure patients.

## Key findings

- The high-high cluster had over twice the odds of 1-year mortality compared to the low-low cluster.
- The high-high cluster had significantly higher rates of prolonged hospital stays and readmissions.
- Combining cross-sectional and longitudinal frailty measures improves risk stratification for heart failure patients.

## Abstract

Frailty is a syndrome associated with increased vulnerability and diminished physiological reserves. Three-quarters (78%) of heart failure (HF) patients are frail. Traditional frailty indices (FIs) assess cross-sectional deficits, while frailty trajectories (FTs) measure changes over time.

This study aims to examine the interaction between FI and FT to enhance risk stratification in hospitalized adults with HF.

This retrospective cohort study utilized data from the Veterans Health Administration, including 143,687 veterans aged >50 admitted for HF from 2005 to 2019. FT measurements were derived from FI calculations for each of the 3 years before index hospitalization. Unsupervised clustering identified 4 clusters based on FI and FT interactions: low-low, low-high, high-low, and high-high. Associations between these clusters and clinical outcomes (ie, 1-year mortality, prolonged hospital stays, emergency department visits, and readmissions) were analyzed.

The study cohort was mostly older (mean age 74 ± 10 years), male (98%), and diverse (55% non-Hispanic White). Survival analysis showed distinct mortality risks across clusters; while the 2 clusters with low FI had the longest survival, the high-high group had the lowest survival probability. Adjusted logistic regression indicated that the high-high cluster had over twice the odds of 1-year mortality compared to the low-low cluster (OR: 2.29; 95% CI: 2.15-2.44). The high-high cluster also had significantly higher rates of prolonged hospital stays, emergency department visits, and readmissions at 30 and 90 days postdischarge.

Integrating cross-sectional FI and longitudinal FT offers a comprehensive assessment of frailty in HF patients, improving risk stratification and disease management.