Clinical Rehabilitation 1998; 12: 11–22

Electromyographic biofeedback for gait training

after stroke
Lesley Bradley Physiotherapy Services, Barry B Hart, Suniti Mandana Department of Psychology, Scunthorpe General
Hospital, Scunthorpe, North Lincolnshire, Kenneth Flowers Department of Psychology, University of Hull,
Hull, Mary Riches Physiotherapy Services and Peter Sanderson Department of Psychology, Scunthorpe General
Hospital, Scunthorpe, North Lincolnshire, UK

Objective: To examine the effects of electromyographic (EMG) biofeedback

training on the recovery of gait in the acute phase post stroke.
Design: Patients were randomly assigned to EMG biofeedback or control
groups. They received treatment three times a week for six weeks. All
patients were assessed prior to treatment, after 18 treatment sessions, and at
three months follow-up.
Setting: The study was carried out at Scunthorpe General Hospital in North
Lincolnshire. The subjects were acute stroke patients who had been admitted
on to the medical and elderly wards.
Interventions: The EMG biofeedback group were treated using EMG as an
adjunct to physiotherapy. The patients were encouraged to facilitate or inhibit
abnormal muscle tone via auditory or visual signals transmitted from
electrodes placed over the appropriate muscles. The control group were
treated using the same techniques, electrodes were used with this group of
patients, but the EMG machine was turned off and faced away from the
patient and the therapist to control the placebo effect.
Outcome measures: A large battery of outcome measures was used for
physical and psychological assessment. The physical measures consisted of
active movement, muscle tone, sensation, proprioception, mobility and
activities of daily living (ADL). The psychological measures included orienta-
tion, memory, spatial performance, language and IQ.
Results: Twenty-one patients were included in the study. Scores were
combined into four groups: mild EMG, severe EMG, mild control and severe
control. Results showed that there was an improvement in physical scores for
active movement, mobility and ADL over time, but there was no significant
difference between the EMG and control groups. Scores on the psychological
tests were within normal limits, and there was no difference in performance
between the EMG and control groups.
Conclusions: This study showed no significant differences in the rate of
improvement after stroke between the two groups. Although EMG biofeed-
back was used as an adjunct to physiotherapy and represented clinical
practice, the results provide little evidence to support the clinical significance
of using EMG biofeedback to improve gait in the acute phase after stroke.

Address for correspondence: Lesley Bradley, Physiotherapy

Services Manager, West Lindsey NHS Trust, John Coupland
Hospital, Ropery Road, Gainsborough, Lincolnshire
DN21 2TJ, UK.
© Arnold 1998 0269–2155(98)CR157OA

Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015

12 L Bradley et al.
Introduction
Electromyographic (EMG) biofeedback is the
process whereby changes in muscle activity are
relayed to a patient in a meaningful way by an
auditory signal and/or in a visual form.1 The lit-
erature on the application of EMG biofeedback
in neuromuscular disorders over the past 25 years
has suggested that its application to stroke
patients can enhance recovery of motor function
and may supplement physiotherapeutic tech-
niques. However, there is no conclusive evidence
for its effectiveness and further investigation has
been suggested.2
Many studies have looked at the effect of
EMG biofeedback training on motor recovery in
chronic stroke patients (i.e. one year or more
post stroke). Basmajian et al.3 compared the
effectiveness of EMG biofeedback to conven-
tional therapy with 20 patients experiencing
chronic footdrop. Improvements in the EMG
group were reported to be twice those made in
the conventional group, but this study was incon-
clusive due to the lack of data analysis. Similar
results were reported by Burnside et al.4 who
found that 11 patients receiving EMG biofeed-
back for footdrop and gait training made signifi-
cant gains in strength and gait compared to 11
patients who received conventional therapy; even
though the treatment in this study was short (two
15-min sessions per week for six weeks) and car-
ried out by an untrained therapist. Prevo et al.5
compared the effectiveness of EMG biofeedback
with conventional therapy on 18 chronic patients
for arm and hand function but found only small
improvements, which were not maintained once
the EMG biofeedback training had stopped.
Other studies with chronic stroke patients have
failed to provide clear evidence to support the
use of EMG biofeedback due to poor study
designs and lack of statistical analysis (see
reviews in refs 6–8).
Two recent meta-analyses of EMG biofeed-
back therapy in stroke rehabilitation produced
conflicting results. Schleenbaker and Mainous9
found an overall positive pooled effect size for
this modality, while Glanz et al.10 did not.
Schleenbacker included studies with nonrandom-
ized controls, used any improvement in neuro-
logical function as an outcome measure, and
Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015
grouped upper and lower extremity data. Thus,
there was little overlap of included studies in
their analyses, and a larger number of subjects
(299) was included, compared to the meta-analy-
sis of Glanz et al. who studied 200 patients. Glanz
et al. looked at eight EMG biofeedback studies
to assess its efficacy in post-stroke rehabilitation.
Change in range of movement of the affected
joint was chosen as the outcome measure. When
the results were pooled there was no significant
effect in favour of the use of EMG biofeedback.
However the authors did find large effect sizes
that represented a potentially important benefit
for patients. Their small number of studies pro-
duced the possibility of a large type II error, lead-
ing them to conclude that incorporating more
future randomized controlled trials could quite
possibly produce a significant result in favour of
EMG biofeedback.
Studying patients in the acute phase post-
stroke is more relevant to therapists as this is
when they have a major input, and when most
of the recovery takes place. Whether EMG
biofeedback is effective at this stage is still not
known as few studies have used early stroke
patients. However, Basmajian et al.11 reported a
controlled study comparing two forms of
therapy for 18 acute severe patients and 11
chronic mild patients using a behavioural
approach including EMG biofeedback and con-
ventional therapy. Both groups improved with
therapy but there were no clinical or statistical
differences between the two. Crow et al.12 com-
pared integrated EMG biofeedback to placebo
EMG biofeedback on the arm for 40 patients
within eight weeks poststroke. The EMG group
improved significantly more than the placebo
group immediately after treatment, but at follow-
up six weeks later there were no differences
between the groups.
Considerable time is spent by physiotherapists
on gait retraining to improve patients’ indepen-
dence after stroke. It has been found that
50–80% of survivors after stroke will walk inde-
pendently, but this apparent success hides the
fact that many independent patients walk slowly
and rarely venture outdoors.13 The velocity of
walking speed as reported by Wade et al.14 is a
simple, valid and reliable measure of gait but
does not describe the whole clinical picture.
 EMG biofeedback for gait training after stroke 13
Added parameters of gait, such as number of
steps, step length, stride width and foot angle,15
along with functional outcomes give a more accu-
rate representation of the abnormalities in gait
encountered after stroke. Despite being simple
objective measures and easy to use within the
clinical context, few EMG biofeedback studies
have used these measures to assess the effective-
ness on gait rehabilitation.
Cognitive deficits following stroke are com-
mon. The recovery of memory, visual, perceptual
and spatial impairments in relation to the recov-
ery of activities of daily living (ADL) function
have been studied previously and found to cor-
relate.16 Assessments of cognitive scores have
been useful in predicting the duration of rehabil-
itation and ADL outcome.17 The impact of cog-
nitive deficits on patients’ ability to learn and
adapt during recovery has not been adequately
studied, and while poor cognitive scores may pre-
dict poor ADL outcome this can not be extra-
polated to the quality of motor recovery. Further
studies are needed to explore the relation-
ship between cognitive impairment and motor
recovery.
In summary, the literature fails to provide con-
vincing evidence that EMG biofeedback can sig-
nificantly facilitate recovery after stroke.
However, little has been done to study its effec-
tiveness in conjunction with physiotherapy in the
acute stages poststroke. The present research was
designed to examine the effects of EMG biofeed-
back training on the recovery of gait in the acute
phase poststroke.
Method
Patients admitted to Scunthorpe General Hospi-
tal with a diagnosis of stroke were registered. At
4–6 weeks poststroke, patients were assessed and
included into the project if they were medically
stable, had unilateral weakness and lived within
Scunthorpe Health District. Patients were
excluded if they had suffered a previous stroke,
exhibited global amnesia, dementia, or had an
underlying neurological disease.
A randomized, stratified controlled design was
used. All patients received conventional physio-
therapy until inclusion into the project at 4–6
Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015
weeks. At this point verbal consent was obtained
to take part and patients were then assessed
physically and psychologically (assessment 1) by
two independent assessors who were blind to
group allocation (M.R. and P.S.). Patients were
stratified according to side of hemiplegia and
their score on the Rivermead Mobility Index18 so
that separate analyses could be carried out on the
mild and severe groups (i.e. 0–3 = severe, 4+ =
mild). Then they were randomly assigned to the
EMG or control group. Inpatients received phys-
iotherapy five days per week, which was the
nature of the physiotherapy service, of which
three sessions were under experimental condi-
tions and two were conventional therapy ses-
sions. Once discharged from hospital outpatients
were seen three times a week under experimen-
tal conditions (i.e. EMG or control); however, we
could not control for length of stay in hospital.
All patients had 18 project treatment sessions
over a period of six weeks by the project phys-
iotherapist (L.B.). Patients were assessed at the
end of the 18 treatment sessions (assessment 2)
and again at follow-up three months later (assess-
ment 3) (M.R.).
The following commonly used measures were
used at all three assessment points to measure the
aspects of physical impairment seen after stroke.
All measures had been previously tested for
validity and reliability.
1) Modified Bobath scale19 (active movement)
2) Modified Ashworth scale20 (muscle tone)
3) Sensation and proprioception by noting the
patient’s response to pinprick, light touch and
passive movement respectively
4) Mobility using the 10-metre walk,14 step
length, stride width and foot angle15 (a group
of normal controls were measured for step
length, stride width and foot angle to com-
pare the results)
5) Rivermead Mobility Index 18 and the Notting-
ham Extended ADL Index.21
A psychological test battery was included so
that any cognitive factors that might affect the
patients’ levels of impairment and recovery at
any stage could be evaluated. In addition, the
scores on the psychological tests were intended
to act as controls both for the level of initial
impairment between groups of patients and for
14 L Bradley et al.
the degree of recovery over the period of the pro-
ject.
1) Mental Status Questionnaire (MSQ)22
(orientation and memory, maximum score
is 5)
2) National Adult Reading Test (NART)23
WAIS-R IQs. (χ = 100, SD = 15)
3) Ravens’ Coloured Progressive Matrices24
WAIS-R IQs. (χ = 100, SD = 15)
4) Rey Auditory Verbal Learning Test
(AVLT)25 (maximum score is 15 words each
trial)
5) Rey Osterrieth Figure Copying Test26
(maximum score for each of the copy and
memory test is 36)
6) Token Test 27 (maximum score is 22).
A pilot study was carried out prior to the main
study in which the inter- and intra-rater reliabil-
ity of the physical assessments were determined.
Five patients were assessed by two physiothera-
pists (L.B. and M.R.) in sequence for active
movement, muscle tone, sensation, propriocep-
tion, gait, Rivermead Mobility Index and the
Nottingham Extended ADL Index. One week
later the patients were assessed by the same phys-
iotherapists in reverse sequence. Spearman rho
correlations were calculated and the inter- and
intra-rater reliability were found to be signifi-
cantly high (p <0.053).
The EMG group of patients were treated using
EMG biofeedback as an adjunct to physiother-
apy. Patients were assessed by the project phys-
iotherapist with particular reference to abnormal
muscle tone. An appropriate problem list and
treatment plan was drawn up for each patient,
following the Bobath28 approach. The EMG
