Measuring fall risk and predicting who will fall: Clinimetric properties of four fall risk assessment tools for residential aged care

Anna L. Barker, Jennifer C. Nitz, Nancy L. Low Choy, Terry Haines

Research output: Contribution to journalArticleResearchpeer-review

27 Citations (Scopus)

Abstract

Background. The purpose of this prospective cohort study was to describe the clinimetric evaluation of four fall risk assessment tools (FRATs) recommended in best practice guidelines for use in residential aged care (RAC). Methods. Eighty-seven residents, mean age 81.59 years(SD ± 10.69), participated. The Falls Assessment Risk and Management Tool (FARAM), Peninsula Health Fall Risk Assessment Tool (PHFRAT), Queensland Fall Risk Assessment Tool (QFRAT), and Melbourne Fall Risk Assessment Tool (MFRAT) were completed at baseline, and 2 and 4 months, and falls occurring in the 6 months after the baseline assessment were recorded. Interrater agreement (kappa), predictive accuracy (survival analysis and Youden Index), and fit to the Rasch model were examined. Twelve-month fall history formed the predictive accuracy reference. Results. Interrater risk classification agreement was high for the PHFRAT (κ =.84) and FARAM (κ =.81), and low for the QFRAT (κ =.51) and MFRAT (κ =.21). Survival analysis identified that 43%-66% of risk factors on each tool had no (p >.10) association with falls. No tool had higher predictive accuracy (Youden index) than the question, "has the resi dent fallen in past 12 months? "(p >. 05). All tools did not exhibit fit to the Rasch model, invalidating summing of risk factor scores to provide an overall risk score. Conclusion. The studied tools have poor clinimetric properties, casting doubt about their usefulness for identifying fall risk factors for those most at risk for falling and measuring fall risk in RAC.

Original languageEnglish
Pages (from-to)916-924
Number of pages9
JournalJournals of Gerontology - Series A Biological Sciences and Medical Sciences
Volume64
Issue number8
DOIs
Publication statusPublished - Aug 2009

    Fingerprint

Cite this