Systematic Review of Progressive Strength Training in Children and Adolescents with Cerebral Palsy Who Are Ambulatory

Abstract

Purpose: 

To capture and analyze the evidence concerning the effects of progressive strength training on function and gait in children and adolescents with cerebral palsy (CP) who are ambulatory.

Method: 

A language-inclusive search was conducted for controlled or noncontrolled studies of strength training for subjects with CP who were ambulatory and aged 4 to 20 years, using objective outcome measures. Quality was assessed with the Maastricht-Amsterdam List. Data were extracted and analyzed.

Results: 

The 13 included articles favored treatment without significant adverse effects. Function and gait improved more following isotonic rather than isokinetic training, and in younger rather than older subjects.

Conclusions: 

Function and gait improvements were greater in preadolescents.

In Brief

This review of the evidence on the effects of strength training in children and adolescents with CP demonstrates that the effects on gait and function are greatest among pre-adolescents.

INTRODUCTION

Cerebral palsy (CP) is defined as “… non-progressive, but often changing, motor impairment syndromes secondary to lesions or anomalies of the brain arising in the early stages of its development.”^1(p9) CP occurs in approximately 2 of every 1000 live births in Europe² and 2.4 of every 1000 live births in the United States.³ Motor abilities develop during childhood but eventually a plateau is reached.⁴ Children risk losing some functional abilities during adolescence because of ongoing secondary impairments such as soft tissue shortening, weakness, and bony deformity.⁵ In particular, teenagers who are ambulatory may lose distance, speed, or quality of walking.⁶ Recent government legislation in the United Kingdom focuses attention on preparing children for adult life⁷; as participation in society is correlated with independent mobility,⁸ it is important to examine gait deterioration.

Strength-training interventions involve effort against progressive resistance. Historically, clinicians attempted to address weakness in subjects with CP by applying the principles of DeLorme in strength-training programs.⁹ With the advent of neurodevelopmental therapy, however, clinical practice largely followed the teaching that weakness is not a major problem in CP.¹⁰ Treatment was focused on inhibiting spasticity,¹¹ and excessive effort was avoided as it was believed to increase spasticity and impair motor control.¹² More recently, experts have questioned these premises, in identifying that both spasticity and weakness are part of the upper motor neurone syndrome¹³ and may coexist in the same muscle.¹⁴ There is some evidence that function is related to both strength and spasticity.¹⁵ Children with CP are significantly weaker than their peers¹⁶ and weakness may be evident in all muscles around a joint.¹⁷

Two previous reviews of strength training in CP found all included studies indicated the intervention increased muscle strength.^18,19 A systematic review²⁰ of literature published up to March 2000 used a thorough search strategy, but was limited to English language papers. The review included upper and lower limb studies of subjects with CP both ambulatory and nonambulatory; most were children and teenagers, but 1 study included adults up to 47 years. The broad inclusion criteria increased clinical heterogeneity and generalizability but limited internal validity. All 23 included studies favored strength training but negative studies may have been missed by exclusion of non-English papers.²¹ The authors concluded strength training improves strength but further research is needed to ascertain activity and participation outcomes, to establish optimum dosages, and to investigate longer-term effects.

The current review included non-English papers, and limited the clinical heterogeneity through a more focused research question. This review aimed to ascertain the current worldwide evidence about progressive strength training for children and adolescents with cerebral palsy who are ambulatory, with particular regard to function and gait outcomes.

METHOD

Search Strategy

A comprehensive language-inclusive search strategy was devised. This was validated by an experienced medical librarian. The search strategy was applied to 8 databases (MEDLINE, AMED, CINAHL, Cochrane Library, EMBASE, PEDro, PsycInfo, and SPORTDiscus) without limitations on publication dates up to March 2007. The keywords were as follows: population cerebral palsy; intervention strength training, strengthening, strength exercise, weight training, weight lifting, resisted exercise, resistance exercise, resisted training, resistance training. A supplementary search attempted to find literature missed by electronic searching. Expert researchers in the field were contacted, in an effort to locate any unpublished studies. Details of the search strategy are available from the authors.

