Electromyographic Biofeedback Applications to the
Transcript of Electromyographic Biofeedback Applications to the
Electromyographic Biofeedback Applications to the Hemiplegic Patient
Changes in Upper Extremity Neuromuscular and Functional Status
STEVEN L. WOLF and STUART A. BINDER-MACLEOD
The effect of a specific EMG biofeedback treatment protocol on quantified changes in neuromuscular measures and functional activities was examined among the upper extremities of 22 chronic stroke patients who each received 60 feedback training sessions. These data were compared with changes measured from a Control Group of 9 (no treatment) patients. Those patients receiving feedback training showed significant improvements in numerous neuromuscular measures but not in functional measures. When the Experimental Group was subdivided into two groups (hand, n = 5; no hand, n = 17) on the basis of acquiring a specific hand function, significant pretreatment differences in neuromuscular status emerged. Based upon these pretreatment differences and outcome measures, characteristics possibly predictive of beneficial outcomes from EMG biofeedback training were exposed. Chronic stroke patients who gained maximal functional benefits from the biofeedback intervention initially had greater active range of motion at all major upper extremity joints and comparatively less hyperactivity within typically "spastic" muscles. Electromyographic biofeedback can lead to substantial improvements among select chronic stroke patients and can be of considerable functional benefit to others.
Key Words: Arm, Biofeedback, Cerebrovascular disorders, Hand, Hemiplegic, Rehabilitation.
A major obstacle that often plagues the best of rehabilitation efforts is restitution of hand function following cerebrovascular accident. Although meaningful upper extremity function clearly hinges on restoration of finger and thumb movements, these activities are frequently limited or nonexistent within the chronic stroke population. We have conducted an informal survey among over one thousand physical therapists throughout the United States and Canada
and have found that only one individual was able to report one case in which a chronic (greater than one year after insult) stroke patient achieved finger extension and thumb abduction-extension movements following treatment when those movements were not previously present. All other manual task achievements were attributed to use of tenodesis mechanisms (unpublished observations).
Concomitantly, we have been impressed and, to some extent, guided by the pioneering work of Brudny and co-workers who have reported extensively on the benefits of EMG biofeedback for the restitution of upper extremity function among stroke patients.1-4 Indeed, well-controlled clinical studies have demonstrated the utility of biofeedback applications to the hemiplegic shoulder5, 6 and wrist,7-9 but clearly most publications have not adequately addressed the relationship between upper extremity functional changes and biofeedback interventions or the issue of predictors of successful applications of biofeedback to the upper extremities of chronic stroke patients.
Dr. Wolf is a senior investigator at the Emory University Rehabilitation Research and Training Center, 1441 Clifton Rd NE, Atlanta, GA 30322 (USA) and Associate Professor, Department of Rehabilitation Medicine, and Assistant Professor, Department of Anatomy, Surgery and Community Health Sciences, Emory University School of Medicine, Atlanta, GA.
Mr. Binder-Macleod was a research associate, Emory University Regional Rehabilitation Research and Training Center when this paper was written, and is now a doctoral student. Department of Physiology, Medical College of Virginia, Virginia Commonwealth University, Richmond, VA 23298.
This work was supported by Grant No. G008003042 awarded to Emory University from the Institute of Handicapped Research, Department of Education, Washington, D.C.
This article was submitted February 23, 1983; was with the author for revision two weeks; and was accepted April 13, 1983.
Volume 63 / Number 9, September, 1983 1393
TABLE 1 Upper Extremity Functional Grades for Biofeedback
Intervention Grade 0 1 + 2 +
3 + 4 +
Definition no change following treatment reduced spasticity or increased strength upper extremity used as an assist but no pre
hension crude prehension, some release skilled prehension with rapid acquisition and
release
With respect to the former, several studies have adapted the functional rating scale described by Brudny and colleagues4 and presented in modified form as Table 1. This scoring system provides little concrete information regarding functional capabilities in activities of daily living. Even less discussion has been devoted to identifying predictors of successful biofeedback interventions among chronic stroke patients. Accumulating evidence seems to indicate that proprioceptive loss, presence of expressive aphasia, age, sex, side of lesion, and duration of previous rehabilitation are not correlated to the patient's ability to interact with the modality or with outcome.10, 11 Specific correlations between changes in neuromuscular measures and function have rarely appeared, however. In this regard, we were particularly impressed by the observation of Gianutsos and co-workers that reacquisition of finger extension is crucial if the chronic stroke patient is to obtain functional prehension using biofeedback treatment.12 This notion is in accord with a follow-up analysis of our own work on EMG biofeedback interventions among a stroke population.13
The present study was undertaken to determine whether EMG biofeedback can produce functional changes among chronic stroke patients compared with a control group. Unlike past studies, quantified measures of function governing shoulder, elbow, wrist, and hand manipulative tasks were used. We further sought to determine predictors that would help to identify those chronic stroke patients for whom application of this modality might prove meaningful.
METHODS
Patients
Thirty-one stroke patients who had 1) sustained cerebrovascular accident at least one year before this study, 2) no previous EMG biofeedback training, and 3) no evidence of receptive aphasia, served as outpatient research subjects for this study. Twenty-two received the experimental protocol, and nine were placed in a Control Group. Basic information on these patients appears in Table 2. The control patients
eventually received the experimental application; however, their data are not reported. The two groups did not differ significantly with respect to age, duration of previous rehabilitation, time since stroke, presence of proprioceptive loss, or existence of expressive aphasia.
