766642

research-article2018
CRE0010.1177/0269215518766642Clinical RehabilitationLi et al.

CLINICAL
Original Article REHABILITATION

Clinical Rehabilitation

Effects of mirror therapy on walking 1­–15

© The Author(s) 2018
Reprints and permissions:
ability, balance and lower limb sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0269215518766642
https://doi.org/10.1177/0269215518766642

motor recovery after stroke: a journals.sagepub.com/home/cre

systematic review and meta-analysis

of randomized controlled trials

Yi Li1,2 , Qingchuan Wei1,2, Wei Gou3 and Chengqi He1,2

Abstract
Objective: To investigate the effects of mirror therapy on walking ability, balance and lower limb motor
recovery in patients with stroke.
Method: MEDLINE, EMBASE, Web of Science, CENTRAL, PEDro Database, CNKI, VIP, Wan Fang,
ClinicalTrials.gov, Current controlled trials and Open Grey were searched for randomized controlled
trials that investigated the effects of mirror therapy on lower limb function through January 2018. The
primary outcomes included were walking speed, mobility and balance function. Secondary outcomes
included lower limb motor recovery, spasticity and range of motion. Quality assessments were performed
with the PEDro scale.
Results: A total of 13 studies (n = 572) met the inclusion criteria. A meta-analysis demonstrated a
significant effect of mirror therapy on walking speed (mean difference (MD) 0.1 m/s, 95% confidence
interval (CI): 0.08 to 0.12, P < 0.00001), balance function (standard mean difference (SMD) 0.66, 95%
CI: 0.43 to 0.88, P < 0.00001), lower limb motor recovery (SMD 0.83, 95% CI: 0.62 to 1.05, P < 0.00001)
and passive range of motion of ankle dorsiflexion (MD 2.07°, 95% CI: 082 to 3.32, P = 0.001), without
improving mobility (SMD 0.43, 95% CI: −0.12 to 0.98, P = 0.12) or spasticity of ankle muscles (MD −0.14,
95% CI: −0.43 to 0.15, P = 0.35).
Conclusion: The systematic review demonstrates that the use of mirror therapy in addition to some
form of rehabilitation appears promising for some areas of lower limb function, but there is not enough
evidence yet to suggest when and how to approach this therapy.

Keywords
Mirror therapy, walking, balance, motor recovery, stroke

Received: 31 August 2017; accepted: 27 February 2018

1Department of Rehabilitation Medicine, West China Hospital, Corresponding author:

