Video Abstract

Video Abstract

Close modal
OBJECTIVES

With the Children with Hemiparesis Arm and Hand Movement Project (CHAMP) multisite factorial randomized controlled trial, we compared 2 doses and 2 constraint types of constraint-induced movement therapy (CIMT) to usual customary treatment (UCT).

METHODS

CHAMP randomly assigned 118 2- to 8-year-olds with hemiparetic cerebral palsy to one of 5 treatments with assessments at baseline, end of treatment, and 6 months posttreatment. Primary blinded outcomes were the assisting hand assessment; Peabody Motor Development Scales, Second Edition, Visual Motor Integration; and Quality of Upper Extremity Skills Test Dissociated Movement. Parents rated functioning on the Pediatric Evaluation of Disabilities Inventory-Computer Adaptive Test Daily Activities and Child Motor Activity Log How Often scale. Analyses were focused on blinded and parent-report outcomes and rank-order gains across all measures.

RESULTS

Findings varied in statistical significance when analyzing individual blinded outcomes. parent reports, and rank-order gains. Consistently, high-dose CIMT, regardless of constraint type, produced a pattern of greatest short- and long-term gains (1.7% probability of occurring by chance alone) and significant gains on visual motor integration and dissociated movement at 6 months. O’Brien’s rank-order analyses revealed high-dose CIMT produced significantly greater improvement than a moderate dose or UCT. All CIMT groups improved significantly more in parent-reported functioning, compared with that of UCT. Children with UCT also revealed objective gains (eg, 48% exceeded the smallest-detectable assisting hand assessment change, compared with 71% high-dose CIMT at the end of treatment).

CONCLUSIONS

CHAMP provides novel albeit complex findings: although most individual blinded outcomes fell below statistical significance for group differences, high-dose CIMT consistently produced the largest improvements at both time points. An unexpected finding concerns shifts in UCT toward higher dosages, with improved outcomes compared with previous reports.

What’s Known on This Subject:

Constraint-induced movement therapy (CIMT) is widely endorsed scientifically, despite large variations in treatments, particularly in dosage and constraint type. It is critically important to assess whether lower CIMT doses and a splint rather than a cast can produce gains above usual customary treatment (UCT).

What This Study Adds:

High-dose CIMT with a splint or cast produced a pattern of greater gains than that of UCT or a moderate dose at both posttreatment assessments (significant for rank-order multiple endpoints). Unexpectedly, UCT children displayed objective gains but not parent-reported improvement.

The Centers for Disease Control and Prevention estimate that cerebral palsy affects 1.5% to 4.0% of US livebirths1,2 ; ∼40% will develop hemiparesis.3  For children with hemiparetic cerebral palsy (HCP), constraint-induced movement therapy (CIMT) is consistently designated highly efficacious to improve arm-and-hand use.4,5  CIMT treatment protocols, however, vary widely in both dosage and constraint type, with uncertainty about the effects of these variations.6 

CIMT involves a high therapy dosage (eg, ≥3-hour sessions, 5 days per week, and ≥2 weeks), constraint of the nonhemiparetic upper extremity (UE), and operant conditioning and motor learning techniques to elicit and shape new skills.7,8  High CIMT doses are costly, however, and may be stressful. Theories of experience-driven neuroplasticity invoke a dose-response principle that higher doses induce larger brain and behavior changes.911  Concerning constraint, a full-time cast (never applied without administering active treatment) encourages using the hemiparetic UE but limits practice of bimanual skills. Alternatively, part-time constraint only during therapy sessions promotes bilateral activities practice outside sessions but may insufficiently increase use of the hemiparetic side. In Pediatrics, results were published from the first randomized controlled trial (RCT) of CIMT12  with 6-hour sessions for 20 days and novel use of a cast. Children showed large-effect size gains in acquiring new skills and using them in typical situations. Follow-up 6 months later revealed high maintenance of benefits.13 

Subsequently, in many RCTs, researchers reported benefits using lower dosages and a splint or mitt or glove part-time.4,5  Unfortunately, cross-study comparisons cannot identify which treatment components produce better outcomes because of varied patient populations and measured outcomes. Thus, we conducted a factorial RCT (the Children with Hemiparesis Arm and Hand Movement Project [CHAMP])14 to systematically compare specific combinations of dose and constraint. We did not know if the lower CIMT dose would produce improvements greater than usual care. Concerning constraint, we hypothesized both constraint types tested could be efficacious, each offering advantages and limitations.

CHAMP’s 3 sites (Charlottesville, VA; Columbus, OH; Roanoke, VA) recruited (January 2015 to December 2018; registration: NCT01895660) from clinics, programs, and Web sites, yielding 118 2-to-8 year olds with HCP, good health, and ability to follow directions. Exclusion criteria were uncontrolled seizures and/or receiving CIMT or botulinum toxin in the previous 6 months. CHAMP’s protocol was institutional review board–approved and monitored by an external data safety and monitoring committee.

CHAMP was a 2 × 2 factorial RCT with all groups assessed at baseline, end of treatment, and 6-months posttreatment. In the Consolidated Standards of Reporting Trials flow diagram (Fig 1), the 4 CIMT groups are identified: high dose with cast, high dose with splint, moderate dose with cast, and moderate dose with splint.

FIGURE 1

Consolidated Standards of Reporting Trials diagram for CHAMP 2 × 2 factorial RCT of variations in CIMT dose and constraint type.

FIGURE 1

Consolidated Standards of Reporting Trials diagram for CHAMP 2 × 2 factorial RCT of variations in CIMT dose and constraint type.

Close modal

CHAMP implemented the CIMT published protocol named ACQUIRE,15,16  on the basis of efficacy12,13,17,18  and clinical-practice effectiveness.16,19  A previous RCT revealed comparable benefits from 6-hour and 3-hour sessions, 5 days per week over 4 weeks.17,18  Accordingly, we tested the 3-hour dose (60 total hours) and a lower dose of 2.5-hour sessions, 3 days per week over 4 weeks (30 hours). This lower dose is similar to forms of “modified CIMT.”14  CHAMP also compares 2 constraints: a lightweight full-arm cast worn continuously and a part-arm splint worn just during treatment sessions (see protocol article for details14 ). Even with the cast, children can use their nonhemiparetic UE in bilateral activities (eg, crawling, holding a large ball).

Occupational or physical therapists received intensive instruction in ACQUIRE, particularly operant conditioning techniques to elicit and shape new UE skills. ACQUIRE sets individualized treatment goals with parents and emphasizes enjoyable play and self-help activities. Treatment occurs in home or homelike settings to promote generalization and maintenance of new skills. Parents participate in sessions weekly. Constraint is removed the last 3 days to promote bimanual skills. Finally, with a transfer package, posttreatment progress is encouraged.

We documented treatment fidelity via weekly video-recordings and daily treatment logs, applying standardized criteria.14  Therapists received feedback and additional training if needed.

No single, widely used assessment tool adequately captures the breadth of the bilateral and unilateral skills and functional outcomes ACQUIRE targets. Accordingly, we selected 3 primary blinded outcomes based on their distinctive domain relevance, psychometric properties, and sensitivity to change: (1) the Assisting Hand Assessment (AHA),2024  which rates use of the hemiparetic UE as a “helper” in bimanual play activities; (2) the modified Peabody Developmental Motor Scales, Second Edition, (PDMS-2)25,26  Visual Motor Integration (VMI) subtest with 72 items about eye-hand coordination (eg, reaching and grasping objects, building blocks, and copying line). Modification involved administering items separately for each UE, yielding an affected side raw-score sum. This subtest avoids floor and ceiling problems; has excellent test-retest (0.90) and interrater (0.98) reliability and high Cronbach’s coefficient α (0.95); and works well with motor-delayed children >6 years old26,27 ; and (3) the Quality of Upper Extremity Skills Test (QUEST)28,29  Dissociated Movement (Affected Side) subtest that measures UE use dissociated from the body trunk. All have been used in CIMT research.

We further identified 2 primary parent-reported outcomes: following Enhancing the Quality and Transparency of Health Research guidelines for patient and proxy–reported outcomes30,31 : (1) the How Often scale of the Child Motor Activity Log (CMAL), a 19-item tool adapted from the Pediatric Motor Activity Log,1215  which reveals 17 of 19 items are correctly ordered (see Supplemental Materials; parents rate the frequency of use of the hemiparetic UE in common play and self-help activities) and (2) the Pediatric Evaluation of Disability Inventory-Computer Adaptive Test (PEDI-CAT)32,33  Daily Activities scale, a widely used, validated scale about eating and mealtime, getting dressed, hygiene, and home tasks. Parents are well-qualified to report these patient-valued functional outcomes.

