Neurorehabilitation and

Research Articles Neural Repair

Volume 23 Number 3
March/April 2009 209-217
© 2009 The American Society of
Mirror Therapy Promotes Recovery From Severe Neurorehabilitation
10.1177/1545968308324786

Hemiparesis: A Randomized Controlled Trial http://nnr.sagepub.com

hosted at
http://online.sagepub.com

Christian Dohle, MD, MPhil, Judith Püllen, Antje Nakaten, Jutta Küst, PhD,
Christian Rietz, PhD, and Hans Karbe, MD

Background. Rehabilitation of the severely affected paretic arm after stroke represents a major challenge, especially in the presence of
sensory impairment. Objective. To evaluate the effect of a therapy that includes use of a mirror to simulate the affected upper extremity
with the unaffected upper extremity early after stroke. Methods. Thirty-six patients with severe hemiparesis because of a first-ever isch-
emic stroke in the territory of the middle cerebral artery were enrolled, no more than 8 weeks after the stroke. They completed a protocol
of 6 weeks of additional therapy (30 minutes a day, 5 days a week), with random assignment to either mirror therapy (MT) or an equiv-
alent control therapy (CT). The main outcome measures were the Fugl-Meyer subscores for the upper extremity, evaluated by indepen-
dent raters through videotape. Patients also underwent functional and neuropsychological testing. Results. In the subgroup of 25 patients
with distal plegia at the beginning of the therapy, MT patients regained more distal function than CT patients. Furthermore, across all
patients, MT improved recovery of surface sensibility. Neither of these effects depended on the side of the lesioned hemisphere. MT
stimulated recovery from hemineglect. Conclusions. MT early after stroke is a promising method to improve sensory and attentional
deficits and to support motor recovery in a distal plegic limb.

Keywords:   Stroke rehabilitation; Arm; Mirror therapy; Randomized clinical trial; Motor recovery; Hemineglect

A mong the different syndromes following stroke, the

severely paretic arm is one of the most devastating.1 For
its alleviation, few effective therapeutic options exist. Basic
pain in the “virtual” limb was often relieved.4 MT was also pos-
tulated to alleviate chronic hemiparesis after stroke.5 In their
pilot study in 9 chronic stroke patients, Altschuler and col-
research demonstrated that the functional deficits after stroke leagues reported effects of this treatment on “patients’ move-
are determined by factors that include the extent of structural ment ability in terms of range of motion, speed, and accuracy,”
damage and the level of cortical stimulation during active or especially for patients with severe hemiparesis.6 Unfortunately,
passive movement of the affected limb.2 This mechanism the effects of the therapy were not described in detail, which
doubly disadvantages patients with severe hemiparesis. First, makes it difficult to understand the specific improvements
the motor impairment regularly prevents active use of the achieved. Subsequently, mainly small scale case studies have
arm for functionally relevant activities, leading to a reduction been published, employing MT in combination with various
of its cortical representation. Second, severe hemiparesis is other therapy approaches.7-9 In a randomized controlled study on
often accompanied by sensory deficits.3 Thus, even when chronic stroke patients, Rothgangel and coworkers reported
limb usage is increased (eg, during therapies), the resulting functional improvement during MT, but the 2 therapy groups
cortical activation is limited. differed at baseline.10 Recently, the benefit of MT for the recov-
As an alternative, mirror therapy (MT) has been proposed as ery of lower limb movements in subacute and chronic stroke
potentially beneficial. For this approach, a mirror is placed in the patients was demonstrated in a high-quality randomized con-
participant’s midsagittal plane, presenting the patient the mirror trolled trial design.11
image of his or her nonaffected arm as if it were the affected one The concept of MT has been further substantiated neuro-
(Figure 1). This approach was first introduced by Ramachandran physiologically. An imaging experiment demonstrated that
and coworkers for arm amputees, where the mirror image of the inversion of the visual image of a hand can elicit lateralized
intact arm was used to simulate its amputated counterpart. By cortical activations.12 In other words, when a right hand is
this procedure, illusory perceptions were induced and phantom used, but perceived as a left hand, this leads to an additional

From the Klinik Berlin, Department of Neurological Rehabilitation, Charite-University Medicine Berlin, Campus Benjamin Franklin, Germany (CD); Godeshöhe
Neurological Rehabilitation Center, Bonn, Germany (CD, JP, AN, JK, HK); and Center for Evaluation and Methods, Department of Psychology, University of
Bonn, Germany (CR, JP). Address correspondence to Christian Dohle, MD, Charite-Universitätsmedizin Berlin, Campus Benjamin Franklin, Abteilung für
Neurologische Rehabilitation, MEDIAN Klinik Berlin, Kladower Damm 223, D-14089 Berlin, Germany. E-mail: dohle.berlin@median-kliniken.de.

209
210   Neurorehabilitation and Neural Repair

Figure 1 pressure, hemicraniectomy or orthopedic, rheumatologic, or

Setup for Mirror Therapy other diseases interfering with their ability to sit or to move
either upper limb. Lesion localization (cortical/subcortical)
was assessed on the basis of the brain scans available (CT or
MRI). Handedness, prior to the stroke, was assessed by self-
report, or report of the family for aphasic patients.
As usual in Germany, the patients’ individual health insur-
ance had the final decision about the duration of inpatient
rehabilitation. Thus, some study patients (see Results section)
could not finish the study intervention because of early discharge.
The study was approved by the local ethics committee and
registered at Current Controlled Trials Ltd (ISRCTN31849226).
All patients or their legal representatives gave written informed
consent prior to the study. Consent was given separately for
participation and videotaping.