Inclusion/Exclusion Criteria

The inclusion criteria for this review were (1) any experimental or quasi-experimental study, or single-group preexperimental studies; involving (2) subjects aged 4 to 20 years with cerebral palsy, ambulatory with/without aids and able to follow simple instructions; (3) any intervention aimed at strengthening one or more lower limb muscles and with some means of exercise progression; (4) any objective measure of walking ability, function, and/or strength. These criteria were narrower than those of the previous review²⁰ as upper limb studies, adults, and subjects who were nonambulatory were excluded.

Study Selection

Initially, the inclusion criteria were applied liberally to the citations, the abstract of each likely citation was read and the full manuscript obtained of each of those possibly meeting the inclusion criteria. If the abstract was unavailable, the full manuscript was obtained. Foreign language papers were translated. The inclusion criteria were stringently applied to each full manuscript by 2 independent reviewers to yield a final set of papers. If a full manuscript or translation was unobtainable, that paper had to be excluded.

Quality Assessment and Levels of Evidence

The Maastricht-Amsterdam List (MAL) quality assessment tool was chosen for this review as it scores both controlled and noncontrolled studies. Two independent reviewers applied the full 19-item MAL to controlled studies and the modified 14-item MAL to noncontrolled studies.²² Assessment reliability was increased by use of Operationalisation Guidelines (Appendix A) devised from the guidelines for the tools from which the MAL had been derived.^23–26 These Guidelines were refined through a 3-stage pilot study by the 2 reviewers. This resulted in the level of agreement, as measured by the unweighted kappa statistic, between the 2 reviewers rising from 0.6162 (standard error 0.0742; 95% confidence interval 0.4708–0.7616) to 0.7194 (standard error 0.1095; 95% confidence interval 0.5047–0.9341). This indicates the strength of agreement increased from “moderate/substantial” to well within the “substantial” range.²⁷ The frequent disagreements over items k and n were eradicated and the items on which they disagreed became more random.

The total score and the subscores for descriptive, internal validity, and statistical criteria were tabulated and compared with the MAL bench-mark scores (Table 1). Sackett’s Levels of Evidence and Grades of Recommendation guidelines were also applied to the included studies²⁸ (Appendix B).

TABLE 1: Maastricht-Amsterdam List Quality Bench-Mark Scores²²

Data Extraction and Analysis

The MAL quality scores were converted to percentages to enable comparisons. Data were extracted from each included study. Meta-analysis was beyond the scope of this review and was likely to be inappropriate because of heterogeneity of studies; descriptive statistics, however, were used to summarize findings. Within-group effect size statistics were calculated where possible and then averaged for each principal outcome type. According to Cohen’s guidelines, an effect size of ≤0.2 is small, 0.5 is medium, and 0.8 is large.²⁷ Confidence intervals for within-group estimates were not calculated, as these are beyond the scope of this review. Subgroup analyses were carried out where possible. The effects of detraining were examined in studies with a postintervention follow-up period.

RESULTS

Eighty-seven studies were identified by abstracts. Thirty-four full articles were reviewed, including 4 non-English studies. Finally 13 studies^29–41 were included in this review, including 1 non-English study.³⁶ No study was excluded because of translation difficulties, but 3 of the 4 unobtainable studies were non-English.

The Maastricht-Amsterdam List quality scores for the 13 included studies are shown in Table 2. Two high-quality RCTs gave level 1b evidence^33,37 and 3 low quality RCTs gave level 2b evidence.^35,36,41 The 8 noncontrolled studies gave level 4 evidence of which 6 met the criteria for “sufficient quality”^28–32,39 and 2 were of lower quality.^34,40 The 5 RCTs yield a grade B recommendation for treatment²⁸ (Appendix B).

