Real-Time Kinematic, Temporospatial, and Kinetic Biofeedback ...

Real-Time Kinematic, Temporospatial, and Kinetic 

Biofeedback During Gait Retraining in Patients: A 

Systematic Review 

Jeremiah J. Tate and Clare E. Milner 

PHYS THER. Published online June 17, 2010 

doi: 10.2522/ptj.20080281 

The online version of this article, along with updated information and services, can be 

found online at: http://ptjournal.apta.org/content/early/2010/06/17/ptj.20080281 

Collections 

E-mail alerts 

This article, along with others on similar topics, appears 

in the following collection(s): 

Electrotherapy 

Gait and Locomotion Training 

Gait Disorders 

Kinesiology/Biomechanics 

Motor Control and Motor Learning 

Systematic Reviews/Meta-analyses 

Sign up here to receive free e-mail alerts 

Online First articles are published online before they appear in a regular issue of 

Physical Therapy (PTJ). PTJ publishes 2 types of Online First articles: 

Author manuscripts: PDF versions of manuscripts that have been peer-reviewed and 

accepted for publication but have not yet been copyedited or typeset. This allows PTJ 

readers almost immediate access to accepted papers. 

Page proofs: edited and typeset versions of articles that incorporate any author 

corrections and replace the original author manuscript. 

Downloaded from http://ptjournal.apta.org/ 

by guest on April 2, 2015

Research Report 

Real-Time Kinematic, Temporospatial, 

and Kinetic Biofeedback During Gait 

Retraining in Patients: 

A Systematic Review 

Jeremiah J. Tate, Clare E. Milner 

Background. Biofeedback has been used in rehabilitation settings for gait 

retraining. 

Purpose. The purpose of this review was to summarize and synthesize the findings 

of studies involving real-time kinematic, temporospatial, and kinetic biofeedback. 

The goal was to provide a general overview of the effectiveness of these forms 

of biofeedback in treating gait abnormalities. 

Data Sources. Articles were identified through searches of the following databases: 

MEDLINE, CINAHL (Cumulative Index to Nursing and Allied Health Literature), 

and Cochrane Central Register for Controlled Trials. All searches were limited to the 

English language and encompassed the period from 1965 to November 2007. 

Study Selection. Titles and abstracts were screened to identify studies that met 

the following requirements: the study included the use of kinematic, temporospatial, 

or kinetic biofeedback during gait training, and the population of interest showed 

abnormal movement patterns as a result of a pathology or injury. 

J.J. Tate, PT, MS, is a doctoral student, 

Department of Exercise, 

Sport, and Leisure Studies, University 

of Tennessee, Knoxville, 

Tennessee. 

C.E. Milner, PhD, is Assistant Professor, 

Department of Exercise, 

Sport, and Leisure Studies, University 

of Tennessee, 1914 Andy 

Holt Ave, HPER 322, Knoxville, TN 

37996-2700 (USA). Address all 

correspondence to Dr Milner at: 

milner@utk.edu. 

[Tate JJ, Milner CE. Real-time kinematic, 

temporospatial, and kinetic 

biofeedback during gait retraining 

in patients: a systematic review. 

Phys Ther. 2010;90:xxx–xxx.] 

© 2010 American Physical Therapy 

Association 

Data Extraction. All articles that met the inclusion criteria were assessed by use 

of the Methodological Index for Nonrandomized Studies. 

Data Synthesis. Seven articles met the inclusion criteria and were included in 

the review. Effect sizes were calculated for the primary outcome variables for all 

studies that provided enough data. Effect sizes generally suggested moderate to large 

treatment effects for all methods of biofeedback during practice. 

Limitations. Several of the studies lacked adequate randomization; therefore, 

readers should exercise caution when interpreting authors’ conclusions. 

Conclusions. Each biofeedback method appeared to result in moderate to large 

treatment effects immediately after treatment. However, it is unknown whether the 

effects were maintained. Future studies should ensure adequate randomization of 

participants and implementation of motor learning concepts and should include 

retention testing to assess the long-term success of biofeedback and outcome measures 

capable of demonstrating coordinative changes in gait and improvement in 

function. 

Post a Rapid Response to 

this article at: 

ptjournal.apta.org 

August 2010 Volume 90 Number 8 Physical Therapy f 1 

Downloaded from http://ptjournal.apta.org/ by guest on April 2, 2015

Real-Time Biofeedback During Gait Retraining 

Biofeedback was first introduced 

in the literature more than 30 

years ago as a training tool used 

in rehabilitation settings to facilitate 

normal movement patterns after injury. 

1 Since then, biofeedback has 

been used primarily in rehabilitation 

settings for the treatment of gait abnormalities 

in adults after stroke. 2–12 

Biofeedback also has been used to 

facilitate the normalization of gait 

patterns in children with cerebral 

palsy 13 and in adults after amputation, 

14 after spinal cord injuries, 15 after 

hip fractures, 14 and after total 

hip 14,16 and knee 14 joint replacements. 

Biofeedback is a technique that typically 

uses electronic equipment to 

provide a client with auditory signals, 

visual signals, or both regarding 

internal physiological events, both 

normal and abnormal (eg, heart rate, 

blood pressure, and level of muscle 

activity). 17 During biofeedback for 

gait retraining, the client is provided 

with augmented information (eg, kinematics, 

kinetics, and electromyography) 

regarding physiological responses. 

Additionally, biofeedback 

provides clinicians with a useful tool 

for giving clients instructions on 

how to modify movement patterns. 

Thus, biofeedback complements the 

already present internal feedback (ie, 

visual, auditory, and proprioceptive 

feedback) and acts as a “sixth 

sense.” 18 Biofeedback typically is 

provided instantaneously to the 

learner (ie, in real time), whereas 

other methods of external feedback 

(eg, verbal and video feedback) are 

provided some time after the movement. 

