Objective: To examine the timed up and go (TUG) and tandem walk test (TWT) as fall prediction assessments in Thai elderly.
Material and Method: Elderly subjects aged between 60 and 86 years and living in Nakhonpathom and Samutsakhon provinces were classified as fallers and non-fallers by self-report in the past six months. The TUG and TWT were used to predict falls. The optimal cutoff score and validity indexes were determined by plotting the receiver operating characteristic (ROC) curve and crosstabs analysis.
Results: One hundred and sixty-one elderly subjects were classified as fallers (7 males and 43 females) and non-fallers (24
males and 87 females). The area under the curve (AUC) of ROC were 0.62 (95% CI = 0.59, 0.76; p = 0.0001) for TUG and 0.605 (95% CI = 0.514, 0.696; p = 0.033) for TWT error score. The cutoff scores were 10.5 seconds for TUG (74% sensitivity and 57.7% specificity) and five scores for TWT error (62% sensitivity and 55% specificity).
Conclusion: TUG and TWT error were useful tools to explain faller status in Thai community-dwelling for the elderly. TWT time was not sensitive enough to detect the elders who were at risk of falls.

Keywords: Falls, Timed up and go test, Tandem walk test, Thai elderly