Two secondary outcomes are the QUEST Grasp28,29  and the CMAL How Well scale (highly correlated with the How Often scale). We excluded the PDMS-2 Object Manipulation subtest about using balls because many items are lower extremity only and unrelated to treatment goals. Finally, parents completed the Perceived Stress Scale34  and reported about their child’s adjustment to treatment.

Randomization involved site stratification. Group assignment had equal probability using a random permutated block design with randomly chosen block sizes of 5 and 10. The study statistician (M.C.) created a computer-generated randomization list, given only to the central study coordinator who revealed assignment to sites after consenting. Local site blinding involved no contact between staff who consented, scheduled, and treated and those who conducted blinded assessments.

Based on AHA logit scores from a previous RCT,17  we sought 27 per group, projecting 10% attrition for a final group size of 24. This results in the F-test having 80% power, with a 5% significance level for a main effect size of 0.58 and interaction size effect of 1.18.

In analyzing continuous-variable outcomes for intention-to-treat participants, we controlled for each child’s baseline and used repeated-measures analyses of covariance (ANCOVAs), with an unstructured covariance matrix applying Bonferroni corrections for multiple comparisons. For the binary AHA outcome of ≥5 logit points (a minimally detectable difference), we used logistic regression. After reviewing obtained results, we identified 2 exploratory statistical methods to estimate the significance of the pattern of greatest improvement occurring among children who received one of the high-dose CIMT treatments. First, we applied a permutation test to answer the following: what is the probability under the null hypothesis that the mean change among groups at each assessment is the same for all outcomes that any 2 groups would have the greatest change on 10 outcomes (5 primary outcomes at both posttreatment assessments)?35  (see Supplemental Materials). Second, we used the O’Brien36  method for exploring group gain differences on all 7 outcomes. With the O’Brien method, one quantifies an individual child’s multidomain profile of gains on identified outcomes, equally weighted. This helps overcome the challenge of no adequate single outcome.3638  O’Brien creates a score using each child’s rank-order in the study population from each outcome ranking. Because this requires data for all outcomes, we applied multiple imputation, creating 250 completed data sets in which participants had observed or imputed values for all outcomes (see Supplemental Materials).39,40  Multiple imputation values made up the data set: individual gain scores on each outcome were assigned rank-orders from 1 (lowest change) to 118 (highest) and then averaged, yielding a mean rank-order score per child.

Table 1 reveals demographic and clinical characteristics. Some group variation in age, Manual Ability Classification System (MACS)4143  or Mini-MACS,44  Gross Motor Functional Classification System (GMFCS),45  and previous CIMT appeared; adjustments by any or all of these did not change any conclusions. At baseline, groups received highly comparable weekly means of 4.5 hours usual customary treatment (UCT) (SD = 4.0), the sum of occupational, physical, and speech and language therapy. UCT doses varied widely: 11% had no weekly treatment, whereas 25% had >7 hours per week. Parent-reported stress levels were below national norms. Table 1 also reveals group mean baseline scores for primary and secondary outcomes.

TABLE 1

Descriptive Characteristics and Baseline (Pretreatment) Scores for 5 Treatment Groups

UCT
(n = 23)
30 h and Splint
(n = 25)
30 h and Cast
(n = 21)
60 h and Splint
(n = 24)
60 h and Cast
(n = 25)
Total Sample
(N = 118)
Demographic, clinical, and previous treatment variablesa       
 Sex, n (%)       
  Male 8 (35) 12 (48) 8 (38) 10 (42) 6 (24) 44 (37) 
  Female 15 (65) 13 (52) 13 (62) 14 (58) 19 (76) 74 (63) 
 Race, n (%)       
  White 20 (87) 19 (76) 14 (67) 21 (88) 18 (72) 92 (78) 
  Black or African-American 1 (4) 2 (8) 4 (19) 1 (4) 3 (12) 11 (9) 
  Asian American 1 (4) 0 (0) 2 (10) 0 (0) 1 (4) 4 (3) 
  Multiracial 1 (4) 2 (8) 1 (5) 1 (4) 2 (8) 7 (6) 
 Ethnicity, n (%)       
  Hispanic 0 (0) 3 (14) 0 (0) 1 (4) 1 (5) 5 (4) 
   Age, mean (SD), y 4.5 (2.1) 4.4 (2.1) 5.3 (2.5) 4.6 (2.5) 3.4 (1.2) 4.4 (2.1) 
 MACS or Mini-MACS,3841 n (%) 13 (57) 9 (36) 15 (71) 13 (54) 15 (60) 65 (55) 
  MACS III 10 (43) 13 (52) 5 (24) 5 (21) 10 (40) 43 (36) 
  MACS IV 0 (0) 3 (12) 1 (5) 6 (25) 0 (0) 10 (85) 
 GMFCS,42 n (%)       
  GMFCS I 14 (61) 11 (44) 11 (52) 11 (46) 11 (44) 58 (49) 
  GMFCS II 9 (39) 11 (44) 8 (38) 13 (54) 13 (52) 54 (46) 
  GMFCS III 0 (0) 2 (8) 2 (10) 0 (0) 1 (4) 5 (4) 
  GMFCS IV 0 (0) 1 (4) 0 (0) 0 (0) 0 (0) 1(<1) 
   Previous CIMT, yes, n (%) 9 (39) 11 (44) 7 (33) 10 (42) 11 (44) 48 (40) 
Affected UE side, n (%)       
  Left 14 (61) 11 (44) 7 (33) 8 (33) 13 (52) 53 (45) 
  Right 9 (39) 14 (56) 14 (67) 16 (67) 12 (48) 65 (55) 
  Parent stress mean (SD) 30.1 (7.1) 31.9 (6.2) 33.3 (10.3) 34.1 (7.6) 32.0 (7.5) 32.3 (7.7) 
 Hours per wk of therapy, pretreatment,
mean (SD) 
4.2 (4.7) 4.4 (3.7) 4.9 (3.6) 4.4 (4.8) 4.7 (3.7) 4.5 (4.1) 
Baseline means on primary outcomes       
 Blinded objective outcomes       
  AHA logit, mean (SD) 43.5 (20.0) 42.9 (22.3) 46.9 (22.3) 40.8 (21.9) 42.0 (26.9) 43.2 (22.7) 
  Peabody VMI (affected side), mean (SD) 33.1 (24.1) 39.6 (24.2) 44.8 (20.3) 33.0 (24.2) 31.7 (21.0) 36.4 (22.8) 
  QUEST DM (Affected side), mean (SD) 12.0 (6.7) 12.8 (7.3) 15.3 (7.0) 11.0 (6.7) 12.6 (7.0) 12.7 (6.9) 
 Parent-Reported Outcomes       
  CMAL how often, mean (SD) 2.3 (1.4) 2.0 (1.2) 2.1 (1.1) 1.9 (0.9) 1.7 (1.0) 2 (1.12) 
  PEDI-CAT Daily Activities, Mean (SD) 50.2 (2.8) 50.1 (3.3) 51.2 (2.2) 50.3 (3.6) 49.3 (2.6) 50.2 (2.9) 
UCT
(n = 23)
30 h and Splint
(n = 25)
30 h and Cast
(n = 21)
60 h and Splint
(n = 24)
60 h and Cast
(n = 25)
Total Sample
(N = 118)
Demographic, clinical, and previous treatment variablesa       
 Sex, n (%)       
  Male 8 (35) 12 (48) 8 (38) 10 (42) 6 (24) 44 (37) 
  Female 15 (65) 13 (52) 13 (62) 14 (58) 19 (76) 74 (63) 
 Race, n (%)       
  White 20 (87) 19 (76) 14 (67) 21 (88) 18 (72) 92 (78) 
  Black or African-American 1 (4) 2 (8) 4 (19) 1 (4) 3 (12) 11 (9) 
  Asian American 1 (4) 0 (0) 2 (10) 0 (0) 1 (4) 4 (3) 
  Multiracial 1 (4) 2 (8) 1 (5) 1 (4) 2 (8) 7 (6) 
 Ethnicity, n (%)       
  Hispanic 0 (0) 3 (14) 0 (0) 1 (4) 1 (5) 5 (4) 
   Age, mean (SD), y 4.5 (2.1) 4.4 (2.1) 5.3 (2.5) 4.6 (2.5) 3.4 (1.2) 4.4 (2.1) 
 MACS or Mini-MACS,3841 n (%) 13 (57) 9 (36) 15 (71) 13 (54) 15 (60) 65 (55) 
  MACS III 10 (43) 13 (52) 5 (24) 5 (21) 10 (40) 43 (36) 
  MACS IV 0 (0) 3 (12) 1 (5) 6 (25) 0 (0) 10 (85) 
 GMFCS,42 n (%)       
  GMFCS I 14 (61) 11 (44) 11 (52) 11 (46) 11 (44) 58 (49) 
  GMFCS II 9 (39) 11 (44) 8 (38) 13 (54) 13 (52) 54 (46) 
  GMFCS III 0 (0) 2 (8) 2 (10) 0 (0) 1 (4) 5 (4) 
  GMFCS IV 0 (0) 1 (4) 0 (0) 0 (0) 0 (0) 1(<1) 
   Previous CIMT, yes, n (%) 9 (39) 11 (44) 7 (33) 10 (42) 11 (44) 48 (40) 
Affected UE side, n (%)       
  Left 14 (61) 11 (44) 7 (33) 8 (33) 13 (52) 53 (45) 
  Right 9 (39) 14 (56) 14 (67) 16 (67) 12 (48) 65 (55) 
  Parent stress mean (SD) 30.1 (7.1) 31.9 (6.2) 33.3 (10.3) 34.1 (7.6) 32.0 (7.5) 32.3 (7.7) 
 Hours per wk of therapy, pretreatment,
mean (SD) 
4.2 (4.7) 4.4 (3.7) 4.9 (3.6) 4.4 (4.8) 4.7 (3.7) 4.5 (4.1) 
Baseline means on primary outcomes       
 Blinded objective outcomes       
  AHA logit, mean (SD) 43.5 (20.0) 42.9 (22.3) 46.9 (22.3) 40.8 (21.9) 42.0 (26.9) 43.2 (22.7) 
  Peabody VMI (affected side), mean (SD) 33.1 (24.1) 39.6 (24.2) 44.8 (20.3) 33.0 (24.2) 31.7 (21.0) 36.4 (22.8) 
  QUEST DM (Affected side), mean (SD) 12.0 (6.7) 12.8 (7.3) 15.3 (7.0) 11.0 (6.7) 12.6 (7.0) 12.7 (6.9) 
 Parent-Reported Outcomes       
  CMAL how often, mean (SD) 2.3 (1.4) 2.0 (1.2) 2.1 (1.1) 1.9 (0.9) 1.7 (1.0) 2 (1.12) 
  PEDI-CAT Daily Activities, Mean (SD) 50.2 (2.8) 50.1 (3.3) 51.2 (2.2) 50.3 (3.6) 49.3 (2.6) 50.2 (2.9) 
a

