Item Distribution in the Berg Balance Scale: A Problem for Use With Community-Living Older Adults

Background and Purpose: The Berg Balance Scale (BBS) is a commonly used clinical test measure to determine balance performance and fall risk. However, a ceiling effect of the BBS has been reported in studies of older adults with relatively higher levels of physical performance. The underlying reason for this ceiling effect may be that the task items in the BBS are insufficiently challenging to discriminate individuals with less severe balance limitations. The purpose of this study was to investigate the unidimensional construct, item difficulty hierarchy, and item distribution of the BBS in order to determine its usefulness among community-living older adults (CLOAs). Methods: CLOA volunteers (N = 112; 34 men, 78 women), mean age 82.4 years (SD = 7.9) (range, 65-99 years), were tested with the BBS by physical therapy faculty and students from Drexel University. Rasch principal component analysis (PCA) was used to investigate the dimensionality of the BBS, and the Rasch rating scale model was used to determine the item difficulty hierarchy and distribution. Results: Rasch PCA confirmed the unidimensional construct of the BBS as a balance ability test. Two items failed to fit the Rasch model, "sitting unsupported" and "standing unsupported with eyes closed." Item difficulty hierarchy indicated that the most difficult test item was "stand on one leg" and the easiest was "sitting unsupported." Item and person measures ranged from -4.35 to 2.66 and -1.77 to 6.58 logits, respectively. Person and item separation indexes were 2.10 and 6.41 (reliabilities of 0.82 and 0.98, respectively). Conclusions: Comparing the item difficulty and person ability, the balance ability of the CLOAs exceeded the difficulty of the test items, rendering it less useful for detecting balance ability and fall risk in CLOAs. More challenging test items, or selection of a different balance test, are recommended for use with this population.