Intervention Protocol
In addition to the standard therapy delivered at the reha-
bilitation center, all patients underwent 6 weeks of study inter-
Note: The patient’s affected arm is hidden behind the mirror. While she is vention (30 minutes a day, 5 days a week) administered by one
moving her unaffected arm, she is watching its mirror image as if it were the of the authors (A. Nakaten or J. Püllen). A standardized ther-
affected one. apy protocol was designed, requiring the execution of arm,
hand, and finger postures in response to verbal instructions.
By variation of the number of different configurations required
activation of the right hemisphere (and vice versa). As recovery simultaneously, this protocol could be scaled according to the
mechanisms are known to be most prominent within the first patients’ actual level of performance (shaping). During MT,
3 months after stroke,13 it is reasonable to assume that MT patients watched the mirror image of the unaffected arm as if
might be most effective when applied within this time win- it were the affected one. During control therapy (CT), no mir-
dow. In summary, there is increasing evidence that MT might ror was present, so patients had direct view of the affected
be an effective method to support recovery from severe hemi- arm. During both therapy interventions, patients were reminded
paresis beyond more established rehabilitation procedures to move their affected limb “as well as possible,” in accor-
based on active or passive movement execution. However, it dance with the initial protocol of Altschuler and coworkers.18
remains unclear which symptoms can be improved. Thus, the Thus, the therapy protocol of both therapy groups did not dif-
following single-blinded randomized trial was designed to fer in motor performance, but only in the type of visual feed-
evaluate the potential beneficial effect of viewing the mirror back. Patients were informed about the existence of 2 therapy
image of the unaffected upper limb on recovery in patients with groups, but not about the study hypothesis. Thus, they were
severe hemiparesis early after stroke. As previous data indicated not aware about their allocation to the experimental group
different degrees of lateralization for proximal and distal motor (MT) or control group (CT). To control for differences in moti-
function,14-16 these aspects were analyzed separately. Preliminary vation and cooperation during the therapy sessions, each treat-
data have been reported in abstract form.17 ment session included an estimation of the patients’ vigilance
(1-3; 2 representing normal) and alertness (1-3; 1 representing
fully alert). The resulting estimates were averaged across ses-
Methods
sions for each patient. Patients were excluded from the study
Patients if they missed more than 4 therapy sessions for any reason.
Standard therapy at the hospital was applied without any
Patients were recruited from all inpatient admissions at the restrictions. The amount of treatment with regular occupa-
Godeshöhe Rehabilitation Center between October 2004 and tional therapy (OT), physiotherapy (PT), and activities of daily
April 2006. Our study was restricted to patients with severe living (ADL) training, as well as the duration of antidepressant
hemiparesis because of a first-ever ischemic stroke confined to medication, was extracted from the patients’ clinical documen-
the territory of middle cerebral artery, occurring no more than tation after discharge.
8 weeks prior to study inclusion. Patients had to be between 25
and 80 years of age, able to follow the therapy instructions,
Therapy Allocation
and capable of participating in 30-minute daily therapy ses-
sions. Patients were excluded if they had experienced previous One of the authors (C. Rietz) created sealed, numbered
strokes, major hemorrhagic changes, increased intracranial envelopes with the randomization sequence, allocating patients
 Dohle et al / Mirror Therapy   211  

either to MT or CT. Others (C. Dohle, J. Püllen, A. Nakaten) by Zimmermann and Fimm were employed. These tests have
selected subjects based on the inclusion and exclusion criteria. floor and ceiling effects at different neglect severities.26,27
The seal was broken after study inclusion and completion of Thus, a 5-point neglect score was defined as follows:
the initial testing procedures (see below).
(0) BIT = impaired (including drawing and copying), TAP =
Assessment clearly impaired, many omissions, complete hemifield