Experimental Group
These patients received 60 EMG biofeedback treatments to the involved upper extremity. Treatment sessions were divided into blocks of 20 with neuromuscular and functional examinations provided before the initial treatment (baseline) and after 20, 40, and 60 sessions. The entire treatment block took approximately six months (range: five to nine months) for each patient. Usually two or three treatment sessions were provided each week.
Control Group
Members of this group underwent two examinations that were identical to those received by the experimental patients. Examinations were separated by approximately four months (range: from three and a half to eight months) and were designed to examine the effects of time and repeated testing. Like the Experimental Group, none of the Control Group were receiving any form of concurrent physical rehabilitation during participation in this study.
Treatment
Feedback training was provided through use of the Hyperion 4080* or the Cyborg J53†. Only EMG biofeedback was provided to experimental patients. All other forms of neuromuscular reeducation were excluded. Treatment followed the protocol described by Kelly and colleagues.14 Briefly, patients were trained in 45 to 60 minute sessions to relax specific hyperactive muscles (upper trapezius, pectoralis major, biceps brachii, wrist and finger flexor mass, and thenar eminence) through the provision of audio and visual feedback detected with surface electrodes. This procedure was followed by attempts to recruit weakened antagonist muscles. Often as antagonist muscles were recruited, patients received feedback from spastic muscles to reinforce previous training directed toward their inhibition. Biofeedback training proceeded in a proximal to distal direction with each patient progressing from isolated joint movements to manipulative efforts requiring voluntary stabilization of proximal musculature. Rarely were these manipu-
* Hyperion 4080 Bioconditioner, Hyperion, Inc, 9579, SW, 168th St, Miami, FL, 33157.
† Cyborg J53, Cyborg Corp, 342 Western Ave, Boston, MA 02135.
1394 PHYSICAL THERAPY
RESEARCH
TABLE 2 Basic Data On All Patients
Group
Experimental (n = 22)
Control (n = 9)
Age ( ± s)
55.3 ± 14.8
55.9 ± 10.9
Men
10
7
Sex
Women
12
2
Duration of Stroke (Yrs)
( ± s)
3.8 ± 3.6
2.4 ± 1.6
Duration of Previous
Rehab (Yr) ( ± s)
1.1 ± 0.98
2.0 ± 1.10
TABLE 3 Baseline Differences Between Experimental and Control Groups
Variablea
ETREX (µV)
FFRTF (sec)
TABRLF (µV)
TEXSS (µV)
CUE4 (sec)
Group
experimental control
experimental control
experimental control
experimental control
experimental control
± s
244.1 ± 124.8 138.1 ± 147.0
29.8 ± 39.5 3.7 ± 3.2 3.0 ± 1.5 6.3 ± 6.3
14.5 ± 15.3 6.0 ± 5.1
1.02 ± 0.1 1.83 ± 1.3
F
1.39
141.65
17.52
9.12
657.70
P (2-tail)
.520
.000
.000
.003
.000
t
- 2 . 0 4
- 2 . 8 5
-2 .21
- 2 . 1 7
- 2 . 1 7
P (2-tail)
.05 (p)b
.01 (s)c
.04 (s)
.04 (s)
.05 (s)
lative efforts identical to assessment tasks. No attempt was made to establish criteria that would necessitate training a specific limb segment within a defined number of treatment sessions. Patients were randomly assigned to one of four physical therapists.
Measurements
Each examination for any one patient was performed by the same therapist from the group of four therapists. The therapist did not treat the patient. As previously reported, examinations included quantified measurements of active and passive upper extremity range of motion, peak EMG during isotonic contractions, time to return a passively lengthened hyperactive muscle or muscle group to the pre-stretched EMG level, and function of the upper extremity.15 Each examination lasted approximately eight hours and was performed over two consecutive patient visits. All measurements were repeated three times. The mean of the three responses was used for subsequent analysis. A rest interval of approximately one minute separated consecutive measurements.
For EMG data, electrode leads were stabilized to minimize movement artifact, and all raw EMG was
examined online to detect and eliminate spurious recordings. Disposable surface electrodes were placed at specific anatomical locations, and skin-electrode impedance was reduced below 15 kΩ. Peak EMG was recorded from the meter of a BFT 401C feedback device.:}: The EMG measurements were made during isotonic contractions of upper trapezius, anterior deltoid, middle deltoid, biceps brachii and triceps brachii muscles and forearm extensor and flexor muscle masses. The forearm extensor and flexor muscle masses were monitored during wrist extension and flexion, respectively, and during attempts at finger extension and flexion with the wrist manually stabilized in flexion and in a neutral position. Electromyographic recordings were also made from the thenar eminence during voluntary efforts at thumb abduction and flexion. The peak EMG was also recorded during slow (approximately 20°/sec) and fast (approximately 200°/sec) stretch of biceps brachii, wrist flexor and finger flexor (wrist stabilized in a neutral position), wrist extensor, and thumb adductor muscles. While the muscle group was held manually in
a Abbreviations: ETREX: peak EMG from triceps brachii muscle—elbow extension. FFRTF: time to return EMG from finger flexor muscles to baseline following fast stretch. TABRLF: EMG resting level of thenar muscles before fast stretch into abduction. TEXSS: peak EMG from thenar muscles when thumb stretched slowly into extension. CUE4: time to abduct shoulder to 90 degrees. b Pooled variance estimate. c Separate variance estimate.
