Real-Time Kinematic, Temporospatial, and Kinetic Biofeedback ...
Real-Time Kinematic, Temporospatial, and Kinetic Biofeedback ...
Real-Time Kinematic, Temporospatial, and Kinetic Biofeedback ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Real</strong>-<strong>Time</strong> <strong>Kinematic</strong>, <strong>Temporospatial</strong>, <strong>and</strong> <strong>Kinetic</strong><br />
<strong>Biofeedback</strong> During Gait Retraining in Patients: A<br />
Systematic Review<br />
Jeremiah J. Tate <strong>and</strong> Clare E. Milner<br />
PHYS THER. Published online June 17, 2010<br />
doi: 10.2522/ptj.20080281<br />
The online version of this article, along with updated information <strong>and</strong> services, can be<br />
found online at: http://ptjournal.apta.org/content/early/2010/06/17/ptj.20080281<br />
Collections<br />
E-mail alerts<br />
This article, along with others on similar topics, appears<br />
in the following collection(s):<br />
Electrotherapy<br />
Gait <strong>and</strong> Locomotion Training<br />
Gait Disorders<br />
Kinesiology/Biomechanics<br />
Motor Control <strong>and</strong> Motor Learning<br />
Systematic Reviews/Meta-analyses<br />
Sign up here to receive free e-mail alerts<br />
Online First articles are published online before they appear in a regular issue of<br />
Physical Therapy (PTJ). PTJ publishes 2 types of Online First articles:<br />
Author manuscripts: PDF versions of manuscripts that have been peer-reviewed <strong>and</strong><br />
accepted for publication but have not yet been copyedited or typeset. This allows PTJ<br />
readers almost immediate access to accepted papers.<br />
Page proofs: edited <strong>and</strong> typeset versions of articles that incorporate any author<br />
corrections <strong>and</strong> replace the original author manuscript.<br />
Downloaded from http://ptjournal.apta.org/<br />
by guest on April 2, 2015
Research Report<br />
<strong>Real</strong>-<strong>Time</strong> <strong>Kinematic</strong>, <strong>Temporospatial</strong>,<br />
<strong>and</strong> <strong>Kinetic</strong> <strong>Biofeedback</strong> During Gait<br />
Retraining in Patients:<br />
A Systematic Review<br />
Jeremiah J. Tate, Clare E. Milner<br />
Background. <strong>Biofeedback</strong> has been used in rehabilitation settings for gait<br />
retraining.<br />
Purpose. The purpose of this review was to summarize <strong>and</strong> synthesize the findings<br />
of studies involving real-time kinematic, temporospatial, <strong>and</strong> kinetic biofeedback.<br />
The goal was to provide a general overview of the effectiveness of these forms<br />
of biofeedback in treating gait abnormalities.<br />
Data Sources. Articles were identified through searches of the following databases:<br />
MEDLINE, CINAHL (Cumulative Index to Nursing <strong>and</strong> Allied Health Literature),<br />
<strong>and</strong> Cochrane Central Register for Controlled Trials. All searches were limited to the<br />
English language <strong>and</strong> encompassed the period from 1965 to November 2007.<br />
Study Selection. Titles <strong>and</strong> abstracts were screened to identify studies that met<br />
the following requirements: the study included the use of kinematic, temporospatial,<br />
or kinetic biofeedback during gait training, <strong>and</strong> the population of interest showed<br />
abnormal movement patterns as a result of a pathology or injury.<br />
J.J. Tate, PT, MS, is a doctoral student,<br />
Department of Exercise,<br />
Sport, <strong>and</strong> Leisure Studies, University<br />
of Tennessee, Knoxville,<br />
Tennessee.<br />
C.E. Milner, PhD, is Assistant Professor,<br />
Department of Exercise,<br />
Sport, <strong>and</strong> Leisure Studies, University<br />
of Tennessee, 1914 Andy<br />
Holt Ave, HPER 322, Knoxville, TN<br />
37996-2700 (USA). Address all<br />
correspondence to Dr Milner at:<br />
milner@utk.edu.<br />
[Tate JJ, Milner CE. <strong>Real</strong>-time kinematic,<br />
temporospatial, <strong>and</strong> kinetic<br />
biofeedback during gait retraining<br />
in patients: a systematic review.<br />
Phys Ther. 2010;90:xxx–xxx.]<br />
© 2010 American Physical Therapy<br />
Association<br />
Data Extraction. All articles that met the inclusion criteria were assessed by use<br />
of the Methodological Index for Nonr<strong>and</strong>omized Studies.<br />
Data Synthesis. Seven articles met the inclusion criteria <strong>and</strong> were included in<br />
the review. Effect sizes were calculated for the primary outcome variables for all<br />
studies that provided enough data. Effect sizes generally suggested moderate to large<br />
treatment effects for all methods of biofeedback during practice.<br />
Limitations. Several of the studies lacked adequate r<strong>and</strong>omization; therefore,<br />
readers should exercise caution when interpreting authors’ conclusions.<br />
Conclusions. Each biofeedback method appeared to result in moderate to large<br />
treatment effects immediately after treatment. However, it is unknown whether the<br />
effects were maintained. Future studies should ensure adequate r<strong>and</strong>omization of<br />
participants <strong>and</strong> implementation of motor learning concepts <strong>and</strong> should include<br />
retention testing to assess the long-term success of biofeedback <strong>and</strong> outcome measures<br />
capable of demonstrating coordinative changes in gait <strong>and</strong> improvement in<br />
function.<br />
Post a Rapid Response to<br />
this article at:<br />
ptjournal.apta.org<br />
August 2010 Volume 90 Number 8 Physical Therapy f 1<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Biofeedback</strong> During Gait Retraining<br />
<strong>Biofeedback</strong> was first introduced<br />
in the literature more than 30<br />
years ago as a training tool used<br />
in rehabilitation settings to facilitate<br />
normal movement patterns after injury.<br />
1 Since then, biofeedback has<br />
been used primarily in rehabilitation<br />
settings for the treatment of gait abnormalities<br />
in adults after stroke. 2–12<br />
<strong>Biofeedback</strong> also has been used to<br />
facilitate the normalization of gait<br />
patterns in children with cerebral<br />
palsy 13 <strong>and</strong> in adults after amputation,<br />
14 after spinal cord injuries, 15 after<br />
hip fractures, 14 <strong>and</strong> after total<br />
hip 14,16 <strong>and</strong> knee 14 joint replacements.<br />
<strong>Biofeedback</strong> is a technique that typically<br />
uses electronic equipment to<br />
provide a client with auditory signals,<br />
visual signals, or both regarding<br />
internal physiological events, both<br />
normal <strong>and</strong> abnormal (eg, heart rate,<br />
blood pressure, <strong>and</strong> level of muscle<br />
activity). 17 During biofeedback for<br />
gait retraining, the client is provided<br />
with augmented information (eg, kinematics,<br />
kinetics, <strong>and</strong> electromyography)<br />
regarding physiological responses.<br />
Additionally, biofeedback<br />
provides clinicians with a useful tool<br />
for giving clients instructions on<br />
how to modify movement patterns.<br />
Thus, biofeedback complements the<br />
already present internal feedback (ie,<br />
visual, auditory, <strong>and</strong> proprioceptive<br />
feedback) <strong>and</strong> acts as a “sixth<br />
sense.” 18 <strong>Biofeedback</strong> typically is<br />
provided instantaneously to the<br />
learner (ie, in real time), whereas<br />
other methods of external feedback<br />