The primary outcome measures were improvements in the (1) BIT = deficits in cancellation and bisection subtests, TAP =
7 upper limb subscores (see below) of the Fugl-Meyer test.18 some omissions, not complete hemifield
For patients with neglect symptoms, the results of the neglect (2) BIT = normal performance, TAP = single omissions, differ-
testing served as secondary outcome measure. Additionally, ences between sides
the Action Research Arm test19 and the motor part (first 13 (3) BIT = normal performance, TAP = reaction time differences
items) of the Functional Independence Measure (FIM)20 were
recorded. The entire assessment was performed before (t1) and (4) No signs of visual hemineglect
after the intervention (t2) and evaluated by independent raters.
The Fugl-Meyer test and the Action Research Arm test were For any given patient and time, this rating was always unam-
videotaped by one of the investigators (either A. Nakaten or J. biguous. For study purposes, it was applied independently by
Püllen) and assessed at the end of the study by 2 out of 3 inde- 2 blinded raters who discussed divergent judgments until they
pendent raters who were not involved in the study. For each agreed on a common score.
single item rating, the average value of the 2 raters’ results was
used for analysis. Motor FIM and neuropsychological testing
were assessed by independent raters, who were not aware of
Data Analysis
the patients’ group assignment, from the sections of OT and Data analysis was performed using SPSS for Windows,
cognitive therapy. Prior to assessment, all raters received spe- version 12.0.1. Only patients who completed the entire therapy
cific training on the tests used. course were included in the analysis. Patients who dropped out
The Fugl-Meyer upper extremity test consists of a total of were lost to follow-up, thus an intention-to-treat analysis was
63 items grouped into 9 parts (A to J), scoring all major neu- not possible. Demographic variables were compared by
rological symptoms on an ordinal scale from 0 to 2, with 2 unpaired t tests or U tests, depending on the results of the
representing no deficit.18 The total upper extremity motor Kolmogorov-Smirnov test for normality of distributions. For
score has been used and evaluated in a number of clinical Fugl-Meyer and Action Research Arm test scoring, Spearman
studies.21 Subdivisions of this score for proximal and distal correlation coefficients for each possible pairing of the 3 raters
function have been employed22 and successfully correlated served as measures of interrater reliability.
with electrophysiological measures.23,24 For this study, 57 Assessment of the therapy effect, on improvement in the
items were utilized, grouped for motor assessment and nonmo- different neurological modalities, was confounded by sponta-
tor assessment. For motor assessment, subscores for proximal neous recovery. Especially, it had to be considered that patients
arm (part A without reflex assessment = 15 items), hand (part scoring better at the time of the initial testing were likely to
B = 5 items), and finger function (part C = 7 items) were used. reach higher final scores than those with worse initial scores.
For assessment of nonmotor signs, the upper extremity sub- Thus, an analysis of covariance (ANCOVA) approach was
scores for surface sensibility (light touch, part Ha = 2 items), used28: final values (measured at t2) in the different scores
proprioception (movement mirroring, part Hb = 4 items), joint were subjected to an analysis of variance (ANOVA) with the
pain during passive movement (part J = 12 items), and range therapy protocol (MT, CT) as the factor and the initial score
of motion (part J = 12 items) were employed. Interrater cor- (measured at t1) as the covariate.
relations served to validate this division. The Action Research The illusory experience during MT (ie, the divergence
Arm test consists of the 4 subscales grasp, grip, pinch, and between the visual impression and the actually performed
gross movement. The test contains 19 movement tasks, with movement) is strongest when patients are not able to move
each task graded on a 4-point scale (total score ranging from their limb at all. This might lead to a greater therapeutic effect
0-57). The motor part of the FIM contains 11 items, measuring in this patient group.6 Thus, the analysis for the 3 motor scores
performance in self-caring and mobility on a 7-point scale was performed separately for the subgroups of patients who
(total score ranging from 7-77). obtained scores of zero at initial testing, ie, those that had no
Patients were classified as aphasic when their Token test motor function at all (initial plegia). For ancillary analysis, the
t value was below 60. For assessment of hemineglect, several side of the lesioned hemisphere and the latency between stroke
subtests of the Behavioral Inattention test (BIT)25 (line cancel- occurrence and study inclusion were included as cofactors
lation, star cancellation, letter cancellation, figure and shape and covariates. As hypotheses were prespecified, no adjust-
copying, line bisection, representational drawing, and article ments were made to the reported P values. Effect sizes were
reading) as well as the omissions and reaction times in each calculated manually, implementing established formulas29 into
visual hemifield in the tests of attentional performance (TAP) Microsoft Excel 2000.
212   Neurorehabilitation and Neural Repair

Table 1
Demographic Data and Characteristics of Standard and Intervention Treatment
Mean (SD) or Numbers

Parameter Normal Distribution CT MT Level of Significance

Age (years) + 58.0 (14.0) 54.9 (13.8) ns

Men/women – 13/5 13/5 ns
Right handed/left handed – 18/0 16/2 ns
CT/MRI – 13/5 14/4 ns
Cortical/subcortical lesion – 15/14 15/14 ns
Lesion of the dominant/nondominant hemisphere –   7/11   4/14 ns
Aphasia – 6 4 ns
Latency between stroke and study inclusion (days) + 27.8 (12.1) 26.2 (8.3) ns
Amount of standard treatment
   Occupational therapy (hours) + 12.3 (5.0) 14.7 (3.4) ns
   Physical therapy (hours) + 23.8 (5.5) 24.7 (5.4) ns
   ADL training (45-minute units) + 11.4 (6.5) 5.9 (6.2) 0.024
   Antidepressive treatment (days) – 21.1 (23.5) 25.0 (23.7) ns
Intervention
   Duration (days) + 47.0 (4.7) 45.8 (2.8) ns
   Number of sessions – 29.0 (1.4) 28.6 (1.4) ns
   Mean vigilance + 1.89 (0.21) 1.92 (0.23) ns
   Mean alertness + 1.15 (0.21) 1.16 (0.15) ns

Abbreviations: SD, standard deviation; CT, control therapy; MT, mirror therapy.

