Action observation with motor simulation of reactive stepping: A randomized study in older adults with a history of falls

Background: Perturbation-based training improves reactive stepping responses to prevent falling following a loss-of-balance. As there is currently no safe and feasible method for home-based practice, this randomized study investigated whether action observation with motor simulation (AOMS) of balance recovery improves reactive stepping in older adults with a history of falls. Additionally, we evaluated whether effects differ between AOMS of a human actor in the same experimental context or of an avatar in a virtual world. Methods: Seventy participants with a history of falls (68.3 ± 5.2y/o;52f) were subjected to 20 balance perturbations eliciting backward reactive steps. The AOMS group was tested after simulation of 20 reactive steps demonstrated by either a human actor (HumanAOMS) or a virtual avatar (AvatarAOMS). The control group was tested without prior observation. The primary outcome was reactive step quality, quantified as the leg angle at stepping-foot contact. Results: Differences between groups in the first perturbation trial were not significant. Upon repeated trials, both AOMS groups improved reactive step quality substantially faster than the control group. AOMS participants required on average five repetitions to achieve a reactive step quality that was no longer different from final performance in the last trial, whereas the control group needed ten. Both HumanAOMS and AvatarAOMS yielded similar gains. Conclusions: The lack of effect in the first trial suggests that AOMS alone may not be sufficient for preventing real-life falls in this population. A next step would be to investigate whether this could be achieved by combining brief real perturbation practice with AOMS.