Observational Gait Scale- Update (3/7/2018)

Posted on: March 7, 2018 | By: cwilck | Filed under: Evaluative, General Information, Observational gait scale, T&M Tools

I have reviewed the previous posts on OGS, and the information remains up-to-date.

Borel S, Schneider P, Newman CJ. Video analysis software increases the interrater reliability of video gait assessments in children with cerebral palsy. Gait and Posture. 2011; 33: 727-729.

Purpose:  This study sought to determine if video analysis software improves the interrater reliability of visual gait assessment, specifically with the Observational Gait Scale (OGS), in children with cerebral palsy (CP).

Study Population: Twelve children (9 males and 3 females) with CP were video recorded in an outpatient rehabilitation clinic. Age of the children ranged from 5 to 14 years old. Five children were diagnosed with bilateral spastic CP and 7 children with unilateral spastic CP. The children were able to ambulate at a self-selected pace over a linear 10 meter walkway. Videos were recorded during ambulation in both the sagittal and frontal planes.

Methods: Twenty of the gait videos were randomly selected to be viewed during this study. Two video observers with differing levels of clinical experience were selected to assess the video recordings: a final year medical undergraduate and a physiotherapist with more than 10 years of clinical experience working with children with cerebral palsy. Both scorers were familiar with OGS before the study, and both had been provided specific instructions on criteria scoring.  Each video was viewed randomly by both scorers, once with Windows Media Player (WMP) and a second time with the video analysis software Dartfish. WMP allowed observers to use slow motion and image pausing, while Dartfish allowed for more sophisticated slow motion and onscreen measurements. Observers utilized the following Dartfish software functions: digital goniometer for measurements of knee and hindfoot position in midstance; timing measurements for initial contact, midstance, and terminal stance; and line drawing for measurements of base of support. Agreement between scorers was measured for each individual video and for overall OGS score, both with and without video analysis software, using weighted Cohen’s kappas.

Interventions: No intervention was applied to participants as part of this study.

Outcome Measure: The OGS was used to score each video based on the following criteria: 1) knee position in midstance, 2) initial foot contact, 3) foot contact at midstance, 4) timing of heelrise, 5) hindfoot at midstance, and 6) base of support. The following criteria were not assessed during the analysis:  7) gait assistive devices and 8) changes in gait over time.

Results: When using WMP to assess one video, the average time was 10 + 2 minutes. When using the Dartfish software, the average time was 18 + 3 minutes to assess one video. Mean total OGS score when using WMP was 13.8 + 4.6 for scorer 1 and 15.1 + 3.4 for scorer 2. Mean total OGS score when using Dartfish was 13.1 + 4.0 for scorer 1 and 13.6 + 3.3 for scorer 2. Interrater agreement was improved with the use of video analysis software for knee (kappa value 0.344 to 0.591) and hindfoot (kappa value 0.160 to 0.346) position during midstance, foot contact (kappa value 0.700 to 0.854) during midstance, timing of heelrise (kappa value 0.769 to 0.835) during terminal stance, and overall OGS score (kappa value 0.778 to 0.809). Little to no change was seen for initial foot contact (kappa value 0.796 to 0.797) or base of support (kappa value 0.366 to 0.371).

Strengths/Limitations: One strength of this study is that the two observers had knowledge of the OGS, but had differing levels of experience with gait analysis. This detail increases the ability to generalize the results of this study to the broad spectrum of expertise that exists in the field of physical therapy.  Additionally, the videos were viewed in random order and the scorers alternated between using WMP and Dartfish first. This prevents the analysis bias of using one video viewing tool over the other. A limitation of the study is that only two scorers were involved in the analysis. Additional scorers should be included to improve the strength of the study.

Conclusion: While the OGS has previously been validated, this study shows that video analysis software can help improve interrater reliability. The OGS items that require direct angle measurements, such as knee flexion and hindfoot position during midstance, saw the greatest improvement in interrater agreement. Items that require precise timing, such as heelrise, generally saw a moderate improvement in interrater agreement, as scorers were able to analyze timing with precision to 20 ms.  Overall, the use of video analysis software can improve the interrater reliability of the OGS without a significant increase in analysis time.

Leave a Reply