This is the first study to examine whether it is possible to modify the normalized ML motion of the axial skeleton during walking gait in people with PD. Our findings support the primary hypothesis and demonstrate that people with PD can decrease ML trunk lean during gait with the assistance of visual biofeedback. The differences observed between the Baseline and Intervention conditions are in line with previous literature that found individuals with PD are able to utilize visual biofeedback to modify trunk position and lateral swaying motions during standing [18, 19].
ML trunk lean was also reduced relative to Baseline in the Post-intervention condition.
This provides evidence of short-term retention of the biofeedback-induced adaptations. No retention of effect was observed during the one-week Follow-up condition however. These findings suggest that one intervention session was insufficient to allow for notable skill retention in individuals with PD. Whilst skill retention in a young healthy population is achievable following a single biofeedback session [
20], it has been reported that repetition over several sessions (e.g. weeks or months) is required for people with PD [
21,
22]. Future research may seek to determine whether longer lasting effects can be achieved with greater exposure to the biofeedback method over an extended time period.
Despite the efficacy of the intervention for reducing ML trunk lean, the provision of biofeedback also resulted in changes in kinematic and spatiotemporal outcomes that may be suggestive of decreased mechanical stability and thus an increased likelihood of falls. Of note, gait velocity and stride length decreased during the Intervention condition relative to all other conditions. As gait velocity is the result of distance covered per unit of time, concurrent decreases in stride length were not surprising and may be suggestive of a decrease in stability [3]. Decreased dynamic stability was also reflected in a decreased COM to base of support distance during the Intervention condition, indicative of greater ML motion of the whole-body COM. Given the influence gait velocity has on several measures of walking stability, the biomechanical consequences of biofeedback interventions on gait velocity and subsequent measures of stability must be given further consideration.
Normalized ML trunk and pelvis motion (i.e. that divided by gait velocity) increased from the Baseline to Intervention condition, before decreasing again during the Post-intervention and Follow-up conditions. Contrary to this, there was no difference in absolute ML trunk and pelvis motion between Baseline and Intervention conditions. Thus, it appears likely that the observed increase in normalized ML trunk and pelvis motion during the Intervention condition occurred primarily as a result of the decrease in gait velocity during this condition. The reduction in gait velocity and subsequent effect upon normalized ML motion measures may be explained by the use of a cueing strategy that sought to provide individuals with greater spatial awareness. As PD individuals are often found to exhibit a decreased postural reserve (i.e. muscle strength, sensory motor integration and higher-level cortical control), increased attention is required to maintain posture and stability. Re-weighting of attention to a separate motor or cognitive task diminishes the attention that can be assigned to controlling mechanical stability and may result in inhibition of the automaticity of gait [23]. In conjunction, a deliberate reduction in gait velocity may have been utilized to allow greater time to adequately process the additional visual information prior to making a postural adjustment. Similar re-weighting of attentional resources has been noted in dual-tasking research where individuals often prioritize all elements of the task equally, rather than giving greatest importance to mechanical stability [24, 25].
It is interesting to note that the increases in normalized ML trunk and pelvis motion were not accompanied by a significant change in normalized ML head motion. These findings may be indicative of successful dampening of motion by the pelvis and trunk, preventing unfavorable ML motion of the head [26, 27]. While both greater head and pelvis motion have been linked with increased falls risk in PD [3], they are strongly correlated with each other. Thus, it is not known which of these measures, or potentially both, are causative of the observed increased falls risk. Regardless, as there was no decrease in either of these measures with the intervention, an alternative approach should be considered in any attempt to translate the observed reduction in ML trunk lean to normalized ML head and pelvis motion.
Visual biofeedback proved able to successfully decrease ML trunk lean, yet the influence of this approach upon falls risk is less clear given the observed changes in ML pelvis and trunk motion. A revised biofeedback intervention approach may therefore be necessary to achieve the desired gait modification, with the consideration of biofeedback effect on gait velocity of importance. Use of a treadmill would allow gait velocity to be controlled across conditions, potentially removing the influence of gait velocity changes on normalized ML motion. This form of intervention would allow for greater exposure to the biofeedback, which may enhance the potential effects of the intervention. The use of treadmill-based interventions is supported by prior literature, which shows it can successful modify step length and gait velocity [28, 29].
There are limitations to this study that must be acknowledged. First, it was beyond the scope of this research to restrict individuals to meet specific gait impairments. Individuals with greater baseline ML motion may have had greater scope to decrease ML motion, potentially impacting the observed responses to the biofeedback. Nevertheless, given the known association between greater ML motion and falls risk, any decrease to ML motion has the potential to benefit all individuals with PD, particularly considering the high incidence of falls in this population. Second, the biofeedback employed a target reduction in trunk lean of 30%. This is less than biofeedback studies in otherwise healthy populations that targeted reductions of 50 and 80% [7, 20]. As the current study was targeting a movement pattern associated with falls in a population at high risk of falling, based on pilot investigations, we considered the 30% target to represent a significant challenge for participant’s without unduly elevating falls risk. It should be noted that a larger target reduction may have elicited a greater change in our outcome variables. Third, despite findings suggesting it was feasible to modify absolute ML head motion through changes to ML trunk lean, resultant changes to falls risk cannot be assumed without prospective falls analysis. The current study serves to inform future prospective research by reporting the immediate and short-term effects of the intervention on biomechanical factors associated with falls risk in PD.