More recently, a resurgence of 

interest in real-time feedback has developed 

because of the expansion of 

technology related to kinematic 19 

and kinetic 14,16 biofeedback. 

Electromyographic biofeedback may 

be the most popular method of providing 

biofeedback and has been 

used frequently in gait retraining after 

stroke. However, meta-analyses 

Table 1. 

Results of Searches of Electronic Databases 

Database 

of electromyographic biofeedback 

studies of people after stroke have 

concluded that little evidence supports 

its use in addition to conventional 

physical therapy. 20,21 Other 

forms of biofeedback, such as kinematic, 

2,8,9 temporospatial, 6,7 and kinetic, 

14,16 also have been developed 

and used in rehabilitation settings for 

gait retraining. It remains unknown 

whether these methods are effective 

in treating gait abnormalities. Therefore, 

the purpose of this systematic 

review was to summarize and synthesize 

the findings of studies involving 

real-time kinematic, temporospatial, 

and kinetic biofeedback. 

This effort provided a general overview 

of the effectiveness of these 

forms of biofeedback in treating gait 

abnormalities. 

Method 

Data Sources and Searches 

Three electronic databases were 

searched in a systematic fashion to 

identify relevant articles: MEDLINE, 

CINAHL (Cumulative Index to Nursing 

and Allied Health Literature), and 

Cochrane Central Register for Controlled 

Trials. All searches were limited 

to the English language and encompassed 

the period from 1965 to 

November 2007. The following key 

words were used in the searches: 

feedback, biofeedback, walking, ambulation, 

and gait. Key words were 

combined to identify studies involving 

biofeedback and gait. Table 1 

shows the specific key words and 

combinations of key words used in 

Key Words 

MEDLINE (1) feedback; (2) biofeedback; (3) 1 or 2; (4) walk a ; (5) ambul a ; 

(6) walking; (7) gait; (8) 4 or 5 or 6 or 7; (9) 3 and 8 

CINAHL (1) feedback; (2) biofeedback; (3) 1 or 2; (4) walk a ; (5) ambul a ; 

(6) walking; (7) gait; (8) 4 or 5 or 6 or 7; (9) 3 and 8 

Cochrane 

a Limit: English language. 

(1) feedback; (2) biofeedback; (3) 1 or 2; (4) walking; 

(5) ambulation; (6) gait; (7) 4 or 5 or 6; (8) 3 and 7 

No. of 

Studies 

504 

162 

85 

each search. Each review article that 

was identified by MEDLINE also was 

manually searched to identify other 

potential articles. 

Study Selection 

One reviewer (J.J.T.) screened titles 

and abstracts identified during electronic 

and manual searches to determine 

eligibility. In the event that the 

title or abstract did not provide 

enough information, the article was 

obtained for full review. Published 

articles were considered for initial 

inclusion in the review if the following 

requirements were met: the 

study included the use of kinematic, 

temporospatial, or kinetic biofeedback 

during gait training and the 

population of interest showed abnormal 

movement patterns as a result of 

a pathology or injury. Figure 1 shows 

a flow diagram of the systematic review 

process. 

Data Extraction and 

Quality Assessment 

All articles that met the inclusion criteria 

were assessed by 2 reviewers 

(J.J.T. and C.E.M.) using the Methodological 

Index for Nonrandomized 

Studies (MINORS). 22 The MINORS instrument 

has been determined to be 

valid and reliable in assessing the 

methodological quality of both randomized 

and nonrandomized studies. 

22 Scores on the MINORS range 

from 0 to 24 for randomized studies 

and 0 to 16 for nonrandomized studies. 

The following methodological 

items are assessed with the MINORS 

2 f Physical Therapy Volume 90 Number 8 August 2010 



instrument: aim of the study, inclusion 

of consecutive participants, prospective 

data collection, appropriate 

outcome assessments, unbiased assessments 

of outcomes, appropriate 

follow-up, attrition, and sample size. 

In addition, the quality of the comparison 

group and statistical analyses 

in randomized trials only are assessed. 

Furthermore, the 2 reviewers clarified 

the intent of the questions concerning 

the inclusion of consecutive 

participants and prospective data 

collection to improve interrater reliability. 

The purpose of the question 

concerning the inclusion of consecutive 

participants was to judge 

whether inclusion and exclusion criteria 

were included in the study. The 

purpose of the question regarding 

prospective data collection was to 

judge whether the study was descriptive 

(score of 0) or prospective 

(score of 1 or 2). The delineation 

between scores of 1 and 2 for prospective 

quality was based on 

whether the study demonstrated 1 

or more specific a priori aims and 

hypotheses. 

The MINORS scores were averaged 

across the 2 reviewers to enable the 

ranking of studies. A histogram of 

the studies and their MINORS scores 

was created to assist in determining a 

cutoff point, based on the natural 

clustering of groups, for inclusion in 

the review (Fig. 2). 

Data Synthesis and Analysis 

A meta-analysis was not performed 

because of the wide variety of study 

designs, methodologies, and outcome 

variables. Effect sizes (ESs) 

were calculated for the primary outcome 

variables for all studies that 

provided enough data. Effect sizes 

were calculated by subtracting the 

mean score at the baseline from the 

mean score after treatment and then 

dividing the result by a standard deviation 

resulting from pooling of the 

baseline and treatment standard de- 

Figure 1. 

Flow diagram of the systematic review process. 

viations. 23 Effect sizes were used in 

this review to provide a means of 

evaluating treatment success because 

they are not directly affected 

by sample size but take into account 

within-group variability. Effect sizes 

were standardized on the basis of the 

work of Cohen (small.02, moderate0.5, 

and large0.8). 23 All ESs 

represented comparisons of the 

mean posttreatment and follow-up 

(if applicable) scores with the mean 

baseline scores for each of the 

biofeedback and control groups. The 

ES for change over time in the 

biofeedback group was then compared 

with that in the control group. 