(eg, verbal <strong>and</strong> video feedback) are<br />
provided some time after the movement.<br />
More recently, a resurgence of<br />
interest in real-time feedback has developed<br />
because of the expansion of<br />
technology related to kinematic 19<br />
<strong>and</strong> kinetic 14,16 biofeedback.<br />
Electromyographic biofeedback may<br />
be the most popular method of providing<br />
biofeedback <strong>and</strong> has been<br />
used frequently in gait retraining after<br />
stroke. However, meta-analyses<br />
Table 1.<br />
Results of Searches of Electronic Databases<br />
Database<br />
of electromyographic biofeedback<br />
studies of people after stroke have<br />
concluded that little evidence supports<br />
its use in addition to conventional<br />
physical therapy. 20,21 Other<br />
forms of biofeedback, such as kinematic,<br />
2,8,9 temporospatial, 6,7 <strong>and</strong> kinetic,<br />
14,16 also have been developed<br />
<strong>and</strong> used in rehabilitation settings for<br />
gait retraining. It remains unknown<br />
whether these methods are effective<br />
in treating gait abnormalities. Therefore,<br />
the purpose of this systematic<br />
review was to summarize <strong>and</strong> synthesize<br />
the findings of studies involving<br />
real-time kinematic, temporospatial,<br />
<strong>and</strong> kinetic biofeedback.<br />
This effort provided a general overview<br />
of the effectiveness of these<br />
forms of biofeedback in treating gait<br />
abnormalities.<br />
Method<br />
Data Sources <strong>and</strong> Searches<br />
Three electronic databases were<br />
searched in a systematic fashion to<br />
identify relevant articles: MEDLINE,<br />
CINAHL (Cumulative Index to Nursing<br />
<strong>and</strong> Allied Health Literature), <strong>and</strong><br />
Cochrane Central Register for Controlled<br />
Trials. All searches were limited<br />
to the English language <strong>and</strong> encompassed<br />
the period from 1965 to<br />
November 2007. The following key<br />
words were used in the searches:<br />
feedback, biofeedback, walking, ambulation,<br />
<strong>and</strong> gait. Key words were<br />
combined to identify studies involving<br />
biofeedback <strong>and</strong> gait. Table 1<br />
shows the specific key words <strong>and</strong><br />
combinations of key words used in<br />
Key Words<br />
MEDLINE (1) feedback; (2) biofeedback; (3) 1 or 2; (4) walk a ; (5) ambul a ;<br />
(6) walking; (7) gait; (8) 4 or 5 or 6 or 7; (9) 3 <strong>and</strong> 8<br />
CINAHL (1) feedback; (2) biofeedback; (3) 1 or 2; (4) walk a ; (5) ambul a ;<br />
(6) walking; (7) gait; (8) 4 or 5 or 6 or 7; (9) 3 <strong>and</strong> 8<br />
Cochrane<br />
a Limit: English language.<br />
(1) feedback; (2) biofeedback; (3) 1 or 2; (4) walking;<br />
(5) ambulation; (6) gait; (7) 4 or 5 or 6; (8) 3 <strong>and</strong> 7<br />
No. of<br />
Studies<br />
504<br />
162<br />
85<br />
each search. Each review article that<br />
was identified by MEDLINE also was<br />
manually searched to identify other<br />
potential articles.<br />
Study Selection<br />
One reviewer (J.J.T.) screened titles<br />
<strong>and</strong> abstracts identified during electronic<br />
<strong>and</strong> manual searches to determine<br />
eligibility. In the event that the<br />
title or abstract did not provide<br />
enough information, the article was<br />
obtained for full review. Published<br />
articles were considered for initial<br />
inclusion in the review if the following<br />
requirements were met: the<br />
study included the use of kinematic,<br />
temporospatial, or kinetic biofeedback<br />
during gait training <strong>and</strong> the<br />
population of interest showed abnormal<br />
movement patterns as a result of<br />
a pathology or injury. Figure 1 shows<br />
a flow diagram of the systematic review<br />
process.<br />
Data Extraction <strong>and</strong><br />
Quality Assessment<br />
All articles that met the inclusion criteria<br />
were assessed by 2 reviewers<br />
(J.J.T. <strong>and</strong> C.E.M.) using the Methodological<br />
Index for Nonr<strong>and</strong>omized<br />
Studies (MINORS). 22 The MINORS instrument<br />
has been determined to be<br />
valid <strong>and</strong> reliable in assessing the<br />
methodological quality of both r<strong>and</strong>omized<br />
<strong>and</strong> nonr<strong>and</strong>omized studies.<br />
22 Scores on the MINORS range<br />
from 0 to 24 for r<strong>and</strong>omized studies<br />
<strong>and</strong> 0 to 16 for nonr<strong>and</strong>omized studies.<br />
The following methodological<br />
items are assessed with the MINORS<br />
2 f Physical Therapy Volume 90 Number 8 August 2010<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Biofeedback</strong> During Gait Retraining<br />
instrument: aim of the study, inclusion<br />
of consecutive participants, prospective<br />
data collection, appropriate<br />
outcome assessments, unbiased assessments<br />
of outcomes, appropriate<br />
follow-up, attrition, <strong>and</strong> sample size.<br />
In addition, the quality of the comparison<br />
group <strong>and</strong> statistical analyses<br />
in r<strong>and</strong>omized trials only are assessed.<br />
Furthermore, the 2 reviewers clarified<br />
the intent of the questions concerning<br />
the inclusion of consecutive<br />
participants <strong>and</strong> prospective data<br />
collection to improve interrater reliability.<br />
The purpose of the question<br />
concerning the inclusion of consecutive<br />
participants was to judge<br />
whether inclusion <strong>and</strong> exclusion criteria<br />
were included in the study. The<br />
purpose of the question regarding<br />
prospective data collection was to<br />
judge whether the study was descriptive<br />
(score of 0) or prospective<br />
(score of 1 or 2). The delineation<br />
between scores of 1 <strong>and</strong> 2 for prospective<br />
quality was based on<br />
whether the study demonstrated 1<br />
or more specific a priori aims <strong>and</strong><br />
hypotheses.<br />
The MINORS scores were averaged<br />
across the 2 reviewers to enable the<br />
ranking of studies. A histogram of<br />
the studies <strong>and</strong> their MINORS scores<br />
was created to assist in determining a<br />
cutoff point, based on the natural<br />
clustering of groups, for inclusion in<br />
the review (Fig. 2).<br />
Data Synthesis <strong>and</strong> Analysis<br />
A meta-analysis was not performed<br />
because of the wide variety of study<br />
designs, methodologies, <strong>and</strong> outcome<br />
variables. Effect sizes (ESs)<br />
were calculated for the primary outcome<br />
variables for all studies that<br />
provided enough data. Effect sizes<br />
were calculated by subtracting the<br />
mean score at the baseline from the<br />
mean score after treatment <strong>and</strong> then<br />
dividing the result by a st<strong>and</strong>ard deviation<br />
resulting from pooling of the<br />
baseline <strong>and</strong> treatment st<strong>and</strong>ard de-<br />
Figure 1.<br />
Flow diagram of the systematic review process.<br />
viations. 23 Effect sizes were used in<br />
this review to provide a means of<br />
evaluating treatment success because<br />
they are not directly affected<br />
by sample size but take into account<br />
within-group variability. Effect sizes<br />
were st<strong>and</strong>ardized on the basis of the<br />
work of Cohen (small.02, moderate0.5,<br />
<strong>and</strong> large0.8). 23 All ESs<br />
