Interrater Reliability of an Observational Rating Scale and Video Analysis of Yoga Poses

Context: Yoga is increasingly popular, not only as a form of recreational exercise but also as a physician-recommended intervention for health conditions. While serious adverse effects accompanying yoga practice are rare, poses that involve upper-extremity weight-bearing have a high risk of discomfort. To better understand factors contributing to adverse effects, there is a critical need for robust instruments that objectively evaluate pose performance. The purpose of this study was to assess the interrater reliability of an observational scale developed to assess the alignment of 3 yoga poses. Design: Cross-sectional experimental study. Methods: Thirty-eight individuals were given standardized instructions and performed 3 poses (Downward Dog, Plank, and Side Plank). Lateral videos were rated by 2 raters. A rating scale evaluating the alignment of 7 regions was developed by the study team with input from yoga teachers. Descriptive statistics were used to summarize the percentage of subjects showing ideal alignment and deviations. Interrater reliability was quantified using Cohen kappa coefficient (κ). Results: In Downward Dog, the prevalence of ideal alignment was 20%, 28%, and 37%, at the neck, shoulder, and back, respectively; κ ranged from .44 to .69. In Plank, the prevalence of ideal alignment was 31%, 45%, and 54% at the neck, shoulder, and back, respectively; κ ranged from .47 to .95. In Side Plank, the prevalence of ideal alignment was 16, 41%, and 24%, at the neck, shoulder, and back, respectively; κ ranged from .20 to .84. Conclusion: The observational scale found a high prevalence of deviations, and demonstrated fair to substantial interrater agreement.