Frailty Assessment in Community-Dwelling Older Adults: A Comparison of 3 Diagnostic Instruments
Y J Lim,Y S Ng,R Sultana,E L Tay,S M Mah,C H N Chan,A B Latib,H M Abu-Bakar,J C Y Ho,T H H Kwek,L Tay
DOI: https://doi.org/10.1007/s12603-020-1396-2
Abstract:Objectives: Compare the diagnostic performance of FRAIL against Fried Phenotype and Frailty Index (FI), and identify clinical factors associated with pre-frailty/frailty. Design: Cross-sectional analysis. Setting: Community-based screenings in Senior Activity Centres, Residents' Corners and Community Centres in northeast Singapore. Participants: 517 community dwelling participants aged >55 years and ambulant independently (with/ without walking aids) were included in this study. Residents of sheltered or nursing homes, and seniors unable to ambulate at least four meters independently were excluded. Measurements: The multidomain geriatric screen included assessments for social vulnerability, mood, cognition, sarcopenia and nutrition. Participants completed a battery of physical fitness tests for grip strength, gait speed, lower limb strength and power, flexibility, balance and endurance, with overall physical performance represented by Short Physical Performance Battery (SPPB). Frailty status was assigned on FRAIL, Fried and 35-item FI. Results: Prevalence of frailty was 1.3% (FRAIL) to 3.1% (FI). Pre-frailty prevalence ranged from 17.0% (FRAIL) to 51.2% (FI). FRAIL demonstrated poor agreement with FI (kappa=0.171, p<0.0001), and Fried (kappa=0.194, p<0.0001). A lower FRAIL cut-off ≥1 yielded significantly improved AUC of 0.70 (95%CI 0.55 to 0.86, p=0.009) against Fried, and 0.71 (95%CI 0.55 to 0.86, p=0.008) against FI. All 3 frailty measures were diagnostic of impaired physical performance on SPPB, with AUCs ranging from 0.69 on FRAIL to 0.77 on Fried (all p values <0.01). Prevalence of low socio-economic status, depression, malnutrition and sarcopenia increased significantly, while fitness measures of gait speed, balance, and endurance declined progressively across robust, pre-frail and frail on all 3 frailty instruments (p <0.05). Conclusions: Our results suggest that different frailty instruments may capture over-lapping albeit distinct constructs, and thus may not be used interchangeably. FRAIL has utility for quick screening, and any positive response should trigger further assessment, including evaluation for depression, social vulnerability and malnutrition.