Sichuan University, Chengdu, People’s Republic of China
2Key Laboratory of Rehabilitation Medicine, West China Hospital,
Sichuan University, Chengdu, People’s Republic of China
3Department of Rehabilitation Medicine, Pidu District People’s
Hospital, Chengdu, People’s Republic of China
Chengqi He, Department of Rehabilitation Medicine, West
China Hospital, Sichuan University, No. 37, Guo-xue-xiang
Street, Chengdu 610065, Sichuan, People’s Republic of China.
Email: hxkfhcq2015@126.com
2 Clinical Rehabilitation 00(0)
Introduction
The lower limbs are essential for executing ambu-
lation, standing and walking, which are prerequi-
sites for performing routine activities of daily life.
The brain regulates motor activation, muscle tone,
selective joint movement and balance,1 while the
spinal cord regulates gait.2 Hence, for people with
stroke, paralysis in the lower limbs can lead to
motor function impairment and balance deficits
associated with gait dysfunction.3,4 Even after
rehabilitation and varying degrees of spontaneous
recovery, as many as 50% of hemiplegic patients
still cannot walk independently.5 Most patients
adopt compensatory strategies such as hip-hiking
and hip-circumduction coupled with a reduced
walking speed,6 which is far below the walking
level allowing them to perform a variety of daily
activities.7 Therefore, restoration of independent
walking to again become active in the community
and increase social involvement is considered a top
priority for rehabilitation in most patients.8
Although various rehabilitation programs have
been shown to be effective approaches to improv-
ing walking ability and balance, such as intensive,
repetitive, mobility-task training, the use of
orthoses, neuromuscular electric stimulation and
others,9 most of them are not performed for a long
duration because of the cost for the required equip-
ment and the nature of labor-intensive one-on-one
therapy. In recent years, mirror therapy, a new, sim-
ple and low-cost technique, has attracted much
attention in stroke rehabilitation. Generally, a mir-
ror is placed in the mid-sagittal plane of the patient,
between the unaffected limb and the affected limb,
so that the former is in front of the reflecting sur-
face while the latter is in its opposite.10 Moving the
healthy limb gives the patients the illusion that the
affected limb is performing the same actions.11
This visual feedback provided by mirror therapy
can compensate for diminished or absent proprio-
ceptive input12 and help to prevent or reduce
“learned nonuse.”10
In previous studies, mirror therapy was mainly
designed for functional recovery of the upper
limbs. As a result, improvement of upper extremity
motor function has been reported.13 However, even
though mirror therapy has been increasingly used
in rehabilitation of paralytic lower limbs, it is not
well established with good evidence owing to
small study samples and different therapeutic regi-
mens. To date, three representative reviews have
described the application of mirror therapy to lower
limbs after stroke.13–15 Although all of the results
showed that mirror therapy may be effective at
improving some aspects of lower limb motor func-
tion, no firm conclusions can be drawn until more
evidence is available. However, most of the reviews
merely performed qualitative analysis, and the
most recent study included in these reviews was
published in 2015. Since then, several randomized
controlled trials (RCTs) have been published.
Therefore, we performed this updated meta-analy-
sis to investigate the effects of mirror therapy on
walking ability, balance and lower limb motor
recovery after stroke.
Methods
Data sources and search strategy
In accordance with the Preferred Reporting Items
for Systematic Reviews and Meta-Analyses
(PRISMA) guidelines, we systematically searched
for studies published through January 2018 in the
following databases: MEDLINE (via Ovid),
EMBASE (via Ovid), Web of Science, CENTRAL
(via The Cochrane Library), Physiotherapy
Evidence Database (via PEDro website), CNKI,
VIP, Wan Fang, ClinicalTrials.gov (https://www.
clinicaltrials.gov/) and Current controlled trials
(www.controlled-trials.com). For the Grey
Literature research, we searched in Open Grey
(http://www.opengrey.eu/). The detailed electronic
search strategies for MEDLINE are provided in
Supplementary Appendix 1. Reference lists of the
relevant studies were manually screened to identify
additional studies for inclusion.
Study selection and eligibility criteria
Two independent reviewers performed the litera-
ture search. Any discrepancy was resolved by dis-
cussion or by achieving a consensus with a third
Li et al. 3
independent reviewer. Studies were eligible if they
met the following inclusion criteria: (1) the study
was a RCT; (2) participants were patients with
stroke who were >18 years of age, regardless of the
duration of illness or initial level of impairment;
(3) more than five patients were recruited; (4) the
study compared mirror therapy to no intervention,
to a different intervention or to a control group that
received the same therapeutic intervention minus
mirror therapy; (5) the study provided original data
or sufficient information about at least one of the
outcomes of walking speed, mobility, balance
function, lower extremity motor recovery, spastic-
ity and range of motion; and (6) the study was pub-
lished in English or Chinese. Articles were
excluded if the study (1) was a cross-over trial,
non-RCT or non-clinical trial; (2) included patients
with lower limb impairment caused by other dis-
eases (Parkinson’s disease, spinal cord injury, etc.);
or (3) was an abstract from meeting proceedings
with no corresponding full article published in a
peer-reviewed journal or no specific data provided
even after contacting the author. In the event of
duplicate publications, the data derived from only
the study with the most up-to-date information
were extracted.
Data extraction and quality assessment
Two experienced reviewers independently assessed
the studies using the inclusion/exclusion criteria
described above and conducted the quality assess-
ment. A third arbitrator was involved in case any
disagreements occurred. From each study, we
extracted general manuscript information (author,
year of publication), demographic and clinical
characteristics of patients (number of patients, age,
gender, stroke type, stroke duration and recovery
stage, paretic side and baseline conditions), study
characteristics (intervention and specific regimens
performed in experimental and control group),
intervention protocols (session length, frequency
and total duration) and outcomes.
We defined the primary outcomes as measures
of walking speed (e.g. quantitative gait analysis or
10-meter walk tests), mobility (e.g. timed up and
go test, Functional Ambulatory Category, Tinetti
test or Rivermead Mobility Index) and balance
(e.g. quantitative evaluations of balance, the Berg
Balance Scale, Brunnel Balance Assessment or the
balance subscale of any scale). We defined the sec-
ondary outcomes as measures of lower extremity
motor recovery (e.g. lower limb subscale of the
Fugl-Meyer Assessment or Brunnstrom stages for
the lower extremity), spasticity (e.g. Modified
Ashworth Scale) and range of motion, which was
assessed using a goniometer. When multiple,
optional scales were used concurrently in a single
study, we extracted data for each outcome from the
scale that was highest on the list. If the results were
merely presented in figures and no reply was
received in response to a data request to the authors,
values were extracted using a plot digitizer
program.16
Quality assessments were performed with the
PEDro scale,17 which is based on the Delphi List
criteria18 and is considered valid and reliable.17,19
The scale contains 11 items. Item 1 reflects exter-
nal validity and is not included in the total PEDro
score. The other 10 items evaluate the internal
validity of a clinical trial.20 One point was given for
each criterion that was satisfied. Therefore, a score
of 0–10 was allocated to each study (9–10: excel-
lent; 6–8: good; 4–5: fair; and ≤ 3: poor).21
Statistical analysis
All statistical analyses were conducted with
Review Manager 5.3 software (The Nordic
Cochrane Center, The Cochrane Collaboration,
Copenhagen, Denmark, 2014). For quantitative
synthesis, the summary effect size was obtained by
calculating the mean and SD of each outcome
between the experimental group and control group
after treatment. If the unit of measurement was
consistent across trials, the results were presented
as the weighted mean difference (MD) with 95%
confidence intervals (CIs) or else replaced by the
standard mean difference (SMD). For studies with
more than one treatment arm, data were calculated
for each arm independently. In order to interpret
the findings, units of measurement were converted
where appropriate. When different scales that
measured the same outcome had opposite changes
4 Clinical Rehabilitation 00(0)
Figure 1.  Flowchart of literature search and study
selection.
in scores, we multiplied the corresponding values
by −1 to ensure an identical directional value.
Statistical heterogeneity was assessed by the
Cochrane Q statistic and estimated I2 value. The
former was based on a P-value, for which less than
0.05 indicated heterogeneity between studies. The
latter quantified the percentage of the heterogene-
ity in the outcome measures. I2 values above 25%
and 50% were considered moderated and high het-
erogeneity, respectively.17 The random-effects
model was applied if high heterogeneity was found
(P < 0.05 or I2 > 50%). Otherwise, the fixed-effects
model was chosen.
We performed a subgroup analysis to establish
the effectiveness relative to the recovery stage
(acute <two  weeks, subacute: two  weeks to
six months and chronic: >six months) and different
intervention characteristics (movement of non-
paretic lower limb/ movement of bilateral lower
limb). The sensitivity analysis was conducted by
deleting each study individually to evaluate the
quality and consistency of the results. When the
number of studies included in each outcome was
fewer than 10, assessment of publication bias was
not performed since the test power was too low to
discern real asymmetry.22 For all outcome varia-
bles, two-tailed P-values of <0.05 were considered
statistically significant.
Results
Study identification
A flowchart of the search process is shown in
Figure 1. A total of 2582 studies were identified in
the initial search. After duplicates were removed
and abstracts screened, 22 studies remained for fur-
ther assessment. After reviewing the full papers to
obtain additional details and excluding studies for
various reasons (no relevant outcomes reported, an
irrelevant study design, non-randomized studies
and publications in languages other than English
and Chinese), 13 studies including 572 patients ful-
filled the inclusion criteria and were included in the
final analysis.23–35
Study characteristics
The features of the included studies are summa-
rized in Table 1. All of the included studies were
RCTs published from 2007 to 2017. The sample
size ranged from 22 to 124, with an average of 56.
About 58% of the participants were male (n = 332),
and the majority of patients had ischemic stroke
(n = 293 ischemic, n = 128 hemorrhagic). The num-
ber of participants who had hemiparesis of the left
side (n = 188) and right side (n = 191) was the same.
The time post-stroke varied greatly, from 6 days to
16 months. With one exception,31 several baseline
conditions were specified, including being able to
walk dependently, Modified Ashworth Scale,
Brunnstrom stages and Functional Ambulatory
Category at the time of recruitment.
Various interventions were administrated in
these trials. All studies had the patient use the mir-
ror set between the lower limbs and watched the
reflection of the non-paretic limb as it moved. In six
of the studies, movements were done bilaterally,
and in the remaining seven, only the non-paretic
limb was moved. Of these studies, three studies
received mirror therapy consisting of ankle move-
ments,23,29,30 one study performed activity-based
Table 1.  Characteristics of included studies in the review.
Study Number Age Gender Stroke Stroke duration Paretic Baseline Intervention Specific regimen Session Frequency Total Outcomes
Li et al.