‡ Biofeedback Technology, Inc, 10612A Trask Ave, Garden Grove, CA 92563.
Volume 63 / Number 9, September, 1983 1395
TABLE 4 Comparison of Experimental and Control Groups for Changes in Variables Between First and Last Examinations
Variablea
EBIFS (µV)
ETRRLEX (µV)
ETREX (µV)
ETRRTEX (sec)
WFRTF (sec)
WEFL (µV)
FFRTF (sec)
TABRLF (µV)
AUE2 (sec)
Group
experimental control
experimental control
experimental control
experimental control
experimental control
experimental control
experimental control
experimental control
experimental control
± s
9.1 ± 17.3 - 8 . 0 ± 28.0 - 4 . 2 ± 13.0
7.0 ± 12.1 - 5 . 2 ± 103.3 71.7 ± 74.0
- 2 1 . 4 ± 26.0 7.8 ± 38.3
- 2 9 . 0 ± 33.3 13.1 ± 31.2
- 1 4 . 9 ± 43.9 24.8 ± 45.0
- 1 5 . 6 ± 29.5 6.7 ± 12.6
- 3 . 2 ± 6.0 2.5 ± 6.3
- 1 . 0 ± 2.5 1.5 ± 0.7
F
2.62
1.15
1.95
2.18
1.14
1.05
5.46
1.10
14.46
P (2-tail)
.07
.88
.33
.16
.89
.80
.02
.80
.05
t
- 2 . 0 8
2.23
2.02
2.38
3.19
2.22
2.80
2.32
2.61
P (2-tail)
.05 (p)b
.03 (p)
.05 (p)
.02 (p)
.004 (p)
.04 (p)
.01 (s)c
.03 (p)
.03 (s)
the lengthened position, the time required for EMG to return to baseline was recorded.
Functional measures, based on the time to complete a task or the amount of resistive weight moved during a task, were recorded and appear as an Appendix.
Statistical Analyses
Among the 31 research patients, measurements on 167 variables at each examination gave a total of over 60,000 data points. Computations were made for mean group changes between results of initial and final examinations. For continuous data, t tests (2-tail) were used to determine if the means between the two groups were significantly different. The homogeneity of the sample variances for each biofeedback and each group comparison were tested using an F statistic. If no significant difference (p > .05) between sample variance was found, a pooled variance estimate was used to determine the t statistic. If a significant difference in sample variance (p < .05) between the Experimental and Control Groups was found for any measurement, the t statistic was determined using separate variance estimates.
When patients were either unable initially to complete a functional task or in timed situations, could
not finish the task within the upper limit of 120 sec, their data were not included in the above parametric testing. To include these patients' data, discrete data analysis was undertaken using a Fisher's exact test (2 x 2 table) to analyze the statistical significance of the difference in the number of patients who showed improvement between the Control and Experimental Groups. This type of analysis, compared to the t test, does not require patient capability to perform the task initially.
In addition, the Experimental Group was divided into two subgroups (see Results). These subgroups were analyzed using similar parametric and nonpara-metric tests for continuous and discrete data, respectively. For the control and experimental subgroups, paired t tests to compare within-group changes between baseline and final measurements were performed.
RESULTS
Because Experimental and Control Groups did not differ among the variables identified in Table 2, t tests (2-tail) were performed on all baseline data. Significantly different group mean baseline values were obtained for only five variables, identified in the
a Abbreviations: EBIFS: peak EMG biceps brachii muscle at fast stretch. ETRRLEX: resting level EMG triceps brachii muscle before isotonic contraction. ETREX: peak EMG of triceps brachii muscle during isotonic contraction. ETRRTEX: time to return triceps brachii muscle EMG to resting level. WFRTF: time to relax wrist flexor muscles after fast stretch. WEFL: peak EMG from wrist extensor muscles during flexion. FFRTF: time to relax finger flexor muscles following fast stretch. TABRLF: resting level EMG thenar muscles before fast stretch into thumb abduction. AUE2: time to make circle with tip of olecranon—facing table. b pooled variance estimate. c separate variance estimate.
1 3 9 6 PHYSICAL THERAPY
RESEARCH
TABLE 5 Within Control Group Comparison of Neuromuscular Changes Between First and Last Examinations (n = 9)
Variable
SABROMA (°)
ETREX (µV)
Description
active shoulder ROM, abduction
peak EMG triceps brachii muscles during elbow extension
Difference ( ± s)
15.7 ± 17.1
71.7 ± 74.0
t
2.75
2.91
df
00 00
p (2-tail)
.025
.020
TABLE 6 Within Experimental Group Comparison of Neuromuscular Changes Between First and Last Examinations (n = 22)
Variable
SABROMA (°) SFLEX (µV) EBIRLFS (µV) EBIFS (/xV) EBIRLFL (µV) EBIRTFL (sec) EBIRTEX (sec) ETRRLSS (µV) ETRRLFS (µV) ETRRTEX (sec) WFRTF (sec) WFROMA (°) WEROMP (°) WERTFL (sec) FFRTS (sec) FFRTF (sec) FFFRTE (sec) TABRLS (µV) TABRLF (µV) TEXRLS (µV) TEXRLF (µV)
Description
active shoulder ROM, abd. peak EMG, shoulder flex. biceps EMG before fast stretch biceps peak EMG, fast stretch biceps EMG before elbow flex. relax time biceps after elbow flex. relax time biceps after elbow ext. triceps EMG before slow stretch triceps EMG before fast stretch relax time triceps after elbow ext. relax time wrist flexors after fast stretch active wrist flex. ROM passive wrist flex. ROM relax time, wrist ext. after wrist flex. relax time finger flex. after slow stretch relax time finger flex. after fast stretch relax time finger flex. after finger ext. thumb EMG before slow stretch to abd. thumb EMG before fast stretch to abd. thumb EMG before slow stretch to ext. thumb EMG before fast stretch to ext.
