Accuracy of the Fullerton Advanced Balance (FAB) scale and a modified FAB model for predicting falls in older adults: A prospective study

Abstract

BackgroundThis study aimed to determine the accuracy of the FAB scale and the accuracy of combining certain FAB scale items into a model for predicting falls in older adults. MethodsEighty older adults were evaluated at baseline using all FAB scale items. A 6-month follow-up period was used to determine fall incidence. The Receiver Operation Characteristic (ROC) curve was used to assess the predictive capability of the total FAB scale for falls in older adults. A model comprising some FAB items was constructed using logistic regression analysis and a forward stepwise method. ROC curve analysis was used to assess the accuracy of the new model. ResultsThe accuracy of the total FAB scale for predicting falls was excellent (AUC = 0.95). The cut-off score for the FAB scale was 25 points, with a sensitivity of 96.5% and a specificity of 80%. Among the ten FAB items, a regression model was identified by combining four items: step up onto and over a 6-inch bench; tandem walk; standing on foam with eyes closed; and reactive postural control. The new model achieved an excellent level of accuracy (AUC = 0.98) with a cut-off score of 11 out of 16 points, a sensitivity of 100%, and a specificity of 87%. ConclusionsBoth the total FAB score and the new FAB model were highly accurate for predicting falls in older adults.