624

Constraint-Induced Movement Therapy for Motor Recovery in Chronic Stroke Patients Annett Kunkel, DiplPsych, Bruno Kopp, PhD, Gudrun Miiller, DiplPsych, Kersten Villringer, MD, Arno Villringer, MD, Edward Taub, PhD, Herta Flor, PhD ABSTRACT. Kunkel A, Kopp B, Miiller G, Villringer K, Villringer A, Taub E, Flor H. Constraint-induced movement therapy for motor recovery in chronic stroke patients. Arch Phys Med Rehabil 1999$0:624-g. Objective: Assessment of the effectiveness of constraintinduced (CI) movement therapy and quantitative evaluation of the effects of CI therapy. Design: Intervention study; case series; pretreatment to posttreatment measures and follow-up 3 months after intervention. Setting: An outpatient department. Patients: Five chronic stroke patients with moderate motor deficit; convenience sample. Interventions: CI therapy consisting of restraint of the unaffected upper extremity in a sling for 14 days combined with 6 hours of training per weekday of the affected upper extremity. Main Outcome Measures: Actual Amount of Use Test (AAUT), Motor Activity Log (MAL), Wolf Motor Function Test (WMFT), and Arm Motor Ability Test (AMAT) Results: There was a substantial improvement in the performance times of the laboratory tests (AMAT, WMFT, p 5 .039) and in the quality of movement (AMAT, WMFT, p % .049; MAL, p = .049), particularly in the use of the extremity in “real world” environments (AAUT, p = .020), supported by results of quantitative evaluation. The effect sizes were large and comparable to those found in previous studies of CI therapy. Conclusions: CI therapy is an efficacious treatment for chronic stroke patients, especially in terms of real world outcome. 0 1999 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation TROKE IS THE LEADING cause of disability in the adult population and is frequently accompanied by substantial S loss of motor function.1,2 Although acute and postacute rehabili-

From the Department of Psychology, Humboldt-University (Drs. Kunkel, Kopp, Miiller, Flor), the Department of Radiology, Free University (Dr. K. Villringer), and the Department of Neurology, CharitC Hospital (Dr. A. Viliringer), Berlin, Germany; and the Department of Psychology, University of Alabama at Birmingham, AL (Dr. Taub). This work is part of Annett Kunkels’ doctoral dissertation submitted to the Faculty of Mathematics and Natural Sciences at Humboldt University. Submitted for publication June 15,1998. Accepted December 17,1998. Supported by the Deutsche Forschungsgemeinschaft (r+%xrch group “Co&al Plasticity” Fl 156/16-l), the German American Academic Council (Transcoop Program), and grant B95-975R from the Rehabilitation Research and Development Service, United States Department of Veterans Affairs. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. Reprint requests to Herta Flor, PhD, Department of Psychology, Clinical Psychology and Behavioral Neuroscience, Humboldt-University, Hausvogteiplatz 5-7, D-10117 Berlin, Germany. 0 1999 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation 0003.9993/99/8006-5079$3.00/O