Percent differences also were calculated 

to aid in the interpretation of 

studies that did not provide enough 

data to calculate ESs. 

Results 

Study Identification 

A search of the MEDLINE database 

identified a total of 504 articles, a 

search of the CINAHL database identified 

162 articles, and a search of 

the Cochrane Central Register of 

Controlled Trials identified 85 articles. 

In total, 666 individual articles 

were identified by these databases. 

Fifty-eight of these articles were 

identified as potentially relevant, and 

their full texts were retrieved. After 

an initial review by the first reviewer 

(J.J.T.), 35 of these articles were excluded 

because they did not meet 




Kinematic Biofeedback 

In 3 studies, kinematic biofeedback 

was provided with electrogoniometers 

(Tab. 2). All 3 studies involved 

analysis of the effect of kinematic 

biofeedback in participants who had 

had a stroke. Ceceli et al 8 and Morris 

et al 9 analyzed the effectiveness of 

providing participants (n41 and 

26, respectively) with kinematic 

biofeedback of the knee compared 

with conventional physical therapy 

in efforts to minimize genu recurvatum. 

Colborne et al 2 investigated the 

effectiveness of providing participants 

(n8) with kinematic biofeedback 

for the ankle in attempts to 

improve ankle control. Outcome 

measures included gait speed, 2,8,9 

number of recurvations, 8 and 

symmetry. 2,9 

Figure 2. 

Histogram of scores on the Methodological Index for Nonrandomized Studies. 22 

the inclusion criteria. Both reviewers 

assessed the remaining 23 studies 

using the MINORS instrument. Of 

the 23 articles assessed, 7 achieved a 

score of 16 or greater out of a possible 

24 and were included in the review 

(Fig. 1). A cutoff point of 16 

was identified qualitatively on observation 

of the histogram as a natural 

breakpoint between clusters of studies 

(Fig. 2). Tables 2, 3, and 4 summarize 

the characteristics of the 

included studies with regard to participant 

population, sample sizes, 

participant ages, treatment protocols, 

presence of masking and retention 

tests, quantitative variables, 

functional outcome measures, and 

conclusions drawn by the authors. 

Biofeedback Protocol 

Generally, there was a large range in 

the structures of the treatment protocols. 

The mean treatment time in 

all studies was 35 minutes, with a 

range of 15 to 50 minutes, and the 

mean treatment frequency was 4 

times per week, with a range of 

twice per week to daily treatments. 

The mean treatment duration was 

3.5 weeks, with a range of 1.5 to 

8 weeks. The mean cumulative 

biofeedback time per study (ie, treatment 

time treatment frequency 

treatment duration) was 397 minutes, 

with a range of 120 to 900 

minutes. 

Ceceli et al 8 assigned participants to 

either an experimental group 

(n26) or a control group (n15). 

Both groups received conventional 

therapy that consisted of pelvis and 

hip control exercises, weight shifting, 

and gait training. In addition, 

participants in the experimental 

group received kinematic biofeedback 

for 10 daily sessions, each lasting 

30 minutes. Posttreatment data 

were collected for participants in the 

experimental group at the end of the 

10 days of biofeedback training; data 

were collected for participants in 

the control group at the time of 

discharge. 

In the study by Morris et al, 9 participants 

received treatment in 2 separate 

phases, lasting 4 weeks each. 

Participants in the experimental 

group (n13) received kinematic 

biofeedback during the first phase 

and conventional physical therapy 

during the second phase. Kinematic 

biofeedback of the involved knee 

was provided to participants during 

standing and gait training for 30 to 

45 minutes per treatment session. 

Participants in the control group 

(n13) received conventional phys- 




Table 2. 

Characteristics of Electrogoniometer Studies 

Study 

No. of 

Participants 

(Experimental 

Study Group/Control 

Population a Group) Age (y) 

Treatment Protocol 

Time 

(min) Frequency Duration 

Masking 

Method of 

Delivery 

Follow-up 

(wk) Quantitative Variables 

Functional 

Outcome Improvement b 

Ceceli Adult CVA 26/15 Median 

et al 8 49.5/54 

30 Daily 10 d No Auditory/ 

visual 

24 Cadence, no of 

recurvations/50 steps 

None Yes 

Morris Adult CVA 13/13 X64/64 

et al 9 (range 

33–74) 

45 5/wk 4 wk/block No Auditory 4 Peak knee hyperextension, 

gait speed, single-limb 

stance symmetry 

Gait recovery 

(Motor 

Assessment 

Scale) 

Yes 

Colborne Adult CVA 8 (repeated 

et al 2 measures) 

Not reported 30 2/wk 4 wk/block No Auditory 4 Gait speed, stride length, 

stride time, knee angle at 

heel-strike, knee angle at 

50% swing, ankle angle 

of motion, push-off 

impulse, swing ratio 

(affected/unaffected), 

weight ratio (affected/ 

unaffected) 

None Yes 

a CVAcerebrovascular accident. 

b Statistically significant improvement in experimental group vs control group, as concluded by study authors. 

Table 3. 

Characteristics of Temporospatial Biofeedback Studies 

Study 

No. of 

Participants 

(Experimental 



Protocol 



Masking 

Method of 

Delivery Follow-up 

Quantitative 

Variables 

Functional 


Aruin et al 6 Adult CVA 8/8 X65 50 Daily 10 d No Auditory None Step width None Yes 

Montoya et al 7 Adult CVA 9/5 X64/60 45 2/wk 4 wk No Auditory/visual None Step length 

(paretic limb) 

None Yes 

a CVAcerebrovascular accident. 

b Statistically significant improvement in experimental group vs control group, as concluded by study authors. 




Table 4. 