Difference ( ± s)
13.8 ± 18.4 42.4 ± 81.8 - 2 . 9 ± 4.9
9.1 ± 17.3 - 3 . 0 ± 6.8
- 1 4 . 9 ± 21.8 - 1 2 . 0 ± 17.7
- 1 . 5 ± 2.8 - 1 . 3 ± 2.5
- 2 1 . 4 ± 25.9 - 2 9 . 0 ± 33.3
6.8 ± 13.8 14.2 ± 25.0
- 5 . 0 ± 8.7 - 1 1 . 3 ± 22.8 - 1 5 . 6 ± 29.5
- 2 . 8 ± 5.5 - 3 . 8 ± 7.3 - 3 . 2 ± 6.0 - 6 . 1 ± 11.5 - 8 . 8 ± 19.5
t
3.51 2.43
- 2 . 6 9 2.47
- 2 . 0 8 - 2 . 9 9 - 2 . 9 5 - 2 . 4 9 - 2 . 5 0 - 3 . 5 9 - 3 . 8 0
2.20 2.67
- 2 . 5 0 - 2 . 1 6 -2 .31 - 2 . 2 4 - 2 . 3 6 - 2 . 3 7 - 2 . 3 6 - 2 . 1 2
dfa
21 21 21 21 21 18 18 21 21 18 18 19 21 18 18 18 18 19 19 19 21
P (2-tail)
.002
.020
.014
.022
.050
.008
.009
.021
.021
.002
.001
.040
.014
.022
.044
.033
.038
.029
.029
.029
.046
left-hand column of Table 3. Initially, the Experimental Group showed significantly 1) greater mean peak EMG of triceps brachii muscle during elbow extension (ETREX), 2) longer times to relax finger flexor muscles following a fast stretch (FFRTF), 3) lower resting levels of EMG from thenar musculature before fast stretch of the thumb into abduction (TABRLF), 4) greater mean peak EMG responses from thenar muscles during slow stretch into thumb extension (TEXSS), and 5) shorter times to abduct the shoulder to 90 degrees (CUE4). Because only 5 of 167 variables revealed significantly different mean baseline measures between the two groups, we proceeded to test for differences between the groups following intervention.
The significant differences between the groups after treatment appear in Table 4. The Experimental Group decreased ETREX (p = .05, pooled variance estimate) but did have a significantly reduced mean resting level of EMG from triceps brachii muscle before extension (ETRRLEX, p = .03, pooled variance estimate) and could relax the triceps more
quickly following elbow extension (ETRRTEX, p = .02, pooled variance estimate). On the other hand, the Experimental Group also showed a significant increase in mean biceps peak EMG of biceps brachii with fast stretch (EBIFS, p = .05, pooled variance estimate).
Experimental patients were able to relax the forearm flexor muscle group more quickly when stretch of the wrist (WFRTF, p = .004, pooled variance estimate) or fingers into extension (FFRTF, p = .01, separate variance estimate) was quickly performed. From a functional perspective, the only task to change significantly was time required for the tip of the olecranon to trace a circle drawn on a table (AUE2, p = .03, separate variance estimate); yet compared with the Control Group, concomitant neuromuscular changes about the shoulder joint to account for changes in this task did not occur.
Within-group analyses were done for both groups by comparing mean neuromuscular changes between the first and final assessments. Table 5 depicts the only two variables for which the Control Group
a Nineteen or 18 df caused by missing data points among two or three patients, respectively.
Volume 63 / Number 9, September, 1983 1397
TABLE 7 Basic Data on Experimental Patients Grouped by Initial or Acquired Hand Function versus No Hand Function
Group
Experimental—hand (n = 5) Experimental—no hand (n = 17)
Age ( ± s)
54.0 ± 18.1
55.6 ± 13.8
Men
4
6
Sex
Women
1
11
Duration of Stroke (yr)
( ± s)
3.4 ± 3.8
4.4 ± 3.5
Duration of Previous Rehab
(yr) ( ± s)
4.8 ± 9.1
1.1 ± 1.0
TABLE 8 Pretreatment Differences Between Experimental Patients Who Had Or Gained Hand Function Versus Those Who Did
Not Gain Hand Function—Shoulder and Elbow
Task Involving:
Shoulder
Elbow
Variable
SABROMA SFLROMA EBIRTS EBIRLFL
EBIF
EBIRTFL EFLROMA ETRRLSS
ETRSS
ETRRLFS
ETRRLEX
EEXROMA ETRRLFL
Description
active ROM, shoulder abduction active ROM, shoulder flexion time to relax, biceps after slow stretch mean EMG biceps before elbow flex
ion mean peak EMG biceps during elbow
flexion time to relax biceps after elbow flexion active ROM elbow flexion resting level EMG triceps before slow
stretch mean peak EMG triceps for slow
stretch resting level EMG triceps before fast
stretch resting level EMG triceps before elbow
extension active ROM, elbow extension resting level EMG triceps before elbow
flexion
P (2-tail)
.04 (p)a
.02 (p)
.04 (p)
.05 (s)b
.02 (p)
.01 (s)
.02 (s)
.0001 (s)
.02 (s)
.0001 (s)
.02 (s)
NS(s) .05 (s)
showed statistically significant changes (paired t test, 2-tail). The Experimental Group underwent significant changes among 21 neuromuscular measurements (Tab. 6, paired t test, 2-tail). All of these changes were harmonious with training strategies and indicative of 1) increased shoulder range of motion (ROM) into abduction with more anterior deltoid muscle activity during voluntary shoulder flexion, 2) reduced biceps brachii muscle hyperactivity with more "control" over triceps brachii muscles, 3) increased ROM about the wrist with reduced hyperactivity within wrist and finger flexor muscle masses, and 4) more relaxed thenar muscles.