Frequencies may not add to group totals because of missing values.

CIMT groups had >95% compliance with the intended dose; 100% correctly used the constraint. One child stopped treatment because of a family emergency. Four adverse events occurred; none were treatment-related. At treatment end, only 2 groups met the originally intended cell size of 24; only 1 did 6 months later. Overall, the final sample was 94% (treatment end) and 89% (6 months posttreatment) of the planned 120.

In Tables 2 and 3, we present posttreatment values for all outcomes. We encountered higher-than-predicted variances within groups; specifically, we estimated a residual SD of 5.78 but encountered 8.10, ∼30% greater. This resulted in repeated-measures ANCOVAs that likely were underpowered to detect true group differences. We nonetheless fully present analytic results and later discuss study limits. We also applied statistical approaches better-suited for smaller sample sizes and considering multiple outcomes, (eg, O’Brien,36 Ramchandani et al,37  and Ristl et al38) as described above.

TABLE 2

Primary Outcomes for 5 CHAMP Groups at the End of Treatment and 6-Months Posttreatment: Changes From Baseline by Treatment Groups

UCT30 h and Splint30 h and Cast60 h and Splint60 h and Cast
AHA      
 Percentage with ≥5 logit units, % (SE)a      
  End of treatment 48 (10.9) 57 (10.3) 60 (12.6) 73 (9.5)b 70 (10.2) 
  6-mo posttreatment 53 (11.5) 50 (11.2) 65 (11.6) 47 (12.1) 71 (11.1)b 
 Mean logit unit gains (SE)c      
  End of treatment 5.5 (1.9) 6.8 (1.9) 6.5 (2.3) 5.7 (2.0) 7.0 (2.0)b 
  6-mo posttreatment 6.4 (1.9) 6.3 (1.8) 7.4 (2.0) 4.4 (2.0) 8.3 (2.0)b 
PDMS-2 VMI, mean point gain (SE)      
 End of treatment 3.5 (2.3) 1.2 (2.3) 2.4 (2.6) 6.2 (2.4)b 4.2 (2.4) 
 6-mo posttreatment 6.8 (3.0) 3.4 (3.5) 1.6 (3.4) 15.6 (3.2)b 11.3 (3.3) 
QUEST Dissociated Movement, mean point gains (SE)      
 End of treatment 1.4 (0.9) 1.5 (0.9) −1.0 (1.0) 2.1 (0.9)b 1.4 (0.9) 
 6-mo posttreatment 1.4 (0.9) 1.6 (0.8) 0.7 (0.9) 2.5 (0.8) 3.1 (0.8)b 
CMAL How Often (parent-reported)      
 End of treatment −0.1 (0.2) 0.7 (0.2) 0.9 (0.2) 0.9 (0.2) 1.3 (0.2)b 
 6-mo posttreatment −0.1 (0.2) 0.6 (0.2) 0.5 (0.2) 0.6 (0.2) 0.7 (0.2)b 
PEDI-CAT Daily Activities (parent-reported)      
 End of treatment 0.0 (0.3) 1.2 (0.3) 1.7 (0.4) 2.0 (0.3)b 0.9 (0.3) 
 6-mo posttreatment 1.4 (0.5) 2.1 (0.5) 1.1 (0.5) 3.1 (0.4)b 2.1 (0.4) 
UCT30 h and Splint30 h and Cast60 h and Splint60 h and Cast
AHA      
 Percentage with ≥5 logit units, % (SE)a      
  End of treatment 48 (10.9) 57 (10.3) 60 (12.6) 73 (9.5)b 70 (10.2) 
  6-mo posttreatment 53 (11.5) 50 (11.2) 65 (11.6) 47 (12.1) 71 (11.1)b 
 Mean logit unit gains (SE)c      
  End of treatment 5.5 (1.9) 6.8 (1.9) 6.5 (2.3) 5.7 (2.0) 7.0 (2.0)b 
  6-mo posttreatment 6.4 (1.9) 6.3 (1.8) 7.4 (2.0) 4.4 (2.0) 8.3 (2.0)b 
PDMS-2 VMI, mean point gain (SE)      
 End of treatment 3.5 (2.3) 1.2 (2.3) 2.4 (2.6) 6.2 (2.4)b 4.2 (2.4) 
 6-mo posttreatment 6.8 (3.0) 3.4 (3.5) 1.6 (3.4) 15.6 (3.2)b 11.3 (3.3) 
QUEST Dissociated Movement, mean point gains (SE)      
 End of treatment 1.4 (0.9) 1.5 (0.9) −1.0 (1.0) 2.1 (0.9)b 1.4 (0.9) 
 6-mo posttreatment 1.4 (0.9) 1.6 (0.8) 0.7 (0.9) 2.5 (0.8) 3.1 (0.8)b 
CMAL How Often (parent-reported)      
 End of treatment −0.1 (0.2) 0.7 (0.2) 0.9 (0.2) 0.9 (0.2) 1.3 (0.2)b 
 6-mo posttreatment −0.1 (0.2) 0.6 (0.2) 0.5 (0.2) 0.6 (0.2) 0.7 (0.2)b 
PEDI-CAT Daily Activities (parent-reported)      
 End of treatment 0.0 (0.3) 1.2 (0.3) 1.7 (0.4) 2.0 (0.3)b 0.9 (0.3) 
 6-mo posttreatment 1.4 (0.5) 2.1 (0.5) 1.1 (0.5) 3.1 (0.4)b 2.1 (0.4) 
a

Permutation test35  results (see Supplemental Materials for details about computation) indicate the probability of 2 groups having all top scores for 5 measures on 2 occasions is 0.017 and the probability for all 10 outcomes coming from a pair of predesignated groups is 0.006.

b

Group with greatest change.

c

The binary AHA outcome of ≥5 logit points was designated as a primary outcome in the original study design,14  reflecting a change above the minimally detectable threshold as recommended by the test authors. The actual AHA logit unit gains are shown just below the percentages for the binary outcome. Note: for computing the permutation test, only the actual logit unit gains were considered.

