Invited Commentary

Patients with Parkinson disease (PD) have motor and cognitive impairments that lead to a heightened risk of falls. The annual incidence may reach 60% to 80%,^1–3 more than twice that of the general elderly population. Traditional treatment approaches in PD generally have focused on symptom relief to maximize function and minimize secondary complications. Indeed, until recently, the assumption has been that motor learning cannot take place in the presence of impaired basal ganglia.^4,5 Exciting evidence from animal models and patient studies suggests, instead, that the pathways involving the basal ganglia in people with PD may be capable of plasticity^6–9 and that their activity patterns may be partly corrected with appropriate intensive training.^10–12 For example, some studies have demonstrated improvements in gait speed and stride length following treadmill training.^10,13,14 Despite promising initial results, many questions remain. It is not fully clear whether training in patients with PD can transfer beyond the task that was specifically trained, whether long-term retention is possible, or whether the risk of falls can be reduced.^9,15

Yen et al¹⁶ describe a different approach to improving motor control and reducing fall risk among patients with PD. They evaluated the potential of using training using virtual reality (VR) to improve postural control. Forty-two patients were randomly assigned to 1 of 3 study arms: VR-augmented balance training, conventional balance (CB) training, or a placebo control (no training). Scores on the Sensory Organization Test (SOT) (ie, postural sway with eyes open on a firm surface and under conditions that challenge vision, vestibular function, or proprioception) were the outcomes. The main findings were that, after 6 weeks of training, individuals in the CB group performed better than those in the other 2 groups in a sensory condition that had vision occluded (SOT-5) and that the VR group performed better in the sensory condition when vision and somato-sensation were distorted (SOT-6). The SOT-6 generally is considered the most challenging condition.

A major strength of this report is the study design. Randomization into 3 study arms allowed the authors to tease apart the added benefits of VR-based training compared with CB training and with no training. The results suggest specificity of training and a potential benefit of using VR. When considering the advantages of VR-based training, it is important to keep in mind that virtual environments are not uniformly superior to conventional training, and each approach may offer specific elements that are critical components to building a comprehensive intervention. There were, however, no differences among the groups with respect to the dual-task or other sensory conditions, nor were there any long-term effects. The authors acknowledge that the study likely was underpowered, that the dual-task administration could be modified, and that the VR system could be enhanced by providing individualized interventions.

When considering VR applications for treating balance and mobility, we suggest a number of issues that should inform the intervention. These issues include dosing intensity, equivalence between real-world and virtual tasks, baseline assessments, and tailored progressions. Dosing intensity can be guided by findings from previous work in the real world and by the increasing body of work on VR. Yen et al speculated that the intensity of their VR training might not have been sufficient. A total of 6 hours of training over 6 weeks is relatively low compared with the body of work on using VR to improve walking for individuals poststroke¹⁷ and about 30% of the training time used in our work on VR and with individuals with PD.¹⁸ Exercise and treadmill studies in PD also suggest that more is better.^10,12–14

In other VR work, care has been taken to address the important issues of equivalence between virtual and real-world tasks. Addressing these issues is particularly challenging, as activities in virtual environments cannot always be perfectly matched with real-world activities. An excellent example of treatment equivalence can be found in the work of Jaffee and colleagues,¹⁹ who used a head-mounted display to deliver a virtual stepping task that was matched with an identical real-world task. Because the 2 interventions were motorically equivalent, they could tease out the specific benefits of the VR approach. In the work by Yen and colleagues, static balance, dynamic weight shifting, and external perturbations were applied in the CB group, but it is unclear how these 3 categories mapped to the VR games and whether the number of movement repetitions and exposure to the exercises were similar in the 2 groups. If the goal of comparing CB and VR interventions is to sort out mechanisms and the added value of VR training, considerations of treatment equivalence should be taken into account in the study design.

One of the potential advantages of VR-based training is the ability to match the level of training to the individual. This matching frequently can be achieved by having a baseline assessment and task progression. In the study by Yen et al, the CB group had a clear progression. It is not clear, however, whether this progression was comparable for the VR group or how the difficulty was set in the virtual games. Did participants have a test each day that set the thresholds for performance? In studies of lower-extremity training to improve walking for people poststroke,^20,21 baseline performance of range of motion and force generation was set for each session, allowing the clinician to set training targets that were flexible and adaptable. Similarly, in a study that used VR in patients with PD, the treadmill gait speed was reassessed once a week, and the training speed was adjusted accordingly.¹⁸

Task difficulty also was updated continuously. For example, the orientation, size, frequency of appearance, and shape of the targets were manipulated according to individual needs following a standardized protocol designed to achieve a success rate of 80% in order to promote engagement and motor learning and to allow for a graded, but individualized, progression. In VR work designed to improve upper-extremity use in individuals poststroke, algorithms used measurements of proximal and distal movement kinematics to adjust and progress difficulty (online and offline) of tasks.²² In the present work, platform sensitivity and movement direction were varied, but it was not fully clear how decisions about training progression were made or standardized. Periodic reassessment of the patients' abilities or extraction of relevant performance parameters to drive an algorithm should help to maximize progression.

An additional beneficial study design feature in the study by Yen et al is the inclusion of preintervention and postintervention testing in a dual-task condition. Parkinson disease typically is considered a motor disease, but cognitive impairment frequently occurs. Reduced dopaminergic input from the ventral tegmental mesencephalic area to the frontal and limbic regions likely contributes to cognitive-behavioral dysfunction in people with PD,²³ mainly affecting executive function and attention. Diminished executive function and attention negatively affect gait and increase the risk of falls in individuals with PD and older adults.^24–33 Recent evidence has shown that interventions aimed at dual-task training in individuals with PD can show immediate benefits and even some retention effects in both motor and cognitive domains.^18,34–36 In the present study, a laudable attempt was made at assessing the effects of attention on balance control. However, the task chosen was likely not ideal—as noted by the authors, it was too simple and not continuous, and the VR and CB training did not specifically manipulate attention or dual-task abilities.

Virtual environments may address multiple facets of motor behavior and cognition. Thus, when considering future intervention studies that build on the findings of Yen et al, it may be better to create systems that address not only postural sway, but also other capabilities that are critical to balance, mobility, and fall risk among patients with PD. Because safe mobility is not simply a motor task,^33,37–39 VR interventions should take advantage of the simulated environment to concomitantly challenge and train cognitive and motor abilities. A VR-based intervention that targets both of these domains and adequately addresses key issues such as dosing, retention, transfer, and biofeedback may yet prove to be more beneficial than CB training for patients with PD.

• © 2011 American Physical Therapy Association