Comparison of overground and treadmill vertical ground reaction forces

Abstract

In the process of 3D-gait analysis interpretation, gait deviations in children with cerebral palsy are identified through comparison with reference data of typically developing children (TD). Generally, TD-data are presented based on averaged normalized curves of numerous strides for different ages and walking velocities. In patients however, often only a limited number of strides are available which are compared to group-averaged reference curves.To investigate the consequences of ignoring stride-to-stride variation when averaged normalized curves are used as a reference paediatric dataset. To illustrate implications for clinical practice, we investigated how many individual strides of TD-children would be classified as abnormal, when compared to averaged normalized curves from the reference group, and how this is affected by age and treadmill versus overground walking.Ninety TD-datasets were collected. Children (4-18y) walked on a 10 m-walkway (n = 49) or instrumented treadmill (n = 41). Joint kinematic and kinetic curves and clinically relevant outcome parameters were established. Individual strides were considered abnormal if they exceeded the group average more than 2SD. In addition, the Edinburgh Visual Gait Score, Gait Profile Score (GPS) and stride-to-stride variability were calculated. Generalized estimation equation analyses were used to investigate effects of age, overground/treadmill and their interaction.Of all 2532 analysed strides, on average 28% were classified as abnormal for joint kinematic curves, 50% for moments, and 51% for powers. Younger children showed a greater percentage of abnormal strides, greater GPS and more variability (p < 0.001). The effect of age was similar between treadmill and overground, but variability was lower on the treadmill.Our findings indicate that due to stride-to-stride variability, even in TD-children a substantial number of strides can be classified as abnormal, when compared to group averaged normalized curves. Consequently, in patients, comparing a single stride to such a reference curve may lead to potential overestimation of gait deviations.