Accuracy, feasibility and predictive ability of different frailty instruments in an acute geriatric setting

Abstract

To analyze the feasibility, accuracy and the ability of different frailty instruments to predict adverse outcomes.

Methods A prospective cohort study was conducted in patients ≥ 70 years admitted to the acute care setting (ACS). Feasibility and prevalence of frailty were assessed by FRAIL, Clinical Frailty Scale (CFS), hand grip strength (HGS) and the Spanish Frailty-VIG. Receiver operator characteristic (ROC) curves and area under the curve (AUC) were performed to identify frailty according to each instrument, setting VIG as the reference. For each instrument, multiple logistic regressions were used to examine the effect of frailty on primary outcome (i.e., three-month mortality) and secondary outcomes (i.e., in-hospital mortality, length of stay, institutionalization, functional decline and 30-day readmission).

Results A total of 185 patients were included, with a median age of 89 years. The feasibility of the instruments was 100%, except for HGS (67%). The prevalence of frailty varied from 65.2% (FRAIL) to 86.7% (VIG). AUCs against VIG ranged from 0.69 (95% confidence interval [CI] 0.57–0.81: FRAIL) to 0.77 (95% CI 63.5–90.2: CFS). Frail patients defined by FRAIL were 2.7times more likely to have a prolonged length of stay than non-frail patients (95% CI 1.385–5.416). Three-month mortality occurred more among frail patients, either defined by FRAIL (OR 2.5; 95% CI 1.072–5.881) or CFS (OR 3.7; 95% CI 1.255–10.812), than in non-frail patients.

Conclusion The four instruments had high feasibility providing variable prevalence of frailty. FRAIL and CFS predicted well for three-month mortality, and FRAIL also for length of stay. However, none of the instruments predicted for the other secondary outcomes of the study.