Comparison of three frailty screening instruments for prediction of adverse outcomes among older adults in the emergency department

Aim To compare the predictive abilities of the FRAIL scale (FS), frailty screening questionnaire (FSQ) and clinical frailty scale (CFS) for adverse outcomes in older adults in the emergency department. Methods In total, 317 older adults aged >= 65 years attending emergency department was screened for frailty using the FS, FSQ and CFS. Outcome measures included all-cause 28-day mortality and intensive care unit readmission. Cox proportional hazards model was used for survival comparison. Logistic regression was used to analyze risk factors for readmissions. In addition, we calculated the C-statistic, net reclassification improvement and integrated discrimination improvement to evaluate the predictive value of three scales. Results The prevalence of frailty was 55.2% (FS), 47.0% (FSQ) and 69.4% (CFS). Cox regression and logistic regression analysis revealed that frailty screening by FS, FSQ and CFS was an independent risk factor for all-cause 28-day mortality and 30- and 90-day readmission after adjustment. Incorporation of FS, FSQ and CFS into a basic model with other risk factors significantly improved C-statistic. For all-cause 28-day mortality, the model including FS had the highest C-statistic from 0.786 (95% confidence interval: 0.706-0.865) to 0.854 (95% confidence interval: 0.802-0.907) and the improvements in risk prediction were also confirmed by category-free net reclassification improvement and integrated discrimination improvement, suggesting FS was significantly better than CFS and FSQ. The three tools had a low predictive ability for readmission (all C-statistics <0.7). Conclusions All three frailty scales showed a predictive ability for 28-day mortality and readmission but FS may be the most valid tool in the emergency department. Geriatr Gerontol Int center dot center dot; center dot center dot: center dot center dot-center dot center dot Geriatr Gerontol Int 2022; center dot center dot: center dot center dot-center dot center dot.