TABLE 2: Maastricht-Amsterdam List Sectional Scores²² and Levels of Evidence²⁸

Internal validity was compromised by lack of blinding. Subjects and therapists cannot be blinded to strength training but only 6 of 13 studies^{30,32,33,37,39,41} blinded the assessors (Table 2). Avoidance of cointerventions was inadequate in 8 of 13 studies^{32,34–36,38–41} and adequate compliance was described in only 7 of 13 studies.^{29–31,33,37–39} The method of randomization was not clear in 3 of 5 RCTs.^35–37 However, most studies either had no drop-outs or described these subjects adequately. The descriptive and statistical criteria of the MAL tool were mostly fulfilled with the exception of a long-term outcome which was carried out in only 4 studies.

A total of 138 subjects participated in strength training, most were described as having spastic-type CP, mainly of hemiplegic or diplegic distribution. Most studies provided the intervention 3 times per week (range 1–5) over 6 weeks (range 4–12). Ten studies involved open and/or closed chain isotonic exercises and 2^35,38 gave isokinetic exercises. Resistance was applied using free weights, body weight, weighted backpacks, elastic bands, weight machines, or an isokinetic machine. Most studies included warm-up and cool-down activities, some included stretches or balance exercises, but all described strength training as the major intervention. The setting varied but all except one³⁴ had adult supervision.

Study data are shown in Tables 3 and 4. Data for function and gait outcomes were incomplete as some studies did not provide sufficient information for effect sizes to be calculated. Data could not be pooled due to heterogeneity between studies and missing data. The 2 reports of the study by Damiano et al^30,31 were considered 1 study to avoid double-counting.

TABLE 3: Summary of Study Data

TABLE 3: (Continued)

TABLE 3: (Continued)

TABLE 4: Study Data—Outcomes

TABLE 4: (Continued)

TABLE 4: (Continued)

Strength

Seven studies^{29–33,37,39,40} measured isometric strength and overall this was increased following isotonic exercising. The 2 studies using isokinetic exercises^35,38 measured results isokinetically and found a significant improvement. One study providing isotonic exercises⁴⁰ found both isokinetic and isometric strength was improved. Measures of effect all favored treatment, with 7 of these regarded as large effects²⁷ (Table 5).

TABLE 5: Within-Group Measures of Effect (Pre/Postintervention) in Subgroups by Age

Two studies^30–32 found strengthening the quadriceps moved the quadriceps-hamstrings ratio further from the normal, but 1 study³⁹ strengthened quadriceps and hamstrings and found the ratio moved nearer to normal. One study⁴⁰ found that isometric quadriceps strength showed greatest improvements at longer muscle lengths but another^30,31 found greatest gains at shorter lengths.

Function

To measure functional changes, 7 studies used the Gross Motor Function Measure. Six^{32,33,35–37,39} of these studies found some improvement at least in dimension E and one³⁸ did not report scores but approximately half the sample improved. One further study²⁹ found significant improvement in 3 weight-bearing functional tests; where effect size could be calculated all favored treatment, but 3 of 6^32,33,35 studies showed small effects²⁷ (Table 5).

Gait

Most studies assessed gait. Four studies^{30–32,35,41} used 3D gait analysis (3DGA) and found improvements in some gait parameters, although this was not always statistically significant. One study³⁴ using visual gait analysis found similar results. Timed walking tests^{29,33,34,36,37,39} found mixed results and timed stair-climbing³³ showed nonsignificant improvements. All studies favored treatment, with small-to-moderate effects²⁷ (Table 5).

Subgroup Results

Preadolescent children showed large effects for strength and moderate effects for gait; adolescents also showed large effects for strength, but smaller effects for gait (Table 5). Subgroup results concerning clinical severity were mostly unidentifiable because of lack of individual or subgroup data in reports. Children with diplegia showed greater strength gains than those with hemiplegia in 2 studies,^32,38 but no difference in 1 study.⁴¹ Two studies^34,36 gave no diagnostic subgroup results. Subjects with diplegia comprised the entire sample in 7 studies.^{29,30,31,33,35,37,39,40}

Detraining

Reassessment after a period of detraining found isometric strength was maintained or slightly increased^29,33,39 but isokinetic strength deteriorated³⁸ (Fig. 1A). Function test results^29,33 were maintained or slightly increased (Fig. 1B). Gait parameters showed a reduction of gains seen post-training, although these did not return to baseline levels (Fig. 1C). The length of detraining for these studies ranged from 4 to 12 weeks.

