Introduction: Balance performance in the elderly is crucial to maintain independence in activities of daily living. Balance impairment results in the likelihood of elderly people falling during these activities. Many tests to assess balance performance have been developed, notably center of pressure (COP) during quiet standing for static balance, timed up-and-go (TUG), and walking speed for dynamic balance, and computerized dynamic posturography, named the EquiTest (NeuroCom) for perturbed balance. However, few reports reveal the structure underlying these balance performance tests between young and older adults. Covariance structure analysis is a tool that is able to test statistically whether the factorial structure fits the data. Therefore, this study examined aging effects on the factorial structure of the balance performance tests.
Methods: Subjects: The subjects were 60 healthy young women (young group) and 60 community-dwelling older women (older group). The mean age of young and older groups was 22.2 and 69.2 years, respectively. The subjects for this study completed all of the balance performance tests.
Test Items: This study consisted of four balance performance tests: COP, TUG, 5-m walking speed, and the EquiTest. In the EquiTest, scores and latencies were measured by perturbing a moving platform and visual surroundings.
Model Specification: As a preliminary step for analysis of factorial invariance, we assumed a 3-balance-factors model on the basis of face validity and the results of factorial analysis for young and older groups. The model was a second-order factor covariance structure model in which BALANCE was the second-order latent variable. Three components of BALANCE were assumed as first-order latent variables with measurement errors: STATIC, DYNAMIC, and PERTURBED. Each first-order variable was assumed to have loadings on BALANCE. Observed variables were area of COP with eyes open and closed and the scores of Condition 2 in the EquiTest for STATIC, preferred and maximum walking speeds and TUG for DYNAMIC, and latencies in the anterior and posterior perturbations of the platform in the EquiTest for PERTURBED.
Results: As a result of the analysis, the goodness-of-fit indexes (GFI) for all data of the models were acceptable (GFI: young group = .931, older group = .923). The same factors in the factorial structure model of BALANCE were revealed for young and older groups. But the factor loading of first-order factors on the second-order factor (BALANCE) were different between groups; for the young group, PERTURBED was the highest factor loading on BALANCE (standardized solutions: STATIC = .21, DYNAMIC = .24, PERTURBED = .76). On the other hand, for the older group, STATIC was highest factor (standardized solutions: STATIC = .71, DYNAMIC = .28, PERTURBED = .43).
Conclusion: This study showed the common factorial structure of balance abilities were static, dynamic, and perturbed balance performance. It was suggested that for young people the perturbed balance performances were highly correlated and explained by a single factor, while for older people it was static balance performance.