Power Calculation patients dropping out were: transfers to acute hospital (CT = 2,
MT = 1); medical worsening (CT = 1, MT = 0); lack of cost
Power calculation is dependent on the type of score that is approval by the health insurance (see methods section; CT =
employed (eg, neurological function or ADL capacity). 1, MT = 4); or withdrawal of patients’ consent (CT = 2, MT =
Previous studies suggested that a specific intervention could 1). Thirty-six patients finished 1 of the 2 therapy protocols.
result in increases on the basic sensorimotor level (such as that There were 18 in each group. Their demographic data and
captured with the Fugl-Meyer subscores) with an effect size of details of their treatment course are depicted in Table 1.
about 0.4.30 For severely affected limbs, effect sizes seem to be The only statistically significant imbalance between both
even higher.31 Thus, supposing an effect size of 0.6, α = 0.05, groups was the amount of ADL treatment, disadvantaging the
1-β = 0.8 and including the increase of power attained by use MT group. No significant differences could be established for
of the ANCOVA,32 a total number of 36 patients was calcu- any other demographic parameter, either in the entire group or
lated to be necessary. Assuming a dropout rate of 33% and in any of the subgroups described below. Patients’ attention
considering a further safety margin, we initially prepared to and vigilance during study performance (as markers for patient
include 60 patients during the recruitment period. The study cooperation and potential treatment bias by the nonblinded
was not powered to detect differences on the Action Research therapists) were similar in both groups.
Arm test or FIM scale, thus these values were not analyzed by
means of the ANCOVA. During the recruitment period, it
turned out that both the recruitment rate and the dropout rate Interrater Reliability
were below expectations. Thus, the recruitment period was The interrater correlation coefficients for each possible
prolonged based on the observed figures, until the targeted pairing of the 2 raters are shown in Table 2. All correlations
figures were attained. No analysis was performed before fin- were significant at P < .0001, thus even higher than those
ishing the entire intervention and assessment. reported previously,33,34 and further justifying the use of the
different Fugl-Meyer subscores in the study.
Results
Therapy Effects
Patient Characteristics The mean values of the different Fugl-Meyer subscores are
During the recruitment period a total of 48 patients met the displayed in Figures 2 and 3. As apparent in the figures, the
inclusion and exclusion criteria, agreed to participate in the mean values of the motor subscores, surface sensibility, and
study, and were randomized. During the course of the study, proprioception improved in both therapy groups because of
12 patients (6 in each group = 25%) dropped out. Reasons for the spontaneous recovery and the standard therapy delivered at
 Dohle et al / Mirror Therapy   213  

Table 2 Figure 2
Interrater Correlations of Fugl-Meyer Subscores Group Data of the Mean Fugl-Meyer Motor Subscores
and Action Research Arm Test (Normalized to 0-2 Points for Each Category)
(Based on Videotaped Observations)
Interrater Spearman
Correlation Coefficients

Rater Rater Rater

1 and 2 2 and 3 1 and 3
(n = 42) (n = 5) (n = 25)

Fugl-Meyer subscores
   Motor arm 0.997 1.000 1.000
   Motor hand 0.991 1.000 0.977
   Motor finger 0.996 1.000 0.996
   Touch 0.947 1.000 1.000
   Proprioception 0.995 1.000 1.000
   Range of motion 0.985 1.000 1.000
   Pain 0.974 1.000 0.979
Action Research Arm test 0.998 1.000 0.998

the hospital. Subscores for range of motion and pain showed a

slight decrease from nearly normal values at t1, suggesting
that these symptoms only occur in the subacute and chronic
stage after stroke. All significant therapy effects reported
below were in favor of the MT group. No adverse events or
side effects were noted in any of the 2 therapy groups.
Regarding motor function, there was no significant therapy
effect in any of the 3 motor subscores across all patients
(Figure 2A). Only the finger motor score revealed a tendency
that failed to reach significance (F [1, 35] = 0.9, ns). This ten-
dency was because of a significant difference in the subgroup Note: Indicated are mean values and standard deviations of the patient groups
of those 25 patients who were initially distal plegic (Figure receiving control therapy (CT) or mirror therapy (MT) before the intervention (t1)
2B; F [1, 24] = 4.4, P = .048, effect size ε = 0.78). In absolute and after it (t2). Upper panel (A) shows group data of all patients, and lower panel
terms, mean improvement of the MT group was 4.4 (95% (B) shows group data of the 3 subgroups of patients with no function at all
at t1 in the different motor categories (see Methods/Results sections).
CI = 2.4-6.4) on the 14-point Fugl-Meyer subscale compared
to a mean improvement of 1.5 (95% CI =  -0.6-3.6) in the
Figure 3
CT group. For the patient subgroups with initially plegic hand
Group Data of Nonmotor Symptoms
(n = 34) and arm (n = 18), no difference between the 2 therapy
groups could be established.
The beneficial effect of MT had also functional conse-
quences for regaining useful reach and grasp movements, as
assessed with the Action Research Arm test (Table 3). Among
the initially distal plegic patients who received CT, only 1 out
of 12 made improvement at the functional level (Action
Research Arm test > 1) where the post-therapy score was 2.5.
In the MT group, this was true for 4 out of 13 patients, where
the maximum score was 21.
Regarding nonmotor symptoms, improvement of surface
sensibility (light touch) was significantly different between the
2 treatment groups (Figure 3; F [1, 35] = 7.7, P = .009, effect
size ε = 0.57). In absolute terms, mean improvement for
MT patients was 0.8 (95% CI = 0.5-1.1) compared to 0.2 (95% Note: Left, shows mean Fugl-Meyer nonmotor subscores (normalized to 0-2
points for each category) of all patients; right, shows neglect scores (see Methods
CI = -0.1-0.5) for CT patients on the 4-point Fugl-Meyer sub-
section) for the 20 patients with neglect symptoms at t1. Indicated are mean values
scale. For proprioception, the final difference between therapy and standard deviations of the patient groups receiving control therapy (CT) or
groups was not because of an effect of therapy (F [1, 35] = mirror therapy (MT) before the intervention (t1) and after it (t2).
214   Neurorehabilitation and Neural Repair

Table 3
Results of Functional Testing for the Entire Group and the 2 Subgroups With Statistical Significant Effects
Mean ARAT (SD) Mean Motor FIM (SD)

