Test retest reliability of computer-based video analysis for the assessment of postural control in individuals with cerebral palsy
MetadataShow full item record
BACKGROUND: Deficits in the control of posture is an important aspect of CP, as defined by SCPE and children with CP often exhibit activity limitations that are associated with postural control problems. However, there is a lack of adequate tools for assessing postural control in individuals with CP, and there is limited documentation of the measurement properties of the existing tools. AIM: The first aim of this study was to explore if variables obtained using a computer-based video analysis software could be used to assess postural control. The second aim of this student thesis was to explore the test-retest reliability properties of these variables in individuals with CP and TD individuals 8-29 years of age. METHOD: Thirteen individuals with CP and 24 typically developing (TD) individuals, 8-29 years, were asked to stand still while three videos recorded their movements from the side. The movements in the video recordings were quantified, using a computer-based video analysis. RESULTS: In the explorative part of this study we found that the variables calculated by the computer-based video analysis that best correlated with the clinical assessment of postural control in the individuals with CP was the mean value of the centroid of motion (Cxmean) and its standard deviation (CxSD) in the horizontal axis. The centroid of motion is the spatial centre of all movements in the picture. In further analyses, the Cxmean showed the best correlations with GMFM-66, a clinical assessment tool for postural control, while the correlations were low for CxSD. In the second part, I found that when we included all participants in the analyses, the ICC values of Cxmean ranged between 0.89 and 0.93, and of CxSD ranged between 0.92 and 0.93. The ICC values of Cxmean and CxSD were higher with more narrow confidence intervals when two video recordings, each of 30 seconds duration, were included, than when a third video recording of two minutes was included in the calculations. The ICC values of Cxmean and CxSD were nearly identical when applying ICC(1,1) and ICC(3,1). The standard error of measurement (SEM) for Cxmean ranged from 2.2 (4 %) to 3.1 (6 %), expressing a small degree of measurement error. The smallest detectable difference (SDD) for Cxmean ranged from 6.0 (10 %) to 8.5 (15 %). However, the SEM values for CxSD ranged from 0.3 (14 %) to 0.7 (27 %), and the SDD values from 0.7 (40 %) to 2.0 (76 %). The Bland-Altman plots for Cxmean verifies graphically the consistency of the 3 video recordings. The Bland-Altman plots for CxSD verifies graphically the consistency of the measures in TD group, while it illustrates a spread in the values in the individuals with CP. The difference between the CxSD values from the two recordings were larger in the individuals with CP than in those with TD. CONCLUSION: We found that the variable Cxmean from the computer-based video analysis software describes certain aspects of postural control in individuals with CP, 8-29 years. The test-retest reliability of this variable is good. However, more studies are required to further evaluate this method and to explore if other variables of the computer-based video analysis may better describe other aspects of postural control.