Abstract
PURPOSE: When screening large populations, performance-based measures can be difficult to conduct because they are time consuming and costly, and require well-trained assessors. The aim of the present study is to validate a set of questions replacing the performance-based measures slowness and weakness as part of the Fried frailty phenotype (FRIED-P). METHODS: A cross-sectional study was conducted among community-dwelling older adults (≥ 60 years) in three Flemish municipalities. The Fried Phenotype (FRIED-P) was used to measure physical frailty. The two performance-based measures of the Fried Phenotype (slowness and weakness) were also measured by means of six substituting questions (FRIED-Q). These questions were validated through sensitivity, specificity, Cohen's kappa value, observed agreement, correlation analysis, and the area under the curve (AUC, ROC curve). RESULTS: 196 older adults participated. According to the FRIED-P, 19.5% of them were frail, 56.9% were pre-frail and 23.6% were non-frail. For slowness, the observed sensitivity was 47.0%, the specificity was 96.5% and the AUC was 0.717. For weakness, the sensitivity was 46.2%, the specificity was 83.7%, and the AUC was 0.649. The overall Spearman correlation between the FRIED-P and the FRIED-Q was r = 0.721 with an observed agreement of 76.6% (weighted linear kappa value = 0.663, quadratic kappa value = 0.738). CONCLUSIONS: The concordance between the FRIED-P and FRIED-Q was substantial, characterized by a very high specificity, but a moderate sensitivity. This alternative operationalization of the Fried Phenotype-i.e., including six replacement questions instead of two performance-based tests-can be considered to apply as screening tool to screen physical frailty in large populations.