Characteristics of Kinetic Biofeedback Studies 

Protocol 

Functional 


Quantitative 

Variables 

Method of 

Delivery Follow-up 

Masking 



No. of 

Participants 

(Experimental 



Study 

24/18 X62/66 30 4 sessions 14 d No Auditory None Weight bearing None Yes 

Isakov 14 Amputation, 

THR, TKR, 

hip fracture 

Yes d 

Harris Hip 

Score 

White and Lifeso 16 THR 12/8/8 c X51/60/ 70 c 15 3/wk 8 wk No Verbal/visual None Symmetry indexes 

for peak force 

during loading, 

loading rate, 

and vertical 

impulse; rate of 

perceived 

exertion 

a THRtotal hip replacement, TKRtotal knee replacement. 

b Statistically significant improvement in experimental groups vs control group, as concluded by study authors. 

c Biofeedback experimental group/no-treatment control group/no-biofeedback experimental group. 

d Authors reported statistically significant improvements in both the biofeedback and no-biofeedback experimental groups. 

ical therapy during both phases. Outcome 

data were collected for all participants 

at baseline and after each 

treatment phase. 

Colborne et al 2 used a 3-period crossover 

design to assess the effectiveness 

of kinematic biofeedback. The 8 

participants received 1 of 3 treatments: 

kinematic biofeedback, electromyographic 

biofeedback, and 

conventional physical therapy. Conventional 

physical therapy was given 

during either the first phase (n4) or 

the last phase (n4). Each treatment 

phase lasted 4 weeks and consisted 

of biweekly treatment sessions, each 

lasting 30 minutes. No washout periods 

were included between treatments. 

Gait speed data were collected 

for all participants at baseline, 

after each treatment phase, and at a 

1-month follow-up. 

Ceceli et al 8 found that participants 

in the experimental group showed a 

statistically significant decrease in 

the number of recurvations compared 

with the control group immediately 

after treatment—from 45 to 8 

recurvations, an 83% decrease, for 

the experimental group, and from 

49 to 39 recurvations, a 19% decrease, 

for the control group. Additionally, 

participants were asked to 

return for a 6-month follow-up. However, 

we did not calculate ESs or 

percent differences for 6-month 

follow-up data because of the poor 

participant follow-up rate (50%); 

with a low follow-up rate, the data 

might not be a good representation 

of the data for all participants. 

Morris et al 9 reported a moderate effect 

for increased gait speed in the 

experimental group (increase of 

0.10 m/s, ES0.50, 33% increase) 

but no effect in the control group 

(increase of 0.01 m/s, ES0.08, 5% 

increase) after the first treatment 

phase. Both groups had similar baseline 

gait velocities (0.33 m/s in the 

experimental group and 0.30 m/s in 




the control group). Both groups 

showed statistically significant reductions 

in peak knee extension during 

stance, although no group difference 

was noted. After the second 

treatment phase, a large increase in 

gait speed was reported in the experimental 

group (increase of 0.23 m/s, 

ES1.45, 75% increase), but only a 

small change was reported in the 

control group (increase of 0.10 m/s, 

ES0.46, 37% increase). Additionally, 


group showed a statistically significant 

decrease in peak knee extension 

compared with participants in 

the control group. Effect sizes and 

percent differences for peak knee 

extension could not be calculated 

because of a lack of data. 

Colborne et al 2 found that kinematic 

biofeedback training resulted in a 

moderate increase in gait speed (increase 

of 0.11 m/s, ES0.51, 23% 

increase) but that conventional therapy 

resulted in only a small improvement 

(increase of 0.08 m/s, 

ES0.37, 17% increase). Participants 

in both of those treatment groups 

had the same baseline gait speed 

(0.48 m/s). We did not interpret the 

follow-up data because carryover effects 

for either or both treatment 

methods might have occurred as a 

result of the crossover design of the 

study. 24 

Temporospatial Biofeedback 

In 2 studies, temporospatial biofeedback 

was provided for participants 

after stroke (Tab. 3). Aruin et al 6 

investigated the effectiveness of 

biofeedback regarding the base of 

support in 16 participants, and Montoya 

et al 7 analyzed the effectiveness 

of biofeedback regarding step length 

in 14 participants. Biofeedback was 

provided in efforts to improve the 

base of support 6 and the symmetry 

of step length. 7 Outcome measures 

included step width 6 and step 

length. 7 

Aruin et al 6 randomly placed participants 

in either an experimental 

group (n8) or a control group 

(n8). Both groups received conventional 

physical therapy twice 

daily for 25 minutes for 10 days. Conventional 

physical therapy consisted 

of weight shifting, trunk stabilization, 

lower-extremity muscle facilitation, 

and gait training. Participants in 

the experimental group received 

biofeedback on their base of support 

through sensors placed around the 

proximal leg. These sensors provided 

participants with an auditory 

signal when their base of support 

was below an established threshold. 

Data on the base of support were 

collected for all participants at baseline 

and 10 days after the start of 

treatment. 

Montoya et al 7 placed participants in 

either an experimental group (n9) 

or a control group (n5). Participants 

in the experimental group received 

biofeedback 2 times per 

week for 4 weeks for a total of 8 

sessions. Participants in the control 

group practiced walking at the same 

dosage but without biofeedback. 

Both groups also participated in a 

standardized rehabilitation program 

while enrolled in the study. Steplength 

biofeedback, with visual and 

auditory signals provided from a 

lighted walkway, was provided to 


group. Step-length data were collected 

for all participants at baseline 

and at the beginning and the end of 

each treatment session (n8). 

Aruin et al 6 reported large improvements 

in step width in both the experimental 

group and the control 

group. However, the experimental 

group showed a larger improvement 

in step width (increase of 0.07 m, 

ES12.7, 78% increase) than the 

control group (increase of 0.03 m, 

ES8.57, 30% increase). Both 

groups had similar baseline step 

widths (0.09 m in the experimental 

group and 0.10 m in the control 

group). Effect sizes for both groups 

were quite large because of very 

small reported standard deviations. 