Patient Population Therapy Group Initial (t1) Final (t2) Initial (t1) Final (t2)

All (n = 36) CT (n = 18) 0.8 (2.1) 3.9 (7.9) 43.9 (13.1) 60.8 (13.0)
MT (n = 18) 0.6 (2.1) 4.7 (12.5) 48.3 (12.3) 66.6 (9.4)
Initially distal plegic (n = 25) CT (n = 12) 0.0 (0.0) 0.4 (0.8) 42.3 (12.9) 58.4 (14.2)
MT (n = 13) 0.0 (0.0) 2.5 (5.8) 47.5 (12.6) 65.8 (9.2)
Former right-handed patients with right CT (n = 9) 1.5 (2.8) 7.4 (10.3) 36.7 (11.2) 50.1 (9.4)
   hemispheric lesions and neglect (n = 20) MT (n = 11) 0.8 (2.7) 6.8 (15.8) 41.3 (10.3) 54.8 (9.8)

Abbreviations: ARAT, Action Research Arm test; SD, standard deviation; FIM, Functional Independence Measure; CT, control therapy; MT, mirror therapy.

0.4, ns), but rather to differences at the baseline level. Range execution.35,36 Apparently, this modulation of excitability con-
of motion and pain showed no therapy effect at all. tributes to motor recovery, even in an initially plegic limb. In
Signs of hemineglect at the beginning of the therapy were our study, this effect is only present for distal arm muscles and
present in 20 out of the 24 right-handed patients with right not for proximal arm muscles. This is in accord with previous
hemispheric lesions (CT = 9/11, MT = 11/13). Among these data, demonstrating a different contribution of both hemi-
patients, improvement of the neglect score was significantly spheres for proximal and distal motor functions.14-16 There is
greater in the MT group (mean = 0.9, 95 % CI = 0.6-1.2) than evidence that the distal component is organized strictly
in the CT group (mean = 0.2, 95% CI = -0.2-0.5) (Figure 3; F unilaterally,37 whereas proximal movements rely more on
[1, 19] = 10.4, P = .005, effect size ε = 0.99). bihemispheric representations.38 Thus, we propose that move-
For the level of ADL capacity (as measured with the motor ment mirroring mainly stimulates lateralized motor represen-
FIM), no difference between both therapy groups could be tations for the distal limb.
established either in the entire group or in the relevant sub- The improvement of sensory deficits further confirms the
groups (Table 3). Small differences favoring the MT group tight coupling of vision and touch. It has been shown that
were already present at the beginning of the therapy. Even movement observation modulates not only motor cortex excit-
though the CT patients received significantly more ADL train- ability, but also cortical somatosensory representations.39
ing, this difference persisted after treatment. Viewing a stimulated body part enhances discrimination abil-
When included as a second factor in the ANCOVA, no ity both in normal and in brain-damaged participants,40 accom-
analysis revealed a significant effect of the side of the lesioned panied by changes in excitability of the primary somatosensory
hemisphere—whether anatomical (right/left) or functional cortex.41 Watching stimulation in a mirror can lead to a referral
(dominant/nondominant). Similarly, inclusion of latency of sensation to the other hand.42 Our results indicate that these
between stroke occurrence and inclusion into the study as a cross-modal processes can also be employed therapeutically
second covariate showed no effect. Furthermore, there was no for long-term enhancement of somatosensory perception. This
obvious effect of lesion locus (cortical or subcortical; Table 1). further supports the hypothesis that patients with sensory defi-
More detailed lesion analysis was not possible because of the cits benefit especially from MT.7 However, our results on
lack of brain scans of equal quality (ie, MRI) for all patients. somatosensory function are only based on the surface sensibil-
ity subscore of the Fugl-Meyer test. Although less detailed
Discussion than the motor subscores, these scores are still sufficiently
valid.32,33 Additional studies are required to explore the effect
We demonstrated, in the present study, that application of of MT on sensory functions more specifically.
MT in the early phase after stroke resulted in functionally The impact of MT on attentional processes is further illus-
relevant improvements in motor, sensory, and attentional trated by its beneficial effect on hemineglect. Interestingly,
domains. These improvements were not because of a nonspe- Ramachandran and coworkers originally proposed alleviation
cific, global, beneficial effect. Besides, as demonstrated by the of hemineglect the other way around. They tried to stimulate
assessment of vigilance and alertness of patients in the MT and awareness for the affected side by placing a mirror on the
CT groups, they cannot be attributed to a treatment bias caused unaffected side of neglect patients.43 In our study, the mirror
by insufficient blinding. The effects are in accord with basic was placed in the neglected hemifield. Apparently, watching a
neurophysiological findings, confirming a role of observing healthy moving arm and hand in the neglected hemifield pro-
mirrored movement in cortical stimulation. vides a stronger stimulus for recovery from neglect than
Regarding improvement of motor functions, it has been watching the attempted movements of a paretic side. One may
demonstrated that observation of mirrored distal movements assume that this improvement of hemineglect promotes recov-
enhances corticospinal excitability, similar to actual movement ery in the motor and sensory domain. In our study, however,
 Dohle et al / Mirror Therapy   215  

