Reliability and validity of visual assessments of gait using a modified physician rating scale for crouch and foot contact.

This study evaluates the visual assessment of gait using portions of the Physicians' Rating Scale (PRS). Thirty children with pathologic gait were evaluated "live" and using full- and slow-speed video. Interobserver reliability (weighted kappa) was 0.57 to 0.74 for foot contact, 0.69 to 0.71 for crouch, 0.30 to 0.40 for hip flexion, 0.57 to 0.65 for knee flexion, and 0.42 to 0.52 for dorsiflexion in stance. Intraobserver reliability (comparing the three conditions) was 0.50 to 0.78 for foot contact, 0.71 to 0.80 for crouch, 0.26 to 0.44 for hip flexion, 0.60 to 0.86 for knee flexion, and 0.39 to 0.61 for dorsiflexion. Observers were correct only 12% to 32% of the time when reporting less than 0 degrees of dorsiflexion and 0% to 29% of the time when reporting more than 20 degrees of hip flexion due to overestimation of hip flexion and underestimation of ankle dorsiflexion. These errors could lead some clinicians to presume the presence of contractures that do not actually exist. Visual assessment using the PRS does not appear to accurately measure what it is most commonly used to assess: ankle position in stance.