Correlation of accelerometry with clinical balance tests in older fallers and non-fallers

Background: falls are a common cause of injury and decreased functional independence in the older adult. Diagnosis and treatment of fallers require tools that accurately assess physiological parameters associated with balance. Validated clinical tools include the Berg Balance Scale (BBS) and the Timed Up and Go test (TUG); however, the BBS tends to be subjective in nature, while the TUG quantifies an individuals functional impairment but requires further subjective evaluation for balance assessment. Other quantitative alternatives to date require expensive, sophisticated equipment. Measurement of the acceleration of centre of mass, with relatively inexpensive, lightweight, body-mounted accelerometers is a potential solution to this problem. Objectives: to determine (i) if accelerometry correlates with standard clinical tests (BBS and TUG), (ii) to characterise accelerometer responses to increasingly difficult challenges to balance and (iii) to characterise acceleration patterns between fallers and non-fallers. Study design and setting: torso accelerations were measured at the level of L3 using a tri-axial accelerometer under four conditions; standing unsupported with eyes open (EO), eyes closed (EC) and on a mat with eyes open (MAT EO) and closed (MAT EC). Older patients (n = 21, 8 males, 13 females) with a mean age of 78 (SD ± 7.6) years who attended a day hospital were recruited for this study. Patients were identified as fallers or non-fallers based on a comprehensive falls history. Measurements: Spearman's rank correlation analysis examined the relationship between acceleration root mean square (RMS) data and the BBS while Pearson's correlation was used with TUG scores. Differences in accelerometer RMS between fallers and non-fallers and between test conditions were examined using t-test and non-parametric alternatives where appropriate. Results: there was a stepwise increase in accelerometer RMS with increasing task complexity, and the accelerometer was able to distinguish significantly between sway responses to all test conditions except between EO and EC (P