Reliability of 3D gait data across multiple laboratories

The aim of this study was to analyze the repeatability of gait analysis studies performed across multiple trials, sessions, and laboratories. Ten healthy participants (6 male/4 female, mean age of 30, mean BMI of 24 kg/m²) were assessed in 3 sessions conducted at each of the three Centers of Excellence for Amputee Care within the Department of Defense. For each test session, kinematic and kinetic parameters were collected during five walking trials for each limb. One independent examiner at each site placed markers on the subjects. Biomechanical data were collected at two walking speeds: self-selected and Froude speed. Variability of the gait data was attributed to inter-trial, inter-session, and inter-lab errors for each subject. These error sources were averaged across all ten subjects to obtain a pooled error estimate. The kinematic errors were fairly consistent at the two walking speeds tested. Median inter-lab kinematic errors were <5.0° (median 2.3°) for all joint angle measurements. However, the kinetic error differed significantly between walking speeds. The median inter-lab kinetic error for the self-selected speed was 0.112 N m/kg (ICR 0.091–0.184) with a maximum of 0.226 N m/kg. The errors were greatly reduced when the subjects walked at their Froude speed. The median inter-lab error was 0.048 N m/kg (ICR 0.025–0.078, maximum 0.086). These data demonstrate that it is possible to get reliable data across multiple gait laboratories, particularly when gait speed is standardized across testing sessions. A key similarity between sites was the use of identical anatomical segment definitions for the respective gait models.