Integrating Frailty and Physical Resilience to Improve Risk Estimation in Aging Populations

Abstract

Background: Physical resilience, broadly defined as the ability to resist or recover from health stressors, is a relatively new concept in the field of aging, yet, anticipated as a potentially game-changing idea. In particular, physical resilience has significant potential to complement the widely used concept of frailty. To date, conceptual and methodological difficulties have hampered such potential. Frailty and physical resilience are broad and closely related concepts with multiple interpretations, many approaches to measure physical resilience have been proposed, and empirical studies have rarely investigated frailty and physical resilience together. Project Aim: Enrich our understanding of the relationship between frailty and physical resilience in the health of aging populations by providing a novel, integrated framework and a concurrent empirical investigation of the two concepts in longitudinal population data. Specific Objectives: 1. Guided by the integrated framework, operationalize multiple specific measures of frailty and physical resilience and provide a descriptive analysis of each measure. 2. Evaluate the relationship between frailty, physical resilience, and mortality. 3. Evaluate the relationship between frailty, physical resilience, and acute functional recovery. Methods: This study uses repeated measurements of a 41-item frailty index (FI) from waves 3-13 of the Health and Retirement Study in the United States to operationalize three indicators of resilience: the frailty-disease mismatch (FM), the rate of aging (RoA), and a dynamical indicator of resilience (DIOR-FI). This study evaluates each measure using descriptive statistics and logistic regression models estimating the probability of 2-year all-cause mortality (n=27,744) and full functional recovery after myocardial infarction (n=1,905). Results: Resilience indicators generally had low agreement (kappa ≤0.24) and moderate to weak correlation (Pearson r ≤0.53). All indicators showed statistically significant associations with mortality and recovery. Despite requiring careful interpretation, FM shows the greatest promise for adding predictive ability beyond age, sex, and frailty. Conclusion: This study demonstrates key insights and lessons learned for future research on frailty and physical resilience. With further refinement of the methods proposed in this thesis, the combination of population data and routinely collected health data offer promising opportunities to improve risk estimation in aging populations.