Fig. 1.: (A) Effects of detraining on strength measures. (A) Effects of detraining on function measures. Macphail and Kramer³⁸ did not reassess function at follow-up; Morton et al³⁹ found significant increase in GMFM-E at follow-up but gave insufficient data to calculate effect size. (C) Effects of detraining on gait measures. Macphail and Kramer³⁸ did not reassess gait at follow-up.

Other Results

No study found an increase in muscle tone; 2 found a significant decrease, as measured by an isokinetic dynamometer³⁵ or by resistance to passive stretch on a myometer.³⁹ A few mild adverse events were recorded which were mostly muscle or joint soreness. The most serious adverse event was development of knee hyperextension in 5 subjects^30,31 who had previously had hamstring-lengthening surgery.

DISCUSSION

Quality of Included Studies

Quality scores indicate sufficient internal validity in more noncontrolled studies (5 of 7) than RCTs (2 of 5). The effects in the noncontrolled studies were mostly larger than in the RCTs, but in the same direction. It is acknowledged that a systematic review including more high quality RCTs would produce more robust evidence, but given the paucity of RCTs in this clinical area the included noncontrolled studies are valuable.⁴² Their findings add strength to the body of evidence.

Effect of Strength on Function and Gait

An intervention should have an effect on the level of disability measured by activity and participation outcomes.⁴³ Knee muscle strength is moderately and significantly correlated with GMFM scores and standing balance in children and young people with CP.^15,44–46 Many neuromuscular and musculoskeletal deficits, however, interfere with motor function in CP and successful intervention depends on identifying which is the most limiting impairment.⁴⁷ Impairments other than strength may have been significant as several included studies^33,39,41 indicated that some subjects had previously had interventions for spasticity, soft tissue contractures, and bony deformities. Selective motor control is infrequently measured but was found to be significantly limiting in younger children.⁴⁸ Teenagers may have more advanced soft tissue contractures and joint deformities that could limit improvements in gait and function, independent of muscle strength. The GMFM may have a ceiling effect in older subjects, thereby masking true functional gains.⁴⁹ As gait deterioration is common in teenagers with CP it may be that nonsignificant gains³⁸ were clinically meaningful, in that subjects did not deteriorate. Here the intervention is fighting the natural course of the condition, whereas preadolescent children are likely to acquire motor skills⁴ and the strength training simply enhances this.

There may have been a ceiling effect in the more able subjects with hemiplegia.⁴⁹ Strength training in adults following stroke showed variable carryover to function⁵⁰ which was possibly partly due to compensation using the uninvolved side. If subjects with hemiplegic CP have developed compensatory ways of functioning, increased strength may not alter or normalize these. The involved side remains weaker than the uninvolved side, especially during short strength-training interventions. Subjects with diplegia are less likely to compensate in this way so weakness may be more clearly reflected in functional scores. The effect of strength on athletic performance in children who are developing typically is inconclusive.⁵¹ If the movement quality of subjects with CP is altered following training, the GMFM does not identify this.⁵²

It has been suggested that strength-training protocols should match functional tasks to facilitate learning⁵³ but this was performed in only 3 studies.^29,36,37 Motor learning is also influenced by an individual’s attitude, environment, and capacity to solve motor problems and harness abilities in different contexts.⁵⁴ Carryover of strength gains to functional activities may not be immediate and 3 studies^29,33,39 showed higher function scores at detraining follow-up. Strength gains may have allowed these children to practice improved functional activities without further training. Croce and DePaepe⁵⁵ argue that children must learn an action to the point of automatic execution before it is incorporated into other activities, which could be related to the intensity and length of studies.