(E/C) (years) (M/F) type and recovery side conditions length duration
stage (R/L)

Sutbeyaz 20/20 63.7 ± 23/17 33I/7H 3.7 ± 1.6 months 13/27 1 ≤ BS ≤ 3 E: MT + Rehab MT: performed Total: 2–5 Five days/ Four weeks BS, MAS,
et al.23 8.7 and Subacute C: Sham + dorsiflexion movement hours/day week FAC
Rehab of non-paretic ankle
while watching mirror
reflection, without
verbal feedback
Sham: performed the
same exercise while
watch non-reflecting
side of the mirror
Rehab: NFT, PT, OT, ST
Mohan 11/11 63.0 ± 12/10 14I/8H 6.4 ± 3.3 days 16/6 BS ≥ 2 E: MT + Rehab MT: performed MT:30 Six days/ Two weeks FMA-L, BBA
et al.35 13.1 and Acute C: Sham + functional movement of minutes/day week
Rehab non-paretic hip, knee Rehab: 1
and ankle while watching hour/day
mirror reflection
Sham: performed the
same exercise while
watch non-reflecting
side of the mirror
Rehab: NFT, sensory
motor reeducation,
active exercise, mobility
training, balance and gait
training
Barnes24 15/18 51.5 ± 21/12 NA NA NA MAS < 3 E: MT + Rehab MT: performed NA NA Two weeks TUG
13.6 C: Rehab functional movement
of bilateral lower
extremities while
watching mirror
reflection
Rehab: mobility training,
electrotherapeutic
modalities, education

(continued)
5
 6
Table 1. (Continued)

Study Number Age Gender Stroke Stroke duration Paretic Baseline Intervention Specific regimen Session Frequency Total Outcomes
(E/C) (years) (M/F) type and recovery side conditions length duration
stage (R/L)

Salem 15/15 59.6 ± 15/15 21I/9H 15.2 ± 1.6 17/13 Able to E: MT + Rehab MT: performed Total: 2–5 Five days/ Four weeks PROM,
and 8.9 months and walk > C: Sham + functional movement of hours/day week MAS, BS,
Huang26 Chronic 10 m Rehab bilateral hip, knee and 10MWT
ankle while watching
mirror reflection
Sham: performed the
same exercise while
watch non-reflecting
side of the mirror
Rehab: NFT, OT, PT,
electrotherapy, gait
training
Ji and 17/17 54.8 ± 19/15 18I/16 4.4 ± 1.4 months 14/20 MAS ≤ 2 E: MT + Rehab MT: performed MT:15 Five days/ Four weeks Gait analysis
Kim25 8.0 H and Subacute C: Sham + functional movement of minutes/day week
Rehab non-paretic hip, knee Rehab: 30
and ankle while watching minutes/day
mirror reflection
Sham: performed the
same exercise while
watch non-reflecting
side of the mirror
Rehab: NFT
Cha and 15/15 60.7 ± 13/17 20I/10H 14.3 ± 2.4 18/12 NA E: MT + rTMS MT: performed MT:20 min/ Five days/ Four weeks Gait analysis
Kim31 12.4 months and C: Sham + functional movement of day week
Chronic rTMS non-paretic hip, knee rTMS: 20
and ankle while watching min/day
mirror reflection
Sham: performed the
same exercise while
watch non-reflecting
side of the mirror
rTMS: 10 Hz rTMS
delivered to the affected
leg motor area with a
figure-of-eight coil
Clinical Rehabilitation 00(0)
Table 1. (Continued)

Study Number Age Gender Stroke Stroke duration Paretic Baseline Intervention Specific regimen Session Frequency Total Outcomes
Li et al.