Arch

Phys

Med

Rehabil

Vol 80, June

1999

tation programs are available to stroke patients, substantial impairment and disability may persist for years. In a long-term follow-up study of stroke: it was found that 56% of the patients tested 5 years after stroke still had pronounced hemiparesis, which was the most common complaint at that time. The literature on the efficacy of rehabilitation programs targeting motor dysfunction (for a comprehensive review, see Duncan4) suggests that few effective methods are available and that their effects on chronic motor disability after stroke may be especially weak and not permanent.5 A possible explanation for the substantial remaining motor deficits in stroke patients might be the occurrence of learned nonuse, a phenomenon first described by Taub.6 Stroke patients who initially attempt to use the affected extremity find themselves unable to do so because the process of spontaneous recovery of function has not yet proceeded sufficiently far. This results in the experience of failure or punishment for attempts to move the extremity and in positive reinforcement for compensatory movements by the unaffected extremity-a learning process that might be supported by the teaching of compensatory activity during rehabilitation. 7 This learned nonuse impedes attempts to further rehabilitate the affected extremity. Based on this theoretical account,6 constraint-induced (CI) movement therapy was developed. It is designed to overcome this learned disability by restraining the unaffected extremity and training the affected extremity, thereby leading to massed practice in the use of the affected extremity for a period of 2 weeks.*-lo One uncontrolled and two controlled treatment outcome studies have tested CI therapy. A study by Wolf and associates10 used unaffected limb restraint without concurrent training for 2.5 subjects with hemiplegia that had resulted from stroke or head injury and observed them until 1 year after treatment. The main change occurred in speed of task execution. Taub and colleagues8 compared a group of four patients who received CI therapy to five subjects who served as attention controls. The treatment group carried out specified tasks under supervision for 6 hours per day for the 10 weekdays of the treatment period; the control group received instructions to focus attention on the affected limb, placebo physical therapy sessions,and self-rangeof-motion exercises. After the 2-week intervention there were substantial decreasesin performance speed, significant gains in quality of movement in the laboratory, and, in particular, very substantial increases in actual amount of use of the affected extremity in the life environment. Moreover, the therapeutic gains persisted at the 2-year follow-up. In a more recent study, 20 chronic stroke patients were treated with restraint of the unaffected, and training of the affected, upper extremity.g The patients showed highly significant improvement of upper extremity function at the posttreatment assessmentas well as at the l-month follow-up. Table 1 contains the characteristics of the reported studies as well as their effect sizes (ESs). Effect sizes were computed by subtracting the posttreatment or follow-up mean of each individual variable from the pretreatment mean and dividing it by the posttreatment or follow-up standard deviation.” The variablelspecific ESs were subsequently averaged to obtain a studywise ES.

CONSTRAINT-INDUCED

Table lirm&cE Authors,

Year

Deficit

Taub and colleagues,8 1993

CVA patients with a paretic upper extremity

>lyr (mean: 3.65yrsl

Taub and cov~orkers,~ 1998

CVA patients with a paretic upper extremiv

zlyr (mean: 3.67~~)

Wolf and associates,‘” 1989

Hemiplegia (stroke or TBI)

s1yr

Kunkel and colleagues, 1999 (this report)

CVA patients with a paretic upper extremity

>lyr 7w)

(mean:

1: Study

MOVEMENT

Characteristics

THERAPY,

and

Effect

Sizes

Type of Intervention

Number of Patients

Group 1: Uninvalved upper extremity restrained plus practice of the involved upper extremity. Group 2: Attention control. Uninvolved upper extremity restrained plus training of the involved upper extremity. Uninvolved upper extremity restrained, but no training. Uninvolved upper extremity restraint plus training ofthe involved upper extremity.

Group 1,4; group 2,5

14 days restraint, 10 days practice for Gh/day

Fol 1,4wk; 2yl3

20

14 days restraint, IO days training for Ghiday

21

14days

5

Duration of Intervention

626

Kunkel

restraint

14 days restraint, 10 days training for Ghlday

F$w;U

p

Outcome MeaSUR?S

fol2,

Effect Size of Study

WMFT, AMAT, MAL

Group 1: prepost, 2.08; pre-fol 1, 2.80*; pre-fol 2,2.95.* Group 2: prepost, .02; pre-fol 1, .31*; pre-fol 2: .Ol.”

Fol 1.4wk

WMFT, MAL

Pre-post, 2.14; pre-fol 1, 2.50.*

Fol 1, Zmo; fol2, 4mo; fol3, 12mo

WMFT

Pre-post, .21; pre-fol 1, .34; prefol2, .48; pre-fol3, .48.

Fol 1.3mo

AAUT, MAL, WMFT, AMAT

Pre-post, 2.38: pm-f01 1, 1.92.f

Abbreviation: Fol, follow-up. * MAL only. + R = 4 subjects only.

The purpose of this study was to (1) replicate the findings of Taub8 and Taub and coworkers9 in a different laboratory, and (2) compare the ESs of the CI treatment carried out in our laboratory to those reported previously. METHODS Subjects Five patients (4 women, 1 man) participated in the study. Their ages ranged from 47 to 66 years, with a median age of 53 years. Time after stroke varied from 3 to 15 years, with a median time after stroke of 6 years). All subjects had premorbid right-arm dominance l*; four were hemiparetic on the right side and one, on the left. They were recruited from a Berlin stroke self-help group. Immediately poststroke, each patient participated in 6 to 8 weeks of rehabilitation; subsequently, they received conventional physical therapy once or twice a week for varying amounts of time. The following exclusion criteria were used: (1) stroke experienced less than 1 year earlier; (2) serious sensory, cognitive, or aphasic deficits (Mini-Mental Testi scores ranged from 26 to 30; Token Test14scores, from -2 to 2) or severely depressed mood (CES-D Scalei scores ranged from 6 to 22); (3) lesion in the primary sensory or motor areas of the cortex; (4) inability to extend at least 10” at the metacarpophalangeal and interphalangeal joints and 20” at the wrist; (5) ability to make extensive use of the involved upper extremity (Motor Activity Log score above 2.5) so that significant further improvement could not be expected; (6) left-arm dominance, and (7) age more than 80 years. For all patients, magnetic resonance imaging (MRI) of the lesion was obtained at the time of the treatment (1..5-Tesla

Siemens Magnetom Visiona Tl-weighed scans, slice thickness lmm, repetition time 22msec, echo time 10msec). Patient 1 showed a lesion in the posterior portion of the left internal capsule; patient 2 had an extended lesion in the left insula and the temporoparietal transition; patients 3 and 4 showed lesions in the left insula and left putamen; and patient 5 had a lesion in the right putamen and caudate nucleus (fig 1). The procedure of the study was approved by the local ethics review board and all subjects signed informed consent after receiving a detailed explanation of the study procedures. Assessment The Actual Amount of Use Test (AAUT) was developed by Taub9 as an implicit measure of actual use of an upper extremity. The test contains 21 items (eg, filling out a form). All patients signed informed consent for videotaping before the assessment phase but were not aware that they were being videotaped while performing AAUT activities. A rating scale with three categories (ranging from nonuse of the affected arm [0] to functional participation of the affected extremity in the activity at whatever level of proficiency [2] is used by the observer to judge the frequency of use of the arm, and a six-category rating scale (ranging from nonuse [0] to normal [5]) is used to judge the quality of movement. The AAUT represents an objective and unobtrusive measure of motor function of the affected upper extremity. A structured interview, the Motor Activity Log (MAL), was used to determine how often the patients used the affected extremity during the activities of daily life (ADL) outside the laboratory. For a specified time period, the patient had to indicate the amount of use of the affected arm for 20 individual Arch

Phys

Med

Rehabil

Vol 80, June

1999

626

CONSTRAINT-INDUCED

MOVEMENT

patient 4

patient 2

Fig 1. Magnetic

resonance

images

of the

lesions

in patients

1 to 4.

ADL (eg, holding a glass while pouring water into it) and the quality of movement; both were noted on separate 6-point scales. Two laboratory motor tests were used. Performance on each was videotaped. The Wolf Motor Function Test (WMFT) lo was developed to quantify motor function after stroke and traumatic brain injury. A modified version of the tests consists of 16 items ranging from simple movements, such as elbow extension and limb flexion for placing the hand on a table; to functional movements, such as lifting a soda can in simulated drinking or turning cards over; to ADL, such as folding a towel. Performance time is measured on 15 items; one item serves to determine the force. The quality and functionality of the test movements are evaluated by a separate 6-point rating from videotapes. The Arm Motor Ability Test (AMAT)8,16,17 assessesthe motor ability of the hand and arm during ADL tasks. The test consists of 13 complex tasks that include one to three components each, with a total of 28 component tasks. Sample items are eating with a spoon, drinking from a cup, putting on a sweater, and buttoning it. Speed of task performance is recorded by a stopwatch, and functional ability and quality of movement are rated on 6-point scales from video recordings of the patient’s performance (identical to the modified WMFT8). The test has been found to have high reliability,i6J7 internal consistency, and validity. l6 The videotapes of all motor tests were rated by an observer unaware of the pre- or posttreatment status of the patient. Procedure The initial evaluation included the psychological and neurologic examination. Before and after the 2-week intervention Arch

Phys

Med

Rehabil

Vol 80, June

1999

THERAPY,

Kunkel

period, as well as 3 months after treatment, the structured interview (MAL) and the laboratory motor tests (WMFT and AMAT) were administered. The AAUT was given only before and immediately after treatment. During the 14 days of intervention, the unaffected upper extremity was restrained by an ensemble consisting of a resting hand splint placed in a sling; this required the patient to carry out all activities with the affected limb. The patient agreed to wear the splint and sling for 90% of waking hours. In a treatment contract, specific exceptions from this rule were listed, such as sleeping, use of water, and any activity where having the unaffected arm restrained might affect safety. The patients kept a diary in which they recorded all the activities that were performed with the affected arm either with the sling and splint in place or removed. On weekends, the patients continued to wear the sling and practiced the tasks they had learned in the laboratory for at least 1 hour per day. In addition to the unaffected arm restraint, the patients were given behavioral training18 of the affected limb for 6 hours per day on each of the 10 weekdays of the 14-day treatment period. Based on the results of the pretreatment motor assessment,12 to 16 specific exercises were chosen for training (shaping) that were focused on improving movements involving the maximum deficit.9 Shaping tasks included such activities as flipping dominos or stacking blocks. In the five patients trained in this study the deficits ranged from abduction and flexion at the shoulder, pronation/supination of the forearm, extension of the wrist or elbow, isolated finger movements, and fine motor skills. Each shaping task was repeated 10 times in a block of trials, and blocks of trials for each task and for different tasks were repeated throughout the day, with appropriate rest intervals between blocks. The difficulty of the tasks was continuously increased in small steps and contingent verbal reinforcement was given for the slightest improvement in performance time or quality of movement. Data Analysis For all the variables in the study Friedman one-way analysis of variance (ANOVA) was used to determine the overall change from pretreatment to posttreatment and to follow-up. These ANOVAprocedures were followed by pretreatment to posttreatment and pretreatment to follow-up one-sided Wilcoxon rank and sign tests. Since four dependent variables were used in the study, we refrained from Bonferroni corrections in order not to further reduce the power of our tests. In addition, ESs were computed for each individual variable for both the pretreatment to posttreatment and the pretreatment to follow-up comparisons. Table 2 contains the means and standard deviations of all the variables used in the study and the ESs. RESULTS Efficacy of the Intervention The AAUT, which is considered to reflect real world behavior, showed very substantial improvements in the use of the affected extremity after the training period. Amount of use of the affected extremity in the laboratory situation almost doubled (increase of 98%) and quality of movement increased by 124% in the pretreatment to posttreatment comparison (both Wilcoxon tests: Z = -2.02, p = .020; table 2). The MAL showed a 166% increase in quality of movement of the affected extremity at posttreatment and 165% at follow-up (Friedman test: x2 = 6, p = .049; individual subject pretreatment to posttreatment and pretreatment to follow-up comparisons, all p values < .OS).A similarly strong improvement occurred on the

CONSTRAINT-INDUCED

Table

2: Pretreatment, Posttreatment, Standard Deviations, and Effect Pretreatment

Posttreatment

Mean

Mean

SD

1.23 2.06

.38 1.10

SD

MOVEMENT

and Follow-Up Means, Sizes for All Patients Follow-Up Mean

Effect SD

Pre-Post

Sizes Pre-Fol

AAUT AOU QOM MAL

.62 .92

.36 .57

AOU

1.97

.55

4.65

.28

QOM AMAT

1.31

.36

3.48

.67

Time

11.40

FA QOM WMFT

2.96 2.77

4.41 .74

9.23 3.54

3.42 .72

.72

3.44

.70

7?me

13.15

FA QOM

3.37 3.17

6.70 .51

7.62 3.93

4.43 .60

.54

3.69

.61

1.61 1.04 4.43 3.54

.32 1.11

9.57 3.24

7.59 1.99

9.36

3.35

.64

.62

3.61 3.49

1.05 1.05

.81 .96

.55 .62

12.15 3.75 3.45

7.09 .83 .82

1.25 .93 .85

.26 .63 .27

Abbreviations: Pre-Post, pretreatment to posttreatment; Pre-Fol, pretreatment to follow-up; AAUT, Actual Amount of Use Test; AOU, amount of use; QOM, quality of movement; MAL, Motor Activity Log; AMAT, Arm Motor Ability Test; FA, functional ability; WMFT, Wolf Motor Function Test.

