Research Report. Limitations. Only studies published in journals and in English were included in the systematic review.

Transcription

1 Research Report H. Huang, MS, OT, is an ScD student in the Department of Physical Therapy and Athletic Training, Boston University, Boston, Massachusetts, and currently is working in the Department of Biokinesiology and Physical Therapy, University of Southern California, 1540 E Alcazar St, CHP 155, Los Angeles, CA (USA). Address all correspondence to Ms Huang at: hsiangha@usc.edu. L. Fetters, PT, PhD, FAPTA, is Professor, Department of Biokinesiology and Physical Therapy, and Department of Pediatrics, Keck School of Medicine, University of Southern California. J. Hale, PT, DPT, is Physical Therapist, The Institute for Rehabilitation and Research (TIRR), Houston, Texas. A. McBride, PT, DPT, is Physical Therapist, Providence Portland Medical Center, Portland, Oregon. [Huang H, Fetters L, Hale J, McBride A. Bound for success: a systematic review of constraintinduced movement therapy in children with cerebral palsy supports improved arm and hand use. Phys Ther. 2009;89: ] 2009 American Physical Therapy Association Bound for Success: A Systematic Review of -Induced Movement Therapy in Children With Cerebral Palsy Supports Improved Arm and Hand Use Hsiang-han Huang, Linda Fetters, Jennifer Hale, Ashley McBride Background. -induced movement therapy (CIMT) is a potentially effective intervention for children with hemiplegic cerebral palsy (CP). Purpose. The objectives of this systematic review are: (1) to investigate whether CIMT is supported with valid research of its effectiveness and (2) to identify key characteristics of the child and intervention protocol associated with the effects of CIMT. Data Sources and Study Selection. A search of MEDLINE (1966 through March 2009), Entrez PubMed (1966 through March 2009), EMBASE (1980 through March 2009), CINAHL (1982 through March 2009), PsychINFO (1887 through March 2009), and Web of Science (1900 through March 2009) produced 23 relevant studies. Data Extraction and Synthesis. The 2 objectives of the review were addressed by: (1) scoring the validity and level of evidence for each study and calculating evidence-based statistics, if possible, and (2) recording and summarizing the inclusion and exclusion criteria, type and duration of constraint, intervention and study durations, and outcomes based on the International Classification of Functioning, Disability and Health (ICF). Limitations. Only studies published in journals and in English were included in the systematic review. Conclusions. Studies varied widely in type and rigor of design; subject, constraint, and intervention characteristics; and ICF level for outcome measures. One outcome measure at the body functions and structure level and 4 outcome measures at the activity level had large and significant treatment effects (d.80), and these findings were from the most rigorous studies. Evidence from more-rigorous studies demonstrated an increased frequency of use of the upper extremity following CIMT for children with hemiplegic CP. The critical threshold for intensity that constitutes an adequate dose cannot be determined from the available research. Further research should include a priori power calculations, more-rigorous designs and comparisons of different components of CIMT in relation to specific children, and measures of potential impacts on the developing brain. Post a Rapid Response or find The Bottom Line: f Physical Therapy Volume 89 Number 11 November 2009

2 Recent studies of constraintinduced movement therapy (CIMT) support the improvement in use of the affected arm for adults with stroke 1 4 and suggest similar effects in children with hemiplegic cerebral palsy (CP) induced movement therapy is based on studies of young monkeys in which somatosensory deafferentation was performed on a single forelimb. 9,10 Following deafferentation, the monkeys failed to use the affected forelimb, a phenomenon termed by Taub as learned non-use. 9,10 When the unaffected forelimb was constrained and could not be used for functional activities, the monkeys overcame the learned non-use. The new functional movements of the deafferented forelimb were sustained, but only if the period of constraint was greater than 7 days. 10 Forced-use therapy for adult humans was derived from this approach and revealed practical implications for rehabilitation. 11,12 Shaping techniques and repetitive practice later were added to constraint. 9,13 With shaping, a behavior was progressively modified toward a goal through successive approximation and reinforcement. This intervention is now termed constraintinduced movement therapy. 3,14 Studies performed on monkeys in utero and shortly after birth supported the idea that forced use or Available With This Article at Invited Commentary by Jeanne Charles and Steven L. Wolf and the Author Response Audio Abstracts Podcast This article was published ahead of print on September 3, 2009, at CIMT also might be effective early in development. 10,15 Following deafferentation both in monkey fetuses and on the day of birth, researchers restrained the less-affected forelimb and forced the use of the affected forelimb with shaping techniques. 15 These newborn monkeys for the first time were developing the use of the affected forelimb in contrast to relearning previous skilled movements, as occurred in the adult monkeys. The young monkeys progressed from a loose, 4-finger grasp to thumb-forefinger prehension and could self-feed with the affected limb. These results supported the hypothesis that children with hemiplegic CP who have not yet developed skilled arm use, or developmental disuse as described by Gordon and colleagues, 16 also might achieve improved upper-extremity (UE) function following forced use or CIMT. Of note is that the central nervous system in these young children is still in the early stages of development. The impact on the developing brain of constraining the unaffected limb and the effects of intense practice on this immature brain require careful study. If true recovery as defined by Krakauer 17 is achieved, then undamaged brain regions may be recruited to support the developing skills. However, this recruitment in the developing brain must be understood in light of future neurobiological development. Moreover, compensatory strategies might be adopted that would allow for improved functional use of the UE, with potentially little impact on the underlying developmental neurobiology. Although the majority of pediatric studies emphasize the effects of CIMT on improving UE function, there remains a significant gap in our understanding of the implications