Experimental patients also improved significantly in four functional tasks, all of which governed shoulder movements. These tasks included AUE1 (t = 2.69; df = 13; p = .019), AUE2 (t = 2.86, df=13, p = .013), EUE1 (t = -2.41, df = 12, p = .033), and EUE9 (t = -2.65, df=11, p = .023). No significant functional changes occurred within the Control
Group. Therefore, a total of only 25 of 167 measured variables were significantly altered after biofeedback training.
Ostensibly, despite an ability to relax wrist and finger flexor muscle groups better, the biofeedback training strategy used in this study did little to change neuromuscular measures when compared in an inter-group analysis with the Control Group. Within the Experimental Group, however, the number of significant neuromuscular improvements increased tenfold compared with the comparable Control Group analysis.
To delineate whether certain neuromuscular measures taken before treatment might be predictive of favorable outcomes with biofeedback training, the Experimental Group was divided into two subgroups based on attaining enough hand and arm function to stack checkers (task EUE7, Appendix). Those patients (n = 5) who possessed (n = 2) or gained (n = 3) this function are designated "Experimental-hand"
a Pooled variance estimate. b Separate variance estimate.
1398 PHYSICAL THERAPY
RESEARCH
TABLE 9 Pretreatment Differences Between Experimental Patients Who Had or Gained Hand Function Versus Those Who Did
Not Gain Hand Function—Wrist and Fingers
Task Involving:
Wrist
Fingers
Thumb
Variable
WFROMA WERTE
WEFL
FFRLSS
FFRTF
FFRLFL
FFRTFL
FFRLEX
FFFRLEX
FERTE
FENROMA
FEFROMA
TABSS
TEXFS
TEXRLAB
TFLRLFL
TFLFL
Description
active ROM, wrist flexion time to relax wrist extensors after wrist
extension mean peak EMG wrist extensors for
wrist flexion resting EMG level forearm flexor mass
before slow stretch of finger flexors time to relax forearm flexor mass after
finger flexion with wrist flexed resting EMG level of forearm flexor
before finger flexion time to relax forearm flexor mass after
finger flexion resting EMG level forearm flexor mass
before finger extension resting EMG level forearm flexor
mass, fingers in flexed position before finger extension
time to relax forearm extensor mass after finger extension
active ROM, finger extension with wrist neutral
active ROM, finger extension with wrist flexed
mean peak EMG thenar muscles slow stretch into thumb abduction
mean peak EMG thenar muscles fast stretch into thumb extension
resting level EMG thenar muscles before thumb abduction
resting level EMG thenar muscles before thumb flexion
mean peak EMG thenar muscles, thumb flexion
Variance Estimate (2-tail probability)
.02 (p)a
.03 (s)b
.01 (p)
.03 (s)
.02 (s)
.01 (s)
.05 (s)
.004 (s)
.04 (s)
.05 (s)
.001 (p)
.02 (s)
.04 (s)
.02 (s)
.01 (s)
.003 (s)
in Table 7, and those patients (n = 17) who did not possess or acquire this capability are denoted as "Experimental-no hand."
A t test showed that these subgroups did not differ significantly for the variables identified in Table 7. Mean differences between these experimental subgroups for neuromuscular measures before feedback treatment were examined for all upper extremity neuromuscular measures. Tables 8 and 9 show those variables for which significant differences were found at pretreatment evaluation of the shoulder, elbow, wrist, and fingers. For each variable, the Experimental-hand subgroup performed better; that is, this group was able to generate more EMG appropriately and show a lower level of resting EMG, a faster time to reduce EMG following stretch, or a greater active ROM. Notably, no member of the Experimental-no hand subgroup could generate any active ROM into finger extension with the wrist in neutral position (FENROMA), and the one member of this subgroup
capable of producing active finger extension with the wrist flexed (FEFROMA) showed no change in hand function after completion of feedback training. In light of the significant pretreatment mean differences between the experimental subgroups among thirty variables, a posttreatment comparative analysis was not attempted.
Changes in functional status among subjects in these subgroups were compared between the Control Group and the total Experimental Group and between the Control Group and each experimental subgroup and are reported in Tables 10-12. Improvement in functional tasks (Appendix) was operationally defined as an ability to complete a task at final examination that could not be completed within a specific time limit at initial evaluation or as a demonstrated reduction in the time to complete a task from initial to final examination. One task, BUE3, dealt with changes in the amount of weight (resistance) that could be moved with elbow extension.
a Pooled variance estimate. b Separate variance estimate.