amount of use scale of the MAL: it showed an improvement of 136% at posttreatment and 122% at the follow-up assessment. Amount of use MAL data were available for only three subjects. The Friedman Test did not reveal a significant difference for the overall change (x2 = 4.6, p = .13), but this result must be interpreted in terms of the very small sample size. When these comparisons were made for the data of individual subjects, all differences were significant beyond the p = .005 level (range p = .004 to .OOOl,Wilcoxon test). Substantial improvements also occurred in all the measures of the laboratory tests of motor ability in all patients at both the posttreatment and the 3-month follow-up assessments. Mean performance time for all subjects for all tasks on the AMAT was significantly faster at posttest (average improvement 19%) and at follow-up (average improvement 18%; Friedman test, x2 = 6.5, p = .039) than at the pretest, as were the performance time scores on the WMFT (average posttreatment improvement = 42%; average follow-up improvement = 13%; table 2; Friedman test, x2 = 8.0, p = .018). Each individual patient showed considerable improvement in mean performance time on both tests and at both the posttreatment and the follow-up assessment (table 2). Quality of movement and functional ability were also significantly improved. The AMAT functional ability rating increased by 20% at posttreatment and 19% at the 3-month follow-up (Friedman test, x2 = 6.5, p = .039), and the quality of movement scores increased by 24% at posttreatment and 23% at follow-up (Friedman test, x2 = 6.0, p = .049). For the WMFT, the improvement in functional ability was 17% at posttreatment and 10% at follow-up (Friedman test, x2 = 7.13, p = .028), and the change in quality of movement was 16% at posttreatment and 7% at follow-up (Friedman test, x2 = 6.5, p = .039). All pretreatment to posttreatment and pretreatment to follow-up comparisons were significant (p values < .05) or showed a significant trend (pretreatment to follow-up comparisons for the time variable in the AMAT and WMET, both p values < .06) except for the pretreatment to follow-up comparison for the quality of movement variable in the WMFT (NS). Comparison of Effect Sizes The ESs in all the CI therapy reports combined, including this one. show that it is an effective method for the treatment of

THERAPY,

627

Kunkel

upper extremity dysfunction in chronic stroke patients. The lowest but still positive ESs were obtained in the Wolf study,lO with ES ranging from .21 to .43 (table 1). The largest ESs were reported in the current study for the pre-post assessment (ES = 2.38) and in the Taub study9 for the l-month follow-up (table 1, ES = 2.50). The mean pretreatment to posttreatment ES of all four studies is 1.70, which is very large by the standards set forth in the meta-analysis literature (ie, .2 = small, .4 to .6 = moderate, and .8 = large).i9 Without the Wolf study,‘Owhich involves only half the protocol used in the other studies, the mean ES was 2.19. Interestingly, the measures that describe “real world” outcome, such as MAL and AAUT, were more improved than variables of motor function (AMAT or WMFT; tables 1 and 2). Overall, the mean ES of the four studies resulted in a substantial 66% improvement (74% without the Wolf studylo) in function for both laboratory and real world measureszOMoreover, the attention-controlled study by Taub* yielded a substantially higher ES for the experimental (ES = 2.08) as compared to the control (ES = .02) group (table l), suggesting that nonspecific effects play a minor role in the efficacy of this approach. DISCUSSION The most striking finding of this study was the dramatic improvement in the amount of use of the affected extremity in real world environments, as shown by the more than 100% increase in use and the large ESs on both the MAL and the AAUT. The agreement of the objective measure (AAUT) and the patients’ subjective report (MAL) indicated that the improvement in the amount of use of the affected limb is transferred from the clinic to the real world situation. Smaller but significant improvements were found on the AMAT and the WMFT tests with respect to speed of task performance, quality of movement, and functional ability. The increased use of the affected arm in the real world environment is the more interesting because quality of movement, although substantially improved, still exhibited a substantial deficit. The gap between the moderate improvement in motor function on the laboratory tests and the massive improvement in actual amount of use of the limb in the life environment is the area within which Cl therapy operates and produces therapeutic effects. There is often a very large difference between what a stroke patient can do and what she or he does2i; CI therapy reduces this difference. As noted above,6,8the disparity between the motor capacity of many stroke patients and their actual use of the limb may be due to learned nonuse that develops in the early poststroke period but which can be overcome by the application of an appropriate technique, such as CI therapy. It is unlikely that the current results are based on spontaneous improvements in motor function. All the patients in this study were several years poststroke, and their condition had not improved for many months to years.Although some (nonsignificant) regression took place, the improvements were retained at the 3-month follow-up, suggesting that the therapy had produced a long-term treatment effect, as noted in previous research.* Taken together, the results from the previous studies and this experiment suggest that CI therapy is a powerful technique for the modification of motor deficit late after stroke for the substantial number of patients for whom it is applicable. This independent replication in another laboratory of CI therapy proved to be as successful as the original study. Larger sample sizes with corrections for multiple tests need to be used in further research to substantiate the findings of this initial replication study. Based on the large effects produced by CI therapy in chronic stroke patients, there is a need for application of CI therapy with acute and subacute stroke patients where it Arch

