K. Bower1, S. Thilarajah2, G. Williams1,3, Y.-H. Pua2, D. Tan2, R. Clark4
1The University of Melbourne, Melbourne, Australia, 2Singapore General Hospital, Singapore, Singapore, 3Epworth HealthCare, Melbourne, Australia, 4University of the Sunshine Coast, Sippy Downs, Australia

Background: Despite frequent research use, static balance centre of pressure (COP) measures following stroke require further validation. A range of COP variables may be obtained from force platform-based testing, with simple measures of velocity and amplitude being most common. However, the findings to support utility for measuring change over time or predicting outcomes such as falls are limited and variable.

Purpose: For a comprehensive range of COP variables derived from static balance testing in people with stroke, this study aimed to examine the:
1) association with clinical gait and balance tests;
2) predictive validity for falls; and
3) changes over three months post discharge from inpatient rehabilitation.

Methods: Seventy-nine individuals (mean age 63 years, 28 days post stroke), recruited from inpatient rehabilitation facilities within Australia and Singapore, were assessed within one week prior to discharge. Falls data were collected in all participants over 12 months via monthly calendars, and 73 (92%) were retested at three months post discharge. In addition to clinical measures of gait speed (6m walk test) and dynamic balance (step test), participants completed two trials of eyes open standing still on a Wii Balance Board. Measures of COP velocity, amplitude, standard deviation, root mean square (RMS), ultralow to moderate frequency wavelet and detrended fluctuation analysis (DFA) were examined in both anteroposterior and mediolateral directions. Correlation analyses were perform using Spearman’s rho. Separate regression analyses were performed for each of the clinical and COP variables, adjusting for country, prior falls and body mass. Odds ratios (ORs) were scaled to the interquartile range (IQR) of each variable. Changes over time were assessed by paired t-tests, standardised response means and effect sizes.

Results: Significant moderate strength correlations (rho=0.51-0.62) with the two clinical measures were shown for 4/48 COP variables (COP velocity, RMS velocity and low-moderate frequency wavelet variables, all in the mediolateral direction). Falls were by reported by 22/79 (28%) participants. Regression analysis demonstrated significant falls prediction for both clinical tests (step test: p=0.001, IQR-OR=4.73; 6m walk test: p=0.022, IQR-OR=3.21) and 2/28 COP variables (anteroposterior DFA: p=0.020 IQR-OR=2.71; mediolateral low frequency wavelet: p=0.028 IQR-OR=2.73). Significant changes over time occurred for both clinical measures and 22/28 COP variables (p<0.001-0.038), with large responsiveness estimates (median-based effect size >0.80) for 7/28 COP variables. Velocity, RMS velocity and higher frequency wavelet variables tended to demonstrate greater responsiveness.

Conclusion(s): Static balance COP variables had mostly insignificant or low associations with dynamic balance and gait speed following stroke. While changes over time in several COP variables were larger than the clinical tests, they were less strongly predictive of falls. Measures of signal frequency and complexity may provide value over more common COP variables and warrant further exploration.

Implications: Based on these findings, it is not recommended to use static balance COP variables over other easy-to-implement clinical tests of dynamic balance for falls prediction post stroke. However, COP variables may be useful in detecting changes in balance performance over time and providing insights into mechanisms of recovery.

Funding, acknowledgements: The Singapore arm of the study was partly funded by Singapore General Hospital Research Grants.

Keywords: Balance, Technology, Assessment

Topic: Neurology: stroke

Did this work require ethics approval? Yes
Institution: (1) Epworth HealthCare; (2) Singapore General Hospital
Committee: (1) Human Research Ethics Committee; (2) Institutional Review Board
Ethics number: (1) 643-14; (2) 2015/2010

All authors, affiliations and abstracts have been published as submitted.

Back to the listing