similar sensorimotor improvements were observed for patients our study to patients early after a first-ever ischemic stroke
with lesions of the dominant and nondominant hemisphere. It confined to the territory of the middle cerebral artery. In prin-
should be pointed out that our neglect rating was based on a ciple, however, the results could be generalized to all neuro-
score that we devised and the validity has not been proven logical conditions with severe hemiparesis because of a
explicitly. Thus, we regard the improvement of hemineglect as unihemispheric lesion.
a positive side effect whose independent therapeutic value Taking all our results together, we found a clear, function-
remains to be proven. At the very least, we have demonstrated ally relevant effect of MT on sensorimotor recovery that is in
that MT is also successfully applicable for patients with severe good accordance with neurophysiological findings. The effect
hemineglect. Again, further studies are required to explore the on regaining ADL capacity was less pronounced. In our study,
interplay between recovery in the attentional and sensorimotor the ADL testing of the 2 therapy groups showed some (not
domain more specifically. significant) difference at the baseline level, slightly advantag-
The contribution of distinct cortical areas to the processes ing the MT group. This difference was not reflected at the
mediating recovery, and thus the precise mechanism of MT, basic sensorimotor level. As only patients with severe hemipa-
remain speculative. Frequently, effects of MT are attributed to resis with very limited functional capacity of the affected limb
“mirror neurons,” ie, neurons in the premotor area of both mon- were included, initial ADL scoring mainly reflected patients’
keys and humans that are active during observation of meaning- greater ability to use compensatory (ie, one hand) techniques.
ful movements.44,45 However, in the only imaging experiment on The higher amount of ADL training that was necessary for the
inverted visual feedback, lateralized activations were not control group supports this interpretation. The final ADL scor-
recorded in the premotor area, but in occipital and posterior ing represents both contributions from regained sensorimotor
parietal regions.12 We assume that the precuneus region (area function in the affected limb and acquired compensatory
V6) plays a decisive role. This area belongs to the neural net- techniques, which are difficult to disentangle.54
work supporting the mental representation of the self.46 It might The effect of MT on recovery of basic motor functions
well be that premotor areas are activated bilaterally, without appears to be most prominent for those patients who have no
lateralization because of the observed body side.47 Thus, the distal function at the beginning of the therapy. The recovery
beneficial effect of MT is possibly mediated by the visual illu- process might be further supported by gains in sensory function
sion that actions carried out by oneself are performed normally. and a possible beneficial effect on hemineglect. This fact is of
It is quite probable that this illusion can prevent, or at least major importance both clinically and economically as many
reduce “learned non-use” of a paretic limb.5 modern rehabilitative concepts, such as constraint-induced
In our study, the effects were observed in the subacute phase movement therapy, can lead to significant functional improve-
after stroke. Within the chosen time frame of 8 weeks after ments, but only when some distal motor function is already
stroke, we found no influence of the latency between onset of present at the beginning of the therapy.55,56 Based on our results,
symptoms and start of the therapy. It remains speculative systematic application of MT in densely hemiplegic patients
whether this result would also be valid for chronic stroke early after stroke might support the recovery of these motor
patients (> 3 months). Recent imaging experiments suggest dif- functions, allowing progress to other forms of therapy. Thus,
ferential involvement of the ipsilateral and contralateral hemi- when integrated into a modern neurorehabilitative program, the
sphere during different phases of recovery from stroke.48 It is long-term effect on arm function and ADL capacity of MT,
not known, however, if this implies different therapeutic strate- applied in the early phase after stroke, might be even greater
gies in different recovery phases.49 We assume that the basic than the immediate effect we recorded in our study.
therapeutic principle of repetitive, effective stimulation of the
lesioned hemisphere remains valid, irrespective of the time
interval between stroke and rehabilitation. However, it should Acknowledgments
be noted that the effects in our study are quite robust, despite the
The study was supported by “refonet”, the rehabilitation
great individual variability in spontaneous recovery from stroke.50
research network of the German Pension Scheme Rhineland
MT therapy is very easy to implement, even in an acute
(www.refonet.de; grant no. 0315). We are indebted to H.
setting, and patients can be instructed to train on their own.7
Pollmann and B. Wild from refonet for methodological advice
However, the optimum procedure with regard to frequency,
in designing the study. We owe special thanks to M. Knebel, I.
duration, and protocol remains to be established.51 In our
Atoudsie and A. Burdorf for patient data processing, to T.
study, we only investigated the effect of the inverted visual
Wullen for fundamental organizational support, and finally to
feedback, thus active movements of the affected side were
many members of the sections of occupational therapy and
those within the patients’ capabilities. However, MT can also
neuropsychology for reliable testing and assessment.
be performed with passive movement of the affected limb,
thus possibly adding the therapeutic value of bilateral arm train-
ing.52 There is clinical and neurophysiological evidence that References
this therapy variant is even more effective than the one we   1. Spieler JF, Lanöe JL, Amarenco P. Costs of stroke care according to handicap
used in our study.36,53 For methodological reasons, we restricted levels and stroke subtypes. Cerebrovasc Dis. 2004;17(2-3):134-142.
216   Neurorehabilitation and Neural Repair

