Abstract
Background Current literature views safe gait as a complex task, relying on motor and cognitive resources. The use of virtual reality (VR) in gait training offers a multifactorial approach, showing positive effects on mobility, balance, and fall risk in elderly people and individuals with neurological disorders. This form of training has been described as a viable research tool; however, it has not been applied routinely in clinical practice. Recently, VR was used to develop an adjunct training method for use by physical therapists in an ambulatory clinical setting.
Objective The aim of this article is to describe the initial clinical experience of applying a 5-week VR clinical service to improve gait and mobility in people with a history of falls, poor mobility, or postural instability.
Design A retrospective data analysis was conducted.
Methods The clinical records of the first 60 patients who completed the VR gait training program were examined. Training was provided 3 times per week for 5 weeks, with each session lasting approximately 1 hour and consisting of walking on a treadmill while negotiating virtual obstacles. Main outcome measures were compared across time and included the Timed “Up & Go” Test (TUG), the Two-Minute Walk Test (2MWT), and the Four Square Step Test (FSST).
Results After 5 weeks of training, time to complete the TUG decreased by 10.3%, the distance walked during the 2MWT increased by 9.5%, and performance on the FSST improved by 13%.
Limitations Limitations of the study include the use of a retrospective analysis with no control group and the lack of objective cognitive assessment.
Conclusions Treadmill training with VR appears to be an effective and practical tool that can be applied in an outpatient physical therapy clinic. This training apparently leads to improvements in gait, mobility, and postural control. It, perhaps, also may augment cognitive and functional aspects.
Gait disturbances are prevalent in aging and contribute to an increased risk of falls, reduced mobility, and poor quality of life.1 The result is a major health problem. Impaired gait and increased fall risk are likely to stem from complex etiology and are frequent in numerous pathologies and neurodegenerative conditions. Approximately 30% of community-dwelling adults over the age of 65 years fall at least once a year.2,3 Among patients with neurodegenerative disorders, falls are even more frequent, with annual incidence rising to 60% to 80%.3,4 The consequences of these falls may be severe, leading to institutionalization, loss of functional independence, disability, fear of falling, depression, and social isolation.5
Normal and safe mobility depends on intact sensory and motor systems, but there is a growing body of research that specifically links the cognitive subdomains of attention and executive function to gait alterations and fall risk.6–8 Executive function apparently plays a critical role in the regulation of gait, especially under challenging conditions where decisions need to be made in real time and constant adaptation is required to manage internal and external factors.9 External factors can include, for example, obstacle crossing or attending to multiple tasks during walking. Both of these tasks heavily rely on the availability of ample cognitive resources due to the need for motor planning and processing of visual stimuli.10,11 The performance during more demanding daily activities, such as walking while performing a simultaneous task (ie, dual tasks or multitask) or obstacle negotiation, plays a key role in the safety and well-being of a variety of individuals with motor and cognitive dysfunctions.12–15 Thus, therapeutic interventions that focus on a combined motor-cognitive approach may improve gait and decrease the risk of falls in older adults.
During the past 15 years, virtual reality (VR) has been used as a research tool for the provision of motor cognitive interventions. Virtual reality technology may enable individualized repetitive practice of motor function, graded in accordance with the needs and level of ability of the person while engaging and stimulating cognitive processes.16 Reports on the use of VR for training of balance, gait, and fall risk in older adults and individuals with neurological disorders have shown positive effects on walking speed, stride time, and step length as well as in the ability to perform dual tasks and obstacle negotiation compared with training in conventional balance training groups.17–19 Mirelman and colleagues developed a VR system that incorporates treadmill training with virtual obstacle negotiation and investigated its effects among patients with Parkinson disease (PD)20 and elderly idiopathic fallers.21 This VR system was designed to provide rich visual and auditory stimuli in an engaging task-oriented training approach that requires planning and decision making as well as motor function. Patients with PD and elderly fallers improved their gait speed and stride length (task-specific effects), but there also was evidence for transfer of training effects and retention, as well as improved cognitive abilities.20,21
These results were obtained from research studies that used fixed protocols and included homogeneous populations based on restricted inclusion criteria. Based on the evidence from our previous research studies and that of other groups, we transformed this approach “from bench to bedside” and implemented the VR training method in an ambulatory clinical service. The service is provided by the physical therapy department at the Tel Aviv Sourasky Medical Center and consists of an intensive progressive training program for individuals with a variety of gait impairments who come for ambulatory physical therapy treatments. Here we report on our experience from the first year of operation of the VR clinical service for improving gait impairments and decreasing the risk of falls. The aim of this analysis was to demonstrate the translational potential of this training approach in an ambulatory clinical service.
