Effect of motor intervention for infants and toddlers with cerebral palsy: systematic review and meta-analysis

Abstract

Purpose:

Conduct a systematic review and meta-analysis on the effect of motor intervention on motor function of infants and toddlers with cerebral palsy (CP).

Methods:

4 databases were searched for randomized controlled trials (RCTs) of motor interventions for children with or at high risk of CP aged less than 36 months. Studies were excluded if less than 50% of children developed CP.

Results:

Eleven RCTs included 363 children; 85% diagnosed with CP. Very low-quality evidence supports that: (1) task-specific motor training was more effective than standard care for improving motor function (small effect), (2) constraint-induced movement therapy (CIMT) may be more effective than bimanual play or massage for improving function of the more-affected hand (moderate effect), and high-intensity treadmill training is no more effective than low-intensity for improving walking.

Conclusions.

Very low-quality evidence supports that task-specific motor training and CIMT may improve motor function of infants and toddlers with CP.

Introduction

Recent advancements in the early detection of cerebral palsy (CP) support that children can be diagnosed with or at high risk of CP as early as 3 to 4 months of age.¹ This breakthrough provides an opportunity for infants and toddlers with or at high risk of CP to participate in CP-specific motor intervention in early infancy² when there is the greatest potential for neuroplastic change to maximize lifelong motor function.³

The potential of early motor intervention to optimize functional outcomes by harnessing the greater neuroplasticity of the developing brain and neuromotor system is particularly relevant to CP. Cerebral palsy is a disorder of movement caused by a malformation or lesion to the developing brain.⁴ It is the most common physical disability in childhood⁵ and affects over 17 million people worldwide.⁶ The functional outcome of CP is the result not only of the direct effects of the malformation or lesion, but also the activity-dependent neural reorganization that occurs in response to environmental experiences, including motor intervention. Recognizing that children with CP reach 90% of their gross motor potential by age 5, with most potential achieved by 3 years,⁷ motor intervention during the early years of life is considered critical to optimize the activity-dependent neuroplasticity and functional outcome of infants and toddlers with CP. Since activity-dependent reorganization can be adaptive or mal-adaptive, one needs to consider not only the positive, adaptive plasticity that can result from early intervention, but also the negative, maladaptive plasticity that can result from not providing intervention early enough or providing suboptimal intervention in terms of content or dose.

To optimize positive activity-dependent neuroplasticity and functional outcomes, it is critical to determine the efficacy of specific types of intervention for infants and toddlers with or at high risk of CP. Three systematic reviews have investigated the efficacy of early motor intervention for infants and toddlers with or at high risk of CP under the age of 3 years.^8–10 One review included infants under the age of 12 months,⁹ one included children under the age of 2 years¹⁰ and one included children under the age of 3 years.⁸ The reviews have 3 limitations. All 3 reviews included studies in which less than 50% of children developed CP by follow-up, which limits the applicability of the findings to children with CP. The reviews did not assess the research literature quantitatively using a meta-analysis to determine the efficacy of specific motor interventions, such as task-specific motor training and constraint-induced movement therapy (CIMT). In addition, the previous reviews did not include the recently published randomized controlled trials (RCTs) on CIMT. This systematic review with meta-analysis addresses these gaps in the literature by quantitatively investigating the effect of specific motor interventions on motor function of infants and toddlers with or at high risk of CP under 3 years of age using the highest quality studies, RCTs, in which more than 50% of children developed CP by follow-up.

Methods

Protocol and Registration

This systematic review and meta-analysis was prepared according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.¹¹ The A MeaSurement Tool to Assess systematic Reviews-2 (AMSTAR-2) checklist was used for the critical appraisal of this review.¹² The protocol was prospectively registered with the PROSPERO International Prospective Register of Systematic Reviews (CRD42021250548).

Search Strategy

A comprehensive literature search was completed from inception to February 2021 and rerun in July 2021 by an information services librarian of 4 databases: CINAHL, Embase, PubMed, and Web of Science. Searches were not restricted by publication date or language. Search terms included: cerebral palsy, children, motor intervention, physical therapy, and occupational therapy. References of included studies were examined and experts in the field were contacted to identify additional relevant studies. The full search strategy by database is provided in Supplemental Digital Content 1.

Eligibility Criteria and Study Selection

Included studies fulfilled the following criteria: (1) RCT and quasi-RCTs; (2) participants were children aged less than 36 months with a diagnosis of CP or at high risk of CP based on neuroimaging, abnormal General Movement Assessment or abnormal Hammersmith Infant Neurological Examination, consistent with the early detection guidelines for CP;¹ (3) intervention targeted motor function and occurred in the outpatient, home or community setting; (4) comparison group consisted of standard care or a different motor intervention; and (5) one or more outcome measures assessed motor function using a standardized test. Only RCTs were included to report findings from the highest quality of evidence.

