Convergent validity of the electronic frailty index

• Published in: Age and ageing Vol. 48; no. 1; pp. 152 - 156
• Main Authors: Brundle, Caroline, Heaven, Anne, Brown, Lesley, Teale, Elizabeth, Young, John, West, Robert, Clegg, Andrew
• Format: Journal Article
• Language: English
• Published: England Oxford University Press 01.01.2019; Oxford Publishing Limited (England)
• Subjects: Aged; Aged, 80 and over; Aging; Cohort analysis; Community; Convergent validity; Cross-Sectional Studies; Electronic Health Records - statistics & numerical data; Electronic medical records; Female; Frail; Frail Elderly - statistics & numerical data; Frailty; Frailty - diagnosis; Geriatric Assessment - methods; Health care; Humans; Male; Older people; Phenotypes; Predictive validity; Primary care; Reproducibility of Results; Residential care; Validation studies; Validity; frailty; electronic frailty index (eFI); older people; validity; correlation; electronic frailty index
• ISSN: 0002-0729; 1468-2834; 1468-2834
• DOI: 10.1093/ageing/afy162

Abstract Background the electronic frailty index (eFI) has been developed and validated using routine primary care electronic health record data. The focus of the original big data study was on predictive validity as a form of criterion validation. Convergent validity is a subtype of construct validity and considered a core component of the validity of a test. Objective to investigate convergent validity between the eFI and research standard frailty measures. Design cross-sectional validation study using data from the Community Ageing Research 75+ (CARE 75+) cohort. Setting multi-site UK community-based cohort study. Subjects three hundred fifty-three community-dwelling older people (median age 80 years, IQR 77–84), excluding care home residents and people in the terminal stage of life. Median eFI score of participants was 0.22 (IQR 0.14–0.31). Methods convergent validities between the eFI and: a research standard frailty index (FI); the phenotype model of frailty; Clinical Frailty Scale (CFS) and Edmonton Frail Scale were assessed using scatter plots and Spearman’s rank tests to estimate correlation coefficients (Spearman’s rho, ρ) and 95% confidence intervals. Results results indicate strong correlation between the eFI and both the research standard FI (ρ = 0.68, 95% CI 0.62–0.74) and Edmonton Frail Scale (ρ = 0.63, 95% CI 0.57–0.69). There was evidence for moderate correlation between the eFI and both the CFS (ρ = 0.59, 95% CI 0.49–0.65) and phenotype model (ρ = 0.51, 95% CI 0.42–0.59). Conclusions This study provides evidence for convergent validity of the eFI, a core component of test validity.