A thorough balance assessment is needed to accurately identify elderly individuals who may be at risk for falling. The purpose of this study was (1) to determine the sensitivity and specificity of performance and impairment-based tests in predicting falls and (2) to determine if a combination of these tests is a better predictor of falls.Methods:
Community-dwelling elderly persons (N=198) were evaluated using both performance-based and impairment-based tests. Fall incidence at 6 months was determined by a follow-up telephone survey.Results:
When compared to each other, the 75% LOS test, an impairment-based test, was most accurate at predicting fallers. The performance-based tests (Tinetti and TUG) were most accurate at predicting nonfallers.Conclusion:
Falling is a complex and multifaceted event. Further research is needed to develop balance assessments at both the impairment and performancebased level that are better at predicting falls in the communitydwelling elderly population.