(E/C) (years) (M/F) type and recovery side conditions length duration
stage (R/L)

Cha and 19/17 58.7 ± 19/17 NA 1.8 ± 0.7 months NA Independ­ E : MT + rTMS MT: performed MT:30 Five days/ Four weeks BBS, TUG
Kim27 8.6 and Subacute ent C: Sham + functional movement of minutes/day week
walking rTMS non-paretic hip, knee rTMS: 10
and ankle while watching minutes/day
mirror reflection,
without verbal feedback
Sham: performed the
same exercise while
watch non-reflecting
side of the mirror
rTMS: 1 Hz rTMS
delivered to the non-
affected motor area with
a 70 mm coil
Kim 17/17 53.5 ± 15/19 26I/8H 4.6 ± 1.2 months 16/18 MAS < 3 E: MT + Rehab MT: performed MT:30 Five days/ Four weeks quantitative
et al.28 8.8 and Subacute C: Sham + functional movement of minutes/day week evaluations
Rehab non-paretic hip, knee Rehab: 30 of balance
and ankle while watching minutes/day
mirror reflection
Sham: performed the
same exercise while
watch non-reflecting
side of the mirror
Rehab: NFT
In et al.29 13/12 55.9 ± 15/10 16I/9H 13.0 ± 4.7 12/13 1 ≤ BS ≤ 4 E: MT + Rehab MT: performed MT:30 Five days/ Four weeks BBS, TUG,
10.8 months and C: placebo + dorsiflexion, minutes/day week 10MWT
Chronic Rehab plantarflexion, adduction Rehab: 30
and abduction of minutes/day
bilateral ankle, foot
and hip while watching
mirror reflection
Placebo: performed the
same exercise while
watch non-reflecting
side of the mirror
Rehab: NFT, PT, OT, ST

(continued)
7
 8
Table 1. (Continued)

Study Number Age Gender Stroke Stroke duration Paretic Baseline Intervention Specific regimen Session Frequency Total Outcomes
(E/C) (years) (M/F) type and recovery side conditions length duration
stage (R/L)

Xu 23/23 54.9 ± 31/15 35I/11H 44.3 ± 6.2 days 28/18 1 < MAS E: MT + Rehab MT: performed MT:30 Five days/ Four weeks PROM,
et al.30 8.6 and Subacute < 4, C: Sham + functional movement minutes/day week MAS, BS,
3 ≤ FAC Rehab of bilateral ankle Rehab: 4 10MWT
≤5 while watching mirror hours/day
reflection with voice
instructions
Sham: performed the
same exercise while
watch non-reflecting
side of the mirror
Rehab: NFT, PT, OT
Arya 19/17 46.5 ± 30/6 27I/9H 15.9 ± 9.0 20/16 FAC ≥ 2 E: MT + Rehab MT: performed activity- MT:30 Three to Three FMA-L,
et al.32 7.5 months and C: Rehab based movement of minutes/day four days/ months 10MWT
Chronic non-paretic lower Rehab: 30 week
extremities while minutes/day
watching mirror
reflection
Rehab: mobility training,
motor training, orthosis
Haiyan 62/62 52.5 ± 72/52 83I/41H 4.0 ± 4.3 months NA BS ≥ 3 E: MT + Rehab MT: performed MT:30 Three Two months FMA-L,
et al.33 5.8 and Subacute C: Rehab functional movement minutes/day days/week quantitative
of bilateral lower Rehab: 30 evaluations
extremities while minutes/day of balance
watching mirror
reflection
Rehab: PT, OT,
acupuncture
Haiyan 41/41 57 ± 8 47/35 NA 31.0 ± 4.0 days 37/45 MAS ≤ 2 E: MT + Rehab MT: performed MT:15 Six days/ Four weeks FMA-L, BBS,
et al.34 and Subacute C: Danhong functional movement minutes/day week Gait analysis
injection + of bilateral lower Rehab: 40
Rehab extremities while minutes/day
watching mirror
reflection
Rehab: PT, OT, NFT