Montoya et al 7 also reported large 

improvements in step length in both 

the experimental group and the control 

group. Likewise, the experimental 

group showed a larger improvement 

in step length (increase of 

0.26 m, ES11.1, 79% increase) than 

the control group (increase of 

0.12 m, ES2.49, 43% increase). 

Both groups had similar baseline 

step lengths (0.34 m in the experimental 

group and 0.28 m in the control 

group). Again, ESs for both 

groups were quite large because of 

small reported standard deviations. 

Neither study included a follow-up 

assessment. 

Kinetic Biofeedback 

Participants were provided with 

kinetic biofeedback in 2 studies 

(Tab. 4). Isakov 14 investigated the effectiveness 

of kinetic biofeedback in 

a disparate group of participants 

(n42) who had undergone a total 

hip or knee replacement or a recent 

amputation or had had a femoral 

neck fracture. White and Lifeso 16 analyzed 

the effectiveness of kinetic 

biofeedback in participants (n28) 

after a total hip replacement. 

Biofeedback was provided either to 

increase the overall amount of 

weight placed on an involved lower 

extremity 14 or to promote kinetic 

symmetry between limbs. 16 Outcome 

measures included changes in 

weight bearing 14 and symmetry for 

the peak force during the loading 

phase, loading rate, and vertical impulse. 

16 White and Lifeso 16 also included 

a functional outcome measure, 

the Harris Hip Score. 25 

Isakov 14 randomly assigned participants 

to an experimental group 

(n24) or a control group (n18). 

Participants in the experimental 

group received biofeedback for 30 




min in 4 physical therapy sessions 

during a 2-week period. Participants 

in the control group received conventional 

gait therapy to promote 

full weight bearing. Participants were 

given kinetic biofeedback through 

an in-shoe device that provided auditory 

feedback to promote an increase 

in weight bearing during gait 

training. Kinetic data were collected 

for all participants at baseline and 

immediately after treatment. 

White and Lifeso 16 assigned participants 

to 1 of 3 groups: a biofeedback 

group (n12), a no-biofeedback 

comparison group (n8), and a notreatment 

control group (n8). Participants 

in the biofeedback and nobiofeedback 

comparison groups 

trained on a treadmill 3 times per 

week for 8 weeks for a total of 24 

sessions. Participants were given kinetic 

biofeedback through a monitor 

that provided real-time visual displays 

of bar graphs representing 

the peak force during the first half 

of stance (ie, the loading response). 

In addition, participants receiving 

biofeedback were given verbal cues 

aimed at improving kinetic symmetry. 

Outcome data were collected for 

all participants at baseline and after 

treatment. 

Isakov 14 reported a statistically significant 

difference between the experimental 

group and the control 

group in improvements in weight 

bearing during gait. The average increase 

in weight bearing in participants 

in the experimental group was 

7.9 kg, but participants in the control 

group did not show a meaningful 

change (increase of 0.7 kg). Effect 

sizes and percent differences could 

not be calculated because of a lack of 

critical variables. 

White and Lifeso 16 reported that participants 

in the biofeedback group 

showed significant improvements in 

symmetry for the loading rate (decrease 

of 15%, ES4.0, 62% improvement 

in symmetry) and vertical impulse 

(decrease of 6.1%, ES4.7, 

80% improvement in symmetry). Participants 

in the no-biofeedback comparison 

group also showed an improvement 

in loading rate symmetry 

(decrease of 9.1%, ES1.46, 35% 

improvement in symmetry) but 

showed no change in peak force or 

vertical impulse. No changes in symmetry 

indices for any of the outcome 

measures were noted for participants 

in the no-treatment control 

group. Neither study included a 

follow-up assessment. 

Discussion 

The specific aim of this review was 

to summarize and synthesize the 

findings of studies involving realtime 

kinematic, temporospatial, and 

kinetic biofeedback. The goal was to 

provide a general overview of the 

effectiveness of these forms of 

biofeedback in treating gait abnormalities. 

Effect sizes and percent differences 

suggested that kinematic, 

temporospatial, and kinetic biofeedback 

resulted in moderate to large 

short-term treatment effects, indicating 

greater success for biofeedback 

than for conventional therapy. Kinematic 

and temporospatial biofeedback 

training interventions were 

used for participants after stroke, 

whereas kinetic biofeedback was 

used for participants who had undergone 

a total hip or knee replacement 

or an amputation or had had a recent 

hip fracture. 

This review highlights the need 

for further research on real-time 

biofeedback to determine whether 

kinematic, temporospatial, or kinetic 

biofeedback or a combination of 

these techniques results in motor 

learning. Only 3 of the 7 studies included 

a long-term follow-up assessment. 

2,8,9 Morris et al 9 reported that 

treatment effects were even larger at 

a 4-week follow-up assessment than 

they were immediately after treatment. 

They believed that this result 

might have been due to participants 

who received biofeedback having 

difficulty transferring gait changes 

from the feedback condition to the 

testing condition (no feedback). 

Morris et al 9 further suggested that 

lower performance at the assessment 

immediately after treatment 

than at the 4-week follow-up assessment 

might have been due to participants 

becoming dependent on receiving 

biofeedback in order to alter 

gait effectively. The control group 

did not experience this effect because 

the testing condition was similar 

to the training condition. It was 

first suggested by Salmoni et al 26 that 

concurrent feedback could impede 

motor learning by preventing the 

processing of other sensory information. 

The follow-up assessment by 

Colborne et al 2 was limited by the 

crossover design of the study, which 

made it difficult to separate the individual 

effects of the treatments. Likewise, 

the follow-up assessment by 

Ceceli et al 8 was limited because of a 

poor response rate (50%), which 

limited the demonstration of any potential 

learning effects. 