Studies were excluded if: (1) they were review papers, abstracts, protocols, conference proceedings, or dissertations; (2) they were published in a language other than English and an adequate English translation could not be obtained; (3) less than 50% of participants had a diagnosis of CP by follow-up; (4) all or part of the intervention occurred in the inpatient setting, including the Neonatal Intensive Care Unit; (5) the study assessed the effect of an adjunct by comparing motor intervention alone to the same motor intervention plus an adjunct, such as electrical stimulation or kinesiotape.

Studies were screened using a web-based screening and data extraction tool, Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia), based on title and abstract, using the inclusion and exclusion criteria. If necessary, a full text review of studies was completed. During the screening and full text review, 2 of 3 authors independently reviewed the studies, data were compared for agreement, and disagreements were resolved by discussion. Studies that were read in full text but excluded are listed in Supplemental Digital Content 2; abstracts are not included in this table.

Level of Evidence and Risk of Bias

Three authors independently appraised the studies using the Oxford Center for Evidence Based Medicine levels of evidence¹³ and the Revised Cochrane risk-of-bias Tool (ROB-2) for randomized trials.¹⁴ Discrepancies were resolved through discussion.

Studies were assigned a level of design rigor using criteria from the Oxford Center for Evidence Based Medicine levels of evidence.¹³ Levels of evidence range from level I through level V, with level I as the highest level of evidence and level V as the lowest level of evidence.

Risk of bias was assessed using the ROB-2.¹⁴ This tool assesses bias resulting from 5 domains: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. The risk of bias was classified as low, some concerns, or high for each domain and for the entire study.

Data Extraction

Data were independently extracted by 2 of 3 reviewers, data were compared for agreement, and discrepancies were resolved through discussion.¹⁵ Sources of funding were assessed to determine if they represented a conflict of interest.

Quantitative Data Synthesis

Vote counting based on statistical significance was used to synthesize results for all studies based on type of intervention. For each type of intervention with multiple studies, 1 motor outcome measure from each study was pooled for meta-analysis. If a study reported multiple motor outcome measures, the primary outcome measure was used. If the primary outcome measure was not a standardized motor assessment, then data from the standardized motor assessment was used. All studies reported outcomes immediately after intervention, therefore, means and standard deviations (SD) of post-intervention scores were used to obtain the pooled estimates. Authors of studies that did not report means and SDs were contacted to obtain this information. For studies reporting medians and quartiles due to violation of normality assumptions, the means and SDs were manually estimated using the method of Wan et al.¹⁶

Using Review Manager (RevMan 5.4.1 Cochrane Collaboration, Copenhagen, Denmark), standard mean differences (SMD) with 95% confidence intervals (CI) were calculated. A random effects model was used with inverse-variance weighting. For SMD effect size interpretation, 0.2 was considered small, 0.5 moderate, and 0.8 large.¹⁷ Heterogeneity was investigated using the I² statistic and the Tau statistic². Heterogeneity was considered substantial when I² was greater than 50%, and the Tau statistic² was greater than 0.05. If heterogeneity was high, a sensitivity analysis was performed to assess the cause of the heterogeneity; this may result in removal of a study from the meta-analysis. If a study was removed from the meta-analysis, the reason for removal was documented. Results of the meta-analysis for each type of intervention were graphically represented using forest plots.

Quality of Evidence

The grading of the quality of evidence was based on the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach for systematic reviews of clinical trials. The quality of evidence refers to the confidence in the estimates of effect. It is graded as high, moderate, low, or very low. The effect of each type of motor intervention with results from RCTs with risk of bias rated as low or some concerns were used. The quality of evidence was assumed to be high, but then was rated down 1 level if there were study limitations, inconsistency of results, indirectness of evidence, and imprecision. The level of evidence was then rated up 1 level if there was a large magnitude of effect or dose-response gradient. Two of 3 authors independently appraised for GRADE and discrepancies were resolved through discussion.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Results

Studies Included

Search strategy and selection details are in Figure 1.¹⁸ Twelve studies were eligible for inclusion, but 2 studies used the same sample thus the results of these studies were combined as 1 study.^19,20 Eleven studies were included in this review.^19–30 All 11 studies were level II RCTs.^19–30 Risk of bias was low for 5 studies,^{19,20,22,23,25,30} some concerns for 5 studies,^24,26–29 and high for 1 study.²¹ The most common reasons for some concerns or high risk of bias included bias due to deviations from intended intervention, selection of reported results and the randomization process. The level of evidence, study design, overall risk of bias, number of participants, interventions, and intervention parameters are in Table 1. Full risk of bias assessments are in Supplemental Digital Content 3. Sources of funding are in Supplemental Digital Content 4.

Figure 1.

PRISMA flow diagram

Table 1.