of CIMT for recovery or compensation in the developing child. Charles and Gordon 5 reviewed 15 studies on the efficacy of CIMT or forced-use therapy in children with either traumatic brain injury (TBI) or CP and suggested that both CIMT and forced-use therapy were effective for improving UE function. However, they did not systematically appraise the validity of the studies or compute evidence-based statistics. Evidence-based statistics refer to the treatment effects calculated from the studies, such as effect size (ES) or control and experimental event rate. A narrative review by Taub et al 8 discussed the origin and mechanism of pediatric and adult CIMT, but did not systematically review each study or compute relevant statistics. A Cochrane review 6 included only 3 randomized controlled trials (RCTs) and excluded studies that were considered less rigorous. In this review, we add to the validity of previous reviews and extend the scope of review by: (1) appraising only research on children with hemiplegic CP (previous reviews included subjects with TBI and children with late-onset stroke); (2) including all published studies regardless of design; (3) applying levels of evidence and validity scores for all studies; (4) calculating the evidence-based statistics, as possible; and (5) applying the International Classification of Functioning, Disability and Health 18 (ICF) levels to all outcome measures. The ICF includes the levels of body functions and structure, activity, and participation 19 and provides a common language to express outcomes across studies that used different measures. Although in the Cochrane review 6 the ICF was used to determine inclusion criteria for studies, it was not used to determine the levels of the outcomes. The specific objectives of this review are: (1) to investigate whether CIMT is supported with valid research of its effectiveness and (2) to identify key characteristics of the child and intervention protocol associated with effects of CIMT. November 2009 Volume 89 Number 11 Physical Therapy f 1127

3 Table 1. Levels of Evidence of Studies a Grade of Recommendation Level of Evidence Method Objective 1: Investigate Whether CIMT Is Supported With Valid Research of Its Effectiveness Data sources and study selection. We searched MEDLINE (1966 through March 2009), Entrez PubMed (1966 through March 2009), EMBASE (1980 through March 2009), CINAHL (1982 through March 2009), PsychINFO (1887 through March 2009), and Web of Science (1900 through March 2009) using the key words hemiplegic, cerebral palsy, constraint-induced movement therapy, and forced-use therapy. References found in other Description A 1a Systematic review with homogeneity of RCT 1b Individual RCT with narrow confidence interval No. of Studies 1c All or none None B 2a Systematic review of cohort studies (with homogeneity) None 2b 3a Individual cohort study (including lowquality RCT; eg, 80% follow-up) Systematic review (with homogeneity) of case-control studies 3b Individual case-control study C 4 Case series (and poorquality cohort and case-control studies) D 5 Expert opinion without explicit critical appraisal on physiology, bench research, or first principles publications also were included, as appropriate. Articles not written in English were excluded. Included studies met the following criteria: (1) participants were children with hemiplegic CP (younger than 18 years of age), (2) CIMT or forced-use therapy was used for intervention, and (3) outcome measures related to the effects of CIMT or forced-use therapy. Twenty-one intervention studies and 2 systematic reviews met the inclusion criteria (Tab. 1). 6,7 The level of evidence ranged from 1a to 5 (Tab. 1). 20 Of the 21 intervention Validity Score (Out of 16 or 11) Study 2 Hoare et al, ,b 2 11/16 10/16 5 8/16 7/16 9/16 9/16 8/16 None 6 5/11 7/11 Hoare et al, ,b Taub et al, DeLuca et al, Willis et al, Eliasson et al, Sung et al, Gordon et al, Charles et al, Crocker et al, Eliasson et al, Naylor et al, Charles et al, Dickerson et al, Wallen et al, Charles et al, Glover et al, Pierce et al, DeLuca et al, Sutcliffe et al, Fergus et al, Martin et al, Cope et al, a Levels of evidence are based on Sackett and colleagues description of levels of evidence and grades of recommendations. 20 RCTrandomized controlled trial. b The Cochrane Review 7 is a summarized version of the Cochrane Systematic Review. 6 studies, 5 were RCTs, were nonequivalent pretest-posttest control group designs, 26,27 3 were onegroup pretest-posttest designs, were single-subject research designs (SSRDs), and 8 were case reports (Tabs. 2, 3, 4, 5, and 6). The results from the study by Juenger et al 42 were not included in the tables of this review due to the age of the subjects (range10 30 years). Assessing validity of studies. The validity of RCTs and nonequivalent pretest-posttest control group de f Physical Therapy Volume 89 Number 11 November 2009

4 signs (ie, the subjects are not assigned to groups randomly) 43 using a between-group comparison was evaluated by 2 reviewers not blinded to authors or journals. We used the scoring protocol developed by Kwakkel et al 44 and Cambach et al 45 containing 16 items, each scored 0/1, which measured study validity for the following methodological categories: (A) randomization, (B) matching, (C) blinding procedure, (D) dropouts and intention-to-treat analysis, (E) characteristics of measurement instruments, (F) control of cointerventions, (G) comparability of group characteristics, and (H) control for dose of therapy (Appendix 1). The protocol was recommended by the Potsdam standards 46 and other investigators 47,48 to identify the potential confounders for the ES of individual studies. Because the scoring protocol was designed for RCTs and nonequivalent pretest-posttest control group designs, we modified the validity scoring for the studies with one-group pretest-posttest designs by deleting categories (D) and (G) (Appendix 2). These studies were scored 0/1 on an 11-point scale that measured aspects of: (A) study population, (B) design, (C) blinding procedure, (D) measurement instruments, (E) control of cointerventions, (F) control for dose of therapy, and (G) appropriate statistical analysis. The kappa coefficient was used to measure scoring agreement between raters. Kappa values for agreement between 2 raters have been categorized as poor (.00), slight (.01.20), fair (.21.40), moderate (.41.60), substantial (.61.80), and almost perfect ( ). 