represented comparisons of the<br />
mean posttreatment <strong>and</strong> follow-up<br />
(if applicable) scores with the mean<br />
baseline scores for each of the<br />
biofeedback <strong>and</strong> control groups. The<br />
ES for change over time in the<br />
biofeedback group was then compared<br />
with that in the control group.<br />
Percent differences also were calculated<br />
to aid in the interpretation of<br />
studies that did not provide enough<br />
data to calculate ESs.<br />
Results<br />
Study Identification<br />
A search of the MEDLINE database<br />
identified a total of 504 articles, a<br />
search of the CINAHL database identified<br />
162 articles, <strong>and</strong> a search of<br />
the Cochrane Central Register of<br />
Controlled Trials identified 85 articles.<br />
In total, 666 individual articles<br />
were identified by these databases.<br />
Fifty-eight of these articles were<br />
identified as potentially relevant, <strong>and</strong><br />
their full texts were retrieved. After<br />
an initial review by the first reviewer<br />
(J.J.T.), 35 of these articles were excluded<br />
because they did not meet<br />
August 2010 Volume 90 Number 8 Physical Therapy f 3<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Biofeedback</strong> During Gait Retraining<br />
<strong>Kinematic</strong> <strong>Biofeedback</strong><br />
In 3 studies, kinematic biofeedback<br />
was provided with electrogoniometers<br />
(Tab. 2). All 3 studies involved<br />
analysis of the effect of kinematic<br />
biofeedback in participants who had<br />
had a stroke. Ceceli et al 8 <strong>and</strong> Morris<br />
et al 9 analyzed the effectiveness of<br />
providing participants (n41 <strong>and</strong><br />
26, respectively) with kinematic<br />
biofeedback of the knee compared<br />
with conventional physical therapy<br />
in efforts to minimize genu recurvatum.<br />
Colborne et al 2 investigated the<br />
effectiveness of providing participants<br />
(n8) with kinematic biofeedback<br />
for the ankle in attempts to<br />
improve ankle control. Outcome<br />
measures included gait speed, 2,8,9<br />
number of recurvations, 8 <strong>and</strong><br />
symmetry. 2,9<br />
Figure 2.<br />
Histogram of scores on the Methodological Index for Nonr<strong>and</strong>omized Studies. 22<br />
the inclusion criteria. Both reviewers<br />
assessed the remaining 23 studies<br />
using the MINORS instrument. Of<br />
the 23 articles assessed, 7 achieved a<br />
score of 16 or greater out of a possible<br />
24 <strong>and</strong> were included in the review<br />
(Fig. 1). A cutoff point of 16<br />
was identified qualitatively on observation<br />
of the histogram as a natural<br />
breakpoint between clusters of studies<br />
(Fig. 2). Tables 2, 3, <strong>and</strong> 4 summarize<br />
the characteristics of the<br />
included studies with regard to participant<br />
population, sample sizes,<br />
participant ages, treatment protocols,<br />
presence of masking <strong>and</strong> retention<br />
tests, quantitative variables,<br />
functional outcome measures, <strong>and</strong><br />
conclusions drawn by the authors.<br />
<strong>Biofeedback</strong> Protocol<br />
Generally, there was a large range in<br />
the structures of the treatment protocols.<br />
The mean treatment time in<br />
all studies was 35 minutes, with a<br />
range of 15 to 50 minutes, <strong>and</strong> the<br />
mean treatment frequency was 4<br />
times per week, with a range of<br />
twice per week to daily treatments.<br />
The mean treatment duration was<br />
3.5 weeks, with a range of 1.5 to<br />
8 weeks. The mean cumulative<br />
biofeedback time per study (ie, treatment<br />
time treatment frequency <br />
treatment duration) was 397 minutes,<br />
with a range of 120 to 900<br />
minutes.<br />
Ceceli et al 8 assigned participants to<br />
either an experimental group<br />
(n26) or a control group (n15).<br />
Both groups received conventional<br />
therapy that consisted of pelvis <strong>and</strong><br />
hip control exercises, weight shifting,<br />
<strong>and</strong> gait training. In addition,<br />
participants in the experimental<br />
group received kinematic biofeedback<br />
for 10 daily sessions, each lasting<br />
30 minutes. Posttreatment data<br />
were collected for participants in the<br />
experimental group at the end of the<br />
10 days of biofeedback training; data<br />
were collected for participants in<br />
the control group at the time of<br />
discharge.<br />
In the study by Morris et al, 9 participants<br />
received treatment in 2 separate<br />
phases, lasting 4 weeks each.<br />
Participants in the experimental<br />
group (n13) received kinematic<br />
biofeedback during the first phase<br />
<strong>and</strong> conventional physical therapy<br />
during the second phase. <strong>Kinematic</strong><br />
biofeedback of the involved knee<br />
was provided to participants during<br />
st<strong>and</strong>ing <strong>and</strong> gait training for 30 to<br />
45 minutes per treatment session.<br />
Participants in the control group<br />
(n13) received conventional phys-<br />
4 f Physical Therapy Volume 90 Number 8 August 2010<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Biofeedback</strong> During Gait Retraining<br />
Table 2.<br />
Characteristics of Electrogoniometer Studies<br />
Study<br />
No. of<br />
Participants<br />
(Experimental<br />
Study Group/Control<br />
Population a Group) Age (y)<br />
Treatment Protocol<br />
<strong>Time</strong><br />
(min) Frequency Duration<br />
Masking<br />
Method of<br />
Delivery<br />
Follow-up<br />
(wk) Quantitative Variables<br />
Functional<br />
Outcome Improvement b<br />
Ceceli Adult CVA 26/15 Median<br />
et al 8 49.5/54<br />
30 Daily 10 d No Auditory/<br />
visual<br />
24 Cadence, no of<br />
recurvations/50 steps<br />
None Yes<br />
Morris Adult CVA 13/13 X64/64<br />
et al 9 (range<br />
33–74)<br />
45 5/wk 4 wk/block No Auditory 4 Peak knee hyperextension,<br />
gait speed, single-limb<br />
stance symmetry<br />
Gait recovery<br />
(Motor<br />
Assessment<br />
Scale)<br />
Yes<br />
Colborne Adult CVA 8 (repeated<br />
et al 2 measures)<br />
Not reported 30 2/wk 4 wk/block No Auditory 4 Gait speed, stride length,<br />
stride time, knee angle at<br />
heel-strike, knee angle at<br />
50% swing, ankle angle<br />
of motion, push-off<br />
impulse, swing ratio<br />
(affected/unaffected),<br />
weight ratio (affected/<br />
unaffected)<br />
None Yes<br />
a CVAcerebrovascular accident.<br />
b Statistically significant improvement in experimental group vs control group, as concluded by study authors.<br />
Table 3.<br />
Characteristics of <strong>Temporospatial</strong> <strong>Biofeedback</strong> Studies<br />
Study<br />
No. of<br />
Participants<br />
(Experimental<br />
Study Group/Control<br />
Population a Group) Age (y)<br />
Protocol<br />
<strong>Time</strong><br />
(min) Frequency Duration<br />
Masking<br />
Method of<br />
Delivery Follow-up<br />
Quantitative<br />
Variables<br />
Functional<br />
Outcome Improvement b<br />
Aruin et al 6 Adult CVA 8/8 X65 50 Daily 10 d No Auditory None Step width None Yes<br />
Montoya et al 7 Adult CVA 9/5 X64/60 45 2/wk 4 wk No Auditory/visual None Step length<br />
(paretic limb)<br />
None Yes<br />
a CVAcerebrovascular accident.<br />
b Statistically significant improvement in experimental group vs control group, as concluded by study authors.<br />
August 2010 Volume 90 Number 8 Physical Therapy f 5<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Biofeedback</strong> During Gait Retraining<br />