Study Characteristics

Author(s) (Year)	Level of Evidence	Study Design	Overall Risk of Bias	Experimental Group vs. Control Group with number of participants	Experimental Frequency Intensity Time Duration	Control Frequency Intensity Time Duration	Home Program
Constraint-Induced Movement Therapy
Chamudot et al. (2018)	II	RCT	Low	CIMT (n=17) vs. BIM (n=16)	Home-based OT: F: 1×/wk I/T: NR D: 8 wks	Home-based OT: F: 1×/wk I/T: NR D: 8 wks	F: 7×/wk I: NR T: 60 min D: 8 wks
Eliason et al. (2018)	II	RCT	Low	CIMT (n=18) vs. IM (n=13)	Home-Based OT: F:1×/week I: NR T: NR D: 12 weeks	Home-Based OT: F: 3 sessions I: NR T: NR D: N/A	F: 6×/wk I: NR T: 30 min (CIMT) T: 5–30 min (IM) D: 12 wks
Maitre et al. (2020)	II	RCT	Low	CIMT (n=37) vs. BIM (n=36)	Home-Based therapy: F: 1×/wk I: NR T: 60 min D: 28 days	Home-Based therapy: F:1×/day I: NR T:20 min D:28 days	CIMT: F: 1×/day I: NR T: 6 hrs (CIMT) T: 20 min (BIM) D: 28 days
Peitruszwski et al. (2020)	II	RCT	Low	CIMT (n=7) vs. BIM (n=6)	Telehealth therapy: F: 1×/wk I: NR T: 15–45 min D: 28 days	Telehealth therapy: F: 1×/wk I: NR T: 15–45 min D: 28 days	F: 5×/wk I: NR T: 6 hours (CIMT) T: 20 min (BIM) D: 28 days
Task-Specific Motor Training Based on Motor Learning Principles
Harbourne et al. (2010)	II	RCT	Some concerns	Perceptual motor intervention (n=15) vs. home program (n=15)	Clinic Based PT: F: 2×/wk I: NR T: 1 hr D: 8 wks	Home-Based Therapy: F: 1×/wk I: NR T: 30–60 min D:8 wks	F: daily I/T: NR D: 8 wks
Hielkema et al. (2019, 2020)	II	RCT	Low	COPCA (n=23) vs. SC (n=20)	Home-Based PT: F: 1×/wk I: NR T: 30–60 min D: 1 yr	Home-Based/Clinic Based PT: F: 1×/wk I: NR T: 30–60 min D: 1 yr	NR
Morgan et al. (2015)	II	RCT	Some concerns	GAME (n=6) vs. SC(n=7)	Home-Based Therapy: F: 1×/wk initially, decreased I: NR T: 60–90 min D: 12 wks	Clinic-Based Therapy: F: 1–2 ×/mo I/T: NR D: 12 wks	F/I/T: NR D: 12 wks
Morgan et al. (2016)	II	RCT	Some concerns	GAME (n=15) vs. SC (n=15)	Home-Based Therapy: F: at least fortnightly I: NR T: 30–90 min D: 16 wks	Home-Based or Clinic Based Therapy: F: NR I: NR T: 15–90 min D: 16 wks	F/I/T: NR D: 16 wks
Palmer et al. (1998)	II	RCT	Some concerns	Learninggames (n=23) vs. NDT (n=25)	Clinic-Based Child-development specialist: F: every 2wk I: NR T: 1 hr D: 6 mo	Clinic-Based PT: F: every 2 wk I: NR T: 1 hr D: 6 mo Clinic-Based PT: F: daily I/T: NR D: until transition to sit independently	F: daily I/T: NR D: 6 mo
Neurofacilitation of Developmental Reaction Approach
Batra et al. (2012)	II	RCT	High	NFDR vs. NDT (total=30*)	Home-Based Therapy: F: 3×/week I: NR T: 40 min D: 3 mo	Home-Based Therapy: F: 3×/wk I: NR T: 40 min D: 3 mo	F/I/T/D: NR
Treadmill Training
Mattern-Baxter et al. (2020)	II	RCT	Some concerns	TT 10×/wk (n=9) vs. TT 2×/wk (n=10)	Home-Based Therapist Instruction: F: 1×/wk I/T: NR D: 6 wks	Home-Based Therapist instruction: F: 1×/wk I/T: NR D: 6 wks	F: 10×/wk (EG) or 2×/wk (CG) I: fastest speed a child could walk T: 20 min D: 6 wks

^*study did not report number of participants per group

vs: versus, RCT: Randomized Control Trial, CIMT: Constraint Induced Movement Therapy, n = number, BIM: Bimanual Training, OT: Occupational Therapy, F: frequency, wk: week, I: intensity, T: time, NR: Not Reported, D: duration, min: minute, IM: infant massage, PT: Physical Therapy, hr: hour, COPCA: Coping with and Caring for Infants with Special Needs, SC: standard care, NFDR: neurofacilitation of developmental reactions, NDT: Neurodevelopmental Treatment, TT: treadmill training, EG: Experimental Group, CG: Control Group,

Participants

A total of 363 children with or at high risk of CP ranging in age from 0 to 32 months participated in the studies; 85% were diagnosed with CP by follow up.^19–30 The characteristics of the participants are in Supplemental Digital Content 5.

Motor Outcome Measures

Results of motor outcome measures are in Table 2. Motor participation was measured in 4 of 11 studies (36%).^19,26–29 Outcomes included the Canadian Occupational Performance Measure, Pediatric Evaluation of Disability Index (PEDI), and Vineland Adaptive Behavior Scales (VABS).