biofeedback electrodes were placed on the motor
point over the selected muscle group on the unaf-
fected side first, and the procedure was demon-
strated and explained to the patient. Then the
electrodes were transferred to the affected side
and treatment commenced. The patients were
asked to do an activity whilst attending to the
auditory tone and the light bar on the EMG
biofeedback machine. They were asked to repeat
the activity several times and by contracting or
relaxing their muscle they could change the level
of auditory tone and/or the level on the light bar.
Treatment consisted of encouraging normal
Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015
alignment in the trunk and hip, facilitating or
inhibiting the presence of abnormal tone in the
trunk and leg, encouraging active control at the
hip, knee and ankle, improving the mechanism of
weight transference through the affected side,
and gait re-education. These techniques were
practised in the positions of lying, sitting and
standing according to the patients’ ability and
needs. For each patient the phases of gait were
analysed whilst walking, problems explained to
the patient and corrected, then practised in con-
junction with the EMG biofeedback machine.
Conventional treatment sessions centred on prac-
tising what had been learned in the EMG
biofeedback sessions, gait training and upper
limb work.
The control group patients were treated using
the same physiotherapy techniques as described
above, but without biofeedback treatment. The
EMG electrodes were applied to the affected
side, the EMG biofeedback machine was
switched on but turned away from the patient
and the physiotherapist so that no visual or audi-
tory feedback was given. This was done to con-
trol for any placebo effects. Again, conventional
treatment sessions were based on what had been
learned in previous sessions, such as gait training
and upper limb therapy.
The EMG biofeedback machine was a Dual
EMG 200 (Biodata, Manchester). The electrodes
were of the self-adhesive and disposable type.
Electrodes were placed on the motor point of the
muscle group wide apart initially, so that the
machine could register a weak contraction, then
closer together as the contraction got stronger to
eliminate overflow signals from another muscle
group. Regular printed output was not a function
of this machine and was not collected. The
machine was selected for its light weight and
portability which enabled it to be incorporated
into the treatment process.
Results
Twenty-three patients were admitted into the
study (12 males and 11 females; mean days post-
stroke 35.6 days). Twenty-one patients completed
a course of physiotherapy and were tested on
three occasions (two died during the course of
 EMG biofeedback for gait training after stroke 15
the project; both were in the control group).
Nineteen patients received 18 treatment sessions
under experimental conditions. One patient had
a total of eight treatment sessions which were
stopped once the patient made a full recovery,
another had a total of 15 treatment sessions
which were drawn out over a period of 15 weeks
due to the patient’s inability to travel. While
there was near equal distribution between the
EMG and control groups, there were more right
side affected patients then left side in both the
mild and severe groups (Table 1).
Preliminary analysis showed that there were
few differences in scores on any measurement
between left- and right-sided cases within the
subgroups, and thus scores were combined into
four groups (mild control, severe control, mild
EMG and severe EMG) for subsequent analysis.
Each set of scores was analysed with a mixed
model analysis of variance (ANOVA) with one
within-subjects factor (assessments 1–2–3) and
two between-subjects factors (mild versus severe
and EMG versus control). One interaction
(assessments × EMG/control) was tested for, as
this was the crucial effect being investigated. That
is, if this was significant it would indicate that the
rate of change across assessments was different
for the EMG and the control groups. All out-
come measures were analysed using the means
from the groups in Table 2.
There was a significant improvement in active
movement for all patients over time (p <0.001) (F
= 11.77), and there was a significant difference
between performance in the mild and severe
groups (p <0.002) (F = 9.01). However there was
no difference in the active movement between
the EMG and the control groups (Figure 1).
Table 1 Number of patients according to side affected and severity of stroke
Left mild Left severe
EMG 1 2
Control 0 2
Table 2 Number of patients and mean age according to final group allocation
Mild control Mild EMG
No. of patients 3 5
Mean age (years) 77 66.6
Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015
There was no significant improvement in muscle
tone over time, but there was a significant dif-
ference between the mild and the severe groups
(p <0. 01, F = 12.71). The severe groups had
increased muscle tone at the hip, knee and ankle.
Similarly, there was no significant improvement
in sensation to pinprick and light touch over time,
the severe group scores were lower than the mild
group scores, indicating greater sensory impair-
ment, but this did not reach significance.
Not all patients were able to walk 10 m at the
assessment points, the numbers of patients
assessed as summarized in Table 3. At assessment
1, 67% (n = 14) of patients were able to walk and
were tested on the 10-metre walking test, 57%
(n = 12) required a walking aid while 10%
(n = 2) walked independently. At assessments 2
and 3, 86% (n = 18) of patients achieved 10 m,
57% (n = 12) of patients used a walking aid and
19% (n = 4) walked independently, the remain-
Figure 1 Comparison of active movement (Bobath scale)
in the EMG (●, mild; ◆, severe) and control (◆, mild;
■, severe) groups.
Right mild Right severe
4 5
3 4
Severe control Severe EMG
6 7
68 72.4
16 L Bradley et al.
Table 3 Number of patients assessed for mobility
Assessment 1
Footprints 10 m test
Normal controls 7
Mild control 2 3
Mild EMG 5 5
Severe control 1 2
Severe EMG 0 4
ing 10% (n = 2) required occasional assistance
from another person and were therefore not clas-
sified as independent. Over assessments 1, 2 and
3 the time taken to walk 10 m and the number of
steps improved for all groups (i.e. the time and
number of steps taken decreased), but there were
no significant differences between the groups
(Figures 2 and 3). Only one patient in the mild
EMG group achieved a walking speed under
10 s which is the norm for the 70 years and above
age group.
Three aspects of gait were measured: step
length, stride width and foot angle. The number
of steps included in the assessment varied from
subject to subject and between assessments, so
for statistical analysis the mean value of each
sample, and the range (largest value minus small-
est value in the samples) were taken for each
measure to give an idea of variability in gait. The
mean of these measures (mean and range +/–
their standard deviation) are given in the Figures
4–6.
Figure 2 Comparison of 10-m walking test (time in
seconds) in the EMG (●, mild; ◆, severe) and control
(▲, mild; ■, severe) groups.
Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015
Assessment 2 Assessment 3
Footprints 10 m test Footprints 10 m test
3 3 2 3
5 5 5 5
4 4 4 4
5 6 5 6
There was no significant difference in step
length between the two mild groups. Although
the mild EMG group did appear to increase step
length over time, while the mild control appeared
to shorten step length, the effect was of no real
importance. The range of step length for both
mild groups increased with time, showing that
patients had an erratic pattern of steps with a
great variety of ranges between subjects. The
severe groups showed no differences between
EMG and control and no improvement over
time. The step lengths for all groups were half the
amount for normals of a similar age. The stride
width for the mild EMG group appeared higher
than the mild control group but did not reach sig-
nificance. There was no evidence of improvement
in stride width over time for the four groups. The
severe groups had higher mean scores for stride
width compared with the normals. There was no
significant difference in foot angle between the
EMG or control groups, and little improvement
over time. However there was no noticeable dif-
Figure 3 Comparison of the 10-m walking test (number
of steps) in the EMG (●, mild; ◆, severe) and control
(▲, mild, ■, severe) groups.
 EMG biofeedback for gait training after stroke
(a)
(b)
Figure 4 Assessment of step length in (a) mild stroke
cases; (b) severe stroke cases (■, control patients group;
▲, EMG patients group; ●, normal controls). Severe
cases could not be tested on assessment 1.
ference between the foot angle in all four groups
compared with the normal controls.
All patients improved over time on the River-
mead and Nottingham ADL scores (p <0.001),
and there was a significant difference between
scores in the mild and severe groups (p <0.001).
However, like active movement, there were no
significant differences between the EMG and
control groups (Figures 7 and 8).
Psychology assessment results
The same analyses were used as for the phys-
ical measures (Table 4). The Mental Status Ques-
tionnaire comparisons were nonsignificant.
Scores were at least average for the National
Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015
17
(a)
(b)
Figure 5 Assessment of stride width in (a) mild stroke
cases; (b) severe stroke cases (■, control patients group;
▲, EMG patients group; ●, normal controls). Severe
cases could not be tested on assessment 1.
Adult Reading Test, and all comparisons were
nonsignificant. All comparisons were nonsignifi-
cant for the Ravens’ Progressive Coloured Matri-
ces. For the Rey Auditory Verbal Learning Test
Trials 1–5 (learning curve):
Assessment 1: improvement over trials 1–5
(F = 18.79, p <0.001)
Assessment 2: improvement over trials 1–5
(F = 36.33, p <0.001)
Assessment 3: improvement over trials 1–5
(F = 16.15, p <0.001)
All other comparisons did not reach significance.
Thus all groups show learning over trials 1–5,
within an assessment, but no other differences
were significant. All scores were within normal
limits, except for the mild control group on some
trials (but only two subjects). Scores for the
18 L Bradley et al.

(a)

Figure 7 Comparison of Rivermead Mobility Index

scores in the EMG (●, mild; ◆, severe) and control
(▲, mild, ■, severe) groups.

(b)

Figure 8 Comparison of Nottingham Extended ADL

Figure 6 Assessment of foot angle in (a) mild stroke
scores in the EMG (●, mild; ◆, severe) and control
cases; (b) severe stroke cases (■, control patients group;
(▲, mild; ■, severe) groups.
▲, EMG patients group; ●, normal controls). Severe
cases could not be tested on assessment 1.

delayed trial (recall) showed improved recall
scores over sessions (F = 10.36, p <0.001). All
other comparisons were nonsignificant (Table 5).
Scores for the Rey Osterrieth Figure Copying
Test showed that all groups improved in scores
over assessments 1, 2 and 3. Memory: sessions
1–2–3 (F = 8.21, p <0.01). The mean scores for
the Token Test for assessments 1, 2 and 3 showed
improvements over time (F = 8.65, p <0.01). The
interaction showed that the two control groups
improved more from assessment 1 to assessments
2 and 3 than the EMG groups did (F = 12.51,
p <0.01).
In summary, the psychometric scores suggested
that there was a normal practice (or recovery)
effect for some measures (i.e. Token Test, AVLT
within-session learning and recall score over ses-
sions). But there were no differences in perfor-
mance between any subgroup of patients (mild
versus severe and EMG versus control). Scores
on all tests were within normal limits, in that
there was no evidence of notable impairment in
any area of mental performance (language, mem-
ory, spatial performance, etc.) as IQ estimates
were near to 100 and other scores were within
normal limits.

Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015

 EMG biofeedback for gait training after stroke 19

Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015

20 L Bradley et al.

Table 5 Rey Auditory Verbal Learning Test (mean scores for five trials over assessments 1, 2 and 3)

T1 T2 T3 T4 T5

Assessment 1
Mild control 4 5.3 5.7 6 5.3
Mild EMG 5 6.8 7.6 8.4 8.2
Severe control 5.2 6.8 7 7.6 9
Severe EMG 4.6 7.1 7.4 8.1 9.4

Assessment 2
Mild control 4.7 6 5.7 7 7.7
Mild EMG 6.2 8 8.4 8.6 9.2
Severe control 6.8 8.6 8.4 8.6 9.6
Severe EMG 5.6 8.4 9.7 10 9.9

Assessment 3
Mild control 3.3 5.7 7 6.7 5.7
Mild EMG 7.6 7.6 9.2 9 9.8
Severe control 8 9.8 10.2 10.6 9.8
Severe EMG 6.6 7.7 8.9 9.4 10.3

Maximum score 15.

Discussion acute phase poststroke was also shown to take

This study has endeavoured to satisfy the list of
suggestions for research posed by De Weerdt and
Harrison6 and Nouri and Sackley8 which they
considered necessary for future evaluation of
EMG biofeedback. The results of this study sug-
gest that there were no significant differences in
treatment effects between the EMG and the con-
trol groups, although the small sample size may
have mitigated against the detection of a treat-
ment effect.
For the physical outcome measures the EMG
and control groups showed improvement over
time but the improvement was not specific to
EMG biofeedback. This indicated that the mea-
sures used were sensitive enough to detect recov-
ery and would have detected any difference
between treatment groups had there been one,
and may be valuable for use in future studies.
The recovery was obviously of clinical signifi-
cance to the patients, 86% were able to walk 10
m at assessment 3, but it is not known whether
the recovery was due to the intervention or to
spontaneous recovery, as a ‘no treatment’ control
group was not used.
The improvements made by patients in this
study were comparable with those in the studies
by Basmajian et al.11 and Crow et al.12 who also
used early stroke patients, where recovery in the
place. It is also important to note that whilst the
improvements in the Rivermead Mobility Index
and Nottingham ADL took place, the assess-
ments did not measure the improvement in qual-
ity of movement, but rather the ability of the
individual to adapt and compensate for their
impairment.
Eight-six per cent of patients achieved inde-
pendent walking at assessment 3 (37 weeks post-
stroke), which is a little higher than has been
found in other studies.14 Nevertheless, clinically
it was found the patients’ quality of gait was slow
and irregular, the mild group patients were
unable to walk outdoors independently, and the
severe group patients could not transfer without
help. On a practical level, the mild group of
patients could manage small domestic tasks but
rarely resumed their leisure activities, whereas
the severe group remained dependent on carers
for everyday activities. These findings are repre-
sentative of the larger studies documenting the
handicap of stroke.13
The psychometric scores showed that there was
no notable impairment in any areas of mental
performance. These scores were included in all
the assessments so that any cognitive factors
which may have affected the patients’ motor
recovery at any stage could be evaluated. In addi-
tion the scores on the psychological tests acted as

Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015

 EMG biofeedback for gait training after stroke 21
controls both for the level of initial impairment
between the groups of patients and for the degree
of recovery over the period of the project.
Studying patients in the acute phase poststroke
was very difficult. One problem encountered was
the variability shown by the patients from one
treatment session to the next, as demonstrated by
the erratic muscle tone measurements, and the
large ranges in the gait scores. Another problem
was patient noncompliance to a rigid regime. This
study was intended to be incorporated into nor-
mal physiotherapeutic clinical practice in a dis-
trict general hospital, rather than removing
patients from the rehabilitation environment.
Problems occurred which reflect the nature of
true clinical practice in that some patients were
unable to attend for three treatment sessions per
week due to illness, transport problems, or other
important engagements. Additionally, adherence
to the normal physiotherapeutic practice for hos-
pital inpatients had a confounding effect on the
results: inpatients received additional conven-
tional treatment compared to the outpatients,
contrary to our original protocol. The inpatients
received physiotherapy daily (five times a week),
of which three sessions were under experimental
conditions and two followed normal physiother-
apeutic practice. The outpatients received phys-
iotherapy three times a week under experimental
conditions. The extra sessions received by the
inpatients may have washed out the effects of the
EMG biofeedback, but it was considered un-
ethical to deny patients access to the normal
routine of daily physiotherapy.
The results of pooling available randomized
control trials to study the efficacy of EMG
biofeedback are not conclusive. Glanz et al.10 sug-
gested that further studies need to look at the
timing of therapy relative to the onset of stroke,
and to consider whether EMG biofeedback may
only be effective within a certain window of time.
Group studies may not be the most appropriate
way to investigate EMG biofeedback, as the
small gains made by individual patients may be
lost by generalization in the group analysis. Ther-
apists may need to address the need for single
case study design for individual patients who
have specific tone problems.
Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015
Conclusion
The results of this study indicated that there were
no significant differences in the rate of improve-
ment after stroke between the two groups.
Although this study used EMG biofeedback as
an adjunct to physiotherapy and represented rel-
evant clinical practice, the results provide little
evidence to support the clinical significance of
using EMG biofeedback to improve gait after
stroke.
Acknowledgements
We thank The Stroke Association for funding
the research, Scunthorpe General Hospital Ethi-
cal Committee for approving the project, the
patients for consenting to take part, and Merrin
Froggatt and Anne Spaight (physiotherapy man-
agers) for their continued support.
References
1 Hurrell M. Electromyographic feedback in
rehabilitation. Physiotherapy 1980; 66: 293–98.
2 Lincoln NB, Sackley CM. Biofeedback in stroke
rehabilitation. Crit Rev Phys Rehabil Med 1992; 4:
37–47.
3 Basmajian JV, Kukulka CG, Narayan MG, Takebe
K. Biofeedback treatment of footdrop after stroke
compared with standard rehabilitation technique:
Effects on voluntary control and strength. Arch Phys
Med Rehabil 1975; 56: 231–36.
4 Burnside IG, Tobias HS, Bursill D. Electro-
myographic feedback in the remobilisation of stroke
patients; a controlled trial. Arch Phys Med Rehabil
1982; 63: 217–22.
5 Prevo AJ, Viser SL, Vogelaar VW. Effect of EMG
feedback on paretic muscles and abnormal co-
contraction in the hemiplegic arm, compared with
conventional physical therapy. Scand J Rehabil Med
1982; 14: 121–31.
6 De Weerdt WG, Harrison MA. The efficacy of
electromyographic feedback for stroke patients: a
critical review of the main literature. Physiotherapy
1986; 72: 106–18.
7 Wolf SL. Electromyographic biofeedback application
to stroke patients: A critical review. Phys Ther 1983;
63: 1448–59.
8 Nouri NB, Sackley CM. Biofeedback in stroke
rehabilitation. Crit Rev Phys Rehabil Med 1992; 4:
37–47.
9 Schleenbaker RE, Mainous AG. Electromyographic
biofeedback for neuromuscular re-education in the
hemiplegic stroke patient: a meta-analysis. Arch
22 L Bradley et al.
Phys Med Rehabil 1993; 74: 1301–304.
10 Glanz M, Klawansky S, Stason W, Berkey C, Shah
N, Phan H, Chalmers C. Biofeedback therapy in
post-stroke rehabilitation: A meta-analysis of the
randomised controlled trials. Arch Phys Med Rehabil
1995; 76: 508–15.
11 Basmajian JV, Gowland CA, Finlayson AJ et al.
Stroke treatment comparison of integrated
behavioural physical therapy vs. traditional physical
therapy programs. Arch Phys Med Rehabil 1987; 68:
267–72.
12 Crow JL, Lincoln NB, Nouri FM, De Weerdt W.
The effectiveness of EMG biofeedback in the
treatment of arm function after stroke. Int Disabil
Studies 1989; 11: 155–60.
13 Wade DT, Langton-Hewer R, Skilbeck CE, David
RM. Stroke: a critical approach to diagnosis,
treatment and management. London: Chapman &
Hall, 1985.
14 Wade DT, Wood VA, Heller A, Maggs J, Langton-
Hewer R. Walking after stroke measurement and
recovery over the first 3 months. Scand J Rehabil
Med 1987; 19: 25–30.
15 Bowker P, Messenger N. The measurement of gait.
Clin Rehabil 1988; 2: 89–97.
16 Lehmann JF, DeLateur BJ, Fowler RS. Stroke
rehabilitation: outcome and prediction. Arch Phys
Med Rehabil 1975; 56: 383–89.
17 Ben-Yishay Y, Gerstman L, Diller L, Haas A.
Prediction of the rehabilitation outcomes from
psychometric parameters in left hemiplegics.
J Consult Clin Psychol 1970; 34: 436–41.
Downloaded from cre.sagepub.com at University of Ulster Library on April 4, 2015
18 Collen FM, Wade DT, Robb GF, Bradshaw CM.
The Rivermead Mobility Index: A further develop-
ment of the Rivermead Motor Assessment. Int
Disabil Studies 1991; 13: 50–54.
19 Guarna F, Corriveau H, Chamberland J, Bertrand
Arsenault A, Dutil E, Drouin G. An evaluation of
the hemiplegic subject based on the Bobath
approach. Scand J Rehabil Med 1988; 20: 1–16.
20 Bohannon RW, Smith MB. Inter-rater reliability of
a modified Ashworth scale of muscle spasticity. Phys
Ther 1987; 67: 206–207.
21 Nouri FM, Lincoln NB. An extended activities of
daily living scale for stroke patients. Clin Rehabil
1987; 1: 301–305.
22 Kahn RL, Miller NE. Assessment of altered brain
function in the aged. In: Storandt M, Siegler I, Ellis
M eds. The clinical psychology of ageing. New York:
Plenum Press, 1978.
23 Nelson HE. The National Adult Reading Test.
Windsor, UK: NFER Nelson, 1982.
24 Raven JC. Guide to the standard progressive
matrices. London: HK Lewis.
25 Rey A. L’Examen clinique en psychologie. Paris:
Presses Universitaires de France, 1964.
26 Osterrieth PA. Le test de copie d’une figure
complexe. Arch Psychol 1944; 30: 206–356.
27 Boller F, Vignolo LA. Latent sensory aphasia in
hemisphere damaged patients: an experimental study
with the Token Test. Brain 1966; 89: 815–31.
28 Bobath B. Adult hemiplegia: evaluation and
treatment, second edition. London: William
Heinemann Medical, 1980.