Table 4.<br />
Characteristics of <strong>Kinetic</strong> <strong>Biofeedback</strong> Studies<br />
Protocol<br />
Functional<br />
Outcome Improvement b<br />
Quantitative<br />
Variables<br />
Method of<br />
Delivery Follow-up<br />
Masking<br />
<strong>Time</strong><br />
(min) Frequency Duration<br />
No. of<br />
Participants<br />
(Experimental<br />
Study Group/Control<br />
Population a Group) Age (y)<br />
Study<br />
24/18 X62/66 30 4 sessions 14 d No Auditory None Weight bearing None Yes<br />
Isakov 14 Amputation,<br />
THR, TKR,<br />
hip fracture<br />
Yes d<br />
Harris Hip<br />
Score<br />
White <strong>and</strong> Lifeso 16 THR 12/8/8 c X51/60/ 70 c 15 3/wk 8 wk No Verbal/visual None Symmetry indexes<br />
for peak force<br />
during loading,<br />
loading rate,<br />
<strong>and</strong> vertical<br />
impulse; rate of<br />
perceived<br />
exertion<br />
a THRtotal hip replacement, TKRtotal knee replacement.<br />
b Statistically significant improvement in experimental groups vs control group, as concluded by study authors.<br />
c <strong>Biofeedback</strong> experimental group/no-treatment control group/no-biofeedback experimental group.<br />
d Authors reported statistically significant improvements in both the biofeedback <strong>and</strong> no-biofeedback experimental groups.<br />
ical therapy during both phases. Outcome<br />
data were collected for all participants<br />
at baseline <strong>and</strong> after each<br />
treatment phase.<br />
Colborne et al 2 used a 3-period crossover<br />
design to assess the effectiveness<br />
of kinematic biofeedback. The 8<br />
participants received 1 of 3 treatments:<br />
kinematic biofeedback, electromyographic<br />
biofeedback, <strong>and</strong><br />
conventional physical therapy. Conventional<br />
physical therapy was given<br />
during either the first phase (n4) or<br />
the last phase (n4). Each treatment<br />
phase lasted 4 weeks <strong>and</strong> consisted<br />
of biweekly treatment sessions, each<br />
lasting 30 minutes. No washout periods<br />
were included between treatments.<br />
Gait speed data were collected<br />
for all participants at baseline,<br />
after each treatment phase, <strong>and</strong> at a<br />
1-month follow-up.<br />
Ceceli et al 8 found that participants<br />
in the experimental group showed a<br />
statistically significant decrease in<br />
the number of recurvations compared<br />
with the control group immediately<br />
after treatment—from 45 to 8<br />
recurvations, an 83% decrease, for<br />
the experimental group, <strong>and</strong> from<br />
49 to 39 recurvations, a 19% decrease,<br />
for the control group. Additionally,<br />
participants were asked to<br />
return for a 6-month follow-up. However,<br />
we did not calculate ESs or<br />
percent differences for 6-month<br />
follow-up data because of the poor<br />
participant follow-up rate (50%);<br />
with a low follow-up rate, the data<br />
might not be a good representation<br />
of the data for all participants.<br />
Morris et al 9 reported a moderate effect<br />
for increased gait speed in the<br />
experimental group (increase of<br />
0.10 m/s, ES0.50, 33% increase)<br />
but no effect in the control group<br />
(increase of 0.01 m/s, ES0.08, 5%<br />
increase) after the first treatment<br />
phase. Both groups had similar baseline<br />
gait velocities (0.33 m/s in the<br />
experimental group <strong>and</strong> 0.30 m/s in<br />
6 f Physical Therapy Volume 90 Number 8 August 2010<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Biofeedback</strong> During Gait Retraining<br />
the control group). Both groups<br />
showed statistically significant reductions<br />
in peak knee extension during<br />
stance, although no group difference<br />
was noted. After the second<br />
treatment phase, a large increase in<br />
gait speed was reported in the experimental<br />
group (increase of 0.23 m/s,<br />
ES1.45, 75% increase), but only a<br />
small change was reported in the<br />
control group (increase of 0.10 m/s,<br />
ES0.46, 37% increase). Additionally,<br />
participants in the experimental<br />
group showed a statistically significant<br />
decrease in peak knee extension<br />
compared with participants in<br />
the control group. Effect sizes <strong>and</strong><br />
percent differences for peak knee<br />
extension could not be calculated<br />
because of a lack of data.<br />
Colborne et al 2 found that kinematic<br />
biofeedback training resulted in a<br />
moderate increase in gait speed (increase<br />
of 0.11 m/s, ES0.51, 23%<br />
increase) but that conventional therapy<br />
resulted in only a small improvement<br />
(increase of 0.08 m/s,<br />
ES0.37, 17% increase). Participants<br />
in both of those treatment groups<br />
had the same baseline gait speed<br />
(0.48 m/s). We did not interpret the<br />
follow-up data because carryover effects<br />
for either or both treatment<br />
methods might have occurred as a<br />
result of the crossover design of the<br />
study. 24<br />
<strong>Temporospatial</strong> <strong>Biofeedback</strong><br />
In 2 studies, temporospatial biofeedback<br />
was provided for participants<br />
after stroke (Tab. 3). Aruin et al 6<br />
investigated the effectiveness of<br />
biofeedback regarding the base of<br />
support in 16 participants, <strong>and</strong> Montoya<br />
et al 7 analyzed the effectiveness<br />
of biofeedback regarding step length<br />
in 14 participants. <strong>Biofeedback</strong> was<br />
provided in efforts to improve the<br />
base of support 6 <strong>and</strong> the symmetry<br />
of step length. 7 Outcome measures<br />
included step width 6 <strong>and</strong> step<br />
length. 7<br />
Aruin et al 6 r<strong>and</strong>omly placed participants<br />
in either an experimental<br />
group (n8) or a control group<br />
(n8). Both groups received conventional<br />
physical therapy twice<br />
daily for 25 minutes for 10 days. Conventional<br />
physical therapy consisted<br />
of weight shifting, trunk stabilization,<br />
lower-extremity muscle facilitation,<br />
<strong>and</strong> gait training. Participants in<br />
the experimental group received<br />
biofeedback on their base of support<br />
through sensors placed around the<br />
proximal leg. These sensors provided<br />
participants with an auditory<br />
signal when their base of support<br />
was below an established threshold.<br />
Data on the base of support were<br />
collected for all participants at baseline<br />
<strong>and</strong> 10 days after the start of<br />
treatment.<br />
Montoya et al 7 placed participants in<br />
either an experimental group (n9)<br />
or a control group (n5). Participants<br />
in the experimental group received<br />
biofeedback 2 times per<br />
week for 4 weeks for a total of 8<br />
sessions. Participants in the control<br />
group practiced walking at the same<br />
dosage but without biofeedback.<br />
Both groups also participated in a<br />
st<strong>and</strong>ardized rehabilitation program<br />
while enrolled in the study. Steplength<br />
biofeedback, with visual <strong>and</strong><br />
auditory signals provided from a<br />
lighted walkway, was provided to<br />
participants in the experimental<br />
group. Step-length data were collected<br />
for all participants at baseline<br />
<strong>and</strong> at the beginning <strong>and</strong> the end of<br />
each treatment session (n8).<br />
Aruin et al 6 reported large improvements<br />