49 Nine of these 21 studies were scored for internal validity (Tabs. 2, 3, and 4) using the scoring protocol for group designs and included 173 subjects ,30 Seven of the 9 studies were scored out of 16 points, and 2 studies were scored out of 11 Table 2. Summary of -Induced Movement Therapy and Forced-Use Therapy in Children With Hemiplegic Cerebral Palsy Randomized Controlled Trial Design a Effect Size or Effect/Direction of Change (/) ICF Levels: Outcome Measures Length of Study Total Amount of Intervention (hr) Frequency and Duration of Intervention Sessions Hours/Day of Type of Age of Subjects Validity Scores (Out of 16) Level of Evidence Study 8 mo PDMS mo Cast 24 PT, OT (no detailed information about time) 2b Willis et al, (N25) Large Large mo Cast 24 PT, OT (6 hr/d for 21 d) mo EBS PMAL TAUT 1b Taub et al, (N18) No effect Small Small or medium 6 6 wk BBT EDPA WeeFIM 9 96 mo Cast 2 OT (ADL) (1 hr/wk for 6 wk) 2b Sung et al, (N31) Body structure No or small Medium Small or large Medium Medium Large y Sling 6 PT, OT (6 hr/d for 10 d) 60 6 mo Body structure Grasp strength TPD MAS J-T Test BOTMP CFUS 2b Charles et al, (N22) mo Cast 24 PT, OT (6 hr/d for 21 d) wk QUEST PMAL EBS 1b DeLuca et al, (N18) a Gray cellnonapplicable item for the study. In the Effect Size column, direction of change is provided when effect size has not been calculated. ICFInternational Classification of Functioning, Disability and Health, PTphysical therapy, OToccupational therapy, PDMSPeabody Developmental Motor Scales, EBSEmerging Behavior Scale, PMALPediatric Motor Log, TAUTToddler Arm Use Test, BBTBox and Block Test, EDPAErhardt Developmental Prehension Assessment, Wee-FIMFunctional Independence Measure for Children, TPD2-point discrimination, MASModified Ashworth Scale, J-T TestJebsen-Taylor Test of Hand Function, BOTMPBruininks-Oseretsky Test of Motor Proficiency, CFUSCaregiver Functional Use Survey, QUESTQuality of Upper Extremity Skills Test. November 2009 Volume 89 Number 11 Physical Therapy f 1129

5 Table 3. Summary of -Induced Movement Therapy and Forced-Use Therapy in Children With Hemiplegic Cerebral Palsy Nonequivalent Pretest-Posttest Control Group Design a Level of Evidence Study Validity Score (Out of 16) Age of Subjects Type of Hours/Day of Frequency and Duration of Intervention Sessions Total Amount of Intervention (hr) Length of Study ICF Levels: Outcome Measures Effect Size 2b Eliasson et al, (N41) mo Splint 2 Functional activities (2 hr/d for 2 mo) mo AHA Large 2b Gordon et al, (N20) y Sling 6 PT, OT (6 hr/d for 10 d) 60 6 m Body structure: Grasp strength TPD MAS J-T Test BOTMP CFUS Body structure No or small b No or small b No or small b Large b Large b Large b a ICFInternational Classification of Functioning, Disability and Health, PTphysical therapy, OToccupational therapy, AHAAssisting Hand Assessment, TPD2-point discrimination, MASModified Ashworth Scale, J-T TestJebsen-Taylor Test of Hand Function, BOTMPBruininks-Oseretsky Test of Motor Proficiency, CFUSCaregiver Functional Use Survey. b The effect size was from the authors report, which did not provide mean, standard deviation, or 95% confidence interval values. The calculation of 95% confidence interval and examination of significance were not applicable for this study. Table 4. Summary of -Induced Movement Therapy and Forced-Use Therapy in Children With Hemiplegic Cerebral Palsy One-Group Pretest-Posttest Designs a Level of Evidence Study Validity Score (Out of 11) Age of Subjects Type of Hours/Day of Frequency and Duration of Intervention Sessions Total Amount of Intervention (hr) Length of Study ICF Levels: Outcome Measures Effect/Direction of Change (/) 4 Eliasson et al, (N9) y Splint 7 Daily recreational activity (7 hr/d for 10 d) 70 5 mo Body structure Grip strength BOTMP AMPS J-T Test Body Structure 4 Charles et al, (N8) y Sling 6 PT, OT (6 hr/d for 10 d) mo J-T Test BOTMP CFUS 4 Wallen et al, (N10) 6 96 mo Mitt 2 Play activities (2 hr/d for 8 wk) mo Body structure Tardieu Scale AHA MAUULF PMAL Participation COPM GAS Body structure Participation a Gray cellnonapplicable item for the study. ICFInternational Classification of Functioning, Disability and Health, PTphysical therapy, OToccupational therapy, BOTMPBruininks-Oseretsky Test of Motor Proficiency, AMPSAssessment of Motor and Process Skills, J-T TestJebsen-Taylor Test of Hand Function, CFUSCaregiver Functional Use Survey, AHAAssisting Hand Assessment, MAUULFMelbourne Assessment of Unilateral Upper Limb Function, PMALPediatric Motor Log, COPMCanadian Occupational Performance Measure, GASgoal attainment scaling f Physical Therapy Volume 89 Number 11 November 2009

6 points. 28,30 This was a sufficient number of studies to achieve substantial interrater reliability. The remaining 11 studies were not scored for internal validity because the items from the scoring protocol did not apply to the study designs (SSRD and case report) (Tabs. 5 and 6). Case reports were systematically reported as the results of clinical practice without the application of inferential statistics. To our knowledge, there is no available validity scoring for this design. Of note is that 3 of the RCTs were assigned level 2b due to the loss of participants (ie, less than 80% in the follow-up phase), 21 lack of statistical control at baseline, 23 or no reports regarding the dropouts in the trial or the lack of blinding of outcome assessors 25 (Tabs. 1 and 2). Evidence-based statistics. If means and standard deviations were included in the published studies, ES was calculated using the Cohen d statistic. 50 This ES formula is readily accepted in the meta-analysis literature and calculated to determine the overall effect of treatment. It is expressed by the following equation: X T X C /sd pooled where X T is the mean result for the treatment group, X C is the mean result in the control group, and sd pooled is the pooled standard deviation. Moreover, when sufficient data (ie, number or percentage improved in each group) were provided in the study, treatment effects were calculated and presented, including the control event rate, experimental event rate, number needed to treat, absolute benefit increase, and relative benefit increase. 20 Table 5. Summary of -Induced Movement Therapy and Forced-Use Therapy in Children With Hemiplegic Cerebral Palsy Single-Subject Research Designs a Effect/Direction of Change (/) ICF Levels: Outcome Measures Length of Study Total Amount of Intervention (hr) Frequency and Duration of Intervention Sessions Hours/Day of Type of Age of Subjects Validity Scores (Out of 16 or 11) Study Design Level of Evidence Study mo PDMS-F A-B-A 2 y Splint 10 OT, ST, PT (1.5 hr/wk for 3 wk) 4 Crocker et al, (N1) wk QUEST A-B-A mo Adult held 1 Functional and play activities (1 hr/d for 4 wk) 4 Naylor et al, (N9) d Behavioral Observation (eg, reach, grasp) A-B-A 2 y Splint 10 OT (6 hr/d for 21 d) 4 Dickerson et al, (N1) a Gray cellnonapplicable item for the study. ICFInternational Classification of Functioning, Disability and Health, OToccupational therapy, STspeech therapy, PTphysical therapy, PDMS-FPeabody Developmental Motor Scales Fine Motor, QUESTQuality of Upper Extremity Skills Test. November 2009 Volume 89 Number 11 Physical Therapy f 1131