Phys

Med

Rehabil

Vol 80, June

1999

628

CONSTRAINT-INDUCED

MOVEMENT

might be possible to avoid completely the development of a portion of the chronic motor deficit that would otherwise develop. References 1. Dennis MS, Bum JP, Sandercock PAG, Bamford JM, Wade DT, Warlow CP. Long-term survival after first-ever stroke: the Oxford Community Stroke Project. Stroke 1993;24:796-800. 2. Ferrucci L, Bandinelli S, Guralnik JM, Lamponi M, Bertini C, Falchini M, et al. Recovery of functional status after stroke: a post rehabilitation follow-up study. Stroke 1993;24:200-5. 3. Wilkinson PR, Wolfe CDA, Warburton FG, Rudd AG, Howard RS, Ross-Russell RW, et al. A long-term follow-up of stroke patients. Stroke 1997;28:507-12. 4. Duncan PW. Synthesis of intervention trials to improve motor recovery following stroke. Top Stroke Rehabil 1997;3: l-20. 5. Wade DT, Collen FM, Robb GF, Warlow CP. Physiotherapy intervention late after stroke and mobility. BMJ 1992;304:609-13. 6. Taub E. Somatosensory deafferentation research with monkeys: implications for rehabilitation medicine. In: Ince LP. editor. Behavioral psychology in rehabilitation medicine: clinical’applications. New York: Williams &Wilkins; 1980. p. 371-401. I. Russo SG. Hemiplegic upper extremity rehabilitation: a review of the forced-use paradigm. Neurol Rep 1995;19: 17-22. 8. Taub E, Miller NE, Novack TA, Cook EW III, Fleming WC, Nepomuceno CS, et al. Technique to improve chronic motor deficit after stroke. Arch Phvs Med Rehabil 1993:74:347-54. 9. Taub E, Crago JE,-Uswatte G. Constr&t induced movement therapy: a new approach to treatment in physical rehabilitation. Rehabil Psvchol1998:43: 152-70. 10. Wolf SL, ‘LeCraw DE, Barton LA, Jann BB. Forced use of hemiplegic upper extremities to reverse the effect of learned

Arch

Phys

Med

Rehabil

Vol 80, June

1999

THERAPY,

Kunkel

nonuse among chronic stroke and head-injured patients. Exp Neurol 1989;104:125-32. 11. Glass GV, McGraw B, Smith MC. Meta-analysis in social research. Beverly Hills (CA): Sage; 1981. 12. Oldfield RC. The assessment and analysis of handedness: the Edinburgh Inventory. Neuropsychologia 1971;9:97-113. 13. Folstein MF, Folstein SE, McHugh PR. Mini-Mental Status Test. Weinheim, Germany: Beltz; 1990. 14. Orgass B. Token test. Weinheim, Germany: Beltz; 1992. 15. Hautzinger M, Bailer M. Allgemeine Depressionskala. Weinheim, Germany: Beltz; 1993. 16. Kopp B, Kunkel A, Flor H, Platz T, Rose U, Mauritz K-H, et al. The Arm Motor Ability Test: reliability, validity, and sensitivity to change of an instrument for assessing disabilities in activities of daily living. Arch Phys Med Rehabil1997;78:615-20. 17. McCulloch K, Cook EW III, Fleming WC, Novack TA, Taub E. A reliable test of upper extremity ADL function [abstract]. Arch Phys Med Rehabil 1988;69:755. 18. Taub E, Pidikiti RD, DeLuca SC, Crago JE. Effects of motor restriction of an unimpaired upper extremity and training on improving functional tasks and altering brain behaviors. In: Toole JF, Good DC, editors. Imaging in neurologic rehabilitation. New York: Demos; 1996. p. 133-53. 19. Cohen J. Statistical power analysis for the behavioral sciences. New York: Academic Press; 1977. 20. Rosenthal R, Rubin DB. A simple general purpose display of magnitude of experimental effect. J Educ Psycho1 1982;74:166-9. 21. Andrews K, Stewart J. Stroke recovery: he can but does he? Rheumatol Rehabil 1979;18:43-8. Supplier a. Siemens AG, Ntirnberger Str. 74, D-91050 Erlangen, Germany.