Accuracy and reliability of temporal gait measurement

This study compared measurements of the temporal gait parameters obtained from videotape using a field counting technique with those obtained from footswitches and force plates. Five subjects traversed a walkway 3 times for each of the 3 conditions, barefoot, in outdoor shoes and in training shoes. Each of the 4 footswitches was connected to a Light Emitting Diode (LED) and these were placed in view of one video camera, a second camera was used to obtain a sagittal plane view of the subject crossing the force plates. The video system had a playback rate of 50 fields/s. The signals from the two cameras were mixed and recorded as a split screen image. The signal from a time code generator was later added to this tape providing information on time to 1/100 s. From the videorecording, 4 individual raters determined the times of heel contact and toe-off for both feet as they crossed the force plates. Inter-rater reliability of detection of the temporal phases of walking was investigated using IntraClass Correlation analysis (ICC) and description of percentage of agreement between raters. Footwear, particularly the training shoes, was found to attenuate the signals from the footswitches leading to differences typically of the order of 0.04 s between the durations of the temporal gait phases calculated from the forceplate data. The technique for visually detecting the time of the key events of make and break of foot/floor contact from videotape was found to be highly reliable with, for example, the barefoot condition demonstrating 95% rater agreement values to within 5% of the time of total support.