van Gelder, Linda
VU University Amsterdam, the Netherlands
dr. van der Krogt, Marjolein & prof. dr. ir. Harlaar, Jaap

Cerebral palsy (CP) is the most common physical disability in childhood affecting 2-2.5 children per 1000 live births. Children with CP often walk in a crouch gait pattern, which is inefficient, and may lead to bone deformities and joint problems in the long term. To reach a more optimal gait pattern in these children, more extension in the knee and hip might be desired, which could possibly be achieved by functional gait training [1]. According to previous studies [2-5] it can be concluded that feedback might be a useful technique to enhance gait training. However, it is unknown whether children walking in crouch can use feedback to walk more upright. Before real-time feedback can be used for training purposes, it is necessary to examine whether children with CP, walking in crouch, can react to feedback. Therefore the main aim of this study was to establish whether children with CP can adjust their gait pattern to real-time feedback in order to reach a more upright gait pattern. If children were capable of reaching more extension, we aimed to determine how this was achieved in terms of specific changes in the gait pattern and we aimed to identify potential responders vs. non-responders to the feedback.

Sixteen children classified with CP (age 6-16; GMFCS I-III) and a flexed-knee gait pattern walked on the Gait Real-time Analysis Interactive Lab, which consists of an instrumented treadmill fronted by a virtual reality screen. All children walked in 3 different conditions of 2 minutes each: regular walking without feedback (NF), feedback on the hip (FH) and on the knee angle (FK), all at set comfortable walking speed. Joint angles were calculated in real-time using the Human Body Model (HBM) [6]. Feedback on their knee or hip angle was presented as a red ball moving across a vertical grey bar. A horizontal grey bar was set as the target to reach. This target was set a few degrees larger than the peak value reached in NF and manually adjusted over time depending on the achievements of the subject. In all trials a three-dimensional gait analysis was performed. The gait profile score (GPS) [7] was calculated to see whether the overall gait pattern improved with feedback. To determine which characteristics could identify potential responders vs. non-responders, all children with CP also received a standardized clinical examination. The effect of feedback (NF, FH and FK) on the different outcome parameters (key kinematic parameters, spatiotemporal parameters, MAPs and GPS) was examined with a one-way repeated measures ANOVA. To define the responders and the non-responders correlation tests were performed between characteristic parameters and the improvement in hip and knee extension.
Almost all children were able to reach more extension in the hip angle in FH (M=5.08°  5.85, p=0.01) and knee angle in FK (M=7.72°  7.10° p=0.002) compared to NF (Figure 1, Table 1). Children used variable changes in their gait pattern in order to reach more extension, but the most common significant changes in FH were: longer strides, more knee extension and more pelvis and trunk tilt movement. In FK almost all significant changes over all participants occurred in the knee parameters itself (Table 1). However, children did not reach a significantly more normal gait pattern (GPS). Out of the correlation tests is followed that children with an overall worse GPS, better selectivity, more motion in the trunk tilt and obliquity angle, less extension in the hip and knee and more dorsiflexion in the ankle angle, all during walking, were more likely to respond to FH. In FK there were no clear responders to the feedback.
Almost all children were able to respond to real-time feedback, resulting in a significant and clinically relevant (>5°) improvement in hip and knee extension. However, children did neither significantly improve, nor worsen in their overall gait pattern. The variable compensation strategies used between patients may be indicative of different underlying causes of the crouch gait pattern. Children with an overall worse gait pattern were more likely to respond to FH. In FK there were no clear responders to the feedback. A limitation of this study was the amount of parameters measured compared to the amount of participants who participated. The best way to determine which parameters would be good predictors for reaching more hip or knee extension would be doing a regression analysis. However, the small amount of children that were measured impeded such an analysis. We did perform many correlation tests so it could be that we did found significant relations by chance. Therefore, the results of the correlation test should mostly be seen as exploration of the data. An improvement for the application would be changing the way of giving feedback. It is possible to let the children follow a graph of the angle, so that they stay within a limit of the norm data, which means that their timing as well as their extension should improve. Further research is needed to investigate this. Even though the application might use some improvements, almost all children were able to adapt their gait pattern and able to respond to real-time feedback, resulting in an improvement in peak hip and knee extension. The various compensation strategies used may be indicative of different underlying causes of the flexed knee gait pattern. These findings show that real-time feedback is a promising tool for advanced gait analysis and might also enable functional gait training in CP.