Background and Purpose: The Timed Up and Go (TUG) is a simple tool used by clinicians to assess functional mobility of their elderly clients. Currently the TUG is only valid as a predictor of activities of daily living status, yet researchers have begun to look at its ability to predict fall occurrence in the elderly. This study investigated the sensitivity and specificity of the TUG, the TUG with a Manual Task (TUGmanual), and the TUG difference (TUGdifference) to predict fall occurrence in two populations of older adults. Cut-off scores were calculated for ease of clinical application.
Subjects: Forty- two elderly (mean age= 80 years, SD=6, range= 67-93) volunteers participated. Subjects were divided into four groups: community non-faller (n= 13), community faller (n= 9), assisted living non-faller (n: 12), and assisted living faller (n= 8). Fall history was determined via a self-reported questionnaire in which a fall within the past year placed a subject in the faller group.
Methods: Subjects were timed during three trials of the TUG and TUGmanual their TUGdifference Was taken from each trial, and the averages of all three trials (TUG, TUGmanual and TUGdifference) were calculated to determine their score. A six month prospective phase followed the initial testing period. During this phase, data was collected every two weeks via a fall occurrence postcard that subjects filled out daily. Subjects were reclassified into appropriate fall groups using the prospective fall occurrence data. Regression analysis was used to identify which testing measures significantly classified subjects into correct fall groups. After a significant testing measure was identified, logistic regression was implemented to formulate a cut-off score.
Results: For all subjects, the TUGmanual was indicated as the most sensitive (73%) and most specific (78%) in identifying those at a fall risk. The TUGdifference was the best predictor of community dwelling subjects (sensitivity= 50%, specificity= 94%), and the TUG was determined significant for classifying assisted living residents (sensitivity= 78%, specificity= 82%). Cut-off scores were calculated for each identified testing measure. Subjects scoring above 14.52 sec on the TUGmanual were more likely to fall (overall prediction rate= 76%), whereas, community dwellers scoring above 1.93 sec on the TUGdifference, and assisted living residents scoring above 12.39 sec on the TUG (overall prediction rate= 82% and 80% respectively) tended to be fallers.
Discussion and Conclusion: When choosing an assessment tool that is predictive of fall occurrence, it is important to know what population of elderly the instrument will be used with. The TUGmanual. in this particular study, was the best predictor of fall occurrence in a mixed population, whereas the TUGdifference was more specific for community elderly. The TUG was found to be most predictive with assisted living residents. Further studies are needed to confirm the results of this study and existing literature. Emphasis should be placed on a comprehensive literature search, and using critical evaluation of research designs, when choosing an objective tool for assessing fall risk.
Files are restricted to Pacific University. Sign in to view.