INTRODUCTION: Different sources of variations, such as electrogoniometer characteristics and procedures, may affect the accuracy and precision of movement measurements during gait. OBJECTIVE: To quantify the variations and compare the effects produced by different sources of variation in electrogoniometric gait recordings: the sensors, procedures (trials) and the knee joint. METHODS: Knee flexion/extension and valgus/varus movements were recorded during gait on the treadmill. The recordings were partitioned into strides and normalized in time using a routine developed in MatLab. Mean curves for the knee during gait were derived from 50 strides, and seven conditions were evaluated: one comparing pairs of sensors; two comparing pairs of different trials (including variations due to sensors); and four comparing the right and left knees (including variations due to sensors and trials). Mean standard deviations of the differences were calculated. To estimate the variations relating to the trial and the knee joint, the compound standard deviations were transformed into variances and split into their components. RESULTS: The variation introduced by pairs of sensors in the same model applied in one trial was smaller than the variation introduced by the same sensor used in two consecutive trials. Furthermore, the variation introduced by the difference between the right and left knees was greater than the variation introduced by the difference between sensors (A and B) and trials (1 and 2). CONCLUSIONS: It is, therefore, preferable to use different sensors in the same data recording (simultaneous) than use the same sensor in two different recordings (consecutive).
movement; measurement error; gait