Crouch Gait

Three studies^30,31,35,41 found significantly improved knee extension during walking. Crouch gait with excessive hip and knee flexion in stance is common in subjects with CP and deteriorates without intervention.⁵⁶ During normal gait, the hip and knee extensors work in a shortened position. Isometric testing identified muscles of subjects with CP appear weaker when working at shorter lengths^30,31,40,57; they may collapse into crouch because they are unable to exert sufficient extensor force in the shortened position. It is unclear whether strength training can alter this length-tension relationship. One study of isotonic open-chain strength training^30,31 identified greater strength gains at the shortest quadriceps length, possibly because the free leg achieves greater knee extension during training without opposition by body weight. The quadriceps work therefore in their weaker shortened range where a training effect may then occur. One study^30,31 found knee extension at heel strike significantly improved: the similarity of this gait action to the training action may account for this. A muscle, however, rarely functions in isolation; the incomplete carryover of strength to function and gait following open-chain exercising may be because of differences in the intensity, frequency, and range of motion used in the weight-bearing studies.⁵⁸ Additionally, as isolated work at one joint is often difficult for children with spasticity, whole-limb activities may be more achievable.^48,59

Training should be specific regarding muscle length, peak force, activation patterns, timing, amplitude, and contraction type.⁵⁸ The 2 isokinetic studies^35,38 found little carryover to function and gait parameters; this indicates strength gained isokinetically may be less functionally useful, as isokinetic movement does not occur in everyday function. Closed-chain isotonic exercises however are similar to the gait parameters the intervention is designed to improve. If strength gains are initially due to improved neural activation, then the training effect from closed-chain exercise is more specific to gait and weight-bearing functions. Additionally Bobath¹² suggested weakness may partly be attributable to tactile or proprioceptive sensory deficits and is counteracted by massive sensory stimulation. Electrical stimulation increases strength perhaps partly because it alters sensory inputs that affect the excitability of interneurones and motor neurones.¹⁷ Closed-chain strength training may enhance sensory inputs by stimulating the proprioceptors in weight-bearing tissues which may improve motor recruitment. Sensory input may also improve body image deficiency, thereby facilitating better voluntary control.⁶⁰

Gait Velocity

The crouched position is inefficient, which reduces gait velocity.⁶¹ Whereas, there is good correlation between strength and velocity in subjects who are developing typically, improved strength in subjects with CP can only increase velocity in the absence of other more limiting impairments.⁶² When subjects with CP walk faster, they tend to increase cadence^32,39 rather than stride length and show increased pelvic excursion, but not increased knee and hip excursion.⁶³ The child relies on proximal trunk muscles to propel the legs forwards, so improved leg strength may have limited effects on velocity.⁶⁴ Significantly increased stride length^{29,30,31,34,35,39} may be indicative of reduced crouch or greater stability of the stance leg.⁴⁶

Most studies using the timed stair climb and timed 10 m walk,^29,33,36,39 or measuring maximum velocity walking as part of 3DGA,^30–32,35 found some improvement. These tests assess maximum-effort performance and may reflect the maximum-effort training protocol used. One study³⁷ found no significant change in maximum gait velocity; this study trained younger subjects by resisted sit-to-stand exercise, which may not have sufficiently influenced activity of the lower limb extensors in an upright walking position.

The 2 or 3-minute walk test at a self-selected pace is more indicative of walking ability for everyday function. Four studies^32,34–36 found significant improvement in velocity, but another 4 found no significant change.^{29–31,38,41} Subjects with CP expend 2 to 3 times the normal amount of energy in walking, which causes excessive fatigue; they tend to select a velocity and cadence that minimizes this fatigue.⁶⁵ Fatigue is identified as significant in many adults with CP⁶⁶ and may be the most limiting factor in the walking test, indicating strength training may not adequately address this factor.

Gait Abnormality

Thelen et al⁶⁷ argue abnormal coupling of hip and knee moments indicates CP gait is inherently altered and is not simply a weaker form of normal gait. Steinwender et al⁶⁸ suggest that as the muