Original Research Article

Clinical Rehabilitation
2022, Vol. 36(9) 1186–1198
Effects of dual-task training on © The Author(s) 2022
Article reuse guidelines:
gait and balance in stroke sagepub.com/journals-permissions
DOI: 10.1177/02692155221097033

patients: A meta-analysis journals.sagepub.com/home/cre

Xueyi Zhang1 , Feng Xu2, Huijuan Shi1, Ruijiao Liu1,

and Xianglin Wan1

Abstract
Objective: To assess the effects of dual-task training on gait and balance in stroke patients.
Data sources: A systematic review of PubMed, Web of Science, Embase and Cochrane Library from their
inception through 20 August 2021.
Review methods: The bibliography was screened to identify randomized controlled trials that applied
dual-task training to rehabilitation function training in stroke patients. Two reviewers independently
screened references, selected relevant studies, extracted data and assessed risk of bias using the
Cochrane tool of bias. The primary outcome was the gait and balance parameters.
Results: A total of 1992 studies were identified and 15 randomized controlled trials were finally included
(512 individuals) were analyzed. A meta-analysis was performed and a beneficial effect on rehabilitation
training was found. Compared to patients who received conventional rehabilitation therapy, those who
received dual-task training showed greater improvement in step length (MD = 3.46, 95% CI [1.01, 5.92],
P = 0.006), cadence (MD = 4.92, 95% CI [3.10, 6.74], P < 0.001) and berg balance scale score (MD =
3.10, 95% CI [0.11, 6.09], P = 0.040). There were no differences in the improvements in gait speed
(MD = 2.89, 95% CI [ − 2.02, 7.80], P = 0.250) and timed up and go test (MD = −2.62, 95% CI [ − 7.94,
2.71], P = 0.340) between dual-task and control groups.
Conclusion: Dual-task training is an effective training for rehabilitation of stroke patients in step length and
cadence, however, the superiority of dual-task training for improving balance function needs further discussion.

Keywords
Stroke, gait, balance, meta-analysis

Received November 11, 2021; accepted April 11, 2022

Introduction 1
Biomechanics Laboratory, Beijing Sport University, Beijing,
China
Stroke can lead to a series of functional disorders of 2
People’s Hospital of Queshan, Henan, China
movement, balance, and cognition, which seriously
affect the quality of life of patients and impose a Corresponding author:
Xianglin Wan, Biomechanics Laboratory, Education and
heavy burden on the family and society.1 The recov- Experiment Building, Beijing Sport University, No. 48 Xinxi
ery of balance and independent walking ability is the Road, Haidian district, Beijing, China, 100084.
basis for improving the quality of life of stroke Email: [email protected]
Zhang et al. 1187

patients, and it is also the primary focus of the framework for reporting systematic reviews and
rehabilitation training programs for patients.2 As a meta-analyses to structure this review.13 The analysis
new rehabilitation training method, dual-task training methods and inclusion criteria were specified in
can be divided into cognitive-motor and motor dual- advance and documented in a protocol. The system-
task training, in which a basic task and another task atic review protocol was registered at PROSPERO
involving cognitive or motor interference are per- (CRD42021281592). The following electronic data-
formed simultaneously.3,4 Cognitive interference bases were searched from their inception through
commonly used in rehabilitation training includes the search date, 20 August 2021: PubMed, Web of
reciprocal, subtraction, the Stroop task, and motor Science, Embase, and Cochrane Library. We also
interventions include carrying a cup full of water or screened the reference lists of the included studies
transferring coins, among other examples. At and related reviews for additional literature.
present, there are still conflicting research results on The inclusion criteria for study selection were (1)
whether dual-task training can improve the walking RCTs that used dual-task training interventions com-
ability more effectively than traditional single-task pared with conventional rehabilitation therapy; (2)
rehabilitation training, along with reducing the risk adult patients with clinically diagnosed stroke (cere-
of falling and improving independence in daily activ- bral hemorrhage and cerebral infarction); (3) trials
ities as well as the quality of life of the patients.5–9 comparing the effects of dual-task training and
Several meta-analyses and systemic reviews have single-task rehabilitation training on the rehabilita-
summarized the application of dual tasks in patients tion period of stroke patients. The dual-task training
with stroke. Plummer et al. showed that dual-task (including motor dual-task training and cognitive-
training improved the walking speed in stroke motor dual-task training) was used in the experimen-
patients,10 but the evidence came from only three tal group, regardless of whether the group underwent
studies. The systematic evaluation results of Wang simultaneous single-task rehabilitation training. The
et al. and Zhou et al. indicated that cognitive- control group underwent single-task rehabilitation
walking dual-task training could improve the gait training only. Single-task rehabilitation training,
and balance functions of stroke patients in the including gait training, balance training, and
short term, while motor dual-task intervention was muscle strength training, was completed in a single-
not included in the studies.11,12 Additionally, the task context. (4) Regarding the outcome measure-
results of recent years were not included. Based on ments, studies were included if they assessed at
these results, an updated meta-analysis of the least one of the following outcomes: (1) gait para-
current evidence is warranted. meters as primary outcome measures, gait speed
This systematic review aimed to identify and (cm/s), step length (cm), and cadence (step/min);
analyze the randomized controlled trials(RCTs) (2) balance parameters as secondary outcome mea-
that examined dual-task training for gait and sures: berg balance scale and timed up and go test
balance rehabilitation in stroke patients, including results. The exclusion criteria were as follows: (1)
gait and balance parameters to evaluate the effective- conference summary or summary; (2) no outcome
ness of dual-task training, and provide a theoretical parameters were included in the criteria; (3) studies
basis for the design of dual-task training programs in which data could not be accurately extracted or
for improving the walking ability of stroke patients original research data were missing; (4)
in the future. This study covers a wide range of dual- non-English research; and (5) the full text could be
task training types and provides more comprehen- obtained.
sive and reliable systematic evaluation results. The following electronic databases were
searched from their inception through the search
date, 20 August 2021: PubMed, Web of Science,
Methods
Embase, and Cochrane Library. We also screened
We adopted the Preferred Reporting Items for the reference lists of the included studies and
Systematic Reviews and Meta-Analyses (PRISMA) related reviews for additional literature.
1188 Clinical Rehabilitation 36(9)

The search strategy was formulated using the fixed-effects model was selected.16 The level of
participants, interventions, comparisons and the meta-analysis was set at α = 0.05. If there was
outcomes(PICO) framework as follows: (P) Adult significant clinical heterogeneity, methods such as
stroke patients with walking dysfunction, (I) dual- subgroup analysis or sensitivity analysis were used
task gait training, whether single-task gait training to treat it, otherwise only descriptive analysis was
was used simultaneously, (C) compared to partici- performed.
pants undergoing single-task rehabilitation train-
ing, another gait training method used for stroke
patients, and (O) change analysis of the gait and
Results
balance outcomes. Specific search strategies are
presented in Appendix 1. In accordance with the PRISMA guidelines,
Two reviewers (X.Z. and R.L.) independently Figure 1 summarizes the process of identifying
read all eligible articles and cross-checked the the eligible studies.17
information. For conflicting evaluations, the third Finally, 15 RCTs were included, involving 512
senior author (X.W.) was consulted to solve the stroke patients.8,9,18–30 Of all the included studies,
dispute, and a final decision was made by the three trials used motor dual-task training,23,29,30 11
majority of the votes. Additional information was trials used cognitive-motor dual-task train-
obtained from the corresponding author of the ing,8,9,18,20–22,24,25,27,28 and one trial used both.26
primary study, if necessary. The following data The basic characteristics of the included studies are
were extracted: patient characteristics, general listed in Table 1. The Cochrane Bias risk assessment
characteristics of the study design, intervention fea- tool was used to evaluate the quality of the literature,
tures, duration, mean value, and standard deviation and the evaluation results are shown in Figure 2.
of the outcome parameters. Eight RCTs used gait speed as an outcome par-
We used the Cochrane tool of bias published in ameter, and included 234 stroke patients (I2 = 0%)
2011 to perform quality assessments for all with high homogeneity among the results.20–22,24–
included studies.14 Random sequence generation, 26,28,29
Therefore, a fixed-effects model was used
allocation concealment, blinding of participants for meta-analysis, and the results showed that
and personnel, blinding of outcome assessments, there was no significant difference in the gait
incomplete outcome data, selective reporting, and speed between the dual-task and control groups
other biases were evaluated. Each criterion was (MD = 2.89, 95% CI [ − 2.02, 7.80], P = 0.250),
rated as having a low, unclear, or high risk of as shown in Figure 3.
bias. Two authors (X.Z. and R.L.) were required Four RCTs used step length as an outcome par-
to complete the assessments independently. If any ameter, and included 159 stroke patients (I2 = 26%)
disagreement persisted, the third senior author with high homogeneity among the results.21,23,24,29
(X.W.) was consulted to reach a consensus. Consequently, a fixed-effects model was used for
RevMan 5.4 software was used for statistical ana- meta-analysis, and the results showed that the
lysis. Since all variables were continuous variables, step length of the dual-task group was significantly
the mean difference (MD) was adopted as the different from that of the control group (MD = 3.46,
effect parameter, and all effect sizes provided 95% 95% CI [1.01, 5.92], P = 0.006), as shown in
confidence intervals (CI). Statistical heterogeneity Figure 3.
was assessed by the Cochrane Q statistic and esti- Seven RCTs used cadence as an outcome param-
mated I2 values. I2 < 25% represented low hetero- eter, and evaluated 244 stroke patients (I2 = 0%)
geneity; I2 > 25% and I2 > 50% were considered with high homogeneity among the results.20,21,23–
moderate and high heterogeneity, respectively.15 If 26,29
Therefore, a fixed-effects model was used for
P < 0.05 or I2 > 50%, the statistical sample was con- meta-analysis, and the results revealed that the
sidered as highly heterogeneous and the cadence of the dual-task group was significantly dif-
random-effects model was applied. Otherwise, the ferent from that of the control group (MD = 4.92,
Zhang et al. 1189

Figure 1. Flowchart of study selection.

95% CI [3.10, 6.74], P < 0.001), as shown in Sensitivity analysis of berg balance scale score
Figure 3. parameters (I2 = 66%) was conducted to exclude
Six RCTs used the berg balance scale score as an the included studies one by one. It was found that
outcome parameter and included 194 stroke the heterogeneity of the results was significantly
patients (I2 = 66%) with high heterogeneity reduced when the study by Aslam et al. was
among the results.9,18,19,22,27,30 Therefore, a excluded (I2 = 38%).18 However, there was no sig-
random-effects model was used for meta-analysis, nificant difference in the berg balance scale scores
and the results showed that the berg balance scale between the dual-task and control groups after the
score of the dual-task group was significantly dif- exclusion (P = 0.17). Therefore, whether dual-task
ferent from that of the control group (MD = 3.10, training can improve berg balance scale scores
95% CI [0.11, 6.09], P = 0.040), as shown in needs to be further verified.
Figure 4. Sensitivity analysis of timed up and go test para-
Five RCTs used timed up and go test time as an meters (I2 = 92%) showed that the heterogeneity of
outcome parameter and analyzed 179 stroke the results was significantly reduced (I2 = 0%)
patients (I2 = 92%) with high heterogeneity when the study by Iqbal et al.23 was excluded,
among the results.8,9,22,23,28 A random-effects and the results remained unchanged after the exclu-
model was used for meta-analysis, and the results sion (P = 0.89).
showed that there was no significant difference in Generally, when there are fewer than 10 studies
timed up and go test time between the dual-task for each parameter, a funnel plot is not recom-
and control groups (MD = −2.62, 95% CI [ − mended for analysis of publication bias.31
7.94, 2.71], P = 0.340), as shown in Figure 4. Therefore, only subjective publication bias analysis
Sensitivity analyses were performed on the out- was performed: (1) the small RCT sample size
comes with high heterogeneity. included in this study may have led to a greater
Table1. Characteristics of included studies.
1190

Mean age(years)
Case (Mean ± SD) Gender Intervention Outcomes
number (male/ Duration
(DG/CG) DG CG female) DG CG Frequency (weeks) Gait Balance

Aslam 15/15 unclear unclear 18/12 exer-gaming training single-task 5 days per 6 berg
et al.18 balance week balance
Pakistan training scale
timed
up and
go test
Aydoğ du 25/28 69.28 ± 71.21 ± 39/14 cognitive-motor single-task gait 5 days a week 8 berg
et al.19 5.03 4.92 dual-task gait training training for 30 min balance
Pakistan each scale
Baek 16/15 56.94 ± 56.13 ± 20/11 dual-task gait training single-task gait 30 min twice 6 Gait
et al.20 8.79 10.25 with treadmill training with per week speed
Korea treadmill Cadence
Cho 11/11 60.00 ± 58.64 ± 7/15 physical and single-task 30 min per day 4 Gait
et al.21 9.38 11.86 occupational physical and 5 times per speed
Korea therapies + virtual occupational week Step
reality treadmill therapies + length
training virtual reality Cadence
treadmill
training）
Choi et al.8 19/18 49.11 ± 49.33 ± 31/6 cognitive-motor single-task gait 15 min per day 4 timed up
Korea 11.93 8.27 dual-task with a training with 5 times per and go
random auditory cue treadmill week test
while walking on a
treadmill
Hong 8/9 56.63 ± 66.22 ± 10/7 cognitive task training single-task 3 times per 4 Gait berg
et al.22 8.78 11.55 training week speed balance
Korea 30 min per scale
time timed
up and
go test
Iqbal 32/32 58.28 ± 58.87 ± 34/30 motor dual-task motor 4 times per 4 Step timed up
7.131 6.131 training week

(Continued)
Clinical Rehabilitation 36(9)
Table1. (Continued)

Mean age(years)
Case (Mean ± SD) Gender Intervention Outcomes
Zhang et al.

number (male/ Duration

(DG/CG) DG CG female) DG CG Frequency (weeks) Gait Balance

et al.23 single-task 40 min per length and go

Pakistan training time Cadence test
Kannan 13/11 57.5 ± 61 ± 13/11 cognitive-motor single-task 5 session for 6 berg
et al.9 8.04 4.6 exergame training training 1-2 weeks balance
USA 3 session for scale
3-4 weeks timed
2 session for up and
5-6 weeks go test
90 min each
session
Kim et al.24 20/20 51.0 ± 48.1 ± 26/14 virtual dual-task single-task gait 30 min per 4 Gait
Korea 13.5 7.5 treadmill training training with session speed
treadmill 3 times per Step
week length
Cadence
Kim et al.25 13/13 52.62 ± 56.15 ± 15/11 progressive treadmill single-task gait 5 days per 4 Gait
Korea 9.84 10.82 cognitive dual-task training with week speed
gait training treadmill 30 min per day Cadence
Liu et al.26 18/10 CDG: 50.8 ± 24/4 CDG： single-task 30 min per 4 Gait
Taiwan (CDG:9; 51.0 ± 13.5 cognitive-motor dual training session speed
MDG:9) 7.1 task gait training 3 sessions per Cadence
MDG: MDG：motor dual week
48.8 ± task gait training
11.7
Park 15/15 56.30 ± 59.75 ± 24/16 dual-task using different single-task 30 min per 6 berg
et al.27 7.14 7.75 cognitive tests training session balance
Korea 3 sessions per scale
week
Plummer 18/19 54.4 ± 59.6 ± 19/17 dual-task gait training single-task gait 30 min per 4 Gait timed up
et al.28 16.4 14.5 training session speed and go
USA 3 times per test
week

(Continued)
1191
 1192

Table1. (Continued)

Mean age(years)
Case (Mean ± SD) Gender Intervention Outcomes
number (male/ Duration
(DG/CG) DG CG female) DG CG Frequency (weeks) Gait Balance

Shim 17/16 65.59 ± 61.56 ± 20/13 motor dual task training single-task traditional 6 Gait
et al.29 5.81 6.17 training physical speed
Korea therapy： Step
30 min, 5 days length
per week Cadence
motor dual
task training：
30 min, 3 days
per week
Song 20/20 55.37 ± 57.10 ± 21/19 motor dual-task single-task 30 min per day 8 berg
et al.30 20.6 7.83 training training 5 times per balance
Korea week scale

DG: dual-task group; CG: control group; CDG: cognitive-motor dual task group; MDG: motor dual-task group .
Clinical Rehabilitation 36(9)
Zhang et al. 1193

Figure 2. Risk of bias summary, green represents low

risk, red represents high risk, and yellow represents
unclear risk.
1194 Clinical Rehabilitation 36(9)

Figure 3. Comparison of gait parameters between dual-task group and control group.

risk of publication bias32 and (2) the inclusion of studies and the number of included studies. This
only English language studies, while excluding analysis included both cognitive-motor dual-task
studies in other languages, may also have intro- and motor dual-task training for analysis, and
duced some publication bias. the included RCTs were published in recent
years, some of which did not obtain positive
results.20,26
Discussion The reasons for the effect on gait improvement
Although previous systematic evaluation may be related to the theory of gait automation.
results10,11 showed that dual-task training can Studies have shown that the dual-task operation
improve the gait and balance function of stroke can reveal the degree of influence of the concentra-
patients, the present study revealed that dual-task tion distribution of action,33 depending on the
training could indeed improve the step length and attention on the primary and secondary task alloca-
cadence of patients compared to single-task tion, action differences, and the effect of task
rehabilitation training, although its improvement change. The dual-task response time is prolonged,
effect on gait speed and balance function was error rate is increased, and the limited attention
limited. The difference in some outcome para- resources cannot be adequately allocated to the
meters can be attributed to the different types of two tasks, resulting in abnormal postural control
Zhang et al. 1195

Figure 4. Comparison of balance parameters between dual-task group and control group.

in stroke patients, which presents as gait and of the dual-task effect. Most studies used a fixed
balance disorders. Repeated training for walking dual-task mode during the duration of the interven-
and training with automated walking devices tion, and the lack of task difficulty at the late stage
reduce the need for attention resources and of intervention may be the reason for the non-
improve the walking performance.34 In addition, significant difference in the parameters. Another
it is believed that dual-tasks affect the gait and reason may be the difficulty in implementing dual-
balance mechanism related to neural plasticity. task interventions. The specific implementation of
After a stroke, excitability of nerves and brain dual-task training is highly dependent on the
activity patterns change,35 and dual-task training skills of the rehabilitation therapists, they need to
alters neural plasticity along with activation of the carefully explain the specific execution process of
brain regions that are associated with the central the dual tasks to ensure that each patient can
executive function.36–39 It promotes endogenous perform each task accurately. In addition, some
neural repair mechanisms, increases the number patients may try to avoid dual tasks. The dual-task
of synapses in the cerebral cortex, promotes the group focused on walking during the dual task,
branching of axons and dendrites,40 and improves reducing the effect of the intervention,42 or they
the ability of the nervous system to control the paused the walking task and focused on answering
body, thus improving the gait and balance function the questions,28 prioritizing cognitive tasks over
of patients. motor tasks.43 In studies including healthy indivi-
The limited improvement in the gait speed and duals, participants given variable priority instruc-
balance function may be related to several factors. tions have been found to learn tasks better than
Brain activity differs during single-task and dual- those given fixed priority instructions,44 which
task walking, with increased activation of the shows that in dual-task training, how therapists
sensorimotor and cerebellar regions and decreased guide the patients’ attention is an important factor.
activation of the prefrontal cortex during dual-task The current review had several limitations. First,
walking.41 These changes in the brain activation most of the included studies did not distinguish the
patterns are associated with reduced gait and cogni- types of strokes, so conducting subgroup analyses
tive performance measurements as well as the size of the types of stroke was not possible. Second,
1196 Clinical Rehabilitation 36(9)

since only a few RCTs were included, the differ- Funding

ences between various dual-task types could not The author(s) disclosed receipt of the following financial
be compared. Third, the small number of RCTs support for the research, authorship, and/or publication of
conferred difficulties in the accurate judgment of this article: This work was supported by the the
the effect of dual-task training on the patient’s Fundamental Research Funds for the Central
rehabilitation status. Further comparative analysis Universities of China, Beijing Outstanding Talents
can be conducted for various types of dual-task Training fund young backbone individual project,
rehabilitation training in the future. (grant number 2021QN012, 2018000020124G093).
In conclusion, current evidence indicates that
dual-task training is superior to single-task rehabili- ORCID iD
tation training in improving the step length and Xueyi Zhang https://orcid.org/0000-0002-0918-8404
cadence of stroke patients. However, its superiority
in improving the gait speed and balance function is References
limited. Clinical practice and future studies are
1. Walker ER, Hyngstrom AS and Schmit BD. Influence of
encouraged to conduct dual-task interventions visual feedback on dynamic balance control in chronic
with a more immersive design to achieve greater stroke survivors. J Biomech 2016; 49: 698–703.
improvements, wherein more attention would be 2. Plummer-D’Amato P, Altmann LJP, Behrman AL, et al.
given to the setting of task difficulty and the Interference between cognition, double-limb support, and
swing during gait in community-dwelling individuals post-
quality of task completion.
stroke. Neurorehabil Neural Repair 2010; 24: 542–549.
Article.
Clinical messages 3. Chuang LL, Lin YH, Lien YS, et al. Comparative effective-
ness research of dual-task and single-task balance training
on gait speed and cognition in individuals with stroke.
• Dual-task training is an effective training Eur J Neurol 2017; 24: 218–218.
for gait rehabilitation of stroke patients in 4. Zheng JJ, Wang XQ, Xu YY, et al. Cognitive dual-task
improving step length and cadence. training improves balance function in patients with stroke.
• Compared with single-task rehabilitation Healthmed 2012; 6: 840–845.
training, the superiority of dual-task train- 5. Yang YR, Wang RY, Chen YC, et al. Dual-task exercise
improves walking ability in chronic stroke: a randomized
ing for improving balance function needs controlled trial. Arch Phys Med Rehabil 2007; 88: 1236–
further discussion. 1240. 2007/10/03.
6. Tisserand R, Armand S, Allali G, et al. Cognitive-motor
dual-task interference modulates mediolateral dynamic sta-
bility during gait in post-stroke individuals. Hum Mov Sci
Acknowledgements 2018; 58: 175–184. 2018/02/16.
7. Bhatt T, Alqahtani S and Patel P. Effect of dual-task on fall
X.W. was responsible for conception and design of the
risk in chronic stroke survivors: examining reactive balance
study. X.Z., F.X., H.S. and R.L. were responsible for
responses to forward perturbations in stance. Archives of
trial screening, data extraction, and data analysis. All Physical Medicine and Rehabilitation 2016; 97: e20.
authors gave final approval of the version to be published Conference Abstract.
and agreed to be accountable for all aspects of the work in 8. Choi W, Lee G and Lee S. Effect of the cognitive-motor
ensuring that questions related to the accuracy or integ- dual-task using auditory cue on balance of surviviors with
rity of any part of the work are appropriately investigated chronic stroke: a pilot study. Clin Rehabil 2015; 29: 763–
and resolved. We would like to thank Editage (www. 770. 2014/11/15.
editage.cn) for English language editing. 9. Kannan L, Vora J, Bhatt T, et al. Cognitive-motor exergam-
ing for reducing fall risk in people with chronic stroke: a
randomized controlled trial. NeuroRehabilitation 2019;
Declaration of conflicting interests 44: 493–510. 2019/07/01.
10. Plummer P and Iyigun G. Effects of physical exercise inter-
The author(s) declared no potential conflicts of interest ventions on dual-task gait speed following stroke: a system-
with respect to the research, authorship, and/or publica- atic review and meta-analysis. Arch Phys Med Rehabil
tion of this article. 2018; 99: 2548–2560.
Zhang et al. 1197

11. Wang XQ, Pi YL, Chen BL, et al. Cognitive motor interfer- 26. Liu YC, Yang YR, Tsai YA, et al. Cognitive and motor dual
ence for gait and balance in stroke: a systematic review and task gait training improve dual task gait performance after
meta-analysis. Eur J Neurol 2015; 22: 555–e537. 2015/01/07. stroke - A randomized controlled pilot trial. Sci Rep 2017;
12. Zhou Q, Yang H, Zhou Q, et al. Effects of cognitive motor 7: 4070. 2017/06/24.
dual-task training on stroke patients: a RCT-based 27. Park MO and Lee SH. Effect of a dual-task program with
meta-analysis. J Clin Neurosci 2021; 92: 175–182. 2021/09/13. different cognitive tasks applied to stroke patients: a pilot
13. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items randomized controlled trial. NeuroRehabilitation 2019;
for systematic reviews and meta-analyses: the PRISMA state- 44: 239–249. 2019/04/23.
ment. Int J Surg 2010; 8: 336–341. 2010/02/23. 28. Plummer P, Zukowski LA, Feld JA, et al. Cognitive-motor
14. Higgins JP, Altman DG, Gotzsche PC, et al. The Cochrane dual-task gait training within 3 years after stroke: a rando-
collaboration’s tool for assessing risk of bias in randomised mized controlled trial. Physiother Theory Pract 2021:
trials. Br Med J 2011; 343: d5928. 2011/10/20. 1–16. Article in Press. DOI: 10.1080/09593985.2021.
15. Hatala R, Keitz S, Wyer P, et al. Tips for learners of evidence- 1872129.
based medicine: 4. Assessing heterogeneity of primary studies 29. Shim S, Yu J, Jung J, et al. Effects of motor dual task train-
in systematic reviews and whether to combine their results. ing on Spatio-temporal gait parameters of post-stroke
CMAJ 2005; 172: 661–665. 2005/03/02. patients. J Phys Ther Sci 2012; 24: 845–848.
16. Jin M, Pei J, Bai Z, et al. Effects of virtual reality in improv- 30. Song GB and Park EC. Effect of dual tasks on balance
ing upper extremity function after stroke: a systematic ability in stroke patients. J Phys Ther Sci 2015; 27: 2457–
review and meta-analysis of randomized controlled trials. 2460.
Clin Rehabil 2021: 2692155211066534. 2021/12/14. DOI: 31. Chang YS, Chu H, Yang CY, et al. The efficacy of music
10.1177/02692155211066534. therapy for people with dementia: a meta-analysis of rando-
17. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA state- mised controlled trials. J Clin Nurs 2015; 24: 3425–3440.
ment for reporting systematic reviews and meta-analyses of 2015/08/25.
studies that evaluate health care interventions: explanation 32. Egger M, Davey Smith G, Schneider M, et al. Bias in
and elaboration. J Clin Epidemiol 2009; 62: e1–34. 2009/07/28. meta-analysis detected by a simple, graphical test Br Med
18. Aslam M, Ain QU, Fayyaz P, et al. Exer-gaming reduces J 1997; 315: 629–634. 1997/10/06.
fall risk and improves mobility after stroke. JPMA The 33. Gray R. Attending to the execution of a complex sensori-
Journal of the Pakistan Medical Association 2021; 71: motor skill: expertise differences, choking, and slumps. J
1673–1675. Journal Article; Randomized Controlled Trial. Exp Psychol Appl 2004; 10: 42–54. 2004/04/01.
19. Aydoğ du YT, Aydoğ du O and Serap İ nal H. The effects of 34. Plummer-D’Amato P, Kyvelidou A, Sternad D, et al.
dual-task training on patient outcomes of institutionalized Training dual-task walking in community-dwelling adults
elderly having chronic stroke. Dement Geriatr Cogn Dis within 1 year of stroke: a protocol for a single-blind rando-
Extra 2018; 8: 328–332. Article. mized controlled trial. BMC Neurol 2012; 12: 129. 2012/
20. Baek CY, Chang WN, Park BY, et al. Effects of dual-task 11/02.
gait treadmill training on gait ability, dual-task interference, 35. Sullivan JE and Hedman LD. Sensory dysfunction follow-
and fall efficacy in people with stroke: a randomized con- ing stroke: incidence, significance, examination, and inter-
trolled trial. Phys Ther 2021; 101, Article. DOI: 10.1093/ vention. Top Stroke Rehabil 2008; 15: 200–217. 2008/07/
ptj/pzab067. 24.
21. Cho KH, Kim MK, Lee HJ, et al. Virtual reality training with 36. Collette F, Olivier L, Van der Linden M, et al. Involvement
cognitive load improves walking function in chronic stroke of both prefrontal and inferior parietal cortex in dual-task
patients. Tohoku J Exp Med 2015; 236: 273–280. 2015/08/01. performance. Brain Res Cogn Brain Res 2005; 24: 237–
22. Hong SY, Moon Y and Choi JD. Effects of cognitive task 251. 2005/07/05.
training on dynamic balance and gait of patients with 37. Dreher JC and Grafman J. Dissociating the roles of the
stroke: a preliminary randomized controlled study. Med rostral anterior cingulate and the lateral prefrontal cortices
Sci Monit Basic Res 2020; 26: e925264. 2020/08/11. in performing two tasks simultaneously or successively.
23. Iqbal M, Arsh A, Hammad SM, et al. Comparison of dual Cereb Cortex 2003; 13: 329–339. 2003/03/13.
task specific training and conventional physical therapy in 38. Schubert T and Szameitat AJ. Functional neuroanatomy
ambulation of hemiplegic stroke patients: a randomized of interference in overlapping dual tasks: an fMRI
controlled trial. J Pak Med Assoc 2020; 70: 7–10. Article. study. Brain Res Cogn Brain Res 2003; 17: 733–746.
24. Kim H, Choi W, Lee K, et al. Virtual dual-task treadmill 2003/10/17.
training using video recording for gait of chronic stroke sur- 39. Wu T and Hallett M. Neural correlates of dual task perform-
vivors: a randomized controlled trial. J Phys Ther Sci 2015; ance in patients with Parkinson’s disease. J Neurol
27: 3693–3697. Neurosurg Psychiatry 2008; 79: 760–766. 2007/11/17.
25. Kim KJ and Kim KH. Progressive treadmill cognitive dual- 40. Overman JJ and Carmichael ST Plasticity in the injured
task gait training on the gait ability in patients with chronic brain: more than molecules matter. Neuroscientist 2014;
stroke. J Exerc Rehabil 2018; 14: 821–828. 20: 15–28. 2013/06/13.
1198 Clinical Rehabilitation 36(9)

41. Szturm T, Kolesar TA, Mahana B, et al. Changes in meta-

bolic activity and gait function by dual-task cognitive game- No. Query Results
based treadmill system in Parkinson’s disease: protocol of a
randomized controlled trial. Front Aging Neurosci 2021; 13: #8 #3 AND #4 AND #7 497
680270. 2021/06/22. #7 #5 OR #6 2,845,636
42. Meester D, Al-Yahya E, Dennis A, et al. A randomized con- #6 balance:ab,ti OR walking:ab,ti OR 2,819,836
trolled trial of a walking training with simultaneous cogni- imbalance:ab,ti OR postural:ab,ti
tive demand (dual-task) in chronic stroke. Eur J Neurol OR stability:ab,ti OR unstable:ab,ti
2019; 26: 435–441. 2018/10/12. OR performance:ab,ti OR
43. Plummer-D’Amato P, Altmann LJ, Saracino D, et al. movement:ab,ti OR fall:ab,ti
Interactions between cognitive tasks and gait after stroke: #5 ‘gait’/exp 62,130
a dual task study. Gait Posture 2008; 27: 683–688. 2007/ #4 ‘dual task’:ab,ti OR ‘motor task’:ab,ti 18,719
10/20. OR ‘cognitive task’:ab,ti OR
44. Kramer AF, Larish JF and Strayer DL. Training for atten- ‘cognitive motor’:ab,ti OR ‘motor
tional control in dual task settings: a comparison of young cognitive’:ab,ti OR ‘additional task’:
and old adults. J EXP PSYCHOL-APPL 1995; 1: 50–76. ab,ti
#3 #1 OR #2 385,289
Appendix 1 Search strategy for each #2 poststroke*:ab,ti 25,114
database #1 ‘cerebrovascular accident’/exp 377,910
Search strategy for PubMed
(“Stroke"[MeSH Terms] OR “cerebral vascular”
OR “poststroke*”) AND “postural” OR “stability” OR “unstable” OR “per-
(“dual task” OR dual-task OR “cognitive task” OR formance” OR “movement” OR “fall”)
“motor task” OR cognitive-task
Search strategy for Embase
OR “cognitive motor” OR “cognitive-motor” OR
“motor cognitive” OR motor-cognitive Search strategy for Cochrane Library
OR “additional task”) AND (“Gait"[MeSH Terms] #1 MeSH descriptor: [Stroke] explode all trees
OR “Postural Balance"[MeSH Terms] #2 (cerebral vascular):ti,ab,kw OR (poststroke*):ti,
OR “balance” OR “walk” OR “walking” OR ab,kw (Word variations have been searched)
“imbalance” OR “postural” OR “stability” #3 #1 AND #2
OR “unstable” OR “performance” OR “movement” #4 (“dual task” OR dual-task OR “cognitive task”
OR “fall”) OR “motor task” OR cognitive-task
OR “cognitive motor” OR “cognitive-motor” OR
Search strategy for Web of Science
“motor cognitive” OR motor-cognitive
((TS = (Stroke OR “cerebral vascular” OR post- OR “additional task”)
stroke*)) AND TS = (“dual task” OR dual-task #5 MeSH descriptor: [Gait] explode all trees
OR “cognitive task” OR “motor task” OR #6 (“balance” OR “walk” OR “walking” OR
cognitive-task OR “cognitive motor” “imbalance” OR “postural” OR “stability”
OR “cognitive-motor” OR “motor cognitive” OR OR “unstable” OR “performance” OR “movement”
motor-cognitive OR “additional task”)) AND TS OR “fall”)
= (“Gait” OR “Balance” OR “balance” OR #7 #5 OR #6
“walk” OR “walking” OR “imbalance” OR #8 #3 AND #4 AND #7