Method
The current retrospective data analysis reviewed the medical records of the first 60 patients attending a gait rehabilitation program at the VR clinic in the Tel Aviv Sourasky Medical Center. The use of patients' records for the current analysis was approved by the local institutional human studies committee.
Participants
In the first year of operation, 82 patients attended the gait rehabilitation program at the VR clinic. The data analysis included a total of 60 participants (mean age=72.18 years, SD=10.39; 50% women) who completed the training program and all pretraining-posttraining evaluations. Patients with incomplete data were excluded from the analysis (Fig. 1).
Flowchart describing inclusion in and exclusion from data analysis. Participants who did not complete the training program or had missing data for the posttraining assessment were excluded from analysis. Common reasons for dropping out of the training included lower back pain and the burden of frequent training sessions in conjunction with working hours. Two participants dropped out of the program because they felt the treatment was ineffective. VR=virtual reality.
All participants were referred to the clinic by their physicians. Indications for referral included recurrent falls, fear of falling, complaints of gait instability, or recent deterioration of gait mainly but not exclusively due to neurological etiology.
Participants were eligible for the training program if they were: (1) able to walk independently for at least 5 minutes with or without walking aids, (2) did not have any cardiac contraindication for moderate training intensity, and (3) did not have severe visual loss that could interfere with their ability to see the VR simulation. Participants who could not follow simple instructions and those with dementia (as per Diagnostic and Statistical Manual of Mental Disorders, 4th ed [DSM-IV]22 guidelines) or diagnosed psychiatric disorders were not eligible for the training program.
Training
Participants came to the clinic 3 times per week for 5 weeks and attended a total of 15 training sessions. Each session lasted about 1 hour and included 3 walking bouts, with rest periods between bouts. Training sessions were provided by experienced clinical therapists and were personalized for each participant's needs. The clinical setting included multiple training stations, allowing simultaneous training of up to 4 individuals.
VR Gait Training System
During the training, participants walked on a treadmill with a safety harness to ensure safety and prevent falls but were not provided with body-weight support. Two light-emitting diodes were attached to the lateral side of each participant's shoes to allow for the gait movement to be captured by a motion capture camera situated on each side of the treadmill (Fig. 2A). This setting allowed transferring the movement of the individual's feet to a computer generating the VR simulation. The VR simulation was projected onto a screen in front of the treadmill.
The training system (A) consists of a treadmill with a safety harness and a computer that generates the virtual reality simulation presented on the screen. The 2 motion capture cameras are situated at the sides of the treadmill and are connected to the computer. Two different virtual obstacles were used. The puddles (B) require the participants to increase their step length and maintain clearance. The hurdles (C) require adaptation in the vertical plane and an increase in step clearance.
Virtual Simulation
The virtual environment (VE) simulates an obstacle course situated along different pathways in an outdoor scene. The various pathways differ in duration, number of intersections, and challenging segments, which include turns and walking on a bridge over a river.
The different virtual obstacles require negotiation in 2 planes: (1) vertical, to increase step clearance, and (2) horizontal, to increase step length (Figs. 2B and 2C). Difficulty levels were graded into 5 levels, ranging from “very easy” to “very hard,” that represent different parameters of obstacle size and frequency of appearance. Obstacles were presented on one side (ie, left or right) of the road, requiring the participants to plan ahead, adapt their steps, and select the correct negotiation strategy to avoid a collision.
To increase the level of difficulty, the VE properties also were manipulated to include a change in the environment's visibility by presenting different lighting conditions (ie, morning, noon, evening, and night) and by adding 3 grades of fog conditions. These changes challenged the participants' prediction of the road and obstacles ahead, offering an interfering effect and increasing the demand for motor planning.
Feedback was provided by the simulation and consisted of knowledge of performance (allowing the participants to see their steps), symmetry, and obstacle clearance using immediate visual and auditory cues. Knowledge of results was presented using cumulative scores of the number of collisions and the number of points accumulated by successful passes over the hurdles without collisions or target over shooting. The feedback was augmented by the therapist with cues such as direction or timing of the movement.
Progression
The training parameters were gradually increased from week 1 to week 5, similar to what was previously described.20 Motor load was increased by adapting the treadmill speed, prolonging walking duration, and decreasing the participants' hand support on the treadmill bars while walking.20 The VE parameters were progressed by presenting a wider range of obstacle sizes, increasing obstacle frequency of appearance, disrupting visual clarity, and the addition of virtual distracters. Therefore, cognitive load progression was achieved by challenging sustained and divided attention, planning, and reaction time.
Clinical Evaluation and Assessment
Prior to the training, all participants underwent a physical evaluation performed by an experienced physical therapist. A similar evaluation was performed during the last training session to assess training effects, motor function, and risk of falls. The assessment included the Timed “Up & Go” Test (TUG), which was used to evaluate functional mobility and dynamic balance. The average score from 2 trials was calculated.23 The TUG is a performance-based measure that has classically been viewed as a measure of mobility and fall risk.24 A score above 13.5 seconds on the TUG reflects an increased risk for falls.23 The distance measured during the Two-Minute Walk Test (2MWT)25 was used to assess endurance. Participants were asked to walk back and forth in a well-lit, 30-m corridor at their comfortable speed for 2 minutes. Gait speed was evaluated using a stopwatch, over the middle 10-m increment of straight line walking in the beginning of the 2MWT. The Four Step Square Test (FSST), a validated measure of fall risk in older adults,26 was used to assess overground obstacle negotiation in a subset of participants (n=15), as this test was added to the clinical evaluation routine at a later stage.
Data Analysis
Data was examined for normality, and descriptive statistics were extracted for all clinical measures. Data were compared across time (ie, before versus after the 5 weeks of training) using paired t tests or the Wilcoxon signed rank test as appropriate. Changes in the outcome measures are reported as mean differences and 95% confidence intervals. Group comparisons were conducted using one-way analysis of variance adjusted for multiple comparisons using Bonferroni correction. We used IBM SPSS version 21 (IBM Corp, Armonk, New York) for all analyses with an alpha level of .05.
Results
Data analysis was based on the first 60 participants who completed the training program and had full records of the clinical assessment before and after the training. Participant characteristics are shown in Table 1. The cohort was heterogeneous and included mainly patients with extrapyramidal disorders (ie, PD), individuals poststroke, those with high-level gait disorders (HLGDs), and elderly idiopathic fallers. The frequency of the different etiologies is presented in Figure 3.
Participant Characteristics (N=60)
Summary of the underlying diagnosis for referral to the virtual reality clinic. The group was heterogeneous in diagnosis and included different pathologies. The majority of participants were patients with Parkinson disease and individuals poststroke. Other pathologies included high-level gait disorders, multiple sclerosis, peripheral neuropathy, vestibular pathology, traumatic brain injury, and above-knee amputation.
Participants excluded from the analysis due to incomplete data showed similar baseline characteristics, including age (P=.20), sex (P=.80), education (P=.67), fall history (P=.72), and gait speed (P=.14). The excluded participants attended an average of 12.64 sessions (SD=3.08) within an average of 5.12 weeks (SD=2.28), whereas participants with full data completed 15 sessions within an average of 5.14 weeks (SD=0.85). Although the number of sessions attended were significantly different between the 2 groups (P=.002), the total duration of participation was not (P=.973).