in step width in both the experimental<br />
group <strong>and</strong> the control<br />
group. However, the experimental<br />
group showed a larger improvement<br />
in step width (increase of 0.07 m,<br />
ES12.7, 78% increase) than the<br />
control group (increase of 0.03 m,<br />
ES8.57, 30% increase). Both<br />
groups had similar baseline step<br />
widths (0.09 m in the experimental<br />
group <strong>and</strong> 0.10 m in the control<br />
group). Effect sizes for both groups<br />
were quite large because of very<br />
small reported st<strong>and</strong>ard deviations.<br />
Montoya et al 7 also reported large<br />
improvements in step length in both<br />
the experimental group <strong>and</strong> the control<br />
group. Likewise, the experimental<br />
group showed a larger improvement<br />
in step length (increase of<br />
0.26 m, ES11.1, 79% increase) than<br />
the control group (increase of<br />
0.12 m, ES2.49, 43% increase).<br />
Both groups had similar baseline<br />
step lengths (0.34 m in the experimental<br />
group <strong>and</strong> 0.28 m in the control<br />
group). Again, ESs for both<br />
groups were quite large because of<br />
small reported st<strong>and</strong>ard deviations.<br />
Neither study included a follow-up<br />
assessment.<br />
<strong>Kinetic</strong> <strong>Biofeedback</strong><br />
Participants were provided with<br />
kinetic biofeedback in 2 studies<br />
(Tab. 4). Isakov 14 investigated the effectiveness<br />
of kinetic biofeedback in<br />
a disparate group of participants<br />
(n42) who had undergone a total<br />
hip or knee replacement or a recent<br />
amputation or had had a femoral<br />
neck fracture. White <strong>and</strong> Lifeso 16 analyzed<br />
the effectiveness of kinetic<br />
biofeedback in participants (n28)<br />
after a total hip replacement.<br />
<strong>Biofeedback</strong> was provided either to<br />
increase the overall amount of<br />
weight placed on an involved lower<br />
extremity 14 or to promote kinetic<br />
symmetry between limbs. 16 Outcome<br />
measures included changes in<br />
weight bearing 14 <strong>and</strong> symmetry for<br />
the peak force during the loading<br />
phase, loading rate, <strong>and</strong> vertical impulse.<br />
16 White <strong>and</strong> Lifeso 16 also included<br />
a functional outcome measure,<br />
the Harris Hip Score. 25<br />
Isakov 14 r<strong>and</strong>omly assigned participants<br />
to an experimental group<br />
(n24) or a control group (n18).<br />
Participants in the experimental<br />
group received biofeedback for 30<br />
August 2010 Volume 90 Number 8 Physical Therapy f 7<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Biofeedback</strong> During Gait Retraining<br />
min in 4 physical therapy sessions<br />
during a 2-week period. Participants<br />
in the control group received conventional<br />
gait therapy to promote<br />
full weight bearing. Participants were<br />
given kinetic biofeedback through<br />
an in-shoe device that provided auditory<br />
feedback to promote an increase<br />
in weight bearing during gait<br />
training. <strong>Kinetic</strong> data were collected<br />
for all participants at baseline <strong>and</strong><br />
immediately after treatment.<br />
White <strong>and</strong> Lifeso 16 assigned participants<br />
to 1 of 3 groups: a biofeedback<br />
group (n12), a no-biofeedback<br />
comparison group (n8), <strong>and</strong> a notreatment<br />
control group (n8). Participants<br />
in the biofeedback <strong>and</strong> nobiofeedback<br />
comparison groups<br />
trained on a treadmill 3 times per<br />
week for 8 weeks for a total of 24<br />
sessions. Participants were given kinetic<br />
biofeedback through a monitor<br />
that provided real-time visual displays<br />
of bar graphs representing<br />
the peak force during the first half<br />
of stance (ie, the loading response).<br />
In addition, participants receiving<br />
biofeedback were given verbal cues<br />
aimed at improving kinetic symmetry.<br />
Outcome data were collected for<br />
all participants at baseline <strong>and</strong> after<br />
treatment.<br />
Isakov 14 reported a statistically significant<br />
difference between the experimental<br />
group <strong>and</strong> the control<br />
group in improvements in weight<br />
bearing during gait. The average increase<br />
in weight bearing in participants<br />
in the experimental group was<br />
7.9 kg, but participants in the control<br />
group did not show a meaningful<br />
change (increase of 0.7 kg). Effect<br />
sizes <strong>and</strong> percent differences could<br />
not be calculated because of a lack of<br />
critical variables.<br />
White <strong>and</strong> Lifeso 16 reported that participants<br />
in the biofeedback group<br />
showed significant improvements in<br />
symmetry for the loading rate (decrease<br />
of 15%, ES4.0, 62% improvement<br />
in symmetry) <strong>and</strong> vertical impulse<br />
(decrease of 6.1%, ES4.7,<br />
80% improvement in symmetry). Participants<br />
in the no-biofeedback comparison<br />
group also showed an improvement<br />
in loading rate symmetry<br />
(decrease of 9.1%, ES1.46, 35%<br />
improvement in symmetry) but<br />
showed no change in peak force or<br />
vertical impulse. No changes in symmetry<br />
indices for any of the outcome<br />
measures were noted for participants<br />
in the no-treatment control<br />
group. Neither study included a<br />
follow-up assessment.<br />
Discussion<br />
The specific aim of this review was<br />
to summarize <strong>and</strong> synthesize the<br />
findings of studies involving realtime<br />
kinematic, temporospatial, <strong>and</strong><br />
kinetic biofeedback. The goal was to<br />
provide a general overview of the<br />
effectiveness of these forms of<br />
biofeedback in treating gait abnormalities.<br />
Effect sizes <strong>and</strong> percent differences<br />
suggested that kinematic,<br />
temporospatial, <strong>and</strong> kinetic biofeedback<br />
resulted in moderate to large<br />
short-term treatment effects, indicating<br />
greater success for biofeedback<br />
than for conventional therapy. <strong>Kinematic</strong><br />
<strong>and</strong> temporospatial biofeedback<br />
training interventions were<br />
used for participants after stroke,<br />
whereas kinetic biofeedback was<br />
used for participants who had undergone<br />
a total hip or knee replacement<br />
or an amputation or had had a recent<br />
hip fracture.<br />
This review highlights the need<br />
for further research on real-time<br />
biofeedback to determine whether<br />
kinematic, temporospatial, or kinetic<br />
biofeedback or a combination of<br />
these techniques results in motor<br />
learning. Only 3 of the 7 studies included<br />
a long-term follow-up assessment.<br />
2,8,9 Morris et al 9 reported that<br />
treatment effects were even larger at<br />
a 4-week follow-up assessment than<br />
they were immediately after treatment.<br />
They believed that this result<br />
might have been due to participants<br />
who received biofeedback having<br />
difficulty transferring gait changes<br />
from the feedback condition to the<br />
testing condition (no feedback).<br />
Morris et al 9 further suggested that<br />
lower performance at the assessment<br />
immediately after treatment<br />
than at the 4-week follow-up assessment<br />
might have been due to participants<br />
becoming dependent on receiving<br />
biofeedback in order to alter<br />
gait effectively. The control group<br />