Therefore, the long-term successes 

of kinematic, temporospatial, and kinetic 

biofeedback methods are unclear 

at present. The results of Morris 

et al 9 might indicate that positive 

changes can be maintained, at least 

for a few weeks. However, it is impossible 

to generalize about whether 

motor learning truly occurred in the 

balance of the studies reviewed because 

of a lack of retention testing 

across the studies. Because the purpose 

of biofeedback is to promote 

the learning of a meaningful and permanent 

change in gait, such 

follow-up testing should be included 

in future research. 

The treatment protocols for each 

method varied in terms of treatment 

time per session, treatment frequency 

(ie, sessions per week), and 

treatment duration (ie, total number 




of weeks). Some investigators chose 

a shorter treatment time per session 

(ie, 15–30 minutes) and a lower 

treatment frequency (ie, 2 or 3 sessions 

per week), 2,14,16 whereas another 

study with a similar treatment 

time per session had daily treatment. 

8 Moreover, some studies with 

a longer treatment time per session 

(ie, 45–50 minutes) had a variety of 

treatment frequencies, ranging from 

as little as 2 times per week to daily 

treatment. 6,7,9 In several studies, 

biofeedback treatment was combined 

with standard physical therapy, 

so that the actual duration of the 

biofeedback treatment was not specifically 

reported. Therefore, we are 

unable to suggest an optimal biofeedback 

treatment protocol on the basis 

of the body of literature included in 

this review. Future studies should 

incorporate treatment protocols 

founded on sound physiological 

principles and reflecting common 

scheduling practices currently used 

in health care. 

Limitations of Existing Studies 

Randomized controlled designs are 

regarded as the gold standard in 

terms of providing the best evidence 

about the effectiveness of a particular 

treatment. The majority of the 

studies included in this review had 

control groups. 6–9,14,16 Despite the 

presence of a control group, the results 

of a few studies should be interpreted 

with caution. The study by 

White and Lifeso 16 had inadequate 

randomization. In that study, differences 

in age among the no-treatment 

control group (60 years), the nobiofeedback 

comparison group (70 

years), and the biofeedback group 

(51 years) could have resulted in age 

being a confounding factor. Furthermore, 

the biofeedback and nofeedback 

comparison groups differed 

from the control group in 

terms of baseline function as defined 

by the Harris Hip Score. 25 Isakov 14 

provided data only for pretest and 

posttest changes and did not provide 

the exact baseline and posttreatment 

data. Therefore, we could not determine 

whether the treatment and 

control groups had similar baseline 

values for the outcome variables. Potential 

baseline differences might be 

confounding factors affecting the 

outcome of retraining studies and 

could lead to misleading conclusions 

about the effects of biofeedback. 

Colborne et al 2 used a crossover design, 

in which participants acted as 

their own control and received both 

biofeedback and conventional therapy 

during the study. This study design 

might have limited the ability to 

fully separate the treatment effects 

of biofeedback and conventional 

therapy and might also have made it 

difficult to identify which treatment 

was the source of any permanent 

changes (ie, motor learning). Interactive 

or additive effects of physical 

therapy, biofeedback, and time 

might have occurred. 

None of the studies explicitly included 

motor learning concepts in 

the study design. Winstein et al 27 reported 

that concurrent feedback led 

to good performance on a partial– 

weight-bearing task during the acquisition 

phase. However, the removal 

of this guidance led to a degradation 

of performance during retention 

testing. This phenomenon is well 

known in the motor learning literature. 

In particular, Swinnen 28 suggested 

that researchers regularly expose 

learners to nonaugmented 

conditions to avoid dependence on 

augmented feedback. White and 

Lifeso 16 were the only investigators 

to expose participants to nonaugmented 

conditions. In that study, 

participants receiving biofeedback 

walked in 5-minute blocks for a total 

of 15 minutes. During each block, 

participants walked for 3 minutes 

with feedback and for 2 minutes 

without feedback. Motor learning 

principles suggest that this type of 

design would be a key component of 

a successful program. Janelle et al 29 

tested a novel method of presenting 

augmented feedback to participants 

by allowing them to self-control 

feedback scheduling. The results of 

that study indicated that participants 

in the self-control group performed 

better on retention tests than participants 

who received different forms 

of scheduled or random feedback. 

They suggested that self-controlled 

feedback provides an environment 

in which participants play a more 

active role in their learning, thus resulting 

in improved motor performance 

and motor learning. 

The majority of the studies included 

only immediate retention testing after 

biofeedback intervention. Therefore, 

their results were limited to 

providing evidence of short-term 

changes in motor performance. It is 

possible that temporary practice effects 

associated with biofeedback 

training had yet to subside and thus 

influenced the results. The inclusion 

of longer-term retention intervals 

(eg, 6-months or 1 year) in biofeedback 

study designs is essential to determining 

whether motor learning 

has occurred as a result of the 

biofeedback. 

In all of the studies assessing gait 

changes after kinematic biofeedback 

in participants after stroke, gait 

speed was used as the primary outcome 

measure. Schmid et al 30 reported 

that significant gains in gait 

speed were associated with meaningful 

improvements in function and 

quality of life. However, gait speed 

alone does not reveal changes in coordination 

in walking or reflect motor 

learning. Gait outcome measures 

such as the Tinetti Performance- 

Oriented Mobility Assessment 31 and 

the Gait Assessment Rating Scale 32 

can provide some additional information 

regarding changes in movement 

patterns on the basis of a clinical 

evaluation. Recent advances in 

technology also have made it possible 

for researchers and clinicians to 




collect temporospatial values more 

easily by using tools such as the 

GAITRite System* or the GaitMat II. † 

A 3-dimensional analysis also can 

provide researchers and clinicians 

with a detailed kinematic and kinetic 

gait analysis. However, this type of 

analysis requires more expensive 

equipment and advanced training. 

