Agreement and predictive power of fall risk assessment methods in the elderly population

Abstract

Introduction: A large number of fall risk assessment methods are available with a variety of performances for screening the risk of falling in older adults, but their agreement for assessing the risk of falling remains unknown. Objective: To describe the agreement and predictive power of methods to classify the risk of falling in older adults using prospective data and published cut-off values. Methods: Observational prospective cohort study using a nonprobabilistic sampling. Participants were assessed at baseline regarding the retrospective fall history and current risk factors for falling. Follow-up interviews were conducted six months after admission for collecting fall occurrence. Fifty-two participants (44 females, 85%) aged 74 years (interquartile range 69–80) were assessed using: Berg Balance Scale (BBS); polypharmacy; Falls Risk Assessment Score (FRAS); Fall Risk Assessment Tool (FRAT-up); Fall Efficiency Scale (FES); and Posturography. Results: Nine participants (17%) reported at least one fall after six months. Cochran’s test showed different proportions of participants classified as at high risk of falling among all FRA methods (Q = 69.560, p < 0.001). A slightly better-then-chance agreement was estimated between all FRA methods (Light’s  = 0.074, 95%CI = 0.021 to 0.142). Pairwise analyses revealed agreements ranging from moderate (FRAS vs. FRAT-up: 73.1%, Cohen’s  = 0.462, p = 0.001) to poor (FRAS vs. Posturography: 38.5%, Cohen’s  = -0.201, p = 0.050). Conclusion: We found both global and pairwise agreement levels that questions the reliability of fall risk assessment methods for screening community-dwelling older adults.