Balance Training in Virtual Reality Promotes Performance Improvement but Not Transfer to Postural Control in People with Chronic Stroke

Introduction

The increasing number of people living with the aftermath of stroke has stimulated the search for novel tools to improve the efficacy of poststroke motor rehabilitation.^1,2

Motor rehabilitation can be characterized as the process of “relearning” how to move to respond satisfactorily to the demands of daily living. One fundamental aspect of motor learning is the capacity to transfer the ability obtained by training to another similar movement, context, or task, saving time and resources spent on movement improvement. ³

The capacity to transfer is a result of the consolidation of gains from one trained task to another similar task. The transfer efficiency is proportional to shared components or characteristics between both tasks.^3–6

Among the several motor sequels of a stroke, the impairments in balance control are considered a crucial obstacle for recovery of independence in daily living activities, return to social participation,^4–9 and consequently, the maintenance of quality of life.

Difficulties in weight bearing through the paretic leg due to the sensory motor alterations in the contralateral side of the neural lesion is considered a main factor for balance control impairment.^4,10,11 In fact, asymmetric weight bearing in standing in people with stroke is associated with disruption in the compensatory balance response, gait, and stair climbing.^9,12,13

The recovery of symmetric weight distribution and weight-shifting ability has been reported as an essential goal for the rehabilitation process for gait and balance improvements in people with chronic stroke.^4,12–15

Several physiotherapy techniques have been used to achieve this purpose, from simple shoe adaptation and strengthening exercises to force platform training. ¹⁶ However, the results from these studies were not enough to build a strong evidence to support the recommendation for clinical practice.^17–21

Thus, the continuous search for new strategies to improve training conditions to facilitate the learning process behind the recovery of symmetric postural control after a stroke is an important goal of physiotherapy. To this end, virtual reality (VR) has been proposed as a novel therapeutic tool for motor rehabilitation.^22–24

Key elements of motor learning are associated with VR therapy such as a large number of repetitions, movement with goal-oriented tasks, external focus of attention, and motivation and feedback about performance.^22–29 On the contrary, there is controversy about the real potential for transferring the gains obtained to similar tasks. This point has been largely discussed for motor and cognitive skills, in both healthy people and people with impairments.^30–38

A recent systematic review, ³⁹ based on results from 16 trials, which investigated the effects of VR intervention on lower limb function and balance and gait improvement for people with stroke, concluded that currently there is insufficient evidence that a VR approach is more effective than conventional therapies for these outcomes.

One factor that may have contributed to this conclusion is the lack of transfer of gain obtained with VR to a real environment. In fact, Yang et al., ³¹ in a study with people with chronic stroke, found transfer of gains obtained in treadmill training associated with VR to symmetry in dynamic tasks, but not to static tasks, suggesting that transfer of gains with VR is limited to trained conditions.

Another factor that might limit the evaluation of transfer of the balance gains obtained with VR to real tasks is the number of sessions provided in studies; between nine and 16 sessions.^30,31,40 This amount of training can lead to improvement in factors such as range of motion, strength, and spasticity that might have interference in the evaluation of transfer itself. Then, the smaller number of session training may be a crucial factor to achieve pragmatic evidence about motor transferability.

The evaluation of specific transfer of gains obtained by balance training with VR was the main goal of a study ³⁶ with young healthy individuals, however, even with the considerable use of VR technology in stroke balance rehabilitation, there is no documentation of transfer of gains after short-term training in this population.

Considering that the main premise for use of VR for rehabilitation is the possibility of transferring the gains to activities that may improve the functionality of people with chronic stroke, new studies that investigate this issue are fundamental to establish the efficacy of VR as a therapeutic tool.

Thus, the purpose of this study was to investigate the transfer of expected improvements obtained by balance training with VR to similar balance tasks performed in a real environment in people with chronic stroke.

Therefore, we conducted a controlled randomized trial, investigating the transfer of gains obtained by VR balance training using a platform-based videogame to balance tests performed using a force platform.

The videogame stimulates symmetrical postural control through alternated changing of body weight between lower limbs.^40–46 In addition, we included a group with no intervention to take into account learning effects on the balance tests.

Materials and Methods

Results

There were no significant differences in the demographic and baseline measures between the groups (Table 2).

Analyses of the score in the virtual tasks for the EG at baseline and at end of the study revealed statistically significant effects of the training session for all games. The post hoc test showed statistically significant improvements in the scores for all games (Table 3).

Analyses of the score in lateral/lateral (ANOVA: F_(1.29) = 0.79, P = 0.38, ES = 0.13) and anteroposterior (ANOVA: F_(1.24) = 0.04, P = 0.82, ES = 0.05) directional control of RWS showed no significant effect of time, group, or interaction. Mean and standard deviation values are presented in Table 4.

Analyses of the score in maximum excursion LOS showed no significant effect of time, group, or interaction among them. As expected, the analyses showed only significant effect for the side (ANOVA: F_(1.29) = 27.80, P < 0.001, ES = 0.99). This was confirmed by Tukey's post hoc test that showed significant inferior values for the affected side (P < 0.001) (Table 5).

Analyses of the score in endpoint excursion LOS showed no significant effect of time, group, or interaction among them. Similarly in the above test, the analysis showed only significant effect for the side (ANOVA: F_(1.27) = 26.96, P < 0.001, ES = 0.99), confirmed by Tukey's post hoc test that showed inferior significant values for the affected side (P < 0.001) (Table 5).

Summarizing, there was a significant difference between affected and nonaffected side, regardless of the group and the time, confirming the presence of asymmetrical postural control in all participants.

Discussion

The aim of the present study was to investigate the transfer of the improvements obtained by balance training with VR to a similar balance task performed in an untrained task with similar balance demands in a group of people with chronic stroke.

The main evidence that emerged from this study is that despite significant learning in the games and the similarity of demands of the games and tests, there was no evidence for the transfer of performance improvement with VR to a similar postural control task.

We suppose that several factors could contribute to these results.

The first factor is that the amount of training was not enough to promote a consolidation of learning. Transfer is considered a result of the learning process and can be achieved only after the stabilization of gain, ³ and the consolidation of improvements in the games was the main premise to achieve transfer in our study.

Despite previous evidence that people with chronic stroke have a slower performance improvement in postural tasks when compared with healthy subjects,^59,60 the analysis of game scores in this study definitely showed that the participants were able to improve their performance in all the five trained games. In fact, for three of them (Basic Step, Torso Twist, Tilt City), the people with chronic stroke were able to reach an asymptote of learning rate after the second training session and this can be considered strong evidence of learning consolidation.

Although most of the previous studies, which found positive effects of balance training with VR, have adopted more extensive training than our study, ^{30,31,40,42,43} several studies have shown that transfer of gain was obtained after only one session of balance training based on a real environment.^61–63

We also believe that a long training can change peripheral conditions, such as range of motion or strength that might be an interference for a specific evaluation of transfer.

The second factor to be considered is the difference in training and evaluation contexts. Efficient transfer depends on similarity; shared components between trained and untrained tasks.^3,51,52

We believed that in this study there were a large number of shared components between the trained games and the balance test. Besides the direction and speed of body center movements, both offered external focus of attention and visual and auditory feedback.

However, other factors such as the level of motivation and division of attention probably were greater in the training with games than in the balance test. In fact, as already mentioned, the potential for gain transfer from virtual to a real task has been the most critical point for validation of this kind of training for therapeutic purposes.^{32,33,36–38}

Therefore, despite the motor similarity between the training games and balance test, other associated factors could have impaired the transfer.

The third factor, which could have contributed to the lack of transfer in the present study, is the high complexity of balance control. Balance control is a complex ability that depends on interacting systems such as sensory strategies, cognitive processing, orientation in space, and postural responses. ⁶⁴

Despite this complexity, people with chronic stroke are able to improve their balance control through specific and challenging therapy. ⁶⁵ Considering that balance training in a VR offers several additional factors to improve the learning process, stimulating the coordination of anticipatory postural response in an appropriated timing,^41,66 which is frequently impaired in people with chronic stroke^9,10,14 and an automatic balance control, we supposed that this kind of training would be able to promote significant improvements even for this complex ability. However, this has not been observed even in young healthy adults. ³⁶ Thus, more studies are needed to validate this kind of training to improve balance control.

Finally, the last factor to be considered is the association between game improvements and compensatory movements.

Several studies based on functional brain activity have shown abnormal excess activation of the motor network in the nonlesioned hemisphere.^67–69 This can explain an individual's tendency to prefer the nonparetic side to perform any task.^25,70

In the present study, during the first session, there was an intervention by a physiotherapist to discourage participants from adopting compensatory movements with the nonparetic side to achieve the best scores in the games.

However, in the other sessions, participants had no external assistance and it was possible that in their best attempt to achieve the game's goal, they used compensatory strategies such as movement of the trunk, which could help the participants to reach the game's purpose. ⁴⁵ This could explain the absence of improvements in the balance control, despite the reliable improvements in the scores of the games. In other words, the compensatory strategies were sufficient to help participants to increase their scores, but not to promote real improvement in the balance control.

Nevertheless, the utilization of trunk compensatory movements in this kind of training should be investigated by further studies.

Small sample size, exclusion of those older than 65 years, or neglect, lack of controlling the compensatory trunk movements may be appointed as limitations of the present study, which should be investigated in furthers studies.

According to our knowledge, this is the first pragmatic study that investigated the transfer in weight-shifting ability in people with chronic stroke.

The possibility of transfer is a basis for the therapeutic use of VR for improving balance control in people with chronic stroke. Several previous studies have concluded that VR intervention is efficient to improve postural control following stroke. However, most of them have adopted clinical measures as outcomes. No doubt, the results from these studies are very relevant; however, they are not enough to exclude the hypothesis that the observed improvements were associated with compensatory strategy development instead of postural control restoration.

Thus, we recommend that the VR environment be used with caution to improve balance control, particularly in terms of weight-shifting ability, in people with chronic stroke due to possible limitations in the ability to transfer the improvements to functional tasks with similar balance demands.