Two of the 7 studies included functional 

outcome measures to assess 

improvements in function related to 

gait retraining. 9,16 In the study by 

Morris et al, 9 participants were assessed 

after stroke with the Motor 

Assessment Scale, 33 and White and 

Lifeso 16 assessed participants after 

total hip replacement by using the 

Harris Hip Score. 25 Functional outcome 

measures specific to a participant 

population may be helpful in 

indicating quality of life and thus 

positive changes in function associated 

with biofeedback training. 

Limitations of This 


The results of this systematic review 

are limited to studies that were written 

in the English language and included 

in the MEDLINE, CINAHL, 

and Cochrane databases or the reference 

lists of review articles identified 

in those databases. Other studies 

may exist outside these resources. 

Selection bias is another potential 

limitation of systematic reviews. 

However, we attempted to minimize 

this limitation by using the MINORS 

instrument, which was designed to 

assess the methodological quality of 

both randomized and nonrandomized 

studies. Although nonrandomized 

studies were not included in 

this review, we did not make a specific 

methodological decision to exclude 

them; they were indirectly 

excluded as a result of their low 

MINORS scores. Publication bias is 

another potential limitation of any 

* EQ Inc, PO Box 16, Chalfont, PA 18914. 

† CIR Systems Inc, 60 Garlor Dr, Havertown, 

PA 19083. 

type of literature review. Frequently, 

only studies demonstrating positive 

results are submitted to peerreviewed 

journals and published. 

This situation may bias results toward 

positive treatment effects. 

However, this limitation is impossible 

to quantify. 

Clinical Implications 

On the basis of the current literature, 

there are insufficient data to make a 

recommendation in the form of a 

clinical practice guideline. Despite 

this shortcoming, we recommend 

that clinicians using real-time 

biofeedback select outcome measures 

capable of revealing changes 

in coordination during walking. Additionally, 

clinicians should consider 

incorporating motor learning principles 

into their treatment protocols to 

provide clients with a practice environment 

that promotes motor learning. 

Clinicians should consider using 

faded feedback schedules along with 

random and variable practice to promote 

improved performance during 

retention testing. 34 Clinicians should 

also consider informing clients about 

the potential negative aspect of concurrent 

feedback to minimize its impact 

on learning. 28 

Future Directions 

Few studies have investigated the effectiveness 

of real-time kinematic, 

temporospatial, and kinetic biofeedback. 

Despite the small number of 

studies, this systematic review indicated 

that these methods of biofeedback 

were capable of providing moderate 

to large short-term treatment 

effects. Therefore, future research is 

warranted to provide clarity regarding 

the potential long-term benefits 

of real-time biofeedback for gait 

retraining. 

On the basis of the results of this 

systematic review, several recommendations 

for future research can 

be made. Future work should include 

adequate randomization to 

ensure that both experimental and 

control groups are equivalent at 

baseline, before biofeedback intervention. 

Researchers should incorporate 

existing motor learning concepts 

into study designs to minimize 

the negative effects of real-time feedback 

while maximizing the benefits. 

Future studies also should include 

multiday retention testing (eg, at 6 

months or 1 year) to assess the learning 

of gait changes. Evidence of permanent 

changes (ie, motor learning) 

is urgently needed to support the 

continued use of biofeedback in rehabilitation 

settings. To obtain such 

evidence, researchers should include 

outcome measures that provide 

more information about coordination 

during gait and overall 

walking function. 

Conclusions 

Real-time kinematic, temporospatial, 

and kinetic biofeedback appears to 

result in short-term moderate to 

large treatment effects. However, it 

is unknown whether treatment 

changes are maintained. Several 

studies lacked adequate randomization, 

a fact that should caution readers 

when interpreting the authors’ 

conclusions. Future studies should 

ensure adequate randomization of 

participants as well as the implementation 

of motor learning concepts 

and the inclusion of retention testing 

to assess the long-term success of 

biofeedback. They also should include 

outcome measures capable of 

demonstrating coordinative changes 

in gait and improvements in function. 

Both authors provided concept/idea/research 

design, writing, and data analysis. Mr 

Tate provided data collection. 

This article was submitted September 10, 

2008, and was accepted April 27, 2010. 

DOI: 10.2522/ptj.20080281 




References 

1 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;56:231–236. 

2 Colborne GR, Olney SJ, Griffin MP. Feedback 

of ankle joint angle and soleus electromyography 

in the rehabilitation of 

hemiplegic gait. Arch Phys Med Rehabil. 

1993;74:1100–1106. 

3 Intiso D, Santilli V, Grasso MG, et al. Rehabilitation 

of walking with electromyographic 

biofeedback in foot-drop after 

stroke. Stroke. 1994;25:1189–1192. 

4 Binder SA, Moll CB, Wolf SL. Evaluation of 

electromyographic biofeedback as an adjunct 

to therapeutic exercise in treating 

the lower extremities of hemiplegic patients. 

Phys Ther. 1981;61:886–893. 

5 Burnside IG, Tobias HS, Bursill D. Electromyographic 

feedback in the remobilization 

of stroke patients: a controlled trial. 

Arch Phys Med Rehabil. 1982;63:217–222. 

6 Aruin AS, Hanke TA, Sharma A. Base of 

support feedback in gait rehabilitation. Int 

J Rehabil Res. 2003;26:309–312. 

7 Montoya R, Dupui P, Pages B, Bessou P. 

Step-length biofeedback device for walk 

rehabilitation. Med Biol Eng Comput. 

1994;32:416–420. 

8 Ceceli E, Dursun E, Cacki A. Comparison 

of joint-position biofeedback and conventional 

therapy methods in genu recurvatum 

after stroke: 6 months’ follow-up. Eur 

J Phys Med Rehabil. 1996;6:141–144. 