did not experience this effect because<br />
the testing condition was similar<br />
to the training condition. It was<br />
first suggested by Salmoni et al 26 that<br />
concurrent feedback could impede<br />
motor learning by preventing the<br />
processing of other sensory information.<br />
The follow-up assessment by<br />
Colborne et al 2 was limited by the<br />
crossover design of the study, which<br />
made it difficult to separate the individual<br />
effects of the treatments. Likewise,<br />
the follow-up assessment by<br />
Ceceli et al 8 was limited because of a<br />
poor response rate (50%), which<br />
limited the demonstration of any potential<br />
learning effects.<br />
Therefore, the long-term successes<br />
of kinematic, temporospatial, <strong>and</strong> kinetic<br />
biofeedback methods are unclear<br />
at present. The results of Morris<br />
et al 9 might indicate that positive<br />
changes can be maintained, at least<br />
for a few weeks. However, it is impossible<br />
to generalize about whether<br />
motor learning truly occurred in the<br />
balance of the studies reviewed because<br />
of a lack of retention testing<br />
across the studies. Because the purpose<br />
of biofeedback is to promote<br />
the learning of a meaningful <strong>and</strong> permanent<br />
change in gait, such<br />
follow-up testing should be included<br />
in future research.<br />
The treatment protocols for each<br />
method varied in terms of treatment<br />
time per session, treatment frequency<br />
(ie, sessions per week), <strong>and</strong><br />
treatment duration (ie, total number<br />
8 f Physical Therapy Volume 90 Number 8 August 2010<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Biofeedback</strong> During Gait Retraining<br />
of weeks). Some investigators chose<br />
a shorter treatment time per session<br />
(ie, 15–30 minutes) <strong>and</strong> a lower<br />
treatment frequency (ie, 2 or 3 sessions<br />
per week), 2,14,16 whereas another<br />
study with a similar treatment<br />
time per session had daily treatment.<br />
8 Moreover, some studies with<br />
a longer treatment time per session<br />
(ie, 45–50 minutes) had a variety of<br />
treatment frequencies, ranging from<br />
as little as 2 times per week to daily<br />
treatment. 6,7,9 In several studies,<br />
biofeedback treatment was combined<br />
with st<strong>and</strong>ard physical therapy,<br />
so that the actual duration of the<br />
biofeedback treatment was not specifically<br />
reported. Therefore, we are<br />
unable to suggest an optimal biofeedback<br />
treatment protocol on the basis<br />
of the body of literature included in<br />
this review. Future studies should<br />
incorporate treatment protocols<br />
founded on sound physiological<br />
principles <strong>and</strong> reflecting common<br />
scheduling practices currently used<br />
in health care.<br />
Limitations of Existing Studies<br />
R<strong>and</strong>omized controlled designs are<br />
regarded as the gold st<strong>and</strong>ard in<br />
terms of providing the best evidence<br />
about the effectiveness of a particular<br />
treatment. The majority of the<br />
studies included in this review had<br />
control groups. 6–9,14,16 Despite the<br />
presence of a control group, the results<br />
of a few studies should be interpreted<br />
with caution. The study by<br />
White <strong>and</strong> Lifeso 16 had inadequate<br />
r<strong>and</strong>omization. In that study, differences<br />
in age among the no-treatment<br />
control group (60 years), the nobiofeedback<br />
comparison group (70<br />
years), <strong>and</strong> the biofeedback group<br />
(51 years) could have resulted in age<br />
being a confounding factor. Furthermore,<br />
the biofeedback <strong>and</strong> nofeedback<br />
comparison groups differed<br />
from the control group in<br />
terms of baseline function as defined<br />
by the Harris Hip Score. 25 Isakov 14<br />
provided data only for pretest <strong>and</strong><br />
posttest changes <strong>and</strong> did not provide<br />
the exact baseline <strong>and</strong> posttreatment<br />
data. Therefore, we could not determine<br />
whether the treatment <strong>and</strong><br />
control groups had similar baseline<br />
values for the outcome variables. Potential<br />
baseline differences might be<br />
confounding factors affecting the<br />
outcome of retraining studies <strong>and</strong><br />
could lead to misleading conclusions<br />
about the effects of biofeedback.<br />
Colborne et al 2 used a crossover design,<br />
in which participants acted as<br />
their own control <strong>and</strong> received both<br />
biofeedback <strong>and</strong> conventional therapy<br />
during the study. This study design<br />
might have limited the ability to<br />
fully separate the treatment effects<br />
of biofeedback <strong>and</strong> conventional<br />
therapy <strong>and</strong> might also have made it<br />
difficult to identify which treatment<br />
was the source of any permanent<br />
changes (ie, motor learning). Interactive<br />
or additive effects of physical<br />
therapy, biofeedback, <strong>and</strong> time<br />
might have occurred.<br />
None of the studies explicitly included<br />
motor learning concepts in<br />
the study design. Winstein et al 27 reported<br />
that concurrent feedback led<br />
to good performance on a partial–<br />
weight-bearing task during the acquisition<br />
phase. However, the removal<br />
of this guidance led to a degradation<br />
of performance during retention<br />
testing. This phenomenon is well<br />
known in the motor learning literature.<br />
In particular, Swinnen 28 suggested<br />
that researchers regularly expose<br />
learners to nonaugmented<br />
conditions to avoid dependence on<br />
augmented feedback. White <strong>and</strong><br />
Lifeso 16 were the only investigators<br />
to expose participants to nonaugmented<br />
conditions. In that study,<br />
participants receiving biofeedback<br />
walked in 5-minute blocks for a total<br />
of 15 minutes. During each block,<br />
participants walked for 3 minutes<br />
with feedback <strong>and</strong> for 2 minutes<br />
without feedback. Motor learning<br />
principles suggest that this type of<br />
design would be a key component of<br />
a successful program. Janelle et al 29<br />
tested a novel method of presenting<br />
augmented feedback to participants<br />
by allowing them to self-control<br />
feedback scheduling. The results of<br />
that study indicated that participants<br />
in the self-control group performed<br />
better on retention tests than participants<br />
who received different forms<br />
of scheduled or r<strong>and</strong>om feedback.<br />
They suggested that self-controlled<br />
feedback provides an environment<br />
in which participants play a more<br />
active role in their learning, thus resulting<br />
in improved motor performance<br />
<strong>and</strong> motor learning.<br />
The majority of the studies included<br />
only immediate retention testing after<br />
biofeedback intervention. Therefore,<br />
their results were limited to<br />
providing evidence of short-term<br />
changes in motor performance. It is<br />
possible that temporary practice effects<br />
associated with biofeedback<br />
training had yet to subside <strong>and</strong> thus<br />
influenced the results. The inclusion<br />
of longer-term retention intervals<br />
(eg, 6-months or 1 year) in biofeedback<br />
study designs is essential to determining<br />