E: experimental group; C: control group; M: male; F: female; y: year; m: month; d: day; I: ischemic; H: hemorrhagic; BS: Brunnstrom stage; MT: mirror therapy; Rehab: rehabili-
tation; PT: Physical therapy; OT: Occupational therapy; ST: Speech therapy; MAS: Modified Ashworth Scale; NFT: Neurodevelopment facilitation; FAC: Functional Ambula-
tion Categories; FMA-L: lower limb subscale of the Fugl-Meyer Assessment; BBA: Brunnel Balance Assessment; NA: not applicable; TUG: timed up and go test; PROM:
Passive range of motion; 10MWT: 10-meter walk test; rTMS: repetitive transcranial magnetic stimulation; BBS: Berg Balance Scale.
Clinical Rehabilitation 00(0)
Li et al. 9
movements like pedaling, ball rolling and wiping,32
and the remaining nine exercised the major joints of
the lower limb, including the hip joint, knee joint
and ankle joint with the following functional activi-
ties: hip-knee-ankle flexion, knee extension with
ankle dorsiflexion, knee flexion beyond 90° or
ankle eversion. Except for two studies that adopted
repetitive transcranial magnetic stimulation (rTMS)
plus mirror therapy,27,31 the experimental group
received some form of rehabilitation plus mirror
therapy, such as neurodevelopmental facilitation
techniques, physical therapy or occupational ther-
apy. Mirror therapy was provided for three or
six days a week for two weeks to three months. Each
session ranged from approximately 15 minutes to
30 minutes.
Seven studies measured walking speed. Four of
these studies used the 10-meter walk test,26,29,30,32
and three used quantitative gait analysis.25,31,34
Four studies evaluated mobility using the timed up
and go test24,27,29 and Functional Ambulatory
Category.23 As for assessment of balance, the
Brunnel Balance Assessment,35 the Berg Balance
Scale27,29,34 and quantitative evaluations of bal-
ance28,33 were used in five studies. Seven studies
assessed lower extremity motor recovery, and four
of them used the lower limb subscale of the Fugl-
Meyer Assessment,32–35 while the other three used
Brunnstrom stages.23,26,30 Three studies measured
the spasticity of the ankle muscle using the
Modified Ashworth Scale,23,26,30 and passive range
of motion of ankle dorsiflexion was reported in two
studies.26,30
Quality assessment
Since it is difficult to blind patients and therapists to
the intervention, the maximum possible total score
was 8 out of 10 points. The PEDro scores of the
included studies ranged from 5 to 8, with a median
score of 5. Six studies were of good quality, while
the other seven were of fair quality. Randomization
was concealed in five trials.23,25,28,30,32 All of the
studies were single assessor-blinded, except for
three studies.26,31,33 Only one study performed an
intention-to-treat analysis.32 A detailed evaluation
of the methodological quality is provided in
Supplementary Table 2.
Primary outcomes
Seven studies (n = 280) provided data on walking
speed. A fixed-effects model was chosen without
evidence of significant heterogeneity (I2 = 0%). A
significant improvement in walking speed of
0.10 m/s (95% CI: 0.08 to 0.12, P < 0.00001, Figure
2(a)) was found for participants in the mirror ther-
apy group compared with the control group.
Supplementary Table 3 shows the raw data on
walking speed. The subgroup mean effect size for
subacute stroke was 0.10 m/s (95% CI: 0.08 to
0.13, P < 0.00001, Figure 2(b)), while that for
chronic stroke was 0.07 m/s (95% CI: 0.00 to 0.14,
P = 0.05). As for trials that performed either bilat-
eral or non-paretic movement of the lower limb,
the pooled results suggested that both of them were
capable of improving walking speed (bilateral: MD
0.1 m/s, 95% CI: 0.07 to 0.12, P < 0.00001; non-
paretic: MD 0.1  m/s, 95% CI: 0.02 to 0.18,
P = 0.010, Figure 2(c)).
Four studies (n = 124) provided data about
mobility. A random-effects model was chosen
with evidence of significant heterogeneity
(I2 = 56%). No significant improvement in mobil-
ity was noted after treatment (SMD 0.43, 95% CI:
–0.12 to 0.98, P = 0.12, Figure 3(a)). The sensitiv-
ity analysis showed that excluding the study con-
ducted by Cha and Kim27 did not change the
direction of the result (SMD 0.15, 95% CI: –0.27
to 0.57, P = 0.48, Figure 3(b)), but the heterogene-
ity decreased (I2 = 0%).
Six studies (n = 322) provided data on balance.
A fixed-effects model was chosen without evi-
dence of significant heterogeneity (I2 = 0%). The
analysis revealed a significant treatment effect for
improving balance after treatment (SMD 0.66,
95% CI: 0.43 to 0.88, P < 0.00001, Figure 4).
Secondary outcomes
Seven studies (n = 373) provided data on lower
limb motor recovery. A fixed-effects model was
chosen with evidence of moderate heterogeneity
(I2 
= 42%). The analysis revealed a significant
effect on accelerating lower limb motor recovery
among individuals with stroke (SMD 0.83, 95%
CI: 0.62 to 1.05, P < 0.00001, Figure 5).
10 Clinical Rehabilitation 00(0)

Figure 2.  Meta-analysis (a) and subgroup analysis based on recovery stage (b) and different intervention
characteristics (c) of mirror therapy on walking speed.

Three studies (n = 109) provided data on the was chosen without evidence of significant hetero-
spasticity of ankle muscles. A fixed-effects model geneity (I2 = 0%). No significant improvement in
Li et al. 11

Figure 3.  Meta-analysis (a) and sensitivity analysis (b) of mirror therapy on mobility.

Figure 4.  Meta-analysis of mirror therapy on balance function.

