THE EFFECT OF INSTRUCTION AND ACTION OBSERVATION ON THE QUALITY OF BALANCE RECOVERY RESPONSES FOLLOWING LARGE BACKWARD PERTURBATIONS

Auteurs: 
Braspenning, Karin
Affiliatie: 
Department of Rehabilitation, Radboud University Medical Centre, Nijmegen, Netherlands
Supervisie: 
Van Bon Geert MSc., Dr. Weerdesteyn Vivian

Falling is a common cause of injuries and reduced quality of life. Especially for populations at risk who often suffer from inappropriate posture control response. The quality of balance recovery responses can be quantified by the leg angle of the first step [1], which increase by repetition of perturbations [2].
Two factors might contribute to the quality of balance reactions. Instruction that supports a good performance is suggested to optimize the learning curve by the feedback it may provide. Weaver et al. showed that an explicit instruction to reach toward a nearby handrail during posterior perturbations resulted in earlier and larger arm reactions in healthy adults and in individuals with PD [3].
Since falls during daily activities are related to unexpected perturbations, improvements at the first trial are highly interesting from a clinical perspective. Action observation (AO) is thought to contribute to a better balance performance on the first trial [4]. Best training effects are expected when AO is combined with motor imagery (MI) [5].
With this study, more insight will be provided concerning the way in which people learn from balance perturbations. This information can be used for optimizing balance training programs.

Methode: 
20 young inexperienced participants were exposed to a sequence of 35 large backward perturbations (T1) which was repeated with one week in between (T2). Ten received the instruction to recover balance with one step, the other ten were not instructed on how to react. For a second experiment, another 20 participants were exposed to 20 identical perturbations within duo-measurements. One participant was instructed to perform AO and MI when watching the balance reactions of another participant prior to their own perturbations. Both received the ‘one-step’ instruction. The perturbations were given by the Radboud Falls Simulator (RFS) which is a moveable platform that can move horizontally in all directions. Participants stood on the platform while it translated in forward direction which resulted in a reaction on a posterior balance perturbation of the participant. The kinetic data of the step response was recorded by two integrated force plates. The 3D positions of skin surface markers (kinematic data) were recorded with the motion analysis system Vicon®. Leg angles of the stepping leg were calculated at first foot contact. In addition, it was noted whether single or multiple stepping recovery responses occurred.
Resultaat: 
Experiment 1: Leg angles gradually increased with repeated trials (main effect of trials p=0.000). The instruction group exhibited on average 5.945 degrees larger leg angles than the no-instruction group (main effect of group, F1,16=5.945, p=0.027, Fig. 1A). This difference remained relatively constant across trials, as indicated by an absent group*trial interaction effect (p=0.507). 88% of all the trials of T1 for the instruction group were performed with a single step compared to 41% for the no-instruction group, but this difference did not reach significance (p=0.272). However, at T2, 99% of the trials were completed with one step for the instruction group compared to again 41% for the no-instruction group, which was significant lower (p=0.012). The no-instruction group needed 3° larger leg angles in reaching the 50% probability of single stepping compared to the instruction group. Experiment 2: In 7 of the 10 pairs, the AO participant showed larger leg angles at the first trial compared to the participant of the control group. The control group showed a mean leg angle of 2.04° ± 8.93° versus 9.17° ± 7.21° in the AO group, which accounts for a difference of 7.14° (p=0.029, Fig. 1B). For the first trial, 50% of the AO group completed in single stepping compared to only 10% of the controls which approached significance between groups (p=0.051). Across all trials, single stepping rates were not significantly different with 95% single step for the AO group versus 76% for the control group (p=0.570).
Conclusie: 
This first experiment shows that instruction contributes to optimizing the way of learning from perturbation based balance training in posterior directions. The effect of more multiple-step balance recovery responses without instruction makes it plausible that a preplanned strategy occurred [6]. This is supported by our finding that the leg angle without instruction has a lower predictive value for single stepping. Another explanation can be that the second steps during responses were not necessary as a compensatory step, but that participants simply felt more comfortable doing so [7]. The lower predictive value for single stepping may also be due to instruction not only having an influence on the leg angle, but on the trunk angle as well. It may be that the trunk angles were less favorable in the absence of instruction which makes it biomechanically harder to react with one step. Action observation combined with motor imagery gives an additional benefit for these balance recovery responses at the very first trial which indicates a promising future within falls prevention. This effect may (at least partly) be explained by a reduced startle reflex. Exaggerated postural responses, likely caused by startle reflex activity counteract the effectiveness of balance recovery [8]. Further research should focus on if the beneficial effects of action observation would generalize across perturbations direction. In addition, it would be interesting to determine the effect when real life observations are absent, in the way of showing balance reactions via a video. This approach would enable providing interventions for improving balance to patients who were otherwise not able to sustain a physical perturbation-based balance training. In conclusion, the results of this study gives more insight concerning the way in which people learn from balance perturbations. This knowledge can be used for optimizing balance training programs.
Afbeelding: