GAMES FOR HEALTH JOURNAL: Research, Development, and Clinical Applications

Volume 6, Number 5, 2017

ª Mary Ann Liebert, Inc.
DOI: 10.1089/g4h.2016.0109

The Effect of a Virtual Reality Game Intervention

on Balance for Patients with Stroke:
A Randomized Controlled Trial

Hsin-Chieh Lee, MSc,1,2 Chia-Lin Huang, BSc,1 Sui-Hua Ho, PhD,1 and Wen-Hsu Sung, PhD2
Downloaded by University of Connecticut e-journal package NERL from www.liebertpub.com at 06/14/18. For personal use only.

Abstract

Objective: The aim of this study was to investigate the effects of virtual reality (VR) balance training conducted
using Kinect for Xbox games on patients with chronic stroke.
Materials and Methods: Fifty patients with mild to moderate motor deficits were recruited and randomly
assigned to two groups: VR plus standard treatment group and standard treatment (ST) group. In total, 12
training sessions (90 minutes a session, twice a week) were conducted in both groups, and performance was
assessed at three time points (pretest, post-test, and follow-up) by a blinded assessor. The outcome measures
were the Berg Balance Scale (BBS), Functional Reach Test, and Timed Up and Go Test (cognitive; TUG-cog)
for balance evaluations; Modified Barthel Index for activities of daily living ability; Activities-specific Balance
Confidence Scale for balance confidence; and Stroke Impact Scale for quality of life. The pleasure scale and
adverse events were also recorded after each training session.
Results: Both groups exhibited significant improvement over time in the BBS (P = 0.000) and TUG-cog test
(P = 0.005). The VR group rated the experience as more pleasurable than the ST group during the intervention
(P = 0.027). However, no significant difference was observed in other outcome measures within or between the
groups. No serious adverse events were observed during the treatment in either group.
Conclusions: VR balance training by using Kinect for Xbox games plus the traditional method had positive
effects on the balance ability of patients with chronic stroke. The VR group experienced higher pleasure than
the ST group during the intervention.

Keywords: Stroke, Virtual reality therapy, Videogames, Postural balance, Activities of daily living

Introduction cause traumatic brain injury, fracture, and fall fear and affect
ADL independence and quality of life (QOL).6 Stroke im-

N eurological symptoms occurring after stroke in-

clude abnormal reflex, muscle hypertonicity, attention
deficiency, hemineglect, and sensorimotor functional impair-
poses huge economic and social burden in the long term and
is among the top three diseases responsible for high medical
expenditures in the National Health Insurance.7 Balance re-
ment. Although patients demonstrate functional improvement covery helps patients with stroke to become independent in
followed by recovery, most of them still experience stroke ADL and reduces the burden of medical costs.
symptoms. Stroke affects independence in activities of daily Nonimmersive virtual reality (VR), which has been used
living (ADL) by reducing postural control, resulting in slow in stroke rehabilitation,8 can improve balance ability through
walking speed and short stride length.1–3 the provision of multisensory feedback and repeat practices.
Common balance problems caused by stroke are decrease Compared with traditional methods, the VR method can
in muscle strength and proprioception, higher loads on the greatly increase motivation. Commercial games such as
nonparetic extremity, and increased postural oscillation.4 Sony ‘‘EyeToy’’, Nintendo Wii, and Microsoft Kinect
Moreover, stroke leads to loss of balance and falls in 56% have been used in nonimmersive VR rehabilitation. Speci-
and 18% of patients every month, respectively5; this may fically, Microsoft Kinect does not require a controller and it

1
Department of Physical Medicine and Rehabilitation, Taipei Medical University Shuang-Ho Hospital, Taipei, Taiwan.
2
Department of Physical Therapy and Assistive Technology, National Yang-Ming University, Taipei, Taiwan.

303
 304 LEE ET AL.

captures movements by using a sensor; thus, patients can see are the same as those obtained using traditional methods;
movements in real time and provide feedback immediately. moreover, the effects could be maintained after 3 months.
Rajaratnam et al. demonstrated the static and dynamic bal-
ance effects of VR-related games by using the Wii balance Materials and Methods
board or Microsoft Kinect for patients with subacute stroke.9
Participants
Subramaniam and Bhatt reported that the introduction of
Microsoft Kinect (‘‘Just Dance 3’’) games to 11 patients after The patients were recruited from February 2015 to January
stroke improved the reaction time, movement velocity, 2016. All outpatients of the Neurorehabilitation Unit of
maximum excursion, and number of steps during dance in- Hospital presenting with residual hemiparesis after stroke
tervention.10 Although studies have supported the effec- were eligible to participate in this study. The inclusion cri-
tiveness of these commercial games, the study designs and teria were as follows: (1) age between 20 and 75 years; (2)
evidence have been vaguely studied. A game is designed for having chronicity >6 months; (3) ability to understand game
leisure goals and not for treatment goals.11 No standard instructions; (4) ability to stand for 15 minutes; and (5)
treatment (ST) integrating rehabilitation and home care is having Brunnstrom stage qIII. The exclusion criteria were
available. Furthermore, few studies have investigated the as follows: (1) having a Montreal Cognitive Assessment
Downloaded by University of Connecticut e-journal package NERL from www.liebertpub.com at 06/14/18. For personal use only.

effects of these games on long-term retention.12 score <16; (2) having visual or auditory impairment with the
In this study, we investigated the effect of the Microsoft inability to clearly see or hear the feedback from the game;
Kinect function for Xbox training plus traditional training on (3) having severe spasticity of lower extremity (Modified
the balance ability of patients with chronic stroke. We hy- Ashworth Scale q3); and (4) having other medical symp-
pothesized that the use of Microsoft Kinect games plus tra- toms that could affect movement.
ditional balance training was feasible and enjoyable for Individuals who met all criteria and agreed to participate
patients and it would show improvements in balance and in the study were provided detailed information. Written
participation performance after 6 weeks of intervention, which informed consent was obtained from all patients. The study

FIG. 1. Flowchart of the experimental protocol. During the recruitment process, a total of 95 outpatients were assessed. Of
these, 25 were excluded, 6 rejected the invitation to participate, and 14 could not meet the time. A total of 50 (52.63%)
patients from the remaining sample met the inclusion criteria and were randomly allocated into the VR group (n = 26) or ST
group (n = 24). After 12 training sessions, the patients’ performance was assessed by a blinded assessor. Three patients in the
ST group dropped out during the study because of relocation to another country (n = 1) and medical reasons (n = 2).
Consequently, data from 47 patients, 26 in the VR group and 21 in the ST group, were included in this study. ST, standard
treatment; VR, virtual reality.
 VIRTUAL REALITY BALANCE TRAINING FOR STROKE 305

was approved by the Institutional Review Board of Taipei the motor learning theory of Gentile and selected the game
Medical University Hospital (No.: 201412023). The pro- according to body stability and stationary conditions for
tocol was registered in ClinicalTrials.gov PRS (No.: initial practice (Fig. 2).13
NCT02735265). The regulatory conditions indicate relevant environmental
features that constrain movement execution and may either
Design be stationary (stationary regulatory conditions) or moving
(in-motion regulatory conditions). With the indicator, inter-
A prospective, randomized controlled design was used.
trial variability is used to differentiate between regulatory
We randomized 50 patients into the VR group or ST group
conditions that change between trials (intertrial variability)
through simple random sampling by using a computer gen-
and those that do not (no intertrial variability). Body orien-
erator. After 12 training sessions, the patients’ performance
tation indicates whether an action requires the performer to
was assessed by a blinded assessor (Fig. 1). An independent
move from one location to another (body transport) or not
occupational therapist who was not blinded to the interven-
(body stability). Object manipulation indicates whether an
tion explained the training procedure to the participants and
object has to be controlled during the action performance
provided technical support.
(object manipulation) or not (no object manipulation). Ac-
Downloaded by University of Connecticut e-journal package NERL from www.liebertpub.com at 06/14/18. For personal use only.

cording to Gentile, the easiest skill category can be found at

Intervention
the top left position. Moving either rightward or downward
Both interventions were administered twice a week for 6 in the table renders the skill category more difficult; thus, the
weeks under the supervision of a qualified occupational most difficult skill category can be found at the bottom right
therapist. The VR group received 45 minutes of ST and 45 of the table.
minutes of interactive VR balance-related games. The ST In each session, 15 minutes of warm up and 15 minutes each
group received ST for 90 minutes. for two selected games were set for patients according to their
ability, requirements, and choice. The therapist demonstrated
VR group. The VR game training system comprised a the game control and instructions to ensure that the patients
television (50LB5800 WiFi Smart TV; LG, Seoul, Korea), understood the controlling of the system, which prevented
Microsoft Kinect (Microsoft, Inc., Redmond, WA), and a com- compensatory movements through oral feedback. For safety,
mercial game. The system was applied in the clinical setting. we placed a handrail in front of the participant if required. We
Nine games were selected from the Kinect Sports, Kinect graded the difficulty by modifying movement challenges, en-
Adventures, and Your Shape Fitness Evolved packages durance time, and feedback frequency during the session. The
that are based on common balance problems experienced patients were asked to rest and stretch muscles for a short time
after stroke (Table 1). We analyzed the tasks on the basis of when their hypertonicity interfered with the game control. The

Table 1. Game Analysis in the Virtual Reality Group

Game name Description Requirement
‘‘Darts’’ (Kinect Small missiles are thrown at a circular dartboard. Unilateral upper limb movement with
Sports: Season 2) The player throws three darts per visit at the static standing.
board with the goal of reducing 501 to 0.
‘‘Golf’’ (Kinect The golf club is swung to hit the ball into the hole. Slight bilateral knee squat, bilateral hand
Sports: Season 2) swing with trunk rotation to one side,
weight shifting.
‘‘Bowling’’ Players are required to reach to their left or right Unilateral upper limb movement with
(Kinect Sports) to pick up a ball before swinging their arm weight shifting, stepping to lunge.
forward to bowl, exaggerating the arm motion
to add a spin if required.
‘‘Virtual smash’’ Virtual bricks are randomly displayed from various Unilateral upper limb movement with
(Your Shape: directions, and a punch or kick across the body weight shifting, single-limb weight
Fitness Evolved) must be performed to hit them. bearing.
‘‘Light race’’ There are five buttons on the ground, and each Stepping to four directions, single-limb
(Your Shape: button randomly lights up. The player is required weight bearing.
Fitness Evolved) to step onto the button to darken it and score.
‘‘Space Pop’’ Players attempt to pop the bubbles by touching Bilateral upper limb movement with
(Kinect them, earning adventure pins. To move upward, locomotion and weight shifting.
Adventures!) the players flap their arms, and to stay at their
current height, they hold their arms out to the sides.
‘‘Rally Ball’’ A ball floats in the air in front of the player. The Bilateral upper limb movement, trunk
(Kinect player can use any body part to reach out and leaning for weight shifting.
Adventures!) whack it.
‘‘Table Tennis’’ A lightweight ball is tossed up and hit back and Unilateral upper limb movement with
(Kinect Sports) forth across a table by using a small paddle. weight shifting.
‘‘River rush’’ The players control the raft with their bodies by Unilateral upper limb movement with
(Kinect moving from side to side and jumping. trunk lean and weight shifting or
Adventures!) stepping.
 306 LEE ET AL.

score was recorded after the game for the next goal, and plete the test while counting backward by 3 from a randomly
feedback was obtained at the end of the therapy. selected number between 20 and 100.18 Participation out-
comes included the Modified Barthel Index (MBI)19 for
ST group. Both groups performed 45 minutes of the ST ADL, Activities-specific Balance Confidence (ABC) scale-
rehabilitative protocol, focusing on strengthening, endurance Taiwan20,21 for balance confidence, and Stroke Impact Scale
training, ambulation, and ADL training on the following: (1) (SIS)22 for QOL. Feasibility outcomes that were recorded after
hip flexor and knee extensor strengthening with resistance each session included the modified Physical Activity Enjoy-
progression by using weight bags or a Thera-band; (2) cycle ment Scale (M-PAES)23 and the incidence of adverse events,
ergometer riding with an increase in speed and resistance; (3) namely soreness, motor sickness, pain, injury, and fall accident.
gait pattern and speed correction through a treadmill and
parallel bar; and (4) hand functional training and strategy Data analysis
teaching for feeding, dressing, and toileting.
Data were analyzed using IBM SPSS statistics (version
In addition, the ST group participated in 15 minutes of
20.0). The significance level was set at 0.05. Descriptive
warm-up exercises for stretching to increase soft tissue
analysis was used, and all data are presented as mean –
Downloaded by University of Connecticut e-journal package NERL from www.liebertpub.com at 06/14/18. For personal use only.

flexibility and range of motion and 30 minutes of functional

standard deviation. Demographic and baseline character-
balance exercises that included the following programs: (1)
istics were evaluated using the chi-square test or indepen-
to facilitate the balance reaction through weight shifting
dent samples t-test. Repeated-measures analysis of variance
exercises by standing on an even to uneven surface, such as a
(ANOVA) with time as a within-participant factor and
tilting board; (2) postural transition, including sit-to-stand,
group as a between-participant factor was performed for the
sit down, reaching to different directions, stepping to dif-
balance, ADL, balance confidence, and QOL outcome
ferent directions with weight transfer, and bending the trunk
measures. The Bonferroni post hoc test was used to inves-
forward and side to side; (3) change in the standing re-
tigate differences between groups at the three time points.
quirement, such as a single-legged stance or lunge stance;
For each repeated-measures ANOVA, we present the par-
and (4) increased perception complications through cogni-
tial g2 as a measure of the effect size. Feasibilities reported
tion or upper extremity tasks to improve dual-task attention.
in both groups were compared using independent sample
The difficulty level of the exercises was adjusted accord-
t-tests.
ing to the progression of each participant. The number of
Data were assessed by using the Kolmogorov–Smirnov
repetitions, height of the exercise step, and the weight were
test to confirm the normality. If the data were not normally
progressively modified. Each set lasted for 15 minutes.
distributed, the Friedman two-way ANOVA test was used for
time effect and the Mann–Whitney U test was used for group
Outcome measures effect. Wilcoxon signed-rank test was used for post hoc
The outcome measures were assessed at three time points analysis.
(before intervention, postintervention, and at 3-month follow-
up) by a therapist who was blinded to the training allocation. Results
The primary outcome measure was the Berg Balance Scale
Patients
(BBS).7,14–16 Other balance outcomes included the Functional
Reach Test (FRT)16,17 and Timed Up and Go-cognition (TUG- A total of 50 patients were recruited and randomly allo-
cog) test. In the TUG-cog test, patients were asked to com- cated into the VR (n = 26) and ST (n = 24) groups. Three

FIG. 2. VR games corresponding to Gentile’s skill categories. The task was analyzed based on motion (body stability or
transport) and condition (stationary or in-motion regulatory conditions). According to Gentile, the easiest skill category can be
found at the top left position. Moving either rightward or downward in the table renders the skill category more difficult; thus, the
most difficult skill category can be found at the bottom right of the table.
 VIRTUAL REALITY BALANCE TRAINING FOR STROKE 307

patients in the ST group dropped out during the study. The Discussion
reasons for exclusion and dropping out are presented in In this study, we observed that both groups exhibited
Figure 1. Finally, we included a total of 47 patients (VR partial balance improvement in the BBS and TUG-cog test
group: 26; ST group: 21). Before treatment, demographic after 12 sessions of balance training and this improvement
characteristics and clinical data did not significantly differ was maintained at the 3-month follow-up; however, no sig-
between the VR and ST groups (Table 2). nificant improvement was observed in balance confidence,
ADL, or QOL. In addition, no significant group effects or
Clinical data group-by-time interactions were observed in all outcome
measures. A higher level of pleasure was experienced in the
The clinical data are listed in Table 3. A significant time VR group than in the ST group during the treatment.
effect was observed in both groups in BBS (P = 0.000) and
TUG-cog (P = 0.009). The post hoc analysis revealed sig-
Balance effectiveness of VR game training
nificantly improved BBS scores from the pre- to post-
intervention (P = 0.000) and follow-up test (P = 0.003). The results revealed significant improvements over time in
Downloaded by University of Connecticut e-journal package NERL from www.liebertpub.com at 06/14/18. For personal use only.

TUG-cog significantly decreased from the pre- to post- the BBS and TUG-cog test, thus supporting the effectiveness
intervention (P = 0.003) and follow-up test (P = 0.006). A of VR-based interventions in improving balance, even for a
nonsignificant time effect was observed in both the groups in long time after stroke. We observed an improvement of 2.84
the FRT (P = 0.187). Although both groups exhibited partial points in the BBS and 3.03 seconds in the TUG test, which
improvement in scores, no significant group-by-time inter- are higher than the previously established minimum detect-
action was observed in any scale. able change of 2.5 points24 and 2.9 seconds,25 for the chronic
stroke population.
VR-based interventions involving motor learning and
Feasibility
plasticity principles focus on high-intensity, repetitive, and
The M-PAES scores significantly differed between the task-specific practices.15 VR-based interventions provide
two groups (P = 0.027). Furthermore, the incidence of ad- multisensory feedback during the game, including proprio-
verse events did not differ significantly between the two ception, vestibular, visual, and auditory feedback. VR com-
groups (6% vs. 10%, P = 0.267; Table 4). The details of bines sensory and cognition stimulation that could benefit a
adverse events in both groups are presented in Figure 3. patient’s dual-task ability, similar to that in the equivalent

Table 2. Comparisons of Characteristics Between Virtual Reality-Based and Standard Therapy Groups
Characteristics VR group (n = 26) ST group (n = 21) Value P
Gender, n (%)
Male 16 (61.5) 18 (85.7) 3.393 0.065
Female 10 (38.5) 3 (14.3)
Age (year) 59.35 – 8.95 55.76 – 9.59 1.322 0.193
Education level, n (%)
<High school 4 (15.4) 5 (23.8) 0.664 0.718
High school 15 (57.7) 10 (47.6)
>High school 7 (26.9) 6 (28.6)
Affected side, n (%)
Left 11 (42.3) 11 (52.4) 0.473 0.491
Right 15 (57.7) 10 (47.6)
Etiology, n (%)
Infarction 16 (61.5) 14 (66.7) 0.132 0.716
Hemorrhage 10 (38.5) 7 (33.3)
Site of stroke, n (%)
Cerebral 22 (84.6) 17 (81.0) 1.266 0.531
Cerebellum 0 (0) 1 (4.8)
Brain stem 4 (15.4) 3 (14.3)
Poststroke duration (day) 839.77 – 719.13 653.24 – 589.70 0.956 0.344
First stroke, n (%) 23 (88.5) 21 (100.0) 2.588 0.108
Regular rehabilitation (min/weeks) 64.62 – 61.01 68.57 – 60.85 -0.221 0.826
FIM of walking ability 6.19 – 0.63 6.29 – 0.64 -0.499 0.620
MoCA 24.46 – 4.49 24.05 – 3.97 0.331 0.742
Br. Stages of LE 4.62 – 1.44 4.05 – 1.28 1.407 0.166
MAS of LE 0.85 – 0.93 0.95 – 0.74 -0.427 0.671
Data are presented as mean – standard deviation or number (%). Value indicates t or v2 according to independent sample t-test or chi-
square test.
Br. stages, Brunnstrom stages; MAS of LE, Modified Ashworth Scale of lower extremity; MoCA, Montreal cognition assessment; ST,
standard treatment; VR, virtual reality.
 308 LEE ET AL.

Table 3. Intra- and Intergroup Comparisons at Three Evaluation Time Points

VR group ST group
(n = 26) (n = 21) Pretest Time · group effect Time effect Group effect
Outcome Partial Partial
measure Mean – SD Mean – SD P F P g2 F/v2 P Partial g2 F/U P g2
BBS
Pretest 43.35 – 6.23 43.48 – 6.62 0.945 0.705 0.497 0.015 10.446 0.000a 0.188 0.117 0.734 0.003
Post-test 46.19 – 5.57 45.71 – 6.64
Follow-up 46.31 – 5.80 45.00 – 5.06
FRT
Pretest 21.43 – 7.62 22.05 – 8.27 0.791 3.066 0.051 0.064 1.707 0.187 0.037 0.601 0.442 0.013
Post-test 22.63 – 5.07 21.84 – 7.46
Follow-up 22.48 – 5.87 18.74 – 5.88
Downloaded by University of Connecticut e-journal package NERL from www.liebertpub.com at 06/14/18. For personal use only.

TUG-cogb
Pretest 27.18 – 14.90 32.13 – 24.63 0.585 — — — 9.329 0.009a — 251.0 0.637 —
Post-test 24.15 – 10.87 28.48 – 21.53
Follow-up 23.52 – 10.96 28.67 – 18.73
MBIb
Pretest 87.31 – 13.80 88.10 – 15.69 0.684 — — — 1.411 0.494 — 247.5 0.575 —
Post-test 88.27 – 14.63 86.90 – 12.79
Follow-up 88.85 – 14.58 89.05 – 12.91
ABC
Pretest 57.00 – 24.12 66.27 – 21.32 0.175 1.269 0.266 0.027 0.038 0.963 0.001 0.644 0.528 0.014
Post-test 59.76 – 25.87 63.96 – 19.21
Follow-up 57.44 – 24.54 65.05 – 21.30
SIS
Pretest 61.06 – 16.36 61.71 – 12.51 0.883 0.247 0.782 0.005 1.295 0.279 0.028 0.003 0.955 0.000
Post-test 62.99 – 18.13 63.10 – 11.61
Follow-up 64.56 – 17.41 63.11 – 14.03
Values are presented as mean – standard deviation.
a
p < 0.05 in comparison between pre- and post-treatment according to two-way ANOVA.
b
The nonparametric test was performed because the data did not exhibit a normal distribution. The Friedman two-way ANOVA test was
used for time effect and the Mann–Whitney U test was used for group effect.
ABC, Activities-specific Balance Confidence Scale; BBS, Berg Balance Scale; FRT, Functional Reach Test; MBI, Modified Barthel
Index; SIS, Stroke Impact Scale; ANOVA, analysis of variance; TUG-cog, Timed Up and Go test-cognitive.

physical world.26 In addition, the score performance and a patient can step forward or move to control the game. It
supervising therapist can motivate patients to learn move- involves whole body movement, which is similar to that in
ments.27 Compared with conventional training, VR training the real world, without the use of a controller or standing on a
provides a more stimulating, enriching, and pleasurable en- narrow board.30 The results of this study reveal a time effect
vironment that motivates patients to completely engage in in balance outcomes, but no significant group effect or
the program.28 group-by-time interaction, which may suggest that both
However, we observed no significant results in the FRT. groups improved in a similar manner. The VR group focused
Hung et al.28 and Gil-Gómez et al.16 have reported positive on weight shifting and stepping, which was similar to the
effects in the FRT executed by using the Wii balance board. functional balance training of the ST group. The results
The FRT requires shifting the weight forward as much as support that VR training plus the traditional method and
possible so that the leg cannot be moved, which is similar to traditional method alone were not significantly different in
the requirement of the Wii balance board training.29 Kinect terms of the improvement of balance effect.
for Xbox does not restrict patients to the same place, and a In this study, although the score tended to improve, par-
ticipation outcomes exhibited no significant improvement on
Table 4. Feasibility Comparisons of Virtual our evaluation tool. A commercial game is designed for
Reality-Based and Standard Therapy Groups leisure and not for ADL function.26 Patients with low ADL
function depended on caregivers for a long time. Although
Outcome VR group ST group the balance function could improve, the improvement is not
measure (n = 26) (n = 21) P adequate to achieve independence in ADL.31 ABC and SIS
are related to self-efficiency in patients after stroke, and
M-PAES 31.76 – 2.670 29.36 – 4.45 0.027a balance function, anxiety, and depression may affect these
Adverse events 0.06 – 0.11 0.10 – 0.13 0.267
scores.20 Shin32 investigated the effectiveness of VR training
a
p < 0.05 according to independent sample t-test. and occupational therapy in SF-36 and reported that the VR
M-PAES, Modified Physical Activity Enjoyment Scale. group exhibited improved scores in role limitation because of
 VIRTUAL REALITY BALANCE TRAINING FOR STROKE 309
Downloaded by University of Connecticut e-journal package NERL from www.liebertpub.com at 06/14/18. For personal use only.

FIG. 3. Adverse events in both groups. In total, 19 adverse events were observed in the VR group and 30 adverse events in
the ST group. Soreness of the upper limbs (n = 6), muscle hypertonicity (n = 5), dizziness (n = 6), and leg pain (n = 2) were
observed in the VR group. Soreness of the lower limbs (n = 10) and upper limbs (n = 5), muscle hypertonicity (n = 7),
dizziness (n = 2), shoulder pain (n = 4), and lower back pain (n = 2) were observed in the ST group.

physical function. The scores could improve because of fa- the game to select a game that matched the rehabilitation
miliarity with the game, resulting in remodeling of cognition goal. Some commercial games are not suitable for patients
to perceive improvement in physical function. Patients feel with stroke. Therapists should be aware of the game design
less stressful and engage themselves in the game.32 and should analyze the game to match the stroke rehabilitation
Hung et al.28 reported that some games (e.g., the basic goal. Patients with motion and cognitive impairment may feel
stepping game) may increase the spasticity and fall risk and frustrated during the game, particularly elderly people with
that patients easily used compensated movements to play the cognitive impairment.8 Therapists should supervise and sup-
game initially and then supervising therapists needed to port these patients to encourage compliance.34
guide them to facilitate optimal movement. Adverse events On the basis of the motor learning theory of Gentile,13 we
observed in our study were soreness, hypertonicity, dizzi- analyzed the tasks and added cognition and motion demand
ness, and shoulder pain because of task requirements or in- after patients became familiar with the system control. Con-
dividual symptoms. All the adverse events recovered after sidering that taxonomy-based skill categories present a
patients took rest and did not continue with the next training structure during the transition from simple to complex motor
session. No serious adverse events were observed in either tasks, training program should involve adjusting the difficulty
group because a therapist supervised the safety and adjusted level, providing the adaptation method, and reducing patients’
challenges to prevent complications. frustration.35 For example, patients with attention impairment,
slow responses, or apraxia require more demonstrations,
Intervention model guidance, and simplification of steps to match their abilities.
Although studies have indicated the effectiveness of the
Study limitations
VR training program, difficulties in performing the training
in clinics because of space limitations, unfamiliarity of the Stroke can cause different levels of paralysis such as
therapist with the setting, and maintenance problems are hemiplegia or hemiparesis that result in a high level of het-
encountered.33 We used a commercial game that could be erogeneity in stroke. For ethical reasons, we did not withhold
easily set up and was available in the market. The therapist routine rehabilitation therapy. The lack of tight control on the
was only required to be familiar with the characteristics of specific tasks performed during routine therapy was another
 310 LEE ET AL.

limitation. Because all the patients were in the chronic stage, 9. Rajaratnam BS, Gui KaiEn J, Lee JiaLin K, et al. Does the
they usually received maintenance exercises. Therefore, the inclusion of virtual reality games within conventional re-
compounding effect of routine training might be small. habilitation enhance balance retraining after a recent epi-
sode of stroke? Rehabil Res Pract 2013; 2013:1–6.
Conclusions 10. Subramaniam S, Bhatt T. Does a virtual reality-based dance
training paradigm increase balance control in chronic
VR balance training by using Kinect for Xbox games plus stroke survivors? A preliminary study. Int J Neurorehabil
the traditional method had positive effects on the balance 2015; 2:1–10.
ability of patients with chronic stroke, and the effects could 11. Lange B, Flynn S, Proffitt R, et al. Development of an
be maintained after 3 months. The level of pleasure experi- interactive game-based rehabilitation tool for dynamic
enced by the VR group was higher than that experienced by balance training. Top Stroke Rehabil 2010; 17:345–352.
the ST group during intervention. Because balance function 12. Laver KE, George S, Thomas S, et al. Virtual reality for stroke
improved only partially, higher dose and frequency of re- rehabilitation. Cochrane Database Syst Rev 2015; 1–107.
habilitation should be used in future studies to increase the 13. Wüest S, van de Langenberg R, de Bruin ED. Design
effectiveness of interventions. considerations for a theory-driven exergame-based reha-
Downloaded by University of Connecticut e-journal package NERL from www.liebertpub.com at 06/14/18. For personal use only.

bilitation program to improve walking of persons with

stroke. Eur Rev Aging Phys Act 2014; 11:119–129.
Acknowledgments
14. Kim JH, Jang SH, Kim CS, et al. Use of virtual reality to
This research was supported by the study projects of enhance balance and ambulation in chronic stroke: A
Taipei Medical University Shuang Ho Hospital (103 SHH- double-blind, randomized controlled study. Am J Phys Med
HCP-001). The authors thank their colleagues from the De- Rehabil 2009; 88:693–701.
partment of Physical Medicine and Rehabilitation, Taipei 15. Peters DM, McPherson AK, Fletcher B, et al. Counting
Medical University Shuang-Ho Hospital, who provided in- repetitions: An observational study of video game play in
sights and the expertise that greatly assisted the research. The people with chronic poststroke hemiparesis. J Neurol Phys
authors thank Dr. Tsan-Hon Liou and Dr. Ya-Ru Yang for Ther 2013; 37:105–111.
their comments that greatly improved the manuscript and 16. Gil-Gómez J-A, Lloréns R, Alcañiz M, Colomer C. Ef-
fectiveness of a Wii balance board-based system (eBaViR)
Wallace Academic Editing for editing the manuscript.
for balance rehabilitation: A pilot randomized clinical trial
in patients with acquired brain injury. J Neuroeng Rehabil
Author Disclosure Statement 2011; 8:30.
No competing financial interests exist. 17. Ustinova KI, Perkins JW, Leonard A, Hausbeck CJ. Virtual
reality game-based therapy for treatment of postural and
co-ordination abnormalities secondary to TBI: A pilot
References
study. Brain Inj 2014; 28:486–495.
1. Gobbens RJJ, van Assen MALM. The prediction of ADL 18. Shumway-Cook A, Brauer S, Woollacott M. Predicting the
and IADL disability using six physical indicators of frailty: probability for falls in community-dwelling older adults
A Longitudinal Study in the Netherlands. Curr Gerontol using the Timed Up & Go Test. Phys Ther 2000; 80:896–
Geriatr Res 2014; 2014:1–10. 903.
2. Cho K, Lee G. Impaired dynamic balance is associated with 19. Collin C, Wade DT, Davies S, Horne V. The Barthel ADL
falling in post-stroke patients. Tohoku J Exp Med 2013; Index: A reliability study. International Disability Studies.
230:233–239. 1998; 10:61–63.
3. Michael KM, Allen JK, Macko RF. Reduced ambulatory ac- 20. Schmid AA, Van Puymbroeck M, Altenburger PA, et al.
tivity after stroke: The role of balance, gait, and cardiovascular Balance and balance self-efficacy are associated with ac-
fitness. Arch Phys Med Rehabil 2005; 86:1552–1556. tivity and participation after stroke: A cross-sectional study
4. Ordahan B, Karahan AY, Basaran A, et al. Impact of ex- in people with chronic stroke. Arch Phys Med Rehabil
ercises administered to stroke patients with balance trainer 2012; 93:1101–1107.
on rehabilitation results: A randomized controlled study. 21. Powell LE, Myers AM. The Activities-specific Balance
Hippokratia 2015; 19:125–130. Confidence (ABC) scale. J Gerontol A Biol Sci Med Sci
5. Gitlin LN. Environmental adaptations for older adults and 1995; 50:28–34.
their families in the home and community. In: International 22. Dunning K. Stroke impact scale. In: Encyclopedia of
Handbook of Occupational Therapy Interventions. Spring- Clinical Neuropsychology. Springer, New York; 2011,
er, New York; 2009, pp. 53–62. pp. 2401–2402.
6. Deutsch J, Robbins D, Morrison J, Guarrera Bowlby P. 23. Vernadakis N, Derri V, Tsitskari E, Antoniou P. The effect
Wii-based compared to standard of care balance and mo- of Xbox Kinect intervention on balance ability for previ-
bility rehabilitation for two individuals post-stroke. In: ously injured young competitive male athletes: A prelimi-
Virtual Rehabilitation International Conference, 2009. nary study. Phys Ther Sport 2013; 30:1–8.
IEEE; 2009, pp. 117–120. 24. Liston RA, Brouwer BJ. Reliability and validity of mea-
7. Cho KH, Lee KJ, Song CH. Virtual-reality balance train- sures obtained from stroke patients using the Balance
ing with a video-game system improves dynamic balance Master. Arch Phys Med Rehabil 1996; 77:425–430.
in chronic stroke patients. Tohoku J Exp Med 2012; 228: 25. Flansbjer UB, Holmbäck AM, Downham D, et al. Relia-
69–74. bility of gait performance tests in men and women with
8. Pichierri G, Wolf P, Murer K, de Bruin ED. Cognitive and hemiparesis after stroke. J Rehabil Med 2005; 37:75–82.
cognitive-motor interventions affecting physical function- 26. Saposnik G, Mamdani M, Bayley M, et al. Effectiveness of
ing: A systematic review. BMC Geriatrics 2011; 11:29. Virtual Reality Exercises in STroke Rehabilitation (EVR-
 VIRTUAL REALITY BALANCE TRAINING FOR STROKE 311

EST): Rationale, design, and protocol of a pilot randomized patients: A randomized, controlled study. Comput Biol
clinical trial assessing the Wii gaming system. Int J Stroke Med 2015; 63:92–98.
2010; 5:47–51. 33. Laver K, George S, Ratcliffe J, Crotty M. Virtual reality
27. Lohse K, Shirzad N, Verster A, et al. Video games and stroke rehabilitation–hype or hope? Aust Occup Ther J
rehabilitation: Using design principles to enhance engage- 2011; 58:215–219.
ment in physical therapy. J Neurol Phys Ther 2013; 37: 34. Celinder D, Peoples H. Stroke patients’ experiences with
166–175. Wii Sports during inpatient rehabilitation. Scand J Occup
28. Hung J-W, Chou CX, Hsieh YW, et al. Randomized com- Ther 2012; 19:457–463.
parison trial of balance training by using exergaming and 35. Lange B, Flynn S, Rizzo A. Initial usability assessment of
conventional weight-shift therapy in patients with chronic off-the-shelf video game consoles for clinical game-based
stroke. Arch Phys Med Rehabil 2014; 95:1629–1637. motor rehabilitation. Phys Ther Rev 2009; 14:355.
29. Jonsson E, Henriksson M, Hirschfeld H. Does the func-
tional reach test reflect stability limits in elderly people?
J Rehabil Med 2003; 35:26–30. Address correspondence to:
30. Sin H, Lee G. Additional virtual reality training using Xbox Wen-Hsu Sung, PhD
Downloaded by University of Connecticut e-journal package NERL from www.liebertpub.com at 06/14/18. For personal use only.

kinect in stroke survivors with hemiplegia. Am J Phys Med Department of Physical Therapy and Assistive Technology
Rehabil 2013; 92:871–880. National Yang-Ming University
31. Hoe ZU, Wang JL, Lo YK, et al. Efficacy of WiiTM-based Taipei 112
virtual reality assisted rehabilitation for chronic stroke patients. Taiwan
Taiwan J Phys Med Rehabil 2010; 38:11–18.
32. Shin J-H, Park SB, Jang SH. Effects of game-based virtual E-mail: whsung@ym.edu.tw;
reality on health-related quality of life in chronic stroke wenhsusung@gmail.com