TABLE 3

Secondary Outcomes for 5 CHAMP Groups at the End of Treatment and 6-Months Posttreatment: Changes From Baseline by Treatment Groups

UCT30 h and Splint30 h and Cast60 h and Splint60 h and Cast
QUEST Grasp, mean point gain (SE)      
 End of treatment 0.1 (0.4) 0.5 (0.4) 0.0 (0.5) 0.3 (0.4) 0.6 (0.4) 
 6-mo posttreatment 0.9 (0.5) 1.0 (0.5) 0.3 (0.6) 0.0 (0.5) 0.8 (0.5) 
CMAL How Well, mean point gain (SE)      
 End of treatment 0.1 (0.2) 0.9 (0.2) 0.9 (0.2) 1.1 (0.2) 1.1 (0.2) 
 6-mo posttreatment 0.2 (0.2) 0.8 (0.2) 0.5 (0.2) 0.9 (0.2) 0.8 (0.2) 
Parent Perceived Social Stress since the baseline, descriptive measure only, mean point change (SE)      
 End of treatment −0.5 (1.7) −1.6 (1.5) −3.4 (1.9) −1.8 (1.6) −1.7 (1.5) 
 6-mo posttreatment −1.5 (1.5) −1.9 (1.4) 0.6 (1.6) −1.9 (1.4) −1.5 (1.3) 
UCT30 h and Splint30 h and Cast60 h and Splint60 h and Cast
QUEST Grasp, mean point gain (SE)      
 End of treatment 0.1 (0.4) 0.5 (0.4) 0.0 (0.5) 0.3 (0.4) 0.6 (0.4) 
 6-mo posttreatment 0.9 (0.5) 1.0 (0.5) 0.3 (0.6) 0.0 (0.5) 0.8 (0.5) 
CMAL How Well, mean point gain (SE)      
 End of treatment 0.1 (0.2) 0.9 (0.2) 0.9 (0.2) 1.1 (0.2) 1.1 (0.2) 
 6-mo posttreatment 0.2 (0.2) 0.8 (0.2) 0.5 (0.2) 0.9 (0.2) 0.8 (0.2) 
Parent Perceived Social Stress since the baseline, descriptive measure only, mean point change (SE)      
 End of treatment −0.5 (1.7) −1.6 (1.5) −3.4 (1.9) −1.8 (1.6) −1.7 (1.5) 
 6-mo posttreatment −1.5 (1.5) −1.9 (1.4) 0.6 (1.6) −1.9 (1.4) −1.5 (1.3) 

Across all primary outcomes and times (10 occasions), the largest gains occurred in one of the High-dose CIMT groups. The permutation test35  revealed the likelihood of this occurring for any 2 groups by chance was P = .017; it was even less likely (P = .006) for 2 prespecified groups, such as high-dose CIMT. At the end of treatment, AHA improvement ≥5 logit points appeared in all groups, ranging from just <50% in UCT to 71% for high-dose CIMT. At 6 months posttreatment, both splint groups revealed declined percentages, whereas CIMT plus cast and UCT did not. AHA gain scores provided a similar result: High-dose CIMT with cast displayed the highest gains of 7.0 (SE: 2.0) and 8.3 (SE: 2.0) units at the end of treatment and 6-months posttreatment, respectively, whereas the lowest gains were for UCT (5.5 [SE: 1.9]) at the end of treatment and for high dose with splint (4.4 [SE: 2.0]) 6 months posttreatment. For PDMS-2 VMI, high-dose groups had a mean gain of 5.3 (SE: 2.4) and 13.4 (SE: 3.3) end of treatment and 6-months posttreatment respectively, more than that of UCT notably at 6-months. Moderate-dose groups and UCT had comparable and low gains. For QUEST DM, the largest mean group gain appeared 6-months posttreatment (3.1 points; SE: 0.8) for high-dose CIMT with cast, more than double the UCT gain of 1.4 (SE: 0.9).

In Tables 4 and 5, we present results of statistical analyses contrasting each dose and constraint component to UCT. For the planned AHA outcome, the largest difference was high-dose versus UCT (mean difference [MD] = 23.7%; 95% confidence interval [CI] −1.6% to 49.1%) at end of treatment. For VMI gains, the largest difference obtained was for high-dose 6-months posttreatment (MD = 6.7; 95% CI −0.8 to 14.2). For DM, the largest group contrast also was high-dose 6-months posttreatment (MD = 1.4; 95% CI −0.6 to 3.5). In Fig 2, we graph the primary blinded outcomes.

FIGURE 2

Primary blinded outcomes for CHAMP treatment groups at end of treatment (left bar) and 6 months posttreatment (right bar with outline). Means and SEs for groups for PMDS-2 VM and QUEST DM reflect gains since the baseline. In Table 1, we provide baseline scores so that final scores can be calculated by adding gain scores. In Table 2, we provide results of statistical analyses contrasting manipulated factors of Dose and Constraint to UCT. Grey indicates UCT; yellow indicates 30-hour moderate-dose CIMT; green indicates 60-hour high-dose CIMT; solid nongrey colors indicate full-time cast; dots indicate part-time splint; no outline indicates end of treatment; black outline indicates 6 months posttreatment. A, AHA: percentage with gain of ≥5 logit points. B, AHA: mean changes (SE) from the baseline. C, VMI (affected side): mean changes (SE) from baseline. D, QUEST Disassociated Movement (affected side): mean changes from baseline.

FIGURE 2

Primary blinded outcomes for CHAMP treatment groups at end of treatment (left bar) and 6 months posttreatment (right bar with outline). Means and SEs for groups for PMDS-2 VM and QUEST DM reflect gains since the baseline. In Table 1, we provide baseline scores so that final scores can be calculated by adding gain scores. In Table 2, we provide results of statistical analyses contrasting manipulated factors of Dose and Constraint to UCT. Grey indicates UCT; yellow indicates 30-hour moderate-dose CIMT; green indicates 60-hour high-dose CIMT; solid nongrey colors indicate full-time cast; dots indicate part-time splint; no outline indicates end of treatment; black outline indicates 6 months posttreatment. A, AHA: percentage with gain of ≥5 logit points. B, AHA: mean changes (SE) from the baseline. C, VMI (affected side): mean changes (SE) from baseline. D, QUEST Disassociated Movement (affected side): mean changes from baseline.

Close modal
TABLE 4

Primary and Secondary Outcomes for 5 CHAMP Groups at the End of Treatment and 6-Months Posttreatment: Primary Outcomes (Selected Contrasts Between Manipulated Treatment Components and UCT)

High-Dose CIMT Minus UCTModerate-Dose CIMT Minus UCTCast CIMT Minus UCTSplint CIMT Minus UCT
MD (SE) or %Lower 95% CIUpper 95% CIMD (SE) or %Lower 95% CIUpper 95% CIMD (SE) or %Lower 95% CIUpper 95% CIMD (SE) or %Lower 95% CIUpper 95% CI
AHA ≥5 logit points, %             
 End of treatment 23.7 −1.6 49.1 10.6 −16.1 33.3 17.4 −9.3 44.0 17.0 −8.4 42.4 
 6-mo posttreatment 6.2 −21.4 33.8 4.7 −22.7 32.2 15.0 −12.4 42.4 −4.1 −31.8 23.6 
PDMS-2 VMI             
 End of treatment 1.6 (2.8) −4.0 7.3 −1.7 (2.9) −7.5 4.0 0.2 (2.8) −5.4 5.8 −0.2 (2.9) −6.0 5.5 
 6-mo posttreatment 6.7 (3.8) −0.8 14.2 −4.3 (3.9) −12.0 3.4 2.7 (3.8) −4.8 10.3 −0.3 (3.8) −7.9 7.3 
QUEST Dissociated Movement             
 End of treatment 0.3 (1.1) −1.8 2.5 −1.2 (1.1) −3.4 1.0 0.4 (1.1) −1.8 2.5 −1.2 (1.1) −3.4 1.0 
 6-mo posttreatment 1.4 (1.0) −0.6 3.5 −0.3 (1.1) −2.4 1.8 0.7 (1.0) −1.4 2.7 0.5 (1.0) −1.6 2.6 
CMAL How Often             
 End of treatment 1.2 (0.2) 0.7 1.7 0.9 (0.2) 0.4 1.4 0.9 (0.2) 0.5 1.4 1.2 (0.2) 0.7 1.7 
 6-mo posttreatment 0.8 (0.3) 0.3 1.3 0.7 (0.3) 0.1 1.2 0.7 (0.3) 0.2 1.2 0.7 (0.3) 0.2 1.2 
PEDI-CAT Ddily Activities             
 End of treatment 1.5 (0.4) 0.7 2.3 1.5 (0.4) 0.6 2.3 1.6 (0.4) 0.8 2.5 1.3 (0.4) 0.5 2.2 
 6-mo posttreatment 1.3 (0.6) 0.1 2.4 0.2 (0.6) −0.9 1.4 1.2 (0.6) 0.1 2.4 0.3 (0.6) −0.9 1.4 
High-Dose CIMT Minus UCTModerate-Dose CIMT Minus UCTCast CIMT Minus UCTSplint CIMT Minus UCT
MD (SE) or %Lower 95% CIUpper 95% CIMD (SE) or %Lower 95% CIUpper 95% CIMD (SE) or %Lower 95% CIUpper 95% CIMD (SE) or %Lower 95% CIUpper 95% CI
AHA ≥5 logit points, %             
 End of treatment 23.7 −1.6 49.1 10.6 −16.1 33.3 17.4 −9.3 44.0 17.0 −8.4 42.4 
 6-mo posttreatment 6.2 −21.4 33.8 4.7 −22.7 32.2 15.0 −12.4 42.4 −4.1 −31.8 23.6 
PDMS-2 VMI             
 End of treatment 1.6 (2.8) −4.0 7.3 −1.7 (2.9) −7.5 4.0 0.2 (2.8) −5.4 5.8 −0.2 (2.9) −6.0 5.5 
 6-mo posttreatment 6.7 (3.8) −0.8 14.2 −4.3 (3.9) −12.0 3.4 2.7 (3.8) −4.8 10.3 −0.3 (3.8) −7.9 7.3 
QUEST Dissociated Movement             
 End of treatment 0.3 (1.1) −1.8 2.5 −1.2 (1.1) −3.4 1.0 0.4 (1.1) −1.8 2.5 −1.2 (1.1) −3.4 1.0 
 6-mo posttreatment 1.4 (1.0) −0.6 3.5 −0.3 (1.1) −2.4 1.8 0.7 (1.0) −1.4 2.7 0.5 (1.0) −1.6 2.6 
CMAL How Often             
 End of treatment 1.2 (0.2) 0.7 1.7 0.9 (0.2) 0.4 1.4 0.9 (0.2) 0.5 1.4 1.2 (0.2) 0.7 1.7 
 6-mo posttreatment 0.8 (0.3) 0.3 1.3 0.7 (0.3) 0.1 1.2 0.7 (0.3) 0.2 1.2 0.7 (0.3) 0.2 1.2 
PEDI-CAT Ddily Activities             
 End of treatment 1.5 (0.4) 0.7 2.3 1.5 (0.4) 0.6 2.3 1.6 (0.4) 0.8 2.5 1.3 (0.4) 0.5 2.2 
 6-mo posttreatment 1.3 (0.6) 0.1 2.4 0.2 (0.6) −0.9 1.4 1.2 (0.6) 0.1 2.4 0.3 (0.6) −0.9 1.4 

—, not applicable.

TABLE 5

Primary and Secondary Outcomes for 5 CHAMP Groups at End of Treatment and 6-Months Posttreatment: Secondary Outcomes

High-Dose CIMT Minus UCTModerate-Dose CIMT Minus UCTCast CIMT Minus UCTSplint CIMT Minus UCT
MD (SE)Lower 95% CIUpper 95% CIMD (SE)Lower 95% CIUpper 95% CIMD (SE)Lower 95% CIUpper
95% CI
MD (SE)Lower 95% CIUpper 95% CI
QUEST Grasp             
 End of treatment 0.4 (0.5) −0.7 1.4 0.1 (0.5) −1.0 1.2 0.3 (0.5) −0.8 1.3 0.2 (0.5) −0.9 1.3 
 6-mo posttreatment −0.5 (0.6) −1.8 0.7 −0.2 (0.7) −1.5 1.1 −0.4 (0.6) −1.7 0.9 −0.4 (0.7) −1.7 0.9 
CMAL How Well             
 End of treatment 1.0 (0.2) 0.5 1.4 0.8 (0.2) 0.3 1.3 0.8 (0.2) 0.4 1.3 0.9 (0.2) 0.4 1.4 
 6-mo posttreatment 0.7 (0.2) 0.2 1.2 0.5 (0.3) 0.0 1.0 0.5 (0.3) 0.0 1.0 0.7 (0.3) 0.2 1.2 
High-Dose CIMT Minus UCTModerate-Dose CIMT Minus UCTCast CIMT Minus UCTSplint CIMT Minus UCT
MD (SE)Lower 95% CIUpper 95% CIMD (SE)Lower 95% CIUpper 95% CIMD (SE)Lower 95% CIUpper
95% CI
MD (SE)Lower 95% CIUpper 95% CI
QUEST Grasp             
 End of treatment 0.4 (0.5) −0.7 1.4 0.1 (0.5) −1.0 1.2 0.3 (0.5) −0.8 1.3 0.2 (0.5) −0.9 1.3 
 6-mo posttreatment −0.5 (0.6) −1.8 0.7 −0.2 (0.7) −1.5 1.1 −0.4 (0.6) −1.7 0.9 −0.4 (0.7) −1.7 0.9 
CMAL How Well             
 End of treatment 1.0 (0.2) 0.5 1.4 0.8 (0.2) 0.3 1.3 0.8 (0.2) 0.4 1.3 0.9 (0.2) 0.4 1.4 
 6-mo posttreatment 0.7 (0.2) 0.2 1.2 0.5 (0.3) 0.0 1.0 0.5 (0.3) 0.0 1.0 0.7 (0.3) 0.2 1.2 

In Fig 3, we display primary parent-reported outcomes. For the CMAL How Often scale, all CIMT groups revealed significant and large gains at both times, whereas UCT revealed low or no gains. The highest gains occurred for high dose with cast at both times. For PEDI-CAT Daily Activities at the end of treatment, all CIMT groups improved, although high-dose with cast improved the least and UCT did not improve. By 6 months posttreatment, however, UCT did reveal some gains, and 3 of the 4 CIMT groups revealed additional improvement. As Table 2 reveals, tests contrasting each treatment component to UCT were statistically significant for the How Often scale at both times. For the Daily Activities scale, the high-dose and cast contrasts but not moderate dose or splint were significantly more than UCT at both times.

FIGURE 3

Primary parent-reported functional outcomes for CHAMP treatment groups at end of treatment (left bar) and 6 months posttreatment (right bar with black outline). Means and SEs for Groups for CMAL how often and PEDI-CAT Daily Activities reflect gains since the baseline. In Table 1, we provide baseline scores so that final scores can be calculated by adding gain scores. In Table 2, we provides results of statistical analyses contrasting manipulated factors of dose and constraint to UCT. Grey indicates UCT; yellow indicates 30-hour moderate-dose CIMT; green indicates 60-hour high-dose CIMT; solid nongrey colors indicate full-time cast; dots indicate part-time splint; no outline indicates end of treatment; black outline indicates 6-mo posttreatment. A, AHA: percentage with gain of ≥5 logit points. B, AHA: mean changes (SE) from the baseline. A, CMAL How Often. B, PEDI-CAT Daily Activities.

FIGURE 3

Primary parent-reported functional outcomes for CHAMP treatment groups at end of treatment (left bar) and 6 months posttreatment (right bar with black outline). Means and SEs for Groups for CMAL how often and PEDI-CAT Daily Activities reflect gains since the baseline. In Table 1, we provide baseline scores so that final scores can be calculated by adding gain scores. In Table 2, we provides results of statistical analyses contrasting manipulated factors of dose and constraint to UCT. Grey indicates UCT; yellow indicates 30-hour moderate-dose CIMT; green indicates 60-hour high-dose CIMT; solid nongrey colors indicate full-time cast; dots indicate part-time splint; no outline indicates end of treatment; black outline indicates 6-mo posttreatment. A, AHA: percentage with gain of ≥5 logit points. B, AHA: mean changes (SE) from the baseline. A, CMAL How Often. B, PEDI-CAT Daily Activities.

Close modal

Parent stress (Table 3) declined modestly over time and comparably across groups. Parents were >95% favorable about their child’s excellent adjustment to full-time cast within 2 days and reported both doses highly acceptable.

Figure 4 reveals boxplots for the 5 groups in terms of their rank-ordered gains. At both assessment times, a similar pattern appeared with the smallest mean rank-order gains for UCT, intermediate for moderate-dose CIMT groups, and largest for high-dose CIMT groups. Figure 4 also reveals the CIs for planned contrasts, indicating the high-dose groups and splint groups had children who ranked statistically significantly more than UCT at both times, whereas moderate dose and cast were significantly more than UCT only at the end of treatment.

FIGURE 4

O’Brien’s composite rank-order outcomes for CHAMP treatment groups at end of treatment and 6-months posttreatment. Panels A and C demarcate the lowest rank of a child in the group, the first quartile, the median group rank-order value, the third quartile, and the highest rank of a child in the group. Sections B and D reveal 95% CIs for contrasts of factors compared with UCT. A, end of treatment. B, end of treatment. C, 6 months posttreatment. D, 6 months posttreatment.

FIGURE 4

O’Brien’s composite rank-order outcomes for CHAMP treatment groups at end of treatment and 6-months posttreatment. Panels A and C demarcate the lowest rank of a child in the group, the first quartile, the median group rank-order value, the third quartile, and the highest rank of a child in the group. Sections B and D reveal 95% CIs for contrasts of factors compared with UCT. A, end of treatment. B, end of treatment. C, 6 months posttreatment. D, 6 months posttreatment.

Close modal

Results of CHAMP data analyses are complex, supporting some predicted and some unexpected findings. CHAMP also reveals challenges likely to be encountered with highly heterogeneous pediatric patient populations.

First, the important finding that UCT produced objective benefits at end of treatment and 6-months later on blinded outcomes was unanticipated. This differs from previous RCT findings of either no or small gains for UCT children.4,46  UCT doses in CHAMP, however, were relatively high with a mean of 4 to 5 hours per week, more than double the 2.1 hours per week reported in the first pediatric CIMT trial12  and later studies.2,17  In fact, some children received UCT dosages similar to the tested moderate-dose CIMT. This suggests that pediatric rehabilitation dosage (at least in these CHAMP sites) has increased, likely because of the ascribed dosage importance in CIMT and other efficacious interventions.4749  We do not know, however, if CHAMP sites reflect nationwide practices. In addition, 40% of CHAMP participants previously received CIMT, which other CIMT trials considered an exclusionary criterion. Future CHAMP analyses need to explore whether previous CIMT, perhaps combined with other child and family variables, moderates subsequent CIMT effects. For example, parents whose children received previous CIMT may subsequently have sought higher-than-usual doses of therapy and increased their expectations for their children’s improvement. This unanticipated UCT finding highlights the importance of including a UCT group, in contradiction to the trend since 2010 revealing a UCT control was excluded in more than one-third of CIMT clinical trials.47 

End-of-treatment blinded outcomes indicate all groups displayed gains, although the pattern of the largest gains occurred among children in 1 of the 2 high-dose groups, extremely unlikely because of chance. Because children’s gain in moderate-dose groups were similar to that of UCT, we did not formally compare moderate- to high-dose CIMT in this article. In contrast, for parent-reported outcomes, children in all CIMT groups revealed significantly greater improvement in how often as well as how well they used their hemiparetic UE in typical activities. Notably, parents whose children received UCT reported almost no functional or real-world improvements in their children’s UE use.

These somewhat complex results reflect mostly differences in statistical significance achieved with different statistical tests. Both the permutation test and the O’Brien36  analysis of rank-order scores on all 7 outcomes affirm statistically significant benefits of high-dose CIMT. For the O’Brien36  analysis, we weighted the 7 outcomes equally because we had no empirical basis for assigning different values. Perhaps a weighted-rank order method that adjusts for treatment goals and baseline performance would be even better. Nonetheless, this equal-weight approach supports the conclusion that high-dose CIMT, regardless of constraint type, produces many more improvements than the lower CIMT dose or UCT. In addition, these significant benefits appear at the end of treatment and endure at least 6 months. (Future reports will explore longer-term outcomes at 12 months posttreatment.) Increasingly, clinical trials methodologists advocate including statistical strategies that recognize the importance of multiple outcomes to accurately capture treatment impact on patients whose clinical conditions affect multiple domains, as HCP does.38 

CHAMP has limits. Key is that the final study was modestly underpowered, largely because of lack of data from recent UCT groups and slightly lower recruitment and cohort maintenance. Nonetheless, CHAMP is the first to directly compare experimentally manipulated dosage and constraint type by using a standardized CIMT intervention protocol and constant outcome measures, thus providing unique findings. A fairly consistent pattern of results, despite differences in reaching statistical significance from alternative approaches, supports the acceptability and multiple benefits of the high-dose CIMT, thus providing relevant evidence for clinical decision-making.

Finally, we acknowledge the debate about including parent-reported outcomes. We justify including these because parents uniquely are able to observe daily behavior and functional outcomes. Blinded assessors seek to elicit a child’s “best performance” on standardized tools to determine if a child can do something, whereas parents can report on real-world “typical” behavior concerning the extent to which their child actually uses skills. In theory, technology advances could be applied to generate blinded real-world functional outcomes (eg, time-sampled video-recordings and body sensors used on multiple days in multiple settings): a methodology achievement we eagerly await. Nonetheless, for CHAMP, we report both blinded assessor and parent outcomes, reasoning that each affords a relevant perspective.

The 60-hour high-dose CIMT, regardless of constraint type (ie, both cast and splint), yielded a predominant pattern of more positive outcomes, with some differences depending on the outcome or time, thus earning the greatest clinical potential for children with HCP. This finding of benefits from the 60-hour dosage level matches that of Sakzewski et al48  on the basis of secondary analyses of 2 independent RCTs. These investigators also had a similar caregiver (nonblinded) finding about children’s typical functioning with benefits appearing for both CIMT doses, whereas only the blinded outcomes supported the higher (but not lower) dose conclusion. In CHAMP, parental stress was not elevated in any treatment group and parents indicated no preference for cast or splint. Given a choice among the 5 treatment groups, with data available on all outcomes, we think most parents and clinicians would select high-dose CIMT. At the same time, we caution that some children in all groups revealed good progress, whereas some did not. In future analyses, researchers need to explore whether differential treatment benefits can be predicted more precisely by clinical and/or environmental variables.

High-doses of CIMT delivered in 3-hour sessions 5 days per week for 4 weeks produced a consistent pattern of gains more than that of UCT on almost all blinded and parent-reported functional outcomes, although the findings are complex. Analysis of multiple end points and the pattern of gains, rather than individual ANCOVAs, provides the strongest support for the overall superiority of high-dose CIMT. A new finding about objective benefits from UCT also is encouraging.

We are immensely grateful to the following individuals who comprise our core CHAMP investigative team at our 3 clinical sites: The Ohio State University: study coordination (Christopher Brown, BS; Andrew Persch PhD, OTR/L); blinded assessor (BethAnn Ball, OTR/L); and treating therapists (Karen Harpster, PhD, OTR/L; Jody Stahr, PT; Cara Whalen PT, DPT, MPH, CHES; Lindy Weaver, PhD, OTR/L). University of Virginia: study coordination (Jodi G. Darring, BS, CCRP; Katheryn Frazier, MD; Michelle J. Haynes, BS, MPA; Gretchen M. Watkins, BA, CIP); blinded assessors (Leonor M. Gett, BS, MA; K. Michelle “Shelley” Knewstep-Watkins, OTD, OTR/L); and treating therapists (Susan J. Berres, BSOT, OTR/L, BSPT; Ruth Goldeen, BSOT, OTR/L, M.Ed; Kathleen A. Gypson, BSOT, OTR/L); Virginia Tech: study coordination (Laura Bateman, BS; Mary Lou Schwarzer, COTA; Melanie Sontheimer, BS); blinded assessors (Lindsay T. Carhart, DPT, CBIS; Vesna Cotic Costello, MS, OTR/L; Jessie Mann, MSc; Libbie Sonnier, PhD); and treating therapists (Kelsey Burke, MS, OTR/L; Mary Rebekah Trucks, MS, OTR/L; Dory Wallace, MS, OTR/L).

We extend grateful appreciation to the families and children who participated in The CHAMP comparative efficacy trial. Finally, we thank the reviewers and Pediatrics for an exceptionally thorough review of the first version of this article, which resulted in major revisions including reformulating the presentation of the complex statistical significance findings.

FUNDING: All phases of this study were supported by National Institutes of Health Grant 1R01HD068345 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Center for Medical Rehabilitation Research, and National Institutes of Health. The National Institutes of Health had no role in the design and conduct of this multisite randomized controlled trial. Funded by the National Institutes of Health (NIH).

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2021-053121

Deidentified individual participant data (including data dictionaries) will be made available, in addition to study protocols, the statistical analysis plan, and the informed consent form. The data will be made available one year after publication to researchers who provide a methodologically sound proposal for use in achieving the goals of the approved proposal. Proposals should be submitted to Sharon Ramey (slramey@vt.edu).

This trial has been registered at www.clinicaltrials.gov (identifier NCT01895660).

Dr Ramey conceptualized and designed the study, helped develop and refine data collection tools, provided guidance for data analysis and presentation strategies, wrote the initial and revised articles, reviewed and edited the manuscript with coauthors, and was the lead multiple principal investigator for CHAMP; Dr DeLuca, multiple principal investigator, helped conceptualize and design the study, led the training of therapists and documentation of fidelity of treatment, oversaw training for data collection, developed data entry and verification systems, worked closely with the statistician (Dr Conaway) and lead multiple principal investigator (Dr Ramey) in delineating data analyses, contributed to preparing and finalizing the manuscript, and directed the clinical site in Roanoke, Virginia; Dr Stevenson, multiple principal investigator, helped conceptualize and design the study, served as the lead developmental pediatrics medical expert, contributed to interpretation of study findings, assisted in review and revision of the manuscript, and directed the clinical site in Charlottesville, Virginia; Dr Conaway, study statistician, established initial study power, created the randomization lists, conceptualized and conducted the data analyses, contributed to understanding the study’s multivariate longitudinal findings, and helped review and revise the manuscript; Dr Darragh directed the clinical site in Columbus, Ohio, contributed to review of the data and the final manuscript, and joined CHAMP after Dr Case-Smith of The Ohio State University, one of the original multiple principal investigators, passed away in 2016; Dr Lo was a coinvestigator, contributed to the clinical oversight regarding pediatric neurology, referred patients for the Columbus, Ohio, clinical site, and assisted in review of findings and the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

AHA

assisting hand assessment

ANCOVA

analysis of covariance

CHAMP

Children with Hemiparesis Arm and Hand Movement Project

CI

confidence interval

CIMT

constraint-induced movement therapy

CMAL

Child Motor Activity Log

GMFCS

Gross Motor Functional Classification System

HCP

hemiparetic cerebral palsy

MACS

Manual Ability Classification System

MD

mean difference

PEDI-CAT

Pediatric Evaluation of Disabilities Inventory-Computer Adaptive Test

PDMS-2

Peabody Motor Development Scales, Second Edition

QUEST

Quality of Upper Extremity Skills Test

RCT

randomized controlled trial

UCT

usual customary treatment

UE

upper extremity

VMI

visual motor integration

1
Centers for Disease Control and Prevention
.
Data and statistics for cerebral palsy
.
Available at: https://www.cdc.gov/ncbddd/cp/data.html. Accessed October 2, 2020
2
Maenner
MJ
,
Blumberg
SJ
,
Kogan
MD
,
Christensen
D
,
Yeargin-Allsopp
M
,
Schieve
LA
.
Prevalence of cerebral palsy and intellectual disability among children identified in two U.S. national surveys, 2011-2013
.
Ann Epidemiol
.
2016
;
26
(
3
):
222
226
3
Odding
E
,
Roebroeck
ME
,
Stam
HJ
.
The epidemiology of cerebral palsy: incidence, impairments and risk factors
.
Disabil Rehabil
.
2006
;
28
(
4
):
183
191
4
Novak
I
,
Morgan
C
,
Fahey
M
, et al
.
State of the evidence traffic lights 2019: systematic review of interventions for preventing and treating children with cerebral palsy
.
Curr Neurol Neurosci Rep
.
2020
;
20
(
2
):
3
5
Novak
I
,
McIntyre
S
,
Morgan
C
, et al
.
A systematic review of interventions for children with cerebral palsy: state of the evidence
.
Dev Med Child Neurol
.
2013
;
55
(
10
):
885
910
6
Eliasson
AC
,
Krumlinde-Sundholm
L
,
Gordon
AM
, et al;
European network for Health Technology Assessment (EUnetHTA)
.
Guidelines for future research in constraint-induced movement therapy for children with unilateral cerebral palsy: an expert consensus
.
Dev Med Child Neurol
.
2014
;
56
(
2
):
125
137
7
Kwakkel
G
,
Veerbeek
JM
,
van Wegen
EE
,
Wolf
SL
.
Constraint-induced movement therapy after stroke
.
Lancet Neurol
.
2015
;
14
(
2
):
224
234
8
Ramey
SL
,
DeLuca
SC
,
Coker-Bolt
P
.
Operationalizing pediatric CIMT: guidelines for transforming basic principles and scientific evidence into clinical practice for individual children
. In:
Ramey
SL
,
Coker-Bolt
P
,
DeLuca
SC
, eds.
Handbook of Pediatric Constraint-Induced Movement Therapy (CIMT): A Guide for Occupational Therapy and Health Care Clinicians, Researchers, And Educators
.
Bethesda, MD
:
AOTA Press
;
2013
:
115
128
9
Cramer
SC
,
Sur
M
,
Dobkin
BH
, et al
.
Harnessing neuroplasticity for clinical applications
.
Brain
.
2011
;
134
(
pt 6
):
1591
1609
10
Dimyan
MA
,
Cohen
LG
.
Neuroplasticity in the context of motor rehabilitation after stroke
.
Nat Rev Neurol
.
2011
;
7
(
2
):
76
85
11
Reid
LB
,
Rose
SE
,
Boyd
RN
.
Rehabilitation and neuroplasticity in children with unilateral cerebral palsy
.
Nat Rev Neurol
.
2015
;
11
(
7
):
390
400
12
Taub
E
,
Ramey
SL
,
DeLuca
S
,
Echols
K
.
Efficacy of constraint-induced movement therapy for children with cerebral palsy with asymmetric motor impairment
.
Pediatrics
.
2004
;
113
(
2
):
305
312
13
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
(
11
):
931
938
14
Ramey
SL
,
DeLuca
S
,
Stevenson
RD
,
Case-Smith
J
,
Darragh
A
,
Conaway
M
.
Children with Hemiparesis Arm and Movement Project (CHAMP): protocol for a multisite comparative efficacy trial of paediatric constraint-induced movement therapy (CIMT) testing effects of dosage and type of constraint for children with hemiparetic cerebral palsy
.
BMJ Open
.
2019
;
9
(
1
):
e023285
15
DeLuca
SC
,
Echols
K
,
Ramey
SL
.
ACQUIREc Therapy: A Training Manual for Effective Application of Pediatric Constraint-Induced Movement Therapy
.
Hillsborough, NC
:
Mindnurture
;
2007
16
DeLuca
SC
,
Ramey
SL
,
Trucks
MR
,
Lutenbacher
R
,
Wallace
DA
.
The ACQUIRE protocol: what we have learned from a decade of delivering a signature form of pediatric CIMT
. In:
Ramey
SL
,
Coker-Bolt
P
,
DeLuca
SC
, eds.
Handbook of Pediatric Constraint-Induced Movement Therapy (CIMT): A Guide for Occupational Therapy and Health Care Clinicians, Researchers, and Educators
.
Bethesda, MD
:
AOTA Press
;
2013
:
129
147
17
DeLuca
SC
,
Case-Smith
J
,
Stevenson
R
,
Ramey
SL
.
Constraint-induced movement therapy (CIMT) for young children with cerebral palsy: effects of therapeutic dosage
.
J Pediatr Rehabil Med
.
2012
;
5
(
2
):
133
142
18
Case-Smith
J
,
DeLuca
SC
,
Stevenson
R
,
Ramey
SL
.
Multicenter randomized controlled trial of pediatric constraint-induced movement therapy: 6-month follow-up
.
Am J Occup Ther
.
2012
;
66
(
1
):
15
23
19
DeLuca
SC
,
Trucks
MR
,
Wallace
DA
,
Ramey
SL
.
Practice-based evidence from a clinical cohort that received pediatric constraint- induced movement therapy
.
J Pediatr Rehabil Med
.
2017
;
10
(
1
):
37
46
20
Krumlinde-Sundholm
L
,
Holmefur
M
,
Kottorp
A
,
Eliasson
AC
.
The Assisting Hand Assessment: current evidence of validity, reliability, and responsiveness to change
.
Dev Med Child Neurol
.
2007
;
49
(
4
):
259
264
21
Holmefur
MM
,
Krumlinde-Sundholm
L
.
Psychometric properties of a revised version of the Assisting Hand Assessment (Kids-AHA 5.0)
.
Dev Med Child Neurol
.
2016
;
58
(
6
):
618
624
22
Louwers
A
,
Krumlinde-Sundholm
L
,
Boeschoten
K
,
Beelen
A
.
Reliability of the Assisting Hand Assessment in adolescents
.
Dev Med Child Neurol
.
2017
;
59
(
9
):
926
932
23
Krumlinde-Sundholm
L
.
Reporting outcomes of the Assisting Hand Assessment: what scale should be used?
Dev Med Child Neurol
.
2012
;
54
(
9
):
807
808
24
Holmefur
M
,
Aarts
P
,
Hoare
B
,
Krumlinde-Sundholm
L
.
Test-retest and alternate forms reliability of the assisting hand assessment
.
J Rehabil Med
.
2009
;
41
(
11
):
886
891
25
Folio
R
,
Fewell
R
.
Peabody Developmental Motor Scales
. 2nd ed.
London, UK
:
Pearson
;
2000
26
Wang
HH
,
Liao
HF
,
Hsieh
CL
.
Reliability, sensitivity to change, and responsiveness of the peabody developmental motor scales-second edition for children with cerebral palsy
.
Phys Ther
.
2006
;
86
(
10
):
1351
1359
27
van Hartingsveldt
MJ
,
Cup
EH
,
Oostendorp
RA
.
Reliability and validity of the fine motor scale of the Peabody Developmental Motor Scales-2
.
Occup Ther Int
.
2005
;
12
(
1
):
1
13
28
Thorley
M
,
Lannin
N
,
Cusick
A
,
Novak
I
,
Boyd
R
.
Reliability of the quality of upper extremity skills test for children with cerebral palsy aged 2 to 12 years
.
Phys Occup Ther Pediatr
.
2012
;
32
(
1
):
4
21
29
Thorley
M
,
Lannin
N
,
Cusick
A
,
Novak
I
,
Boyd
R
.
Construct validity of the Quality of Upper Extremity Skills Test for children with cerebral palsy
.
Dev Med Child Neurol
.
2012
;
54
(
11
):
1037
1043
30
Calvert
M
,
Kyte
D
,
Mercieca-Bebber
R
, et al;
the SPIRIT-PRO Group
.
Guidelines for inclusion of patient-reported outcomes in clinical trial protocols: the SPIRIT-PRO extension
.
JAMA
.
2018
;
319
(
5
):
483
494
31
Gilmore
R
,
Sakzewski
L
,
Boyd
R
.
Upper limb activity measures for 5- to 16-year-old children with congenital hemiplegia: a systematic review
.
Dev Med Child Neurol
.
2010
;
52
(
1
):
14
21
32
Haley
SM
,
Coster
WJ
,
Dumas
HM
, et al
.
Accuracy and precision of the Pediatric Evaluation of Disability Inventory computer-adaptive tests (PEDI-CAT)
.
Dev Med Child Neurol
.
2011
;
53
(
12
):
1100
1106
33
Fragala-Pinkham
M
,
Shore
B
,
Kramer
J
.
Introduction to the Pediatric Evaluation of Disability Inventory Computerized Adaptive Test (PEDI-CAT) for children with cerebral palsy: A new option for measuring function in daily activities, mobility, social/cognitive and responsibility
. In:
American Academy for Cerebral Palsy & Developmental Medicine 69th Annual Meeting
;
October 21–24, 2015
;
Austin, TX
34
Cohen
S
,
Kamarck
T
,
Mermelstein
R
.
A global measure of perceived stress
.
J Health Soc Behav
.
1983
;
24
(
4
):
385
396
35
Good
P
.
Permutation, Parametric, and Bootstrap Tests of Hypotheses
. 3rd ed.
Berlin, Germany
:
Springer
;
2005
36
O’Brien
PC
.
Procedures for comparing samples with multiple endpoints
.
Biometrics
.
1984
;
40
(
4
):
1079
1087
37
Ramchandani
R
,
Schoenfeld
DA
,
Finkelstein
DM
.
Global rank tests for multiple, possibly censored, outcomes
.
Biometrics
.
2016
;
72
(
3
):
926
935
38
Ristl
R
,
Urach
S
,
Rosenkranz
G
,
Posch
M
.
Methods for the analysis of multiple endpoints in small populations: a review
.
J Biopharm Stat
.
2019
;
29
(
1
):
1
29
39
Beretta
L
,
Santaniello
A
.
Nearest neighbor imputation algorithms: a critical evaluation
.
BMC Med Inform Decis Mak
.
2016
;
16
(
Suppl 3
):
74
40
Chen
J
,
Shao
J
.
Nearest neighbor imputation for survey data
.
J Off Stat
.
2000
;
16
(
2
):
113
131
41
Plasschaert
VF
,
Ketelaar
M
,
Nijnuis
MG
,
Enkelaar
L
,
Gorter
JW
.
Classification of manual abilities in children with cerebral palsy under 5 years of age: how reliable is the Manual Ability Classification System?
Clin Rehabil
.
2009
;
23
(
2
):
164
170
42
Eliasson
AC
,
Krumlinde-Sundholm
L
,
Rösblad
B
, et al
.
The Manual Ability Classification System (MACS) for children with cerebral palsy: scale development and evidence of validity and reliability
.
Dev Med Child Neurol
.
2006
;
48
(
7
):
549
554
43
Wagner
LV
,
Davids
JR
.
Assessment tools and classification systems used for the upper extremity in children with cerebral palsy
.
Clin Orthop Relat Res
.
2012
;
470
(
5
):
1257
1271
44
Eliasson
AC
,
Ullenhag
A
,
Wahlström
U
,
Krumlinde-Sundholm
L
.
Mini-MACS: development of the Manual Ability Classification System for children younger than 4 years of age with signs of cerebral palsy
.
Dev Med Child Neurol
.
2017
;
59
(
1
):
72
78
45
Palisano
R
,
Rosenbaum
P
,
Walter
S
,
Russell
D
,
Wood
E
,
Galuppi
B
.
Development and reliability of a system to classify gross motor function in children with cerebral palsy
.
Dev Med Child Neurol
.
1997
;
39
(
4
):
214
223
46
Ramey
SL
,
DeLuca
SC
.
Pediatric CIMT: History and definition. (Appendix)
. In:
Ramey
SL
,
Coker-Bolt
P
,
DeLuca
SC
, eds.
Handbook of Pediatric Constraint-Induced Movement Therapy (CIMT): A Guide for Occupational Therapy and Health Care Clinicians, Researchers, and Educators
.
Bethesda, MD
:
AOTA Press
;
2013
:
19
40
47
Canadian Agency for Drugs and Technologies in Health
.
Constraint induced movement therapy for children: update of clinical effectiveness and guidelines
.
48
Sakzewski
L
,
Provan
K
,
Ziviani
J
,
Boyd
RN
.
Comparison of dosage of intensive upper limb therapy for children with unilateral cerebral palsy: how big should the therapy pill be?
Res Dev Disabil
.
2015
;
37
:
9
16
49
Charles
JR
,
Gordon
AM
.
A repeated course of constraint-induced movement therapy results in further improvement
.
Dev Med Child Neurol
.
2007
;
49
(
10
):
770
773

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data