Abstract
OBJECTIVES: The reliability of observational gait analyses is generally considered poor or moderate. This is considered due to the drawbacks in the experimental designs and measurement methods used in related studies. This study examined the reliability and validity of physical therapists' (PTs) visual joint angle estimation in gait analysis with a rigorous design. METHODS: Twenty PTs were divided into 2 groups of 10 each based on their experience: <10 years and >10 years. They observed videos and still images of patients with gait disorders and visually estimated the joint angles at specific points during the gait cycle. Reliability was assessed using the intraclass correlation coefficient, and validity was assessed by a correlation analysis, which compared their estimates with reference values obtained from the digital video analysis. RESULTS: Intra-rater reliability for visual estimation using video tended to exceed 0.6 for most indicators, whereas inter-rater reliability ranged from 0.04 to 0.59. For validity, correlation coefficients indicated an accuracy of 0.06-0.62 for video and 0.08-0.68 for still images. CONCLUSIONS: Although intra-rater reliability tended to be higher than inter-rater reliability, neither reached levels considered sufficiently reliable, and validity showed only limited agreement with reference values. These results highlight the limitations of visual estimation in joint angle assessment. Future efforts should focus not only on improving observational consistency but also on improving agreement with reference values through corrective training using still images.