Average time to complete the TUG decreased by 10.3%, from 15.33 seconds (SD=6.05) to 13.47 seconds (SD=4.85) (P<.001), indicating a significant reduction in fall risk. Significant improvement was observed in endurance, as measured by an increase in the distance walked during the 2MWT (P<.001), and in decreased time to complete the FSST (P=.041). Gait speed also improved; however, this change was not significant. Table 2 presents the results of the assessment measures.
Scores of Assessment Measures Before and After Traininga
Training effects were compared between patient subgroups, based on the 4 main diagnoses that lead to referral to the VR clinic (Fig. 4). The 4 groups were: (1) participants with extrapyramidal disorders (n=24), elderly idiopathic fallers (n=6), participants poststroke (n=10), and participants with HLGD (n=6). No significant interaction effects were found for the TUG (P=.961) or gait speed (P=.476). However, a significant time × group interaction was observed in the distance walked during the 2MWT (P=.004). In post hoc tests, the elderly fallers had significantly larger improvements in their endurance test compared with the extrapyramidal group (P=.004) and the HLGD group (P=.007), covering, on average, an additional 31.82 m (SD=8.70) during the 2MWT after the training.
Training effects on functional mobility, gait speed, and endurance were compared among different subgroups based on the underlying pathology. The groups were participants with extrapyramidal disorders (EXP group, n=24), elderly idiopathic fallers (IF group, n=6), participants poststroke (cardiovascular accident [CVA] group, n=10), and participants with high-level gait disorders (HLGD group, n=6). Groups did not differ on the TUG (P=.86) or gait speed (P=.743). However, a significant interaction effect was observed in endurance, as measured with the Two-Minute Walk Test (2MWT) (P=.004). The IF group showed a significantly larger improvement in the distance walked compared with the EXP group (P=.004) and the HLGD group (P=.007).
After completing 15 training sessions, participants reported high satisfaction from the training. Forty percent of the participants expressed their intention to attend additional maintenance training sessions, which are held once a week. Subjective anecdotal reports by the participants and by their caregivers reflected that the participants were able to “walk better,” “pay more attention to hazards,” and “focus better” and, in general, were “feeling safer while walking outdoors.” The training itself was regarded as highly engaging, presenting a challenge within a game context, which contributed to high motivation and to 95% patient adherence to the program.
Discussion
To our knowledge, this is the first time that VR has been used as a tool in physical therapy to enhance gait and functional mobility in a diverse group of adults in a clinical service. The VR gait training uses innovative technology that was originally investigated as a research tool and transitioned into a therapeutic tool. This report gives an added value to the current literature, demonstrating the applicability and efficacy of the previously investigated method in the context of an ambulatory clinical service addressing a heterogeneous group of individuals.
After 5 weeks of intensive treadmill training with VR, the participants performed the TUG and the FSST faster, suggesting improved functional mobility. The average change in TUG scores was consistent with the minimal clinically important difference described in the literature,27 and the average time to perform the TUG dropped below 13.5 seconds after training, reflecting a decrease in fall risk at the group level. However, as the data collected for this study were part of a routine clinical service, information on the effects on future fall frequency in this cohort is not available.
In contrast, gait speed did not reach a statistically or clinically significant change28,29 after training. This finding could be explained by the fact that the goal of the training was not to improve usual walking speed but rather to improve gait under complex conditions and to improve obstacle negotiation. Indeed, the improvement in obstacle negotiation, as observed in the FSST, directly relates to the task-specific training. On the other hand, the changes in TUG scores reflect a transfer of enhanced abilities to an untrained functional task. This fact further highlights the intricacy of what led to improvement in function-based measures. Because TUG performance has been associated with cognitive function,30–32 this finding suggests the interesting possibility that training affected cognitive function, in addition to motor abilities.
Indeed, the training with VR differs from usual treadmill training, as it contains cognitive aspects of planning, with constant adaptation and shifting of attention under challenging motor conditions. As a combined approach, it promotes motor learning through problem solving, thus enhancing executive function.20,33 Previous studies have demonstrated evidence for transfer to cognitive abilities (ie, significant improvements in performing the Trail Making Test, parts A and B, and significant reduction in dual-task cost while performing complex tasks after the training).20,21 Unfortunately, we did not directly assess cognitive function or retention effects in this cohort. Tests of cognitive function have recently been added to the current clinical service protocol and will be further explored.
As in previous studies using VR gait training,20,21 the majority of patients referred to the clinic had a background of neurological pathology. Nevertheless, other participants also improved during the training and gained substantial achievements in gait quality, confidence, and mobility. The program is individually tailored and contains various adjustable features, which allow extending the range of patients who can benefit from this training and personalizing the training to fit individual needs. The use of different virtual obstacles, for instance, can promote greater clearance and increased step length to improve gait pattern in a graded progressive manner. The VE enables challenging training in a functional context while maintaining patient safety, which is valuable for the patient and the trainer. In addition, the engagement in the training may result in longer duration of actual performance. This improvement, in turn, may lead to higher muscular and cardiovascular endurance levels. A future analysis of a larger cohort, including participants with additional musculoskeletal pathologies, may help to identify further indications for this program.
A limitation of the study is that objective cognitive assessment was not used. Therefore, our speculation regarding the cognitive effects of this training is based on previous work and remains restricted. As noted, this limitation has been rectified, and the relevant data are being collected as part of the clinical service routine. In addition, formal satisfaction questionnaires were not used. However, the minimal dropout and the drive to continue the training reflect high motivation and adherence to the training. The participants' positive attitude toward the training serves as an important impetus for this unique exercise program aimed at fall risk reduction. Another limitation of this initial experience is that analysis was not planned in advance and used retrospective data collected as part of the VR clinic routine. Therefore, we did not have an active control or comparison group, and our ability to assess the added value of the VR, compared with traditional approaches, is partial and based largely on previous work.
Our plan is to expand the current clinical assessment, instrument the physical evaluation to better assess subtle kinematic changes, and include cognitive measures. This plan might provide better insight into the training effects, help present more feedback to the patients, and fine-tune the training program. In parallel, we are conducting a multicenter randomized controlled trial with more than 300 older adults to compare the effects of treadmill training with VR and treadmill training alone.34 The results of this trial should provide information as to the unique advantages of the VR training and help to directly assess retention and the potential of using VR to improve mobility and reduce fall frequency in a diverse group of older adults.
In conclusion, the findings in the current work demonstrate that an innovative VR gait training program can serve as a practical and effective clinical service whose aim is to enhance gait in a variety of patients with gait instability. This VR gait training program can be used as a clinical service provided by physical therapists.
Footnotes
Dr Giladi, Dr Hausdorff, and Dr Mirelman provided concept/idea/research design. Ms Shema, Dr Hausdorff, and Dr Mirelman provided writing. Ms Shema, Ms Brozgol, Ms Dorfman, Ms Maidan, Ms Sharaby-Yeshayahu, Ms Malik-Kozuch, and Dr Mirelman provided data collection. Ms Shema, Ms Brozgol, Ms Dorfman, Dr Hausdorff, and Dr Mirelman provided data analysis. Ms Yannai, Dr Giladi, and Dr Mirelman provided project management. Dr Giladi provided fund procurement, participants, and institutional liaisons. Dr Giladi and Dr Hausdorff provided facilities/equipment and consultation (including review of manuscript before submission). Dr Mirelman takes full responsibility for the analysis and interpretation of the data and the conduct of the work. The authors thank Mr Gal Sasson for his help in the initial design of the VR system and Mr Eran Gazit for his technical support.
The retrospective analysis of patients' clinical records was approved by the Ethical Committee of Tel Aviv Sourasky Medical Center.
This research was supported, in part, by the European Commission (FP7 project V-TIME- 278169) and by the Tel Aviv Sourasky Medical Center Grant of Excellence.
- Received July 15, 2013.
- Accepted April 20, 2014.
- © 2014 American Physical Therapy Association
References
Issue highlights