7 Table 6. Summary of -Induced Movement Therapy and Forced-Use Therapy in Children With Hemiplegic Cerebral Palsy Case Report Design a Level of Evidence Study Validity Scores (Out of 16 or 11) Age of Subjects Type of Hours/Day of Frequency and Duration of Intervention Sessions Total Amount of Intervention (hr) Length of Study ICF Levels: Outcome Measures Effect/Direction of Change (/) 5 Charles et al, (N3) 9 13 y Sling 6 Functional and play activities (6 hr/d for 14 d) 64 6 mo Body structure Grasp strength TPD J-T Test Body structure 5 Glover et al, (N2) 19 and 39 mo Splint, cast 24 PT, OT (2 hr/d for 2 wk) 28 6 wk Clinical observation Parent report Clinical observation 5 Pierce et al, (N1) 12 y Mitt 24 PT, OT (4 hr/wk for 3 wk) 12 8 mo Body structure: Grip strength WMFT AMPS Body structure 5 DeLuca et al, (N1) 15 mo Splint, cast 24 PT, OT (6 hr/d for 36 d) mo and 1 wk PDMS-F DDST PMAL TAUT 5 Sutcliffe et al, (N1) 8 y Cast 24 OT (1 hr/wk for 3 wk) 3 6 mo Body structure fmri PMAL QUEST AHA Participation COPM Body structure Participation 5 Martin et al, (N1) 35 mo Cast 7.5 OT (4 hr/d for 12 d) 48 3 mo Body structure Grip and pinch strength MAUULF PEDI Self-Care Participation COPM Body structure Participation 5 Cope et al, (N1) 12 mo Cast 24 PT, OT (8 hr/wk for 2 wk) 16 6 mo PDMS PMAL TAUT Knox Parent Questionnaire 5 Fergus et al, (N1) 13 mo Mitt 6 hr/d for 3 wk, then5wkof decreasing by 1 hr/wk PT, OT (6 hr/d for 3 wk, then 5 wk of decreasing by 1 hr/wk) mo Modified PMAL, interview a Gray cellnonapplicable item for the study. ICFInternational Classification of Functioning, Disability and Health, PTphysical therapy, OToccupational therapy, TPD2-point discrimination, J-T TestJebsen-Taylor Test of Hand Function, WMFTWolf Motor Function Test, AMPSAssessment of Motor and Process Skills, PDMS-FPeabody Developmental Motor Scales Fine Motor, DDSTDenver Developmental Screening Tool, PMALPediatric Motor Log, TAUTToddler Arm Use Test, QUESTQuality of Upper Extremity Skills Test, AHAAssisting Hand Assessment, COPMCanadian Occupational Performance Measure, MAUULFMelbourne Assessment of Unilateral Upper Limb Function, PEDIPediatric Evaluation of Disability Inventory f Physical Therapy Volume 89 Number 11 November 2009

8 Objective 2: Identify Key Characteristics of the Child and Intervention Protocol Associated With Effects of CIMT In addition to assessing the validity of the studies, we recorded: (1) inclusion and exclusion criteria, (2) type of constraint, (3) duration of constraint, (4) intervention duration, (5) study duration, and (6) outcomes measures at the ICF levels. We included the definitions of these features for each study and the rationale for choosing the definitions. We believe that examining the rationale for the choices for each component of CIMT will give the reader insight into which choices might warrant inclusion in future studies and how the specific children in their clinical caseloads compare with research participants. Results Reliability of Scoring Each RCT and nonequivalent pretestposttest control group study was scored by 2 reviewers on the 16 validity items, totaling 112 items scored across 7 studies Reviewers agreed on 82 of the 112 items scored, for a kappa coefficient of.66. Reviewers disagreements were resolved easily with discussion. There were 22 items scored for the 2 studies that used a one-group, pretest-posttest design. 28,30 Reviewers agreed on 20 of the 22 items scored, for a kappa coefficient of.76. Objective 1: Investigate Whether CIMT Is Supported With Valid Research of Its Effectiveness Validity scores and study characteristics are shown in Tables 2 through 6. Validity scoring. The 7 RCTs and nonequivalent pretest-posttest control group designs had validity scores between 7 and 11 of the maximum 16 points (Tabs. 2 and 3). The 2 one-group designs had validity scores between 5 and 7 of the maximum 11 points 28,30 (Tab. 4). Effect size. Only 4 studies provided means and standard deviations to enable computation of Cohen d and the 95% confidence interval (CI), for a total of 12 outcome measures 22,25,26,51 (Tab. 7). Cohen 49 defined ES as no effect (d0.2), small (0.2d0.5), medium (0.5d0.8), and large (d0.8). We also included in Table 7 the ES (eta values) published in the studies by Charles et al 23 and Gordon et al. 27 There were 5 ES values at the body functions and structure level, but only 1 was significant (ie, the CI did not include zero), and there were 14 ES values at the activity level, with 4 being significant (Tab. 7). All significant ES values were medium to large (ie, d ). Of particular note is the significant treatment effect reported by Eliasson et al 26 favoring the CIMT on the Assisting Hand Assessment (AHA) immediately posttreatment (large ES) and sustained at 6 months (medium ES). The AHA is currently the only reliable and valid measure used in pediatric CIMT studies that measures functional use of the affected UE in bimanual tasks. 6 The norm-referenced tests evaluating fine motor development (eg, Peabody Developmental Motor Scale) were used in several studies. 28,31,38 However, these tests do not measure the functional use of the affected hand in bimanual tasks. Some tests (eg, Pediatric Motor Log [PMAL], Emerging Behavior Scale [EBS]) were developed specifically to examine the effects of CIMT, but the psychometric properties of these tests have not been established. A significant treatment effect favoring CIMT with a large ES was reported by Charles et al 23 on the frequency of use portion of the Caregiver Functional Use Survey. Taub et al 22 also found significant and large ES values for amount of use on the PMAL immediately posttreatment and 3 weeks posttreatment and for the EBS. Only the study by Eliasson et al 26 provided sufficient data for calculating control event rate (0.78), experimental event rate (1), number needed to treat (5), absolute benefit increase (0.22), and relative benefit increase (0.28). The control event rate and experimental event rate are determined in order to compute number needed to treat, which is the number of children who would need to be treated to achieve one additional favorable outcome. 20 To increase the score on the AHA for children with hemiplegic CP, 5 children with hemiplegic CP must be treated with CIMT (2 hours of constraint per day and 2 hours of intervention per day for 2 months). The absolute benefit increase is the difference in rates of positive outcomes between experimental and control subjects in the same trial, 20 and it was 22% in the study by Eliasson et al. 26 The absolute benefit increase is divided by the control event rate to give the relative benefit increase, which is the proportional increase in rates of positive outcomes between experimental and control subjects in the same trial (28%). 20 Direction of change results. For the 16 studies for which ES could not be calculated, outcome measures were primarily at the activity level and supported positive change immediately and up to 12 months postintervention in fine motor and functional activities (Tabs. 2, 3, 4, 5, and 6). Change at the body structures and function level was mostly positive. Three studies measured changes at the participation level, with positive changes reported. 29,39,41 November 2009 Volume 89 Number 11 Physical Therapy f 1133

