Development and Evaluation of a Modification of the Fried Frailty Criteria Using Population‐Independent Cutpoints

Abstract

Objectives To develop and evaluate a modification of the Fried frailty assessment using population‐independent cutpoints and to determine frailty prevalence of community‐dwelling elderly people in a German population. Design Cross‐sectional analysis of 8‐year follow‐up data of a large German cohort study. Setting Saarland, Germany. Participants Three thousand one hundred twelve community‐dwelling adults aged 59 and older. Measurements Frailty was operationalized using modified Fried frailty criteria. Criteria were categorized according to quintiles (lowest‐quintile approach) or using population‐independent cutpoints derived from the literature (population‐independent approach). Agreement and construct validity of frailty classification according to both approaches were evaluated according to weighted kappa (κ) and Spearman rank correlation (r_Sp). Associations between frailty and covariates were assessed using multiple logistic regression models. Results Although more participants were identified as frail according to the population‐independent index (8.9%) than the lowest‐quintile index (6.5%), agreement and correlation of frailty classification using both approaches was high (κ = 0.75 and r_Sp = 0.84). Sex differences in frailty prevalence were more pronounced when the population‐independent approach was used (women 11.4%; men 6.1%). Similarly strong significant associations with sociodemographic, lifestyle, and medical factors such as older age, female sex, smoking, and obesity were seen for both approaches. Conclusion The modified Fried index using literature‐derived cutpoints independent from the frailty criteria distributions in the underlying study population showed good correlation with the lowest‐quintile approach and enables prevalence estimates that are directly comparable between different populations.