whether motor learning<br />
has occurred as a result of the<br />
biofeedback.<br />
In all of the studies assessing gait<br />
changes after kinematic biofeedback<br />
in participants after stroke, gait<br />
speed was used as the primary outcome<br />
measure. Schmid et al 30 reported<br />
that significant gains in gait<br />
speed were associated with meaningful<br />
improvements in function <strong>and</strong><br />
quality of life. However, gait speed<br />
alone does not reveal changes in coordination<br />
in walking or reflect motor<br />
learning. Gait outcome measures<br />
such as the Tinetti Performance-<br />
Oriented Mobility Assessment 31 <strong>and</strong><br />
the Gait Assessment Rating Scale 32<br />
can provide some additional information<br />
regarding changes in movement<br />
patterns on the basis of a clinical<br />
evaluation. Recent advances in<br />
technology also have made it possible<br />
for researchers <strong>and</strong> clinicians to<br />
August 2010 Volume 90 Number 8 Physical Therapy f 9<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Biofeedback</strong> During Gait Retraining<br />
collect temporospatial values more<br />
easily by using tools such as the<br />
GAITRite System* or the GaitMat II. †<br />
A 3-dimensional analysis also can<br />
provide researchers <strong>and</strong> clinicians<br />
with a detailed kinematic <strong>and</strong> kinetic<br />
gait analysis. However, this type of<br />
analysis requires more expensive<br />
equipment <strong>and</strong> advanced training.<br />
Two of the 7 studies included functional<br />
outcome measures to assess<br />
improvements in function related to<br />
gait retraining. 9,16 In the study by<br />
Morris et al, 9 participants were assessed<br />
after stroke with the Motor<br />
Assessment Scale, 33 <strong>and</strong> White <strong>and</strong><br />
Lifeso 16 assessed participants after<br />
total hip replacement by using the<br />
Harris Hip Score. 25 Functional outcome<br />
measures specific to a participant<br />
population may be helpful in<br />
indicating quality of life <strong>and</strong> thus<br />
positive changes in function associated<br />
with biofeedback training.<br />
Limitations of This<br />
Systematic Review<br />
The results of this systematic review<br />
are limited to studies that were written<br />
in the English language <strong>and</strong> included<br />
in the MEDLINE, CINAHL,<br />
<strong>and</strong> Cochrane databases or the reference<br />
lists of review articles identified<br />
in those databases. Other studies<br />
may exist outside these resources.<br />
Selection bias is another potential<br />
limitation of systematic reviews.<br />
However, we attempted to minimize<br />
this limitation by using the MINORS<br />
instrument, which was designed to<br />
assess the methodological quality of<br />
both r<strong>and</strong>omized <strong>and</strong> nonr<strong>and</strong>omized<br />
studies. Although nonr<strong>and</strong>omized<br />
studies were not included in<br />
this review, we did not make a specific<br />
methodological decision to exclude<br />
them; they were indirectly<br />
excluded as a result of their low<br />
MINORS scores. Publication bias is<br />
another potential limitation of any<br />
* EQ Inc, PO Box 16, Chalfont, PA 18914.<br />
† CIR Systems Inc, 60 Garlor Dr, Havertown,<br />
PA 19083.<br />
type of literature review. Frequently,<br />
only studies demonstrating positive<br />
results are submitted to peerreviewed<br />
journals <strong>and</strong> published.<br />
This situation may bias results toward<br />
positive treatment effects.<br />
However, this limitation is impossible<br />
to quantify.<br />
Clinical Implications<br />
On the basis of the current literature,<br />
there are insufficient data to make a<br />
recommendation in the form of a<br />
clinical practice guideline. Despite<br />
this shortcoming, we recommend<br />
that clinicians using real-time<br />
biofeedback select outcome measures<br />
capable of revealing changes<br />
in coordination during walking. Additionally,<br />
clinicians should consider<br />
incorporating motor learning principles<br />
into their treatment protocols to<br />
provide clients with a practice environment<br />
that promotes motor learning.<br />
Clinicians should consider using<br />
faded feedback schedules along with<br />
r<strong>and</strong>om <strong>and</strong> variable practice to promote<br />
improved performance during<br />
retention testing. 34 Clinicians should<br />
also consider informing clients about<br />
the potential negative aspect of concurrent<br />
feedback to minimize its impact<br />
on learning. 28<br />
Future Directions<br />
Few studies have investigated the effectiveness<br />
of real-time kinematic,<br />
temporospatial, <strong>and</strong> kinetic biofeedback.<br />
Despite the small number of<br />
studies, this systematic review indicated<br />
that these methods of biofeedback<br />
were capable of providing moderate<br />
to large short-term treatment<br />
effects. Therefore, future research is<br />
warranted to provide clarity regarding<br />
the potential long-term benefits<br />
of real-time biofeedback for gait<br />
retraining.<br />
On the basis of the results of this<br />
systematic review, several recommendations<br />
for future research can<br />
be made. Future work should include<br />
adequate r<strong>and</strong>omization to<br />
ensure that both experimental <strong>and</strong><br />
control groups are equivalent at<br />
baseline, before biofeedback intervention.<br />
Researchers should incorporate<br />
existing motor learning concepts<br />
into study designs to minimize<br />
the negative effects of real-time feedback<br />
while maximizing the benefits.<br />
Future studies also should include<br />
multiday retention testing (eg, at 6<br />
months or 1 year) to assess the learning<br />
of gait changes. Evidence of permanent<br />
changes (ie, motor learning)<br />
is urgently needed to support the<br />
continued use of biofeedback in rehabilitation<br />
settings. To obtain such<br />
evidence, researchers should include<br />
outcome measures that provide<br />
more information about coordination<br />
during gait <strong>and</strong> overall<br />
walking function.<br />
Conclusions<br />
<strong>Real</strong>-time kinematic, temporospatial,<br />
<strong>and</strong> kinetic biofeedback appears to<br />
result in short-term moderate to<br />
large treatment effects. However, it<br />
is unknown whether treatment<br />
changes are maintained. Several<br />
studies lacked adequate r<strong>and</strong>omization,<br />
a fact that should caution readers<br />
when interpreting the authors’<br />
conclusions. Future studies should<br />
ensure adequate r<strong>and</strong>omization of<br />
participants as well as the implementation<br />
of motor learning concepts<br />
<strong>and</strong> the inclusion of retention testing<br />
to assess the long-term success of<br />
biofeedback. They also should include<br />
outcome measures capable of<br />
demonstrating coordinative changes<br />
in gait <strong>and</strong> improvements in function.<br />
Both authors provided concept/idea/research<br />
design, writing, <strong>and</strong> data analysis. Mr<br />
Tate provided data collection.<br />
This article was submitted September 10,<br />
2008, <strong>and</strong> was accepted April 27, 2010.<br />
DOI: 10.2522/ptj.20080281<br />
10 f Physical Therapy Volume 90 Number 8 August 2010<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Biofeedback</strong> During Gait Retraining<br />