9 Table 7. Effect Sizes and 95% Confidence Intervals (CI) of 5 Studies From the Outcome Measures Based on 3 ICF Levels a Effect Size (95% CI) Study Outcome Measure Posttreatment 1-Week Posttest 3-Week Posttest 1-Month Posttest 5-Month Posttest 6-Month Posttest Charles et al, (N22) Body structure Grasp strength 0.05 (0.88 to 0.79) 0.42 (1.24 to 0.44) 0.28 (0.57 to 1.11) 2-point discrimination b 0.74 (0.15 to 1.57) 0.70 (0.19 to 1.53) 0.61 (0.26 to 1.44) MAS-Shoulder b 1.69 (2.59 to 0.66) c 0.60 (1.43 to 0.28) 1.18 (2.03 to 0.23) c MAS-Elbow b 0(0.84 to 0.84) 0.15 (0.69 to 0.98) 0.46 (1.29 to 0.40) MAS-Wrist b 0.45 (1.28 to 0.41) 0.18 (1.01 to 0.66) 0.50 (1.32 to 0.37) J-T Test b 0.47 d 0.11 (0.94 to 0.73), 0.32 d 0.04 (0.80 to 0.88) 0.11 (0.94 to 0.73) BOTMP 0.55 d 0.55 (0.32 to 1.38) 0.49 (0.37 to 1.32) 0.13 (0.71 to 0.97) CFUS-How Frequently 1.15 (0.21 to 2.00) c 1.26 (0.31 to 2.13) c 0.96 (0.05 to 1.80) c CFUS-How Well 0.20 (0.65 to 1.03) 1.48 (0.48 to 2.35) c 0.90 (0.01 to 1.74) Gordon et al, (N38) J-T Test b 0.43 d BOTMP 0.41 d CFUS-How Frequently 0.37 d CFUS-How Well 0.49 d Sung et al, (N31) Box and Block Test 0.15 (0.57 to 0.86) EDPA 0.37 (0.36 to 1.08) WeeFIM-Motor 0.38 (0.35 to 1.09) WeeFIM-Self-Care 0.60 (0.15 to 1.31) WeeFIM-Communication 0.23 (0.94 to 0.49) WeeFIM-Cognitive 0.19 (0.90 to 0.53) Eliasson et al, (N41) Assisting Hand Assessment 1.16 (0.46 to 1.78) c 0.72 (0.10 to 1.37) c Taub et al, (N18) EBS 1.28 (0.21 to 2.22) c PMAL-Amount of Use 1.61 (0.48 to 2.59) c 1.36 (0.28 to 2.31) c PMAL-Quality of Use 0.76 (0.23 to 1.68) 0.70 (0.28 to 1.62) a ICFInternational Classification of Functioning, Disability and Health, MASModified Ashworth Scale, J-T TestJebsen-Taylor Test of Hand Function, BOTMPBruininks-Oseretsky Test of Motor Proficiency, CFUSCaregiver Functional Use Survey, EDPAErhardt Developmental Prehension Assessment, Wee-FIMFunctional Independence Measure for Children, EBSEmerging Behavior Scale, PMALPediatric Motor Log. b Lower scores on the test correspond to better performance. c The CI does not include zero, which corresponds to a significant treatment effect. d The numbers are from the authors reports. The authors used eta square for calculating effect size instead of Cohen d. Eta square values of 0.01 to 0.03, 0.06 to 0.09, and 0.14 represent small, medium, and large effect sizes, respectively. Cohen 50 defined effect size as no effect (d.02), small (0.2d0.5), medium (0.5d0.8), and large (d0.8) f Physical Therapy Volume 89 Number 11 November 2009

10 Objective 2: Identify Key Characteristics of the Child and Intervention Protocol Associated With Effects of CIMT Common features of studies. The following features are summarized by study in Tables 2 through 6: (1) inclusion and exclusion criteria, (2) type of constraint, (3) duration of constraint, (4) intervention duration, (5) study duration, and (6) outcome measures at the ICF levels. Inclusion and exclusion criteria. Hemiplegic CP was the only consistent criterion across all studies. There was heterogeneity across the studies regarding the inclusion criteria (ie, the required minimal wrist range of motion and UE movement abilities). Four studies required subjects to have at least 20 degrees of active extension at the wrist and 10 degrees of active extension at the metacarpophalangeal joints. 23,27,30,41 These were the same criteria for inclusion used in the studies of adults with stroke, specifically in the EX- CITE trial. 3,14,52 54 One study recruited children who were able to actively extend the affected wrist at least 10 degrees with finger extension and lift the arm from resting on a table and place it on a 13-cm-high box, 35 and 2 studies required that subjects be able to use the affected extremity only as a gross functional assist in tasks (eg, reaching) or have some active wrist or finger extension of the affected UE. 29,31 Movement ability was not an inclusion criterion in the other 14 studies. Three studies excluded children with severe paralysis of the UE. 23,25,27 Only 4 studies addressed the issue of sensory impairments and excluded children with major sensory disorders or vision problems. 23,27,30,31 Positive outcomes were reported in all 4 studies following implementation of CIMT. No specific criteria for sensory abilities were included in the other studies. Age and CP severity varied across studies. The association between subject characteristics and the effects of CIMT has not been systematically evaluated through research. Type of constraint. There was wide variability in the type of constraint, ranging from a full-arm cast to gentle parental restraint (Tabs. 2, 3, 4, 5, and 6). Nine studies included a cast as the constraint, 21,22,24,25,34,36,38,39,41 referencing the literature on adults with stroke in justifying their selection. Five of the studies used bivalved casts, 22,24,25,38,41 and Cope et al 34 used a nonremovable cast and determined it was the best method of restraint for the particular child in their study. Willis et al 21 did not bivalve the cast and reported that several subjects dropped out of the study due to the irritability from casting. Two studies did not provide information regarding the cast. 36,39 Wallen et al 29 and Pierce et al 37 selected a fabric mitt, with no specified reason provided. Fergus et al 40 reported that using a soft, removable mitt provided a simple, easily manipulated constraint. Researchers in 5 studies chose a splint, 26,28,31,33,36 citing previous literature. Glover et al 36 used 2 types of restraint: cast and splint. Eliasson et al 26 reported that the splint restricted use of the unaffected hand but allowed for necessary balance reactions. Four studies used a sling strapped to the child s trunk to prevent bimanual tasks. 23,27,30,35 Gordon and colleagues choice 27 was derived from positive results of their pilot studies. The parent gently restrained the child s unaffected extremity in the study by Naylor and Bower. 32 Duration of constraint. duration ranged from 1 to 24 hours per day (Tabs. 2, 3, 4, 5, and 6). The unaffected UE was constrained for at least 10 hours per day in 11 studies. 21,22,24,25,31,33,34,36 39 In 9 of these studies, the subjects wore the restraint for 24 hours. 21,22,24,25,34,36 39 Two studies used a splint continuously except for bathing, sleeping, and short rests. 31,33 The remaining 10 studies used a constraint only during treatment ,32,35,40,41,51 Treatment duration in these studies varied from 1 to 7.5 hours per day. The authors justified the duration of constraint based on previous studies of adult and pediatric CIMT literature and in order to create a child-friendly intervention. Martin et al 41 casted children s affected UE for 4 hours during therapy and for an additional 3 to 5 hours per day without providing a rationale. Eliasson et al 26 had children wear a glove for 7 hours, 28 and Eliasson and collegues 26 and Wallen et al 29 had children wear a glove for 2 hours per day. Intervention duration. Intervention duration ranged from 1 hour per week to 7 hours per day (Tabs. 2, 3, 4, 5, and 6). Ten studies included 6 to 7 hours of therapy per day ,27,28,30,33,35,38,40 This high-intensity intervention was based on the interventions used in studies of adult subjects. 2,55 58 DeLuca et al 37 chose this intensity, stating that the prior pediatric literature on CIMT had yet to match the intensity of the studies of adults with stroke. Furthermore, the authors noted that the intensity used for adults with stroke was exceeded in order to maximize cortical reorganization, although no supporting data in a pediatric population were provided. Treatment in the studies by Eliasson et al 26 and Wallen et al 29 consisted of 2 hours of therapy per day within the child s usual environment (ie, at home or in the preschool setting). The authors noted that a rich natural environment was important to facilitate the learning process. Glover et al 36 applied 2-hour sessions of occupational therapy or physical therapy as weekday treatment without pro- November 2009 Volume 89 Number 11 Physical Therapy f 1135

11 viding a rationale for intervention duration. Naylor and Bower 32 also provided therapy within the home setting for 1 hour per day, stating that treatment in the child s natural environment increased adherence among parents and teachers, thus decreasing the need for an intense training schedule in clinical settings. Pierce et al 37 included two 1-hour sessions of physical therapy and two 1-hour sessions of occupational therapy per week (ie, a total of 4 hours of treatment per week), noting that the high intensity of treatment in studies by Taub and colleagues 59,60 was not feasible in a managed care environment. Similarly, Cope et al 33 applied a total of 8 hours of treatment per week. Four studies provided little intervention (eg, 1 hour of physical therapy or occupational therapy per week) and assessed the effectiveness of constraint alone. 20,24,30,38 The optimal durations of constraint and intervention have not been systematically evaluated. Study duration. Study durations in 19 of the studies ranged from 6 weeks to 18 months (Tabs. 2, 3, 4, 5, and 6), 21 24,26 35,37 41 with 2 studies without a follow-up period, only evaluation immediately after intervention. 25,36 Eight of the 19 studies used a 6-month intervention, 23,26,27,29,31,34,35,39 and the intervention was of 8 months duration in 4 studies. 21,22,30,37 Positive effects were demonstrated in most studies up to 6 to 8 months after intervention. Outcome measures. Eight studies used outcome measures with multiple ICF levels (Tabs. 2, 3, 4, 5, and 6). 23,27 29,35,37,39,41 Eight studies had outcome measures at the body functions and structure level, 23,27 29,35,37,39,41 including functional magnetic resonance imaging (fmri), muscle tone (velocity-dependent resistance to stretch), sensation, grip strength (forcegenerating capacity), 2-point discrimination, and pinch gauge, and 20 studies measured change at the activity level ,37 41 There was little consistency in these 20 studies as to which activities were assessed. Only 3 studies 26,29,39 used the AHA. In addition, 3 studies measured outcomes at the participation level by using the Canadian Occupational Performance Measures (COPM). 29,39,41 A systematic evaluation of shared and valid outcome measures is critical to understanding the impact of CIMT. Discussion The Cochrane review, 6 which represents the highest level of research evidence (ie, 1a), determined that there was insufficient evidence to either support or refute the use of CIMT for children with hemiplegic CP and that more-rigorous research was needed and specifically research with valid outcome measures. However, if we consider studies at all levels of evidence, as we did in this systematic review, there is positive support for the use of CIMT to improve the frequency of use of the UE for children with hemiplegic CP. The strongest evidence for this improvement comes from the 3 RCTs and 1 nonequivalent pretestposttest study, 27 but an increased frequency of use of the UE following CIMT also was supported in studies with less-rigorous designs. 30,38 Taub et al 22 and Charles et al 23 demonstrated large ES values for the frequency of use, with the findings from the study by Charles et al 23 extended to 6 months posttherapy. Of particular note is that both Charles et al 23 and Gordon et al 27 had only half of the total time for intervention (60 hours) and a fourth of the time wearing the restraint (6 hours per day) in comparison with the study by Taub et al 22 (total intervention time120 hours; restraint time24 hours per day). It is not apparent from the available literature whether even less intervention time and less time in restraint also could have a positive and large impact on the frequency of use of the UE. This remains an empirical question. Although improving the frequency of use of the affected extremity is an excellent starting point, improving the quality of use, particularly for functional activities, also is an important goal. Charles et al 23 reported a large and significant ES for the quality of use of the UE on the Caregiver Functional Use Survey (CFUS), a caregiver questionnaire of UE skills, although subjects in this study showed these effects only at 1 month posttreatment and not at 1 week or 6 months posttreatment. Subjects in the study by Gordon et al 27 showed a large ES on the CFUS 1 week posttreatment. Taub et al 22 showed a significant and large ES following