spasticity was noted after treatment (MD –0.14,
95% CI: –0.43 to 0.15, P = 0.35, Figure 6).
Two studies (n = 76) provided data on PROM of
ankle dorsiflexion. A random-effects model was
chosen with evidence of significant heterogeneity
(I2 = 52%). A significant improvement in PROM of
2.07° (95% CI: 082 to 3.32, P = 0.001, Figure 7)
was found for participants in the mirror therapy
group compared with the control group.
Discussion
The primary finding of this systematic review and
meta-analysis indicates that patients with stroke
who received mirror therapy had significant
improvements in walking speed, balance, lower
limb motor recovery and passive range of motion
of ankle dorsiflexion. Furthermore, both recovery
stage and different intervention characteristic sub-
group analyses show improvement in walking
speed after treatment. However, there was no evi-
dence to support that mirror therapy promoted
rehabilitation of mobility or spasticity of ankle
muscles.
The results of this meta-analysis are consistent
with the results of some earlier reviews. However,
previous reviews of mirror therapy have either
focused on qualitative synthesis or have mainly
reported its effect on lower limb motor function.
Moreover, compared with previous reviews, this
systematic review and meta-analysis only included
RCTs rather than non-RCTs or case studies, and
lower limb function was further subdivided into
walking speed, mobility, balance function, motor
12 Clinical Rehabilitation 00(0)
Figure 5.  Meta-analysis of mirror therapy on lower limb motor recovery.
Figure 6.  Meta-analysis of mirror therapy on spasticity.
recovery, spasticity and range of motion, which
would contribute to uncovering the specific func-
tional improvements. Also, most of the included
studies were newly published since 2015.
Meanwhile, subgroup analysis was performed to
investigate the best therapeutic regime and optimal
stage to receive mirror therapy after stroke. The
above-mentioned items are the strengths of this
systematic review and meta-analysis.
Although subgroup analysis showed that mirror
therapy had a positive effect on walking speed for
patients in the chronic stage of stroke, the results
were unstable. Three of four studies were not sta-
tistically significantly different when compared
with the control group. Therefore, it seemed that
mirror therapy was more effective for patients in
the subacute stage of stroke. On one hand, this
finding could be explained by the neural plasticity
of the brain in the early phase after stroke. On the
other hand, according to the underlying mechanism
of mirror therapy, observing mirrored movements
could cause additional neural activities in motor
areas of the affected hemisphere, leading to corti-
cal reorganization and improved function, which
reinforce its effect during this period. With regard
to the subgroup analysis of different intervention
characteristics, the result was also unstable when
patients received non-paretic movement of the
lower limb since the direction of outcome changed
after excluding the study conducted by Cha and
Kim.31 Therefore, it seems that mirror therapy was
more effective when patients performed bilateral
movement of the lower limb. The reason may be
explained, in part, by the fact that bilateral move-
ments of limbs activated the primary motor cortex
area located in the affected hemisphere, which
resulted in greater improvement.36,37 Notably,
although a favorable effect of mirror therapy was
observed in this meta-analysis, this finding seemed
to have little clinical significance. The average
walking speed of subjects in seven of the included
studies was about 0.5 m/s before and 0.64 m/s after
treatment. According to the classification criteria
proposed by Perry et al.4 (<0.4 m/s: house-hold
ambulation, 0.4–0.8 m/s: limited community ambu-
lation and >0.8 m/s: full community ambulation),
these individuals would have been confined to lim-
ited community ambulation before and after
Li et al. 13
Figure 7.  Meta-analysis of mirror therapy on range of motion.
treatment. Moreover, except for one study,31 none
of the studies achieved a transition to a higher level
following treatment with mirror therapy
(Supplementary Table 3).
According to our analysis, although no signifi-
cant improvement in mobility was found between
the mirror therapy group and control group, the
trend of change was positive. However, there are
limitations to the evidence, which may result from
the diversity of patient recruitment, treatment regi-
mens and outcome evaluation that cannot be
ignored. All of these characteristics have the poten-
tial to increase heterogeneity and decrease validity,
and high statistical heterogeneity was found.
Intriguingly, after excluding the study conducted
by Cha and Kim,27 the heterogeneity dropped to
zero. It is worth noting that all of the included stud-
ies had negative results except for this one, which
adopted rTMS plus mirror therapy. However, fair
research quality and small sample sizes made it dif-
ficult to verify whether rTMS coupled with mirror
therapy has a synergistic effect that may eventually
facilitate improvements in mobility. By contrast,
the simple ankle movements (dorsiflexion, plantar-
flexion, abduction or adduction) performed and
short treatment period (two weeks) applied may be
the reason for the lack of improvement in mobility
in the other three studies.
Lower limb motor recovery has been found to
correlate significantly with walking speed.38,39 The
analysis revealed a significant effect on accelerat-
ing lower limb motor recovery among individuals
with stroke. However, the heterogeneity was mod-
erate. The reason can be inferred from the obvious
outlier35 and two studies with negative results.26,32
The former started mirror therapy in the acute
phase of stroke, which was within the maximum