  2. Lindberg PG, Schmitz C, Engardt M, Forssberg H, Borg J. Use-dependent 24. Hendricks HT, Pasman JW, Merx JL, van Limbeek J, Zwarts MJ.
up- and down-regulation of sensorimotor brain circuits in stroke patients. Analysis of recovery processes after stroke by means of transcranial
Neurorehabil Neural Repair. 2007;21:315-326. magnetic stimulation. J Clin Neurophysiol. 2003;20(3):188-195.
  3. Broeks JG, Lankhorst GJ, Rumping K, Prevo AJ. The long-term outcome 25. Wilson B, Cockburn J, Halligan P. Development of a behavioral test of
of arm function after stroke: results of a follow-up study. Disabil Rehabil. visuospatial neglect. Arch Phys Med Rehab. 1987;68(2):98-102.
1999;21(8):357-364. 26. Azouvi P, Samuel C, Louis-Dreyfus A, et al. Sensitivity of clinical and
  4. Ramachandran VS, Rogers-Ramachandran D, Cobb S. Touching the behavioural tests of spatial neglect after right hemisphere stroke. J Neurol
phantom limb. Nature. 1995;377(6549):489-490. Neurosurg Psych. 2002;73:160-166.
  5. Ramachandran VS. Phantom limbs, neglect syndromes, repressed memo- 27. Buxbaum LJ, Ferraro MK, Veramonti T, et al. Hemispatial neglect: sub-
ries, and Freudian psychology. Int Rev Neurobiol. 1994;37:291-333. types, neuroanatomy, and disability. Neurology. 2004;62(5):749-756.
  6. Altschuler EL, Wisdom SB, Stone L, et al. Rehabilitation of hemiparesis 28. Vickers AJ, Altman DG. Statistics notes: analysing controlled trials with
after stroke with a mirror. Lancet. 1999;353(9169):2035-2036. baseline and follow up measurements. BMJ. 2001;323(7321):1123-1124.
  7. Sathian K, Greenspan AI, Wolf SL. Doing it with mirrors: a case study of 29. Bortz J. Statistik für Sozialwissenschaftler. Berlin, Germany: Springer; 1993.
a novel approach to neurorehabilitation. Neurorehabil Neural Repair. 30. Ottenbacher KJ, Jannell S. The results of clinical trials in stroke rehabili-
2000;14(1):73-76. tation research. Arch Neurol. 1993;50(1):37-44.
  8. Stevens JA, Stoykov ME. Using motor imagery in the rehabilitation of 31. Kwakkel G, Wagenaar RC, Twisk JW, Lankhorst GJ, Koetsier JC.
hemiparesis. Arch Phys Med Rehabil. 2003;84(7):1090-1092. Intensity of leg and arm training after primary middle-cerebral-artery
  9. Stevens JA, Stoykov ME. Simulation of bilateral movement training through stroke: a randomized trial. Lancet. 1999;354(9174):191-196.
mirror reflection: A case report demonstrating an occupational therapy tech- 32. Mulligan NW, Wiesen C. Using the analysis of covariance to increase the
nique for hemiparesis. Top Stroke Rehabil. 2004;11(1):59-66. power of priming experiments. Can J Exp Psychol. 2003;57(3):152-166.
10. Rothgangel A, Morton A, van der Hout J, Beurskens A. Spiegeltherapie 33. Sanford J, Moreland J, Swanson LR, Stratford PW, Gowland C.
in der Neurologischen Rehabilitation: Effektivität in Bezug auf die Arm- Reliability of the Fugl-Meyer assessment for testing motor performance
und Handfunktion bei chronischen Schlaganfallpatienten. Neurol Rehabil. in patients following stroke. Phys Ther. 1993;73(7):447-454.
2007;13(5):271-276. 34. Gladstone DJ, Danells CJ, Black SE. The Fugl-Meyer assessment of
11. Sütbeyaz S, Yavuzer G, Sezer N, Koseoglu BF. Mirror therapy enhances motor recovery after stroke: a critical review of its measurement proper-
lower-extremity motor recovery and motor functioning after stroke: ties. Neurorehabil Neural Repair. 2002;16(3):232-240.
a randomized controlled trial. Arch Phys Med Rehab. 2007;88(5): 35. Garry MI, Loftus A, Summers JJ. Mirror, mirror on the wall: viewing a
555-559. mirror reflection of unilateral hand movements facilitates ipsilateral M1
12. Dohle C, Kleiser R, Seitz RJ, Freund HJ. Body scheme gates visual pro- excitability. Exp Brain Res. 2005;163(1):118-122.
cessing. J Neurophysiol. 2004;91(5):2376-2379. 36. Fukumura K, Sugawara K, Tanabe S, Ushiba J, Tomita Y. Influence of mirror
13. Jorgensen HS, Nakayama H, Raaschou HO, Vive-Larsen J, Stoier M, therapy on human motor cortex. Int J Neurosci. 2007;117(7):1039-1048.
Olsen TS. Outcome and time course of recovery in stroke. Part II: time 37. Parsons LM, Gabrieli JD, Phelps EA, Gazzaniga MS. Cerebrally lateral-
course of recovery. The Copenhagen Stroke Study. Arch Phys Med ized mental representations of hand shape and movement. J Neurosci.
Rehabil. 1995;76(5):406-412. 1998;18(16):6539-6548.
14. Müller F, Kunesch E, Binkofski F, Freund HJ. Residual sensorimotor 38. Bawa P, Hamm JD, Dhillon P, Gross PA. Bilateral responses of upper
functions in a patient after right-sided hemispherectomy. Neuropsychologia. limb muscles to transcranial magnetic stimulation in human subjects. Exp
1991;29(2):125-145. Brain Res. 2004;158(3):385-390.
15. Binkofski F, Dohle C, Posse S, et al. Human anterior intraparietal area 39. Rossi S, Tecchio F, Pasqualetti P, et al. Somatosensory processing during
subserves prehension: a combined lesion and functional MRI activation movement observation in humans. Clin Neurophysiol. 2002;113(1):16-24.
study. Neurology. 1998;50(5):1253-1259. 40. Serino A, Farnè A, Rinaldesi ML, Haggard P, Làdavas E. Can vision of
16. Dohle C, Ostermann G, Hefter H, Freund HJ. Different coupling for the the body ameliorate impaired somatosensory function? Neuropsychologia.
reach and the grasp components in bimanual prehension movements. 2007;45(5):1101-1107.
Neuroreport. 2000;11(17):3787-3791. 41. Schaefer M, Flor H, Heinze HJ, Rotte M. Dynamic modulation of the
17. Dohle C, Püllen J, Nakaten A, et al. The effect of mirror therapy in the primary somatosensory cortex during seeing and feeling and touched
acute phase of stroke recovery. Paper presented at: Society for hand. Neuroimage. 2006;29(2):587-592.
Neuroscience 36th Annual Meeting; 2006; Atlanta, GA. 42. Sathian K. Intermanual referral of sensation to anesthetic hands.
18. Fugl-Meyer AR, Jääskö L, Leyman I, Olsson S, Steglind S. The post- Neurology. 2000;54(9):1866-1868.
stroke hemiplegic patient. 1. a method for evaluation of physical perfor- 43. Ramachandran VS, Altschuler EL, Stone L, Al-Aboudi M, Schwartz E,
mance. Scand J Rehabil Med. 1975;7(1):13-31. Siva N. Can mirrors alleviate visual hemineglect? Med Hypotheses.
19. Yozbatiran N, Der-Yeghiaian L, Cramer SC. A standardized approach to 1999;52(4):303-305.
performing the Action Research Arm test. Neurorehabil Neural Repair. 44. Rizzolatti G, Fadiga L, Gallese V, Fogassi L. Premotor cortex and the recog-
2008;22:78-90. nition of motor actions. Brain Res Cogn Brain Res. 1996;3(2):131-141.
20. Granger CV, Hamilton BB, Linacre JM, Heinemann AW, Wright BD. 45. Buccino G, Binkofski F, Fink GR, et al. Action observation activates
Performance profiles of the functional independence measure. Am J Phys premotor and parietal areas in a somatotopic manner: an fMRI study. Eur
Med Rehabil. 1993;72(2):84-89. J Neurosci. 2001;13(2):400-404.
21. Platz T, Pinkowski C, van Wijck F, Johnson G. ARM - Arm Rehabilitation 46. Cavanna AE, Trimble MR. The precuneus: a review of its functional
Measurement. Baden-Baden, Germany: Deutscher Wissenschafts-Verlag; anatomy and behavioural correlates. Brain. 2006;129:564-583.
2005. 47. Shmuelof L, Zohary E. A mirror representation of others’ actions in the
22. Feys HM, De Weerdt WJ, Selz BE, et al. Effect of a therapeutic interven- human anterior parietal cortex. J Neurosci. 2006;26(38):9736-9742.
tion for the hemiplegic upper limb in the acute phase after stroke: a 48. Feydy A, Carlier R, Roby-Brami A. Longitudinal study of motor recovery
single-blind, randomized, controlled multicenter trial. Stroke. 1998;29(4): after stroke: recruitment and focusing of brain activation. Stroke.
785-792. 2002;33(6):1610-1617.
23. Hendricks HT, Pasman JW, van Limbeek J, Zwarts MJ. Motor evoked 49. Page SJ, Gater D, Bach-Y-Rita P. Reconsidering the motor recovery
potentials in predicting recovery from upper extremity paralysis after plateau in stroke rehabilitation. Arch Phys Med Rehab. 2004;85(8):
acute stroke. Cerebrovasc Dis. 2003;16(3):265-271. 1377-1381.
 Dohle et al / Mirror Therapy   217  

50. Duncan PW, Goldstein LB, Matchar D, Divine GW, Feussner J. 54. Nakayama H, Jorgensen HS, Raaschou HO, Olsen TS. Compensation in
Measurement of motor recovery after stroke. Outcome assessment and recovery of upper extremity function after stroke: the Copenhagen stroke
sample size requirements. Stroke. 1992;23(8):1084-1089. study. Arch Phys Med Rehabil. 1994;75:852-857.
51. Dohle C, Nakaten A, Püllen J, Rietz C, Karbe H. Grundlagen und 55. Wolf SL, Winstein CJ, Miller JP, et al. Effect of constraint-induced
Anwendung des Spiegeltrainings. In: Minkwitz K, Scholz E, eds. movement therapy on upper extremity function 3 to 9 months
Standardisierte Therapieverfahren und Grundlagen des Lernens in der after stroke: the EXCITE randomized clinical trial. JAMA. 2006;296(17):
Neurologie. Idstein, Germany: Schulz-Kirchner-Verlag; 2005:59-68. 2095-2104.
52. Cauraugh JH, Summers JJ. Neural plasticity and bilateral movements: A 56. Fritz SL, Light KE, Patterson TS, Behrman AL, Davis SB. Active
rehabilitation approach for chronic stroke. Prog Neurobiol. 2005;75(5): finger extension predicts outcomes after constraint-induced movement
309-320. therapy for individuals with hemiparesis after stroke. Stroke. 2005;36(6):
53. Miltner R, Simon U, Netz J, Hömberg V. Bewegungsvorstellung in der 1172-1177.
Therapie von Patienten mit Hirninfarkt. Neurol Rehabil. 1999;5(2):66-72.