CIMT on the EBS, which is a scale that quantifies the number of different movement patterns that are observed. This same trend, but without available ES, was demonstrated by DeLuca et al 24 on the EBS. The most compelling data on the positive impact of CIMT comes from the study by Eliasson et al 26 with a large and sustained ES for the AHA. 6,61,62 The AHA measures the frequency of use of the affected UE but also captures aspects of quality of use. 62 This study, 26 however, had a rather low validity score (7 of 16) because it was not an RCT and the control of cointervention was not adequate. Of importance is that the constraint was worn for only 2 hours per day and intervention was 2 hours per day during the constraint period, but the treatment was implemented over a 60-day period. This again raises the question of what is the appropriate dose for intervention and constraint. The critical threshold for intensity in terms of both number of hours per day and number of weeks that constitute an adequate dose cannot be determined from the available research f Physical Therapy Volume 89 Number 11 November 2009

12 We cannot identify key characteristics of the child and intervention protocol associated with effects of CIMT. The reviewed studies have few common features regarding the method. However, based on the current findings, all future CIMT studies could calculate power a priori in order to have a sufficient sample size. Although no a priori power calculations were reported or provided in the reviewed studies, 22,23,25,26 the actual statistical power can be computed from the numbers of subjects and the reported means and standard deviations in the studies. The results show that outcomes at the body functions and structure and activity levels have power between 50% and 70% and 80% and 95%, respectively. Thus, for example, to provide 80% statistical power and a medium to large ES (d0.8) at the activity level, a CIMT RCT design study would require at least 20 subjects in the treatment group and 20 subjects in the control group. Impact on the Developing Brain As mentioned in the introduction to this review, Krakauer 17 described true recovery in contrast to compensation in motor learning. True recovery involves recruiting the undamaged brain regions to control the muscles previously used for a movement. Compensation, in contrast, involves the use of different muscle groups to achieve the same movement goal. 17 The use of inappropriate compensatory strategies may limit development or recovery in children or adults after neurological impairments. 63,64 The effects of CIMT on the brain that result in either true recovery or compensation require more-detailed analysis, particularly for the developing brain. The impact of CIMT on undamaged brain regions during development remains largely unknown, and the potential impact may differ with the stage of development during which CIMT is applied. 5 To our knowledge, only 2 studies have investigated the potential cortical changes from CIMT in children. 39,42 Sutcliffe et al 39 reported the effects of cortical reorganization in an 8-year-old child with hemiplegic CP following a 3-week CIMT intervention (24 hours of constraint per day, 1 hour of therapy per week). The results of fmri indicated increased bilateral cortical activation in the sensorimotor cortex of the affected UE after therapy and a shift in laterality from the ipsilateral hemisphere to the contralateral hemisphere. Modified CIMT led to increased activity of the contralateral somatosensory cortex. Sutcliffe and colleagues 39 also found increased cortical activation in the ipsilateral motor cortex and contralateral somatosensory cortex after therapy. These cortical changes were associated with improvements in clinical measures and persisted at 6 months. Juenger et al 42 also reported increased fmri activations in the primary sensorimotor cortex of the lesioned hemisphere in 3 children with congenital hemiparesis (age range12 16 years) after a 12-day CIMT intervention (10 hours of constraint per day, 2 hours of therapy per day), suggesting that these results supported the mechanism of cortical reorganization for improvements of UE function after modified CIMT. Eyre et al 65 suggested that due to the loss of competition from the lesioned hemisphere in children with hemiplegic CP, the unlesioned hemisphere might have acquired bilateral organization. They found that transcranial magnetic stimulation of the unlesioned hemisphere evoked bilateral motor responses in 10 children with hemiplegic CP (median age11.5 years), demonstrating persistence of ipsilateral and contralateral corticospinal projections from the unlesioned hemisphere. There was only a sparse contralateral corticospinal projection from the lesioned hemisphere after transcranial magnetic stimulation. The organization of the corticospinal system in children with hemiplegic CP is characterized by the loss of motor responses of the lesioned side and emergence of bilateral responses from the unlesioned side. However, the authors did not specify whether the unlesioned hemisphere was the dominant hemisphere. Preliminary work by Kuhnke and colleagues 66 with adolescents with congenital hemiparesis suggests that individuals with preserved crossed cortical projections versus ipsilateral projections may have different outcomes from CIMT. Although both groups of subjects had significant improvements of UE function, subjects with preserved crossed cortical projections performed the relatively short tasks of the Wolf Motor Function Test more rapidly compared with those with ipsilateral projections. induced movement therapy might influence interhemispheric inhibition by reducing cortical activity in the unlesioned hemisphere (by constraining the unaffected UE), while increasing cortical activity in the lesioned hemisphere (by implementing the intensive intervention). 66 Nevertheless, it is still unclear whether similar results will be found in children with congenital hemiparesis. Thus, Hoare 67 recommends stratification on cortical-spinal organization for future studies of congenital hemiplegia. Future Research There have been insufficient studies systematically comparing the critical variables of duration and type of constraint and the duration and type of intervention using valid and reliable outcome measures. The future direction for CIMT studies is to develop further good-quality research with a priori power calculations to examine the different features of CIMT, including the effectiveness of the re- November 2009 Volume 89 Number 11 Physical Therapy f 1137

13 straint and the frequency and duration of the intervention sessions. Measures at the participation level of the ICF also should be included in future studies, as well as an elucidation of the relationships among ICF levels. Finally, there has been insufficient investigation of the potential central nervous system changes that may ensue from constraining the extremity of a developing child to support the use of prolonged constraint of an unaffected extremity. Research using CIMT should incorporate outcome measures of the effects on the developing brain to guide best physical therapist practice. Ms Huang and Dr Fetters provided concept/ idea/research design and project management. All authors provided writing, data collection and analysis, and consultation (including review of manuscript before submission). This article was received April 14, 2008, and was accepted July 15, DOI: /ptj References 1 Bonaiuti D, Rebasti L, Sioli P. The constraint induced movement therapy: a systematic review of randomised controlled trials on the adult stroke patients. Eura Medicophys. 2007;43: Taub E, Uswatte G, Pidikiti R. induced movement therapy: a new family of techniques with broad application to physical rehabilitation a clinical review. J Rehabil Res Dev. 1999;36: Wolf SL. Revisiting constraint-induced movement therapy: are we too smitten with the mitten? Is all nonuse learned? and other questions. Phys Ther. 2007;87: Winstein CJ, Prettyman MG. induced therapy for functional recovery after brain injury:unraveling the key ingredients and mechanisms. In: Baudry M, Bi X, Schreiber S, eds. Synaptic Plasticity. New York, NY: Marcel Dekker Inc; 2005: Charles J, Gordon AM. A critical review of constraint-induced movement therapy and forced use in children with hemiplegia. Neural Plast. 2005;12: Hoare BJ, Imms C, Carey L, Wasiak J. -induced movement therapy in the treatment of the upper limb in children with hemiplegic cerebral palsy: a Cochrane systematic review. Clin Rehabil. 2007;21: Hoare BJ, Wasiak J, Imms C, Carey L. -induced movement therapy in the treatment of the upper limb in children with hemiplegic cerebral palsy. Cochrane Database Syst Rev. 2007;2:CD Taub E, Griffin A, Nick J, et al. Pediatric CI therapy for stroke-induced hemiparesis in young children. Dev Neurorehabil. 2007; 10: Taub E. Movement in nonhuman primates deprived of somatosensory feedback. Exerc Sports Sci Rev. 1976;4: Taub E. Somatosensory deafferentation research with monkeys: implications for rehabilitation medicine. In: Ince LP, ed. Behavioral Psychology in Rehabilitation Medicine: Clinical Applications. Baltimore, MD: Williams & Wilkins; 1980: Wolf SL, Lecraw DE, Barton LA, Jann BB. Forced use of hemiplegic upper extremities to reverse the effect of learned nonuse among chronic stroke and head-injured patients. Exp Neurol. 1989;104: Ostendorf CG, Wolf SL. Effect of forced use of the upper extremity of a hemiplegic patient on changes in function. Phys Ther. 1981;61: Knapp HD, Taub E, Berman AJ. Movements in monkeys with deafferented limbs. Exp Neurol. 1963;7: Wolf SL, Winstein CJ, Miller JP, et al. Effects of constraint-induced movement therapy on upper extremity function 3 to 9 months after stroke: the EXCITE randomized clinical trial. JAMA. 2006;296: Taub E, Perrella PN, Miller EA, Barro G. Diminution of early environmental control through perinatal and prenatal somatosensory deafferentation. Biol Psychiatry. 1975; 10: Gordon AM, Charles J, Wolf SL. Methods of constraint-induced movement therapy for children with hemiplegic cerebral palsy: development of a child-friendly intervention for improving upper-extremity function. Arch Phys Med Rehabil. 2005;86: Krakauer JW. Motor learning: its relevance to stroke recovery and neurorehabilitation. Curr Opin Neurol. 2006;19: International Classification of Functioning, Disability and Health: ICF. Geneva, Switzerland: World Health Organization; Jette AM. The changing language of disablement. Phys Ther. 2005;85: Sackett DL, Straus SE, Richardson WS, et al. Evidence Based Medicine: How to Practice and Teach EBM. 2nd ed. New York, NY: Churchill Livingstone Inc; Willis JK, Morello A, Davie A, et al. Forced use treatment of childhood hemiparesis. Pediatrics. 2002;110: Taub E, Ramey SL, DeLuca SC, Echols K. Efficacy of constraint-induced movement therapy for children with cerebral palsy with asymmetric motor impairment. Pediatrics. 2004;113: Charles JR, Wolf SL, Schneider JA, Gordon AM. Efficacy of a child-friendly form of constraint-induced movement therapy in hemiplegic cerebral palsy: a randomized control trial. Dev Med Child Neurol. 2006; 48: DeLuca SC, Echols K, Law CR, Ramey SL. Intensive pediatric constraint-induced therapy for children with cerebral palsy: randomized, controlled, crossover trial. J Child Neurol. 2006;21: Sung IY, Ryu JS, Pyun SB, et al. Efficacy of forced-use therapy in hemiplegic cerebral palsy. Arch Phys Med Rehabil. 2005;86: Eliasson A-C, Krumlinde-Sundholm L, Shaw K, Wang C. Effects of constraintinduced movement therapy in young children with hemiplegic cerebral palsy: an adapted model. Dev Med Child Neurol. 2005;47: Gordon AM, Charles J, Wolf SL. Efficacy of constraint-induced movement therapy on involved upper-extremity use in children with hemiplegic cerebral palsy is not agedependent. Pediatrics. 2006;117: Eliasson A-C, Bonnier B, Krumlinde- Sundholm L. Clinical experience of constraint-induced movement therapy in adolescents with hemiplegic cerebral palsy: a day camp model. Dev Med Child Neurol. 2003;45: Wallen M, Ziviani J, Herbert R, et al. Modified constraint-induced therapy for children with hemiplegic cerebral palsy: a feasibility study. Dev Neurorehabil. 2008;11: Charles JR, Gordon AM. A repeated course of constraint-induced movement therapy results in further improvement. Dev Med Child Neurol. 2007;49: Crocker MD, MacKay-Lyons M, McDonnell E. Forced use of the upper extremity in cerebral palsy: a single-case design. Am J Occup Ther. 1997;51: Naylor CE, Bower E. Modified constraintinduced movement therapy for young children with hemiplegic cerebral palsy: a pilot study. Dev Med Child Neurol. 2005; 47: Dickerson AE, Brown LE. Pediatric constraint-induced movement therapy in a young child with minimal active arm movement. Am J Occup Ther. 2007;61: Cope SM, Forst HC, Bibis D, Liu XC. Modified constraint-induced movement therapy for a 12-month-old child with hemiplegia: a case report. Am J Occup Ther. 2008;62: Charles J, Lavinder G, Gordon AM. Effects of constraint-induced therapy on hand function in children with hemiplegic cerebral palsy. Pediatr Phys Ther. 2001;13: Glover JE, Mateer CA, Yoell C, Speed S. The effectiveness of constraint-induced movement therapy in two young children with hemiplegia. Pediatr Rehabil. 2002;5: f Physical Therapy Volume 89 Number 11 November 2009