Table 2.

Summary of Results

Author(s) (Year)	Experimental vs Control Adherence	Timing of Outcome Measures	ICF Model	Outcome Measures	Between Group Differences EG vs CG	Within Group Differences EG & CG		Clinical Implications
Constraint-Induced Movement Therapy
Chamudot et al. (2018)	CIMT: 75% hrs	Pre, Post	A	Mini-AHA	NS, ES: .17*	NS	NS	For infants with CP, 8 wks of CIMT compared to BIM resulted in similar improvements in hand and gross motor function.
			A	FI-GMS	NS	NS	NS
	BIM: 81% hrs		A	FI-UHU	NS, ES: .00*	NS	NS
			A	FI-BHU	NS	NS	NS
Eliason et al. (2018)	CIMT: 97% hours	Pre, 6 weeks, 12 weeks, post, 18mo of age	A	HAI-A	↑, ES: .43*	↑	NS	For infants with UCP, 12-wks of CIMT compared to infant massage resulted in improved function of the affected hand. During follow up at 18 weeks, results were maintained.
			A	HAI-NA	NS	NR	NR
	Massage: 72% sessions		A	HAI-B	NS, ES: .57*	NS	NS
Maitre et al. (2020)	CIMT: 38 hours/wk	Pre, post	A	BSITD-UFM-A	↑, ES: .35	NR	NR	For infants with asymmetric CP, 28 days of CIMT compared to a wait list control resulted in improved fine motor skills of the more affected hand.
			A	BSITD-UFM-NA	NS	NR	NR
	BIM: NR		A	BSITD-BI	NS, ES: .02	NR	NR
			A	BSITD-GM	NS	NR	NR
Peitruszwski et al. (2020)	CIMT: 21h/wk	Pre, post	A	BSITD UFM	↑, ES: 3.91*	NR	NR	For infants with hCP or asymmetric CP, 28 days of CIMT compared to a wait list control resulted in improved fine motor skills of the more affected hand.
	BIM: NR		A	BSITD BFM	NS, ES: 2.1*	NR	NR
			A	BSITD GM	NS	NR	NR
Task-Specific Training Based on Motor Learning Principles
Harbourne et al. (2010)	Perceptual motor intervention: NR	Pre, 1 mo post	A	GMFM-sitting	NS, ES: .07*	NR	NR	For infants with or at high risk of CP, 8 wks of perceptual-motor intervention compared to a home program had resulted in similar effects on sitting function.
	Home program: NR
Hielkema et al. (2019, 2020)	COPCA: median 3.0 session/mo	Pre, 3 mo, 6 mo, post, at 21 mo CA	A	IMP-perf	NS, ES: 0.29*	NR	NR	For infants with or at high risk of CP, 1 year of COPCA intervention compared to typical infant physiotherapy resulted in similar developmental outcomes.
			A	GMFM-88	NS, ES: .11*	NR	NR
	TIP: median 2.5 sessions/mo		A	GMFM-66	NS, ES: .36*	NR	NR
			A	GMFM (adapted)	NS	NR	NR
			A	AIMS	NS, ES: .08*	NR	NR
			A	BSID-II: PDI	NS, ES: .10*	NR	NR
			A	PEDI	NS	↑	↑
			P	VABS	NS	↑	↑
Morgan et al. (2015)	GAME: mean 140.58 hrs	Pre, post	A	GAS	NS	NR	NR	For infants with or at high risk of CP, 12-wks of GAME intervention compared to standard care resulted in improved motor skills.
			P	COPM: perf	NS	NR	NR
	SC: mean 54.17 hrs		A	PDMS-2: TMQ	↑, ES: .68 *	NR	NR
			A	PDMS-2 total motor SS	↑, ES: .67*	NR	NR
Morgan et al. (2016)	GAME: mean 21.91 hrs	Pre, post, 12 mo of age	A	PDMS-2: TMQ	NS, ES: .44*	NR	NR	For infants with or at high risk of CP, 16 wks of GAME intervention compared to standard care resulted in improved motor skills.
	SC: mean 12.82 hrs		A	PDMS-2: RS	↑, ES: −.10*	NR	NR
			P	COPM: perf	NS, ES: .25*	NR	NR
Palmer et al. (1998)	Learninggames: >90% compliance	Pre, post, 6 mo post	A	BSID-PDI	↑, ES: .57*	NR	NR	For infants with CP, 6 months of Learninggames intervention compared to NDT intervention resulted in improved motor skills.
	NDT: >90% compliance
Neurofacilitation of Developmental Reaction Approach
Batra et al. (2012)	NFDR: NR	Pre, post	BSF	MAS	↑	NR	NR	For infants with CP, 3-mo of NFDR program compared to NDT resulted in improved motor function.
			BSF	PR	NS	NR	NR
	NDT: NR		A	GMFM	↑	NR	NR
Treadmill Training
Mattern-Baxter et al. (2020)	TT 2x/wk: 183 +/− 46 m	Pre, post, 1 mo and 4 mo post	A	GMFM D	NS, ES: 0.34*	↑	↑	For toddlers with CP, 6-wks of TT intervention delivered 2x/wk or 10x/wk resulted in similar improvement of walking skills.
			A	GMFM E	NS, ES: 0.25*	NS	↑
	TT 10x/week: 762 +/− 213 m		A	PDMS-2 loc	NS, ES: 0.28*	↑	↑
			A	PEDI mob	NS	↑	↑
			A	10 MWT	NS	↑	↑
			BSF	1MWT	NS	↑	↑
			A	Avg steps	NS	NS	NS
			A	Avg active min	NS	NS	NS