recovery period (one week to one month) for lower
limb recovery. Notably, significant differences
were found on the lower limb subscale of the Fugl-
Meyer Assessment at baseline in this study.
Moreover, although Mohan et al. had administered
the therapeutic program for six days/week, which
was longer than others’ five days/week, the total
duration only lasted for two weeks, and previous
studies with significant motor recovery of lower
extremities generally performed the intervention
from three to six weeks. In contrast, the latter two
recruited patients mainly in the chronic stage of
stroke. Therefore, more trials are needed to investi-
gate the effect of mirror therapy on post-stroke
patients at different recovery stages and explore the
optimal therapeutic regimen. As for balance, a sig-
nificant positive correlation was shown to exist
between lower extremity motor recovery and bal-
ance impairment in patients with stroke.40
Consequently, recovery of lower limb function was
accompanied by improvement in balance.
Appropriate ankle motion is a prerequisite for
functional gait. Given that the ankle muscles have
more connective tissue and type I muscle fibers than
other muscles,41 patients with stroke are more vul-
nerable to increased spasticity and decreased range
of motion of ankle dorsiflexion.42 In this meta-anal-
ysis, although favorable effects on passive range of
motion of ankle dorsiflexion were found between
the mirror therapy group and the control group, the
high heterogeneity (I2 = 52%) made the result uncon-
vincing. Two studies with positive results were
included in the analysis. One of these studies
recruited ambulatory patients in the chronic stage,
whereas the other enrolled patients with poorer
mobility in the subacute stage. The great heteroge-
neity may come from the different baseline charac-
teristics. Moreover, no significant improvement in
spasticity was observed between the mirror therapy
14 Clinical Rehabilitation 00(0)
group and control groups. This result was in line
with several previous studies.12,43 It is possible that
spasticity involves multiple aspects of pathophysiol-
ogy, and only providing visual feedback may be
inadequate for triggering change.
This meta-analysis has the following limitations
that must be considered: (1) total numbers of
included studies and patients were small and might
not offer enough statistical power to support the
results. Therefore, caution is needed when inter-
preting the results. (2) Although all of the included
studies were considered high quality, lack of con-
cealed allocation and intention-to-treat analysis in
most of the studies cannot be ignored. (3) Even
after a comprehensive literature search, we still
cannot rule out the possibility that some other rel-
evant RCTs might have been omitted. (4) Some
studies published in languages other than English
and Chinese were excluded. (5) The diversity of
the studies (therapeutic regimen, phase of stroke
recovery and evaluation indicators) may have
caused variation in the results.
Clinical Messages
•• Mirror therapy was associated with
improvements in walking speed and pas-
sive range of movement at the ankle, but
it was not clinically significant.
•• Mirror therapy appears to be promising
for some areas of lower limb function,
but there is not enough evidence yet to
suggest when and how to approach this
therapy.
Acknowledgements
Yi Li and Qingchuan Wei contributed equally to this
study.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest
with respect to the research, authorship and/or publica-
tion of this article.
Funding
The author(s) received no financial support for the
research, authorship and/or publication of this article.
ORCID iD
Yi Li https://orcid.org/0000-0002-6714-3028
References
1. Arene N and Hidler J. Understanding motor impairment
in the paretic lower limb after a stroke: a review of the
literature. Top Stroke Rehabil 2009; 16: 346–356.
2. Arya KN and Pandian S. Interlimb neural coupling: impli-
cations for poststroke hemiparesis. Ann Phys Rehabil Med
2014; 57: 696–713.
3. Cunha IT, Lim PA, Henson H, et al. Performance-based
gait tests for acute stroke patients. Am J Phys Med Rehabil
2002; 81: 848–856.
4. Perry J, Garrett M, Gronley JK, et al. Classification of
walking handicap in the stroke population. Stroke 1995;
26: 982–989.
5. Jørgensen HS, Nakayama H, Raaschou HO, et al.
Recovery of walking function in stroke patients: the
Copenhagen Stroke Study. Arch Phys Med Rehabil 1995;
76: 27–32.
6. Boudarham J, Roche N, Pradon D, et al. Variations in
kinematics during clinical gait analysis in stroke patients.
PLoS ONE 2013; 8: e66421.
7. Flansbjer UB, Holmback AM, Downham D, et al.
Reliability of gait performance tests in men and women
with hemiparesis after stroke. J Rehabil Med 2005; 37:
75–82.
8. Cress RH and Fleming WC. Treatment goals and selection
of patients for rehabilitation with hemiplegia. Ala J Med
Sci 1966; 3: 307–311.
9. Winstein CJ, Stein J, Arena R, et al. Guidelines for adult
stroke rehabilitation and recovery: a guideline for health-
care professionals from the American Heart Association/
American Stroke Association. Stroke 2016; 47: e98–
e169.
10. Dohle C, Pullen J, Nakaten A, et al. Mirror therapy pro-
motes recovery from severe hemiparesis: a randomized
controlled trial. Neurorehabil Neural Repair 2009; 23:
209–217.
11. Deconinck FJ, Smorenburg AR, Benham A, et al.
Reflections on mirror therapy: a systematic review of the
effect of mirror visual feedback on the brain. Neurorehabil
Neural Repair 2015; 29: 349–361.
12. Yavuzer G, Selles R, Sezer N, et al. Mirror therapy
improves hand function in subacute stroke: a randomized
controlled trial. Arch Phys Med Rehabil 2008; 89: 393–
398.
13. Thieme H, Mehrholz J, Pohl M, et al. Mirror therapy
for improving motor function after stroke. Cochrane