Volume 63 / Number 9, September, 1983 1399
TABLE 10 Fisher's Exact Test (2x2 Table) for Functional
Improvement: Experimental Versus Control Patients— Shoulder
Task
AUE2b I Nl
Comparison Groupsa
Et
13C
8
Enh
10d
6
C
1 8
TABLE 11 Fisher's Exact Test (2x2 Table) for Functional
Improvement: Experimental Versus Control Patients— Elbow
Taskb
BUE1
BUE3
BUE4
I Nl I Nl I Nl
Et
11 c
3 11
6 10d
9
Comparison Groupsa
E n h
10d
3
8d
7
Eh
3 d
0 2 2
C
1 7 1 6 0 9
Within Tables 10-12, the number of patients reported for each group or subgroup does not represent the total for that component because any patient who could initially perform any task in less than 1.5 sec was omitted from the analyses. Thus, for example, in Table 10, while the total Experimental Group (ET) was 22 for task AUE 2, discrete data (I, Nl) are provided for only 21 patients because one patient could initially complete AUE 1 in less than 1.5 seconds. Inspection of these tables reveals that one or
TABLE 12 Fisher's Exact Test (2x2 Table) for Functional
Improvement: Experimental Versus Control Patients— Hand
Taskb
EUE3
EUE4
EUE5
EUE6A
I Nl I Nl I Nl I Nl
Comparison
Eh
4c
1 4c
1 3 d
2 4c
1
Groupsa
C
0 9 0 9 0 9 0 9
more experimental subgroups were able to improve to an extent approximating or achieving statistical significance compared with the Control Group. None of the nine control patients was able to improve in hand tasks (Tab. 12), despite the fact that two of these patients had voluntary wrist and finger extension initially.
DISCUSSION
This biofeedback study is the first to use a true control (no treatment) chronic stroke group that was not receiving concomitant treatment. Baseline measurements (Tab. 3) comparing the groups indicated that although the experimental patients had significantly more activity in the triceps brachii muscle, they also had greater hyperactivity in the finger flexor and thumb muscles—major deterents to upper extremity functional reacquisition. Following treatment, our Experimental Group showed less triceps activity but had an increased biceps brachii muscle response to stretch (Tab. 4) compared with the Control Group. The decreased triceps activity may be indicative of decreased cocontraction (biceps activity) about the joint, thus requiring decreased triceps activity to extend the elbow. Unfortunately, recording from the biceps during extension showed no statistical decrease (mean 71 µV, pretreatment; mean 59 µV, posttreat-ment). Thus, the persistent biceps activity during
a Abbreviations: Et: total experimental group. Enh: experimental-no hand function group. C: control group. I: improved. Nl: not improved. b Shoulder control—time to make circle with tip of
olecranon. c p < .02. d p < .05.
a Abbreviations: Et: total experimental group. Enh: experimental no hand function subgroup. Eh: experimental hand function subgroup. C: control group. I: improved. Nl: not improved b Tasks: BUE1 —Elbow extension—time to achieve full elbow
extension with arm resting on table at 90 degrees elbow flexion.
BUE3—Elbow extension—amount of weight around wrist that can be applied during full elbow extension from 90 degrees of elbow flexion.
BUE4—Elbow extension—time to retrieve 0.5 kg weight without lifting elbow off table.
c p < .05. d p < .01.
a Abbreviations: Eh: experimental hand function subgroup. C: control group. I: improved. Nl: not improved. b Tasks:
EUE3—Functional pinch—time to lift pencil. EUE4—Functional grasp—time to grasp cup
around its side. EUE5—Refined pinch—time to lift paper clip. EUE6A—Pronation-supination—time to flip five 3 x
5 note cards; involved hand. c p < .01. d p < .05.
1400 PHYSICAL THERAPY
RESEARCH
elbow extension efforts may be indicative of continuing cocontraction about the elbow joint and, if so, parallel the observations made by Prevo and colleagues who used similar feedback training strategies.16 The significant increase in biceps activity during fast stretch may contribute to functional disadvantages.
Within-group analyses revealed that experimental patients were able to demonstrate significant improvement in neuromuscular measures for 21 variables (Tab. 6) compared to only 2 variables for which such improvement occurred within the Control Group (Tab. 5). Although these differences are superficially impressive, their value is obscure because 1) such changes are few compared with the 167 variables analyzed, 2) only one significant functional change was noted between the groups, and 3) only four significant functional changes occurred within the Experimental Group. These last changes all centered about the shoulder joint and, in part, are explained by significant increases in active shoulder abduction ROM and anterior deltoid EMG activity. Thus, as a statistical exercise, comparative changes are of interest, but from a clinical perspective, specific functional improvements appear minute.
For this reason, we felt compelled to scrutinize our Experimental Group more carefully. The ability to perform task EUE7 (stacking checkers) was selected to segregate experimental patients into "hand" and "no hand" subgroups because this maneuver requires prehensile ability with shoulder stabilization and elbow extension but without accompanying wrist and finger movements, (such as those in task EUE6 A in the Appendix). On the basis of this criterion, 30 variables appeared significantly different between the experimental subgroups before treatment (Tabs. 8 and 9). Following feedback training, the "hand" subgroup showed improvement in most functional tasks, which included use of the hand (Tab. 12).