References<br />
1 Basmajian JV, Kukulka CG, Narayan MG,<br />
Takebe K. <strong>Biofeedback</strong> treatment of footdrop<br />
after stroke compared with st<strong>and</strong>ard<br />
rehabilitation technique: effects on voluntary<br />
control <strong>and</strong> strength. Arch Phys Med<br />
Rehabil. 1975;56:231–236.<br />
2 Colborne GR, Olney SJ, Griffin MP. Feedback<br />
of ankle joint angle <strong>and</strong> soleus electromyography<br />
in the rehabilitation of<br />
hemiplegic gait. Arch Phys Med Rehabil.<br />
1993;74:1100–1106.<br />
3 Intiso D, Santilli V, Grasso MG, et al. Rehabilitation<br />
of walking with electromyographic<br />
biofeedback in foot-drop after<br />
stroke. Stroke. 1994;25:1189–1192.<br />
4 Binder SA, Moll CB, Wolf SL. Evaluation of<br />
electromyographic biofeedback as an adjunct<br />
to therapeutic exercise in treating<br />
the lower extremities of hemiplegic patients.<br />
Phys Ther. 1981;61:886–893.<br />
5 Burnside IG, Tobias HS, Bursill D. Electromyographic<br />
feedback in the remobilization<br />
of stroke patients: a controlled trial.<br />
Arch Phys Med Rehabil. 1982;63:217–222.<br />
6 Aruin AS, Hanke TA, Sharma A. Base of<br />
support feedback in gait rehabilitation. Int<br />
J Rehabil Res. 2003;26:309–312.<br />
7 Montoya R, Dupui P, Pages B, Bessou P.<br />
Step-length biofeedback device for walk<br />
rehabilitation. Med Biol Eng Comput.<br />
1994;32:416–420.<br />
8 Ceceli E, Dursun E, Cacki A. Comparison<br />
of joint-position biofeedback <strong>and</strong> conventional<br />
therapy methods in genu recurvatum<br />
after stroke: 6 months’ follow-up. Eur<br />
J Phys Med Rehabil. 1996;6:141–144.<br />
9 Morris ME, Matyas TA, Bach TM, Goldie<br />
PA. Electrogoniometric feedback: its effect<br />
on genu recurvatum in stroke. Arch<br />
Phys Med Rehabil. 1992;73:1147–1154.<br />
10 Cozean CD, Pease WS, Hubbell SL.<br />
<strong>Biofeedback</strong> <strong>and</strong> functional electric stimulation<br />
in stroke rehabilitation. Arch Phys<br />
Med Rehabil. 1988;69:401–405.<br />
11 M<strong>and</strong>el AR, Nymark JR, Balmer SJ, et al.<br />
Electromyographic versus rhythmic positional<br />
biofeedback in computerized gait<br />
retraining with stroke patients. Arch Phys<br />
Med Rehabil. 1990;71:649–654.<br />
12 Bradley L, Hart BB, M<strong>and</strong>ana S, et al. Electromyographic<br />
biofeedback for gait training<br />
after stroke. Clin Rehabil. 1998;12:<br />
11–22.<br />
13 Colborne GR, Wright FV, Naumann S.<br />
Feedback of triceps surae EMG in gait of<br />
children with cerebral palsy: a controlled<br />
study. Arch Phys Med Rehabil. 1994;75:<br />
40–45.<br />
14 Isakov E. Gait rehabilitation: a new<br />
biofeedback device for monitoring <strong>and</strong> enhancing<br />
weight-bearing over the affected<br />
lower limb. Eura Medicophys. 2007;43:<br />
21–26.<br />
15 Petrofsky JS. The use of electromyogram<br />
biofeedback to reduce Trendelenburg<br />
gait. Eur J Appl Physiol. 2001;85:491–495.<br />
16 White SC, Lifeso RM. Altering asymmetric<br />
limb loading after hip arthroplasty using<br />
real-time dynamic feedback when walking.<br />
Arch Phys Med Rehabil. 2005;86:<br />
1958–1963.<br />
17 Basmajian JV. Introduction: principles <strong>and</strong><br />
background. In: Basmajian JV, ed. <strong>Biofeedback</strong>:<br />
Principles <strong>and</strong> Practice for Clinicians.<br />
3rd ed. Baltimore, MD: Williams &<br />
Wilkins; 1989:1–4.<br />
18 Bottomley JM. <strong>Biofeedback</strong>: connecting<br />
the mind <strong>and</strong> body. In: Davis CM, ed. Complementary<br />
Therapies in Rehabilitation.<br />
2nd ed. Thorofare, NJ: Slack Inc; 2003:<br />
131–156.<br />
19 Davis IS. Gait retraining in runners. Orthopedic<br />
Physical Therapy. 2005;17:8–13.<br />
20 Woodford H, Price C. EMG biofeedback<br />
for the recovery of motor function after<br />
stroke. Cochrane Database Syst Rev.<br />
2007;2:CD004585.<br />
21 Morel<strong>and</strong> JD, Thomson MA, Fuoco AR.<br />
Electromyographic biofeedback to improve<br />
lower extremity function after<br />
stroke: a meta-analysis. Arch Phys Med Rehabil.<br />
1998;79:134–140.<br />
22 Slim K, Nini E, Forestier D, et al. Methodological<br />
Index for Non-r<strong>and</strong>omized Studies<br />
(MINORS): development <strong>and</strong> validation of<br />
a new instrument. ANZ J Surg. 2003;73:<br />
712–716.<br />
23 Cohen J. Statistical Power Analysis for<br />
the Behavioral Sciences. 2nd ed. Hillsdale,<br />
NJ: Lawrence Erlbaum Associates; 1988.<br />
24 Woods JR, Williams JG, Tavel M. The<br />
two-period crossover design in medical<br />
research. Ann Intern Med. 1989;110:<br />
560–566.<br />
25 Harris WH. Traumatic arthritis of the hip<br />
after dislocation <strong>and</strong> acetabular fractures:<br />
treatment by mold arthroplasty—an endresult<br />
study using a new method of result<br />
evaluation. J Bone Joint Surg Am. 1969;<br />
51:737–755.<br />
26 Salmoni AW, Schmidt RA, Walter CB.<br />
Knowledge of results <strong>and</strong> motor learning:<br />
a review <strong>and</strong> critical reappraisal. Psychol<br />
Bull. 1984;95:355–386.<br />
27 Winstein CJ, Pohl PS, Cardinale C, et al.<br />
Learning a partial-weight-bearing skill: effectiveness<br />
of two forms of feedback. Phys<br />
Ther. 1996;76:985–993.<br />
28 Swinnen SP. Information feedback for motor<br />
skill learning: a review. In: Zelaznik<br />
HN, ed. Advances in Motor Learning <strong>and</strong><br />
Control. Champaign, IL: Human <strong>Kinetic</strong>s;<br />
1996:37–65.<br />
29 Janelle CM, Kim J, Singer RN. Subjectcontrolled<br />
performance feedback <strong>and</strong><br />
learning of a closed motor skill. Percept<br />
Mot Skills. 1995;81:627–634.<br />
30 Schmid A, Duncan PW, Studenski S, et al.<br />
Improvements in speed-based gait classifications<br />
are meaningful. Stroke. 2007;38:<br />
2096–2100.<br />
31 Tinetti ME. Performance-oriented assessment<br />
of mobility problems in elderly patients.<br />
J Am Geriatr Soc. 1986;34:119–126.<br />
32 Wolfson L, Whipple R, Amerman P, Tobin<br />
JN. Gait assessment in the elderly: a gait<br />
abnormality rating scale <strong>and</strong> its relation to<br />
falls. J Gerontol. 1990;45:M12–M19.<br />
33 Carr JH, Shepherd RB, Nordholm L, Lynne<br />
D. Investigation of a new motor assessment<br />
scale for stroke patients. Phys Ther.<br />
1985;65:175–180.<br />
34 Rose DJ, Christina RW. Augmented Feedback<br />
<strong>and</strong> Motor Learning: A Multilevel<br />
Approach to the Study of Motor Control<br />
<strong>and</strong> Learning. 2nd ed. San Francisco, CA:<br />
Benjamin Cummings; 2006:291–320.<br />
August 2010 Volume 90 Number 8 Physical Therapy f 11<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015
<strong>Real</strong>-<strong>Time</strong> <strong>Kinematic</strong>, <strong>Temporospatial</strong>, <strong>and</strong> <strong>Kinetic</strong><br />
<strong>Biofeedback</strong> During Gait Retraining in Patients: A<br />
Systematic Review<br />
Jeremiah J. Tate <strong>and</strong> Clare E. Milner<br />
PHYS THER. Published online June 17, 2010<br />
doi: 10.2522/ptj.20080281<br />
Subscription<br />
Information<br />
http://ptjournal.apta.org/subscriptions/<br />
Permissions <strong>and</strong> Reprints http://ptjournal.apta.org/site/misc/terms.xhtml<br />
Information for Authors<br />
http://ptjournal.apta.org/site/misc/ifora.xhtml<br />
Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015