Database Syst Rev 2012; 14: CD008449.
14. Hung GKN, Li CTL, Yiu AM, et al. Systematic review:
effectiveness of mirror therapy for lower extremity post-
stroke. Hong Kong J Occupat Ther 2015; 26: 51–59.
15. Rothgangel AS, Braun SM, Beurskens AJS, et al. The
clinical aspects of mirror therapy in rehabilitation: a
Li et al. 15
s­ ystematic review of the literature. Int J Rehabil Res 2011;
34: 1–13.
16. Huwaldt J. Plot digitizer, http://plotdigitizer.sourceforge.
net/ (2001, accessed 24 October 2015).
17. Maher CG, Sherrington C, Herbert RD, et al. Reliability
of the PEDro scale for rating quality of randomized con-
trolled trials. Phys Ther 2003; 83: 713–721.
18. Verhagen AP, De Vet HC, De Bie RA, et al. The Delphi
list: a criteria list for quality assessment of randomized
clinical trials for conducting systematic reviews devel-
oped by Delphi consensus. J Clin Epidemiol 1998; 51:
1235–1241.
19. De Morton NA. The PEDro scale is a valid measure of the
methodological quality of clinical trials: a demographic
study. Aust J Physiother 2009; 55: 129–133.
20. Sherrington C, Herbert RD, Maher CG, et al. PEDro: a
database of randomized trials and systematic reviews in
physiotherapy. Man Ther 2000; 5: 223–226.
21. Teasell R, Foley N, Salter K, et al. Evidence-based review
of stroke rehabilitation: executive summary, 12th edition.
Top Stroke Rehabil 2009; 16: 463–488.
22. Sterne JA, Sutton AJ, Ioannidis JP, et al. Recommendations
for examining and interpreting funnel plot asymmetry in
meta-analyses of randomised controlled trials. BMJ 2011;
343: d4002.
23. Sutbeyaz S, Yavuzer G, Sezer N, et al. Mirror therapy
enhances lower-extremity motor recovery and motor
functioning after stroke: a randomized controlled trial.
Arch Phys Med Rehabil 2007; 88: 555–559.
24. Barnes LJ. Effect of mirror therapy on lower extrem-
ity motor control and gait in patients with stroke.
ClinicalTrials.gov, 2014, https://clinicaltrials.gov/ct2/
show/study/NCT01574079?cond=effect+of+mirror+ther
apy+on+lower+extremity+motor+control&rank=1
25. Ji SG and Kim MK. The effects of mirror therapy on the
gait of subacute stroke patients: a randomized controlled
trial. Clin Rehabil 2015; 29: 348–354.
26. Salem HMA and Huang X. The effects of mirror therapy
on clinical improvement in hemiplegic lower extremity
rehabilitation in subjects with chronic stroke. Int Scholar
Sci Res Innovat 2015; 9: 163–166.
27. Cha HG and Kim MK. Therapeutic efficacy of low fre-
quency transcranial magnetic stimulation in conjunc-
tion with mirror therapy for sub-acute stroke patients. J
Magnet 2015; 20: 51–56.
28. Kim MK, Ji SG and Cha HG. The effect of mirror therapy
on balance ability of subacute stroke patients. Hong Kong
Physiother J 2016; 34: 27–32.
29. In T, Lee K and Song C. Virtual reality reflection therapy
improves balance and gait in patients with chronic stroke:
randomized controlled trials. Med Sci Monitor 2016; 22:
4046–4053.
30. Xu Q, Guo F, Salem HMA, et al. Effects of mirror ther-
apy combined with neuromuscular electrical stimulation
on motor recovery of lower limbs and walking ability of
patients with stroke: a randomized controlled study. Clin
Rehabil 2017; 31: 1583–1591.
31. Cha HG and Kim MK. The effects of repetitive transcra-
nial magnetic stimulation integrated mirror therapy on
the gait of chronic stroke patients. J Magnetics 2015; 20:
133–137.
32. Arya KN, Pandian S and Kumar V. Effect of activity-
based mirror therapy on lower limb motor-recovery and
gait in stroke: a randomised controlled trial. Neuropsychol
Rehabil. Epub ahead of print 26 September 2017. DOI:
10.1080/09602011.2017.1377087.
33. Haiyan W, Zhijun Z, Ping J, et al. Effect observattion of
mirror therapy combined with routine rehabilitation train-
ing on limb motor function and balance of patients with
post-stroke hemiplegia. Med Pharm J Chin PLA 2017; 29:
34–37.
34. Haiyan W, Zhijun Z and Ping J. Study on rehabilitation
effect of Danhong injection combined with mirror therapy
on lower limb motor dysfunction in stroke patients with
hemiplegia. Shanxi Med J 2017; 46: 2178–2181.
35. Mohan U, Babu SK, Kumar KV, et al. Effectiveness of
mirror therapy on lower extremity motor recovery, bal-
ance and mobility in patients with acute stroke: a rand-
omized sham-controlled pilot trial. Ann Ind Acad Neurol
2013; 16: 634–639.
36. Staines WR, McIlroy WE, Graham SJ, et al. Bilateral move-
ment enhances ipsilesional cortical activity in acute stroke: a
pilot functional MRI study. Neurology 2001; 56: 401–404.
37. Cunningham CL, Stoykov ME and Walter CB. Bilateral
facilitation of motor control in chronic hemiplegia. Acta
Psychol (Amst) 2002; 110: 321–337.
38. Hsu AL, Tang PF and Jan MH. Analysis of impairments
influencing gait velocity and asymmetry of hemiplegic
patients after mild to moderate stroke. Arch Phys Med
Rehabil 2003; 84: 1185–1193.
39. Brandstater ME, De Bruin H, Gowland C, et al.
Hemiplegic gait: analysis of temporal variables. Arch
Phys Med Rehabil 1983; 64: 583–587.
40. Bowden MG, Clark DJ and Kautz SA. Evaluation of
abnormal synergy patterns poststroke: relationship of
the Fugl-Meyer Assessment to hemiparetic locomotion.
Neurorehabil Neural Repair 2010; 24: 328–337.
41. Given JD, Dewald JP and Rymer WZ. Joint dependent
passive stiffness in paretic and contralateral limbs of spas-
tic patients with hemiparetic stroke. J Neurol Neurosurg
Psychiatry 1995; 59: 271–279.
42. Gao F, Ren Y, Roth EJ, et al. Effects of repeated ankle
stretching on calf muscle-tendon and ankle biomechani-
cal properties in stroke survivors. Clin Biomech (Bristol,
Avon) 2011; 26: 516–522.
43. Samuelkamaleshkumar S, Reethajanetsureka S,
Pauljebaraj P, et al. Mirror therapy enhances motor per-
formance in the paretic upper limb after stroke: a pilot ran-
domized controlled trial. Arch Phys Med Rehabil 2014;
95: 2000–2005.