9 Morris ME, Matyas TA, Bach TM, Goldie 

PA. Electrogoniometric feedback: its effect 

on genu recurvatum in stroke. Arch 

Phys Med Rehabil. 1992;73:1147–1154. 

10 Cozean CD, Pease WS, Hubbell SL. 

Biofeedback and functional electric stimulation 

in stroke rehabilitation. Arch Phys 

Med Rehabil. 1988;69:401–405. 

11 Mandel AR, Nymark JR, Balmer SJ, et al. 

Electromyographic versus rhythmic positional 

biofeedback in computerized gait 

retraining with stroke patients. Arch Phys 

Med Rehabil. 1990;71:649–654. 

12 Bradley L, Hart BB, Mandana S, et al. Electromyographic 

biofeedback for gait training 

after stroke. Clin Rehabil. 1998;12: 

11–22. 

13 Colborne GR, Wright FV, Naumann S. 

Feedback of triceps surae EMG in gait of 

children with cerebral palsy: a controlled 

study. Arch Phys Med Rehabil. 1994;75: 

40–45. 

14 Isakov E. Gait rehabilitation: a new 

biofeedback device for monitoring and enhancing 

weight-bearing over the affected 

lower limb. Eura Medicophys. 2007;43: 

21–26. 

15 Petrofsky JS. The use of electromyogram 

biofeedback to reduce Trendelenburg 

gait. Eur J Appl Physiol. 2001;85:491–495. 

16 White SC, Lifeso RM. Altering asymmetric 

limb loading after hip arthroplasty using 

real-time dynamic feedback when walking. 

Arch Phys Med Rehabil. 2005;86: 

1958–1963. 

17 Basmajian JV. Introduction: principles and 

background. In: Basmajian JV, ed. Biofeedback: 

Principles and Practice for Clinicians. 

3rd ed. Baltimore, MD: Williams & 

Wilkins; 1989:1–4. 

18 Bottomley JM. Biofeedback: connecting 

the mind and body. In: Davis CM, ed. Complementary 

Therapies in Rehabilitation. 

2nd ed. Thorofare, NJ: Slack Inc; 2003: 

131–156. 

19 Davis IS. Gait retraining in runners. Orthopedic 

Physical Therapy. 2005;17:8–13. 

20 Woodford H, Price C. EMG biofeedback 

for the recovery of motor function after 

stroke. Cochrane Database Syst Rev. 

2007;2:CD004585. 

21 Moreland JD, Thomson MA, Fuoco AR. 

Electromyographic biofeedback to improve 

lower extremity function after 

stroke: a meta-analysis. Arch Phys Med Rehabil. 

1998;79:134–140. 

22 Slim K, Nini E, Forestier D, et al. Methodological 

Index for Non-randomized Studies 

(MINORS): development and validation of 

a new instrument. ANZ J Surg. 2003;73: 

712–716. 

23 Cohen J. Statistical Power Analysis for 

the Behavioral Sciences. 2nd ed. Hillsdale, 

NJ: Lawrence Erlbaum Associates; 1988. 

24 Woods JR, Williams JG, Tavel M. The 

two-period crossover design in medical 

research. Ann Intern Med. 1989;110: 

560–566. 

25 Harris WH. Traumatic arthritis of the hip 

after dislocation and acetabular fractures: 

treatment by mold arthroplasty—an endresult 

study using a new method of result 

evaluation. J Bone Joint Surg Am. 1969; 

51:737–755. 

26 Salmoni AW, Schmidt RA, Walter CB. 

Knowledge of results and motor learning: 

a review and critical reappraisal. Psychol 

Bull. 1984;95:355–386. 

27 Winstein CJ, Pohl PS, Cardinale C, et al. 

Learning a partial-weight-bearing skill: effectiveness 

of two forms of feedback. Phys 

Ther. 1996;76:985–993. 

28 Swinnen SP. Information feedback for motor 

skill learning: a review. In: Zelaznik 

HN, ed. Advances in Motor Learning and 

Control. Champaign, IL: Human Kinetics; 

1996:37–65. 

29 Janelle CM, Kim J, Singer RN. Subjectcontrolled 

performance feedback and 

learning of a closed motor skill. Percept 

Mot Skills. 1995;81:627–634. 

30 Schmid A, Duncan PW, Studenski S, et al. 

Improvements in speed-based gait classifications 

are meaningful. Stroke. 2007;38: 

2096–2100. 

31 Tinetti ME. Performance-oriented assessment 

of mobility problems in elderly patients. 

J Am Geriatr Soc. 1986;34:119–126. 

32 Wolfson L, Whipple R, Amerman P, Tobin 

JN. Gait assessment in the elderly: a gait 

abnormality rating scale and its relation to 

falls. J Gerontol. 1990;45:M12–M19. 

33 Carr JH, Shepherd RB, Nordholm L, Lynne 

D. Investigation of a new motor assessment 

scale for stroke patients. Phys Ther. 

1985;65:175–180. 

34 Rose DJ, Christina RW. Augmented Feedback 

and Motor Learning: A Multilevel 

Approach to the Study of Motor Control 

and Learning. 2nd ed. San Francisco, CA: 

Benjamin Cummings; 2006:291–320. 



Real-Time Kinematic, Temporospatial, and Kinetic 

Biofeedback During Gait Retraining in Patients: A 


Jeremiah J. Tate and Clare E. Milner 

PHYS THER. Published online June 17, 2010 

doi: 10.2522/ptj.20080281 

Subscription 

Information 

http://ptjournal.apta.org/subscriptions/ 

Permissions and Reprints http://ptjournal.apta.org/site/misc/terms.xhtml 

Information for Authors 

http://ptjournal.apta.org/site/misc/ifora.xhtml

Real-Time Kinematic, Temporospatial, and Kinetic Biofeedback ...

Create successful ePaper yourself

Delete template?

Save as template?