↑, statically significant increase; ↓, statistically significant decrease

^*indicates that effect size was calculated by authors (Hedges’ g)

vs: versus; ICF: International Classification of Functioning, EG: experimental Group, CG: control group, CIMT: Constraint Induced Movement Therapy, hrs: hours, n: number, BIM: Bimanual Training, A: activity, Mini-AHA: mini Assisting Hand Assessment, FI-GMS: Functional Inventory gross motor subtest, FI-UHU: Functional Inventory unilateral hand use, FI-BHU: Functional Inventory bilateral hand use, NS: non-significant, ES: effect size, CP: cerebral palsy, mo: months, HAI-A: Hand Assessment of Infants of the more affected hand, HAI-NA: Hand Assessment of Infants of the less-affected hand, HAI-B: Hand Assessment of Infants of both hands, S: significant, NR: not reported, BSITD-UMF-A: Bayley Scales of Infant and Toddler Development unilateral motor function of the more affected hand, BSITD-UFM-NA: Bayley Scales of Infant and Toddler Development unilateral motor function of the less-affected hand, BSITD-BI: Bayley Scales of Infant and Toddler Development bilateral hand function, BSITD-GM: Bayley Scales of Infant and Toddler Development gross motor, GMFM: Gross Motor Functional Measure, COPCA: Coping With and Caring for Infants With Special Needs, TIP: Traditional Infant Physical Therapy, P: participation, IMP-perf: Infant Motor Profile-performance subtest, AIMS: Alberta Infant Motor Scale, BSID-II: Bayley Scales of Infant Development- second edition, PDI: Pain Disability Index, PEDI: Pediatric Evaluation of Disability Inventory, VABS: Vineland Adaptive Behavior Scales, GAME: Goals Activity Motor Enrichment, SC: standard care, GAS: Goal Attainment Scale, COPM: Canadian Occupational Performance Measure, PDMS: Peabody Developmental Motor Scales, TMQ: Total Motor Quotient; SS: standard score, NDT: Neurodevelopmental Treatment, NFDR: Neurofacilitation of Developmental Reactions, BSF: Body Structure Function, MAS: Modified Ashworth Scale, PR: primitive reflex, TT: treadmill training, GMFM D: Gross Motor Function Measure- subscale D, GMFM E: Gross Motor Function Measure- subscale E, 10 MWT: 10-Meter Walk Test, 1 MWT: 1-Meter Walk Test, Avg: Average

Activity was measured in 11 of 11 studies (100%).^19–30 Fine motor outcomes included the Assisting Hand Assessment (AHA) and Mini-AHA; Bayley Scales of Infant and Toddler Development fine motor subscale (BSITD-fm); Functional Inventory bilateral hand use (FI-bhu) and unilateral hand use (FI-uhu); Hand Assessment of Infants of the more-affected hand (HAI more-affected), less-affected hand (HAI less-affected), and both hands (HAI-bim). Gross motor outcomes included the Alberta Infant Motor Scale (AIMS), average active minutes, BSITD-gross motor composite, Functional Inventory gross motor (FI-gm), Gross Motor Function Measure (GMFM), Infant Motor Profile (IMP), Peabody Developmental Motor Scale, 2^nd edition (PDMS-2), Pediatric Evaluation of Disability Index mobility score (PEDI-m) and 10-meter walk test (10-MWT). Motor outcomes were measured using the Bayley Scales of Infant Development motor composite (BSID-pdi) and Goal Attainment Scale.

Body structure and function (BSF) was measured in 2 of 11 studies (18%).^21,26,29 Outcomes included the Modified Ashworth Scale (MAS), primitive reflexes, and 1-Minute Walk Test (1MWT) to assess endurance.

Motor Interventions

Motor interventions were categorized into 4 groups: (1) task-specific motor training based on motor learning principles (n=5 studies), (2) CIMT (n=4 studies), (3) neurofacilitation of developmental reaction approach (n=1 study), and (4) treadmill training (n=1 study). Supplemental Digital Content 6 includes principles of each experimental intervention, including goals, motor intervention principles, and family coaching.

Task-Specific Motor Training based on Motor Learning Principles

Five RCTs assessed outcomes of task-specific motor training based on motor learning principles compared to standard care or another motor intervention.^{19,20,24,27–29} Task-specific motor training based on motor learning principles was defined as intensive, targeted motor training of task and context-specific goals that incorporated motor learning principles of trial-and-error learning, movement exploration, and embedding practice into daily routines. Two studies assessed outcomes of Goals Activity Motor Enrichment (GAME),^27,28 One study assessed outcomes of Coping with and Caring for Infant with Special Needs (COPCA),^19,20 1 study assessed Learninggames,²⁹ and 1 study assessed outcomes of a perceptual motor intervention.²⁴ The participants ranged in age from 0 to 24 months; 73% of infants were diagnosed with CP. Types of CP included unilateral and bilateral spastic CP and athetoid CP with function ranging from GMFCS level I-V. The experimental intervention dose ranged from 8 weeks to 12 months with a frequency of 2 times per week to every other week for 30 to 90 minutes. Two of 5 studies used control interventions that were dose-matched with the experimental groups,^19,20,29 the other 3 used a lower dose.^24,27,28 No adverse events were reported.

The 5 task-specific motor training interventions reported participation and activity outcomes.^{19,20,24,27–29} Between the experimental and control groups, 17 outcomes were compared after intervention with 0 of 3 participation outcomes and 4 of 14 activity outcomes (PDMS-2, BSID-pdi) reaching statistical significance. One study reported within group comparisons.^19,20 Within the experimental and control groups, 2 of 2 participation outcomes (PEDI, VABS) reached statistical significance in the intended direction in both groups.

Four studies, with a total of 121 children with or at high risk of CP reported sufficient data on motor function to be pooled for meta-analysis.^24,27–29 The meta-analysis supported that infants and toddlers with or at high risk of CP who received task-specific motor training based on motor learning principles (n=59) scored a SMD of 0.41 (95% CI 0.05 to 0.78, p=0.03; small effect) higher on motor function compared to children who received standard care or another motor intervention (n=62). The effect was small, the meta-analysis was significant, and heterogeneity was low (I²=0%, Tau²=0). Figure 2 is the forest plot. An initial meta-analysis was performed with five studies, refer to Supplemental Digital Content 7 and 8 for details.

Figure 2.

Task-Specific Motor Learning based on Motor Learning Principles Forest Plot. Infants and toddlers with or at high risk of cerebral palsy who received task-specific motor training based on motor learning principles (n=59) scored a SMD of 0.41 (95% CI 0.05 to 0.78, p=0.03; small effect) higher on motor function compared to children who received standard care or another motor intervention (n=62). The meta-analysis was significant, and heterogeneity was low. GMFM, Gross Motor Function Measure; PDMS-2, Peabody Developmental Motor Scales 2^nd Edition; TMQ, Total motor quotient; BSID, Bayley Scales of Infant Development; SD, Standard deviation; Std, standard

There is very low-quality evidence (downgraded for risk of bias, indirectness, imprecision) that, immediately post intervention, task-specific motor training based on motor learning principles may be more effective than standard care for improving motor function of infants and toddlers with or at high risk of CP. Supplemental Digital Content 9 is the GRADE Evidence Profile.

Constraint-Induced Movement Therapy

Four RCTs assessed outcomes of CIMT compared to bimanual play or infant massage.^22,23,25,30 CIMT was defined as intensive, structured motor training of the more-affected upper extremity while restricting use of the less-affected upper extremity. In all 4 studies a soft custom mitt was used to restrict use of the less-affected upper extremity. The participants ranged in age from 3 to 24 months, and all were diagnosed with unilateral or bilateral asymmetric spastic CP. The experimental intervention dose ranged from 28 days to 12 weeks, 1 to 6 times per week for 30 minutes to 6 hours. 2 of 4 studies used control interventions that were dose matched with the experimental groups,^22,23 the other 2 used a lower dose.^25,30 No adverse events were reported.

The 4 CIMT studies reported activity outcomes.^22,23,25,30 Between the experimental and control groups, 13 outcomes were compared after intervention with 3 of 13 activity outcomes (BSITD-fm more-affected, HAI more-affected) reaching statistical significance. 2 studies reported within group comparisons.^22,23 Within the experimental and control groups, 7 activity outcomes were compared after intervention with 6 outcomes reaching statistical significance in both groups (Mini AHA, FI-bhu, FI-uhu, FI-gmu, HAI more-affected, HAI less-affected) and 1 reaching statistical significance in the experimental group only (HAI-bim).

All 4 studies with a total of 150 children with CP reported sufficient data on unimanual function of the more affected hand to be pooled for meta-analysis.^22,23,25,30 The meta-analysis supported that children with unilateral CP who received CIMT (n=79) scored a SMD of 0.59 (95% CI −0.18 to 1.37, p=.13; moderate effect) higher on unimanual function of the more affected hand compared to children with unilateral CP who received bimanual play or infant massage (n=71). Although the combined pooled effect was moderate, it only trended towards significance and there was substantial heterogeneity (I²=75%, Tau²=0.42). Figure 3 is the forest plot. A sensitivity analysis was performed and the Pietruszewski³⁰ study contributed to the high heterogeneity. Refer to Supplemental Digital Content 7 for details.

Figure 3.

