Item Distribution of the Berg Balance Scale in Older Adults with Hip Fracture: a Rasch Analysis

Introduction Balance impairment occurs after a hip fracture, but the characteristics of the impairment are not clear. Objective To investigate the uni-dimensionality, fit statistics, and item difficulty of the Berg Balance Scale (BBS) in older adults with hip fracture by conducting a Rasch analysis. Methods This was an observational cross-sectional study. The 254 participants were all >= 65 years old and had been hospitalized for rehabilitation after a unilateral hip fracture incurred during a fall. We collected their BBS scores at the time of hospital discharge and conducted a Rasch analysis to examine the uni-dimensionality, fit statistics, and item difficulty. Results The principal component analysis (PCA) of the Rasch model demonstrated that the BBS is uni-dimensional. The information-weighted mean square (MnSq) fit statistic was within the range of fit criteria for all items. The underfit item of the outlier-sensitive MnSq fit statistics was "Standing unsupported eyes closed" with the MnSq of 2.06. The difficult items were in order of logits: "Standing on one leg" (logits = 4.01); "Step tool" (logits = 2.74); and "Turn 360 degrees" (logits = 2.61). Conclusion The BBS is uni-dimensional and conforms with the Rasch model. The BBS most difficult items for older adults with a hip fracture required one-legged support and dynamic balance.