Therefore, a tentative description of neuromuscular measures taken before feedback training emerges as a predictor for those chronic stroke patients who will benefit most from this therapeutic intervention. More specifically, these measurements may help to identify those patients with the greatest potential for improving hand function. These measures are defined in Tables 8 and 9 and can generally be summarized to include 1) demonstrable active ROM for the shoulder, elbow, wrist, and fingers; 2) ability to generate substantial biceps EMG but also to reduce biceps activity rapidly following passive stretch or shortening contraction; and 3) ability to relax thenar and finger flexor musculature during passive stretch. In short, these patients appear to have more active range of motion and less "spasticity."
Of all these measures, we sense that the most clinically significant is active finger extension. All five
Experimental-hand subgroup patients had this ability before treatment and improved it during biofeedback training. In contrast, only 1 of 17 Experimental-no hand patients possessed some active finger extension before treatment (1° of active index finger extension measured at the MP joint) with no changes resulting after 60 training sessions. Equally compelling for the consideration of including EMG feedback as part of a therapeutic protocol is the observation that the two control patients with active finger extension initially were unable to improve finger extension or to enhance any function in an examination after a four-month interval.
It has been argued that chronic stroke patients can gain voluntary finger extension if many feedback sessions, sometimes lasting years, are provided or if frequent voluntary efforts are undertaken (Brudny, personal communication, 1982). Although this observation might be valid, it represents a clinical situation that is totally unrealistic for most patients and the professionals responsible for rendering physical therapy services.
Five of 22 experimental patients, who had voluntary joint movements and comparatively less hyperactivity within spastic musculature, gained considerable functional improvement, especially in tasks associated with hand activities (Tab. 12). This 22.8 percent compares favorably with less objective data provided by Brudny and co-workers who observed at least a reacquisition of crude prehension in 11 of 70 hemiparetic patients (15.7 percent) and a gain of full prehensile capabilities in 7 out of 8 patients selected from the group of 11. A reasonable final comparison would, therefore, indicate that 7 of 70 (10%) patients studied by Brudny gained prehension using far more sophisticated equipment and many more treatment sessions than employed in the present study.
Because many patients were unable to complete functional tasks before treatment, analyses of improvement were undertaken using a discrete data base. These improvements are not necessarily correlated to changes in neuromuscular measures. For example, a within-group analysis of Experimental-no hand patients revealed significant improvement on 22 of 167 variables (unpublished data), including some measures governing the wrist and fingers, but none of these changes was of enough magnitude to augment hand function. Although the Experimental-no hand subgroup initially had significantly poorer neuromuscular measures than the Experimental-hand subgroup (Tabs. 8 and 9), their within-group improvements clearly enabled them to enhance only some shoulder and elbow functions (Tabs. 10 and 11). Therefore, patients are capable of augmenting assistive functional capacity of the involved upper extremity even if prehension is not achieved.
Although patients were examined four times, we
Volume 63 / Number 9, September, 1983 1401
chose to present differences between only the initial and final examinations. Differences between measures comparing baseline to first (20 sessions) and second (40 sessions) did not reveal any point along the 20 session interval segments where the greatest change could be consistently observed; that is, the rate of improvement seemed to vary among experimental patients. This occurrence may be due to location of lesion, motivation, type or duration of previous rehabilitation, or a myriad of circumstances that would require sizable numbers of patients to decipher. Therefore, when the maximal benefits from this intervention accrue remained an unresolved issue, but more frequent measurements interspersed between treatments to a larger sample could provide an answer. Within our constraints, however, given the complexity and time required for each examination, this possibility was unrealistic.
We chose to report peak EMG values rather than integrated data over the time for each measurement. To pursue the latter would have required computer capabilities beyond our resources. Since completion of this study, we have compared peak EMG to integrated values recorded simultaneously from the same spastic muscle groups of spinal cord injured patients and have found peak EMG to be an accurate reflection of integrated values for the same maneuver.
Taken in total perspective, findings from this study suggest that EMG biofeedback can be beneficial in restoring improved upper extremity function among chronic stroke patients, and specific neuromuscular characteristics may be indicative of patients who will gain the greatest success. Inevitably, more precise patient characteristics, such as specificity and extent of lesion, may prove to have greater predictive merit.
Until such time, however, clinicians now have at their disposal some unique neuromuscular attributes to make a reasonable decision regarding implementation of this modality to the upper extremities of chronic stroke patients.
At the same time, it should be noted that the feedback application was offered without any other therapeutic treatment. Hopefully, if the modality is used as originally intended, as an informational resource during other treatment approaches, the outcomes may be even better. Until such time as this speculation is tested, clinicians are encouraged to replicate the present findings. More important, however, they should seek self-initiative to augment them.
SUMMARY
Applications of EMG biofeedback to the upper extremities of chronic stroke patients can favorably alter neuromuscular activity and enhance function when a specific treatment protocol is followed. Those patients who achieve the most substantial improvement in manipulative abilities initially possess voluntary finger extension; comparative greater active ROM about the shoulder, elbow, and wrist; and comparatively less hyperactivity in muscles usually considered as major contributors to the typical flexor synergy.
Acknowledgments. The authors are grateful to James Kelly, Marilyn Nacht, and Marianne van Lun-teren-Hudson for their tireless efforts to acquire data and treat patients. The graphic skills provided by James Perry and the typing talents of Gloria Bassett are very much appreciated. Dr. Ray Bain offered invaluable assistance in advising us on appropriate statistical designs.
REFERENCES
1. Brudny J, Korein J, Levidow L, et al: Sensory feedback therapy as a modality of treatment in central nervous system disorders of voluntary movement. Neurology 24:925-932, 1974
2. Brudny J, Korein J, Grynbaum BB, et al: EMG feedback therapy: Review of treatment of 114 patients. Arch Phys Med Rehabil 57 :55-61 , 1976
3. Grynbaum BB, Brudny J, Korein J, et al: Sensory feedback therapy for stroke patients. Geriatrics 31:43-47, 1976
4. Brudny J, Korein J, Grynbaum BB, et al: Helping hemipar-etics to help themselves. JAMA 241:814-818, 1979
5. Lee K-H, Hill E, Johnston R, et al: Myofeedback for muscle retraining in hemiplegic patients. Arch Phys Med Rehabil 57 :588-591 , 1976
6. Skelly AM, Kenedi RM: EMG biofeedback therapy in the reeducation of the hemiplegic shoulder in patients with sensory loss. Physiotherapy 68:34-38, 1982
7. Mroczek N, Halpern D, McHugh R: Electromyographic feedback and physical therapy for neuromuscular retraining in hemiplegia. Arch Phys Med Rehabil 59:258-267, 1978
8. Bowman BR, Baker LL, Water RL: Positional feedback and electrical stimulation: An automated treatment for the hemiplegic wrist. Arch Phys Med Rehabil 60 :497-501 , 1979
9. Marsh RW: Electromyographic feedback treatment of hemiplegia. New Zealand Medical Journal 91:96-97, 1980
10. Middaugh SJ, Miller MC: Electromyographic feedback: Effect on voluntary contractions in paretic subjects. Arch Phys Med Rehabil 61:24-29, 1980
11. Wolf SL, Baker MP, Kelly JL: EMG biofeedback in stroke: Effect of patient characteristics. Arch Phys Med Rehabil 60:96-102, 1979
12. Gianutsos J, Eberstein A, Krasilowsky G, et al: EMG feedback in the rehabilitation of upper extremity function: Single case studies of chronic hemiplegics. International Neuropsychological Society Bulletin, Symposium Issue, 1979, p 12
13. Wolf SL, Baker MP, Kelly JL: EMG biofeedback in stroke: A 1 -year follow-up on the effect of patient characteristics. Arch Phys Med Rehabil 61:351 -355 , 1980
14. Kelly JL, Baker MP, Wolf SL: Procedures for EMG biofeedback training in involved upper extremities of hemiplegic patients. Phys Ther 59:1500-1507, 1979
15. Wolf SL: Developing specific treatment strategies for muscle biofeedback in stroke patients (Project R-64). Emory University Regional Rehabilitation Research and Training Center Annual Progress Report, Grant No. 10/P-56808/4-16, Atlanta, GA, Grant period 4 /1 / 79 -3 /31 / 8 0 , pp 134-172
16. Prevo AJH, Visser SL, Vogelaar TW: Effect of EMG feedback on paretic muscles and abnormal co-contraction in the hemiplegic arm, compared with conventional physical therapy. Scand J Rehabil Med 14:121-131, 1982
1402 PHYSICAL THERAPY
RESEARCH
APPENDIX
Functional Tasks
CODE NAME
AUE1 AUE2 AUE3
AUE4 BUE1
BUE2
BUE3
BUE4
CUE1
CUE2
CUE3
CUE4 DUE1 DUE2
DUE3
DUE4
EUE1
EUE2A1 EUE2A2 EUE2B1 EUE2B2 EUE3 EUE4 EUE5 EUE6A EUE6B EUE7 EUE8A
EUE8B
DESCRIPTION
shoulder flexion—time to place forearm on table shoulder control—time to make circle with tip of olecranon shoulder stabilization—hold a weight from sliding 1 cm using forearm;
seated facing table shoulder flexion—time to flex to 90 degrees elbow extension—time to achieve full elbow extension with arm resting
on table at 90 degrees elbow flexion forearm assist—time to remove a middle sheet of three pages held down
with involved arm elbow extension—amount of weight around wrist that can be applied
during full elbow extension from 90 degrees of elbow flexion elbow extension—time to retrieve 0.5 kg weight without lifting elbow off
table shoulder abduction—time to place elbow on table from lap when seated
parallel to table shoulder abduction—time to make continuous circle with olecranon when
seated parallel to table shoulder abduction—hold a weight from sliding 1 cm using forearm;
seated parallel to table shoulder abduction—time to abduct shoulder to 90 degrees elbow extension—time to extend elbow on table seated parallel to table shoulder stabilization—time to remove middle sheet of three papers held
down with involved arm; patient seated parallel to table elbow extension—amount of weight around wrist that still permits full
elbow extension from 90 degrees elbow flexion; patient seated parallel to table
elbow extension—time to retrieve 0.5 kg weight without lifing elbow off table; patient seated parallel to table
shoulder flexion—elbow extended—amount of weight that can be placed on table starting arm at side
grip strength—maximum force to squeeze dynanometer, uninvolved hand grip strength—time to release dynanometer, uninvolved hand grip strength—maximum force to squeeze dynamometer, involved hand grip strength—time to release dynanometer, involved hand functional pinch—time to lift pencil functional grasp—time to grasp cup around its side refined pinch—time to lift paper clip pronation-supination—time to flip five 3 x 5 note cards; involved hand pronation-supination—time to flip five 3 x 5 note cards; uninvolved hand fine motor—time to stack three checkers reciprocal movement—time to move hand to mouth and back to lap x 3;
involved arm reciprocal movement—time to move hand to mouth and back to lap x 3;
uninvolved arm
Volume 63 / Number 9, September, 1983 1403