Constraint-Induced Movement Therapy Forest Plot. Infants and toddlers with or at high risk of cerebral palsy who received constraint induced movement therapy (n=79) scored a SMD 0.59 (95% CI −0.18–1.37, p=.13, moderate effect). higher on unimanual function of the more affected hand compared to children with unilateral CP who received bimanual play or infant massage (n=71). Although the combined pooled effect was moderate, it only trended towards significance and there was substantial heterogeneity. SD, Standard deviation; Std, standard

There is very low-quality evidence (downgraded for inconsistency, indirectness, imprecision) that, immediately post intervention, CIMT may be more effective than bimanual play or infant massage for improving unimanual function of the more affected hand in infants with unilateral CP.

Neurofacilitation of Developmental Reaction Approach

One RCT assessed outcomes of the neurofacilitation of developmental reaction approach compared to neuro-development treatment.²¹ Participants ranged in age from 6 months to 2 years and all were diagnosed with CP. The intervention dose was 3 months, 3 times per week for 40 minutes. No adverse events were reported. Between the experimental and control groups, 3 outcomes were compared after intervention with 1 of 1 activity outcomes (GMFM) and 1of 2 BSF outcomes (MAS) reaching statistical significance. The outcomes of this RCT were not assessed using GRADE because the study had a high risk of bias.

Treadmill Training

One RCT assessed outcomes of high versus low-intensity treadmill training.²⁶ Participants ranged in age from 14 to 32 months; 74% were diagnosed with spastic diplegic CP. The intervention dose was 6 weeks of 20-minute sessions, implemented 10 sessions per week for the high-intensity group and 2 sessions per week for the low-intensity group. No adverse events were reported. Between the experimental and control groups, no outcomes reached statistical significance. Within the experimental and control groups, 8 outcomes were compared after intervention with 5 outcomes reaching statistical significance in both groups: 4 activity outcomes (GMFM-standing dimension, PDMS-2 locomotion subtest, PEDI-m, 10-MWT) and 1 BSF outcomes (1MWT). One activity outcome (GMFM-walking, running and jumping dimension) reached statistical significance in the control group only.

There is very low-quality evidence (downgraded for risk of bias, imprecision) that, immediately post intervention, high-intensity treadmill training is no more effective than low-intensity treadmill training for improving gait function of toddlers with spastic diplegic CP.

Discussion

This systematic review found 11 RCTs that investigated the effects of motor intervention on motor outcomes of infants and toddlers with or at high risk of CP; 85% of children developed CP. Interventions with the strongest evidence were task-specific motor training based on motor learning principles and CIMT although the evidence was of very low-quality. This review compares findings from this review to other systematic reviews published in the literature. Recommendations to strengthen the quality of future research evidence and to improve clinical application of this evidence are discussed.

This review found very low-quality evidence from 4 RCT’s with moderate risk of bias that, immediately post intervention, task-specific motor training based on motor learning principles may improve motor function of children with or at high risk of CP compared to standard care or another motor intervention (SMD 0.41, 95% CI 0.05 to 0.78, p=0.03; small effect). This finding is consistent with the 3 systematic reviews on motor intervention for infants and toddlers with or at high risk of CP ^8–10 and the motor recommendations from the international clinical practice guideline.² All 4 reviews supported motor intervention that included task and context specific goals, child-initiated movement, task-specific motor training with high repetition and practice, trial-and-error learning with encouragement of movement exploration and variability, and avoiding passive handling techniques. They also supported prioritizing the child-parent relationship, embedding practice into daily routines, and enriching the environment. All studies on task-specific motor training based on motor learning principles included in this review, documented the use of at least 60% of these principles.

This review found very low-quality evidence from 4 RCTs with low risk of bias that, immediately post intervention, CIMT compared with bimanual therapy or infant massage may improve unimanual function of the more-affected hand of children with unilateral CP (SMD 0.59, 95% CI −0.18–1.37, p=.13, moderate effect). This finding is consistent with a recent systematic review and meta-analysis on CIMT for children with unilateral CP ages 3 months to 19 years.³¹ Similar to this review, the review reported low to very low-quality evidence that CIMT was more effective than low-dose occupational therapy for improving unimanual capacity of the more-affected hand of children with CP. However, the review found no difference in unimanual capacity when CIMT was compared to high-dose or dose-matched occupational therapy or bimanual therapy.³¹ Consistent with this finding, 2 studies in this review found significant differences in unimanual capacity of the more affected hand when CIMT was compared to low-dose bimanual therapy.^25,30 One study in this review found no difference in unimanual capacity of the more-affected hand when CIMT was compared to dose-matched bimanual therapy.²² However, when CIMT was dose-matched with infant massage, CIMT was more effective for improving unimanual capacity of the more affected hand.²³ This supports that active child-initiated motor interventions, such as CIMT and bimanual therapy, may be more effective than passive interventions such as massage. Further research is needed to determine the effect of dose versus content of the intervention, such as CIMT or bimanual therapy, on unimanual and bimanual outcomes of infants and toddlers with unilateral CP.

