Frailty in a Global Population: Should Geographic Region Influence Frailty Definitions?

Abstract

Introduction: The frailty phenotype is a commonly used tool to study frailty. Two characteristics evaluated as part of the frailty phenotype are “low” grip strength and “low” physical activity, defined by the lowest quintile thresholds for age and sex. In studies of frailty in different geographic regions of the world, it is not established whether these thresholds should be applied universally or whether region-specific thresholds of grip strength and physical activity should be applied. This study aims to determine which way of defining frailty is more appropriate. Methods: Using data from the Prospective Urban Rural Epidemiology study, two variations of the frailty phenotype were defined: universal frailty in which thresholds for low grip strength and physical activity were taken to be the lowest quintile of the entire study population and region-specific frailty, in which these thresholds were calculated separately for each region. Frailty prevalence was calculated for each definition and Cox proportional hazards modelling was used to determine which definitions predicted mortality. Likelihood ratio tests statistics, area under the receiver operating characteristics curve, and the net reclassification improvement index were also calculated. Results: Overall frailty prevalence was 5.6% using universal definitions of frailty and 5.8% for region-specific definitions of frailty. Across regions, universal frailty prevalence ranged from 2.4% (North America/Europe) to 20.1% (Africa), while region-specific frailty ranged from 4.1% (Russia and Central Asia) to 8.8% (Middle East). The hazards ratios for all-cause mortality were 2.66 (95% CI: 2.47-2.86) and 2.09 (95% CI: 1.94-2.26) for universal frailty and region-specific frailty respectively (adjusted for age, sex, education, smoking status and alcohol consumption); statistical tests indicated that universal frailty better fit survival data and predicted mortality slightly better. Conclusions: Frailty prevalence varies greatly across regions depending on how the thresholds for low physical activity and grip strength are calculated. Using region-specific thresholds does not help improve the predictive value of frailty when measuring frailty in heterogenous populations using the frailty phenotype.