Evaluating the convergent and discriminant validity of three versions of the frailty phenotype in heart failure: results from the FRAME-HF study.

Abstract

Frailty is an important predictive measure of mortality and rehospitalisation in people with heart failure. To date, there are no frailty instruments validated for use in people with heart failure. The aim of this study was to evaluate the convergent and discriminant validity of three versions of the frailty phenotype in those with heart failure. A single site, prospective cohort study was undertaken among individuals with a confirmed diagnosis of heart failure. Frailty was assessed concurrently using three versions of the frailty phenotype: the original frailty phenotype and two modified versions; the Survey of Health, Ageing and Retirement in Europe frailty instrument (SHARE-FI) and the St Vincent's frailty instrument. Convergent and discriminant validity were assessed by reporting the correlations between each version and related heart failure subconstructs, and by evaluating the ability of each version to discriminate between normal and abnormal scores of other physical and psychosocial scales specific to heart failure-related subconstructs. The New York Heart Association classes were moderately correlated with the St Vincent's frailty instrument (r=0.47, P⩽0.001), SHARE-FI (r=0.42, P⩽0.001) and the frailty phenotype (r=0.42, P⩽0.001). The SHARE-FI and the St Vincent's frailty instrument were both able to discriminate consistently between normal and abnormal scores in three out of five of the physical and psychosocial subconstructs that were assessed. The SHARE-FI was also able to discriminate between inpatients and outpatients who were classified as frail. Both the SHARE-FI and the St Vincent's frailty instrument displayed good convergent and discriminant validity.