This review found very low-quality evidence from one RCT with moderate risk of bias that high-intensity treadmill training is no more effective than low-intensity treadmill training for improving gait function of toddlers with spastic diplegic CP. This finding is consistent with a systematic review and meta-analysis on treadmill training for children under 6 years of age at risk for neuromotor delay that found no difference of high and low-intensity treadmill training on onset of independent walking for children with Down syndrome,³² however, they did find moderate-quality evidence that treadmill training compared to no treadmill training promoted earlier independent walking attainment for children with Down syndrome.³² An area for future research may be to determine if treadmill training or another type of intensive walking training promotes earlier independent walking attainment for children with CP.

In limiting the inclusion criteria to only RCTs in which over 50% of infants and toddlers developed CP by follow-up, the applicability of the results of this systematic review to children with CP is strengthened, but many motor interventions were not included. Other promising motor interventions for infants and toddlers with or at high risk of CP include: the CareToy,³³ iMOVE with dynamic weight support, ^34,35 Self-Initiated Prone Progression Crawler (SIPPC),³⁶ Sitting Together And Reaching to Play (START-Play),³⁷ Small Step Program,³⁸ and whole body vibration.³⁹ In addition, large efficacy trials of many interventions included in this review or noted above are ongoing and likely to provide more information within the next 5 years on the efficacy of motor intervention on motor outcomes of infants and toddlers with or at high risk of CP. The trials include: GAME, ACTRN12617000006347; SIT-PT, NCT04230278; Early locomotor training, NCT04561232; CIMT/I-AQUIRE, NCT03910075; and CIMT/APPLES, NCT02567630.

Implications for Research

The following 4 recommendations are proposed to improve the quality of research on early motor intervention for infants and toddlers with or at high risk of CP. First, to increase the number of infants with CP in research studies, it is recommended that infants at high risk of CP be identified using the early detection guidelines¹ and followed for the first 2 years to confirm that the infants developed CP and to quantify the type, topography, and functional level of CP. This may allow for subgroup analyses that could strengthen applicability of the results to children with specific types and functional levels of CP.

Second, it is recommended to quantify the content, dose, fidelity, and adherence of not only the experimental and control interventions, but also the standard care that the child is receiving and the parents’ implementation of the motor intervention into the children’s daily routines. Most studies in this review (81%) reported content, dose, and adherence in the experimental groups, but this was only reported in 63% of studies for the control groups. Far fewer studies assessed fidelity to the intervention or quantified the parents’ implementation of intervention principles into daily routines. This is critical since there is a growing appreciation of the importance of dose and the families’ potential role in supporting an adequate dose of daily motor practice for positive activity-dependent neuroplasticity and improved functional outcomes.

Third, it is recommended to use primary outcome measures that are responsive to change for infants and toddlers with or at high risk of CP. The GMFM is the only outcome measure developed and validated to evaluate change in the gross motor function of children with CP,⁷ yet it has not been validated for infants under 5 months. New outcome measures may need to be developed or current outcome measures modified to better quantify change with intervention for infants and toddlers with CP.

Fourth, it is recommended that characteristics of the family that may impact the outcome of intervention are collected and analyzed to determine barriers or opportunities for improved infant and family outcomes. The recommended home programs in the included studies ranged from twice a week to daily and up to 6 hours a day. Contextual factors such as maternal and paternal stress, age, socio-economic status, or education^19,20,27,28 may contribute to a family’s ability to incorporate intervention principles into daily routines. Understanding these contextual factors may allow improved ability to determine the optimal interventions for specific infant and family contexts and the most effective ways to coach parents to support the child-parent relationship, strengthen the infant and parents’ sense of efficacy, and organize the environment to feasibly incorporate motor practice into daily activities.

Implications for Clinical Practice

This review supports the clinical recommendations from the 3 previous systematic reviews and the recent international clinical practice guideline for infants and toddlers at high risk of CP: initiate early intervention at the time of suspected diagnosis, focus on task-specific motor training based on motor learning principles, initiate CIMT and/or bimanual training when a unilateral CP diagnosis is suspected, and avoid the use of passive handling techniques.^2,8–10 This review adds that there is very low-quality evidence that task-specific training based on motor learning principles and CIMT may have, respectively, a small and moderate effect on motor function.

Limitations

This review’s ability to quantify the efficacy of motor intervention for infants and toddlers with or at high risk of CP was limited by the small number of high-quality studies, small sample sizes within studies, and heterogeneity of interventions and outcome measures which contributed to a meta-analysis with substantial heterogeneity. In addition, less than 10 studies were included for each meta-analysis which precluded an assessment for publication bias.

Conclusions

A growing body of evidence supports the feasibility, safety, and efficacy of motor intervention for infants and toddlers with CP. Task-specific training based on motor learning principles and CIMT have the strongest quality of evidence, however, the overall quality of evidence is very low. Further rigorous research is needed.

What this paper adds.

• For infants and toddlers with or at high risk of cerebral palsy, very low-quality evidence supports that:

  • task-specific motor training based on motor learning principles may have a small effect on motor function
  • constraint-induced movement therapy may have a moderate effect on function of the more-affected hand
  • high-intensity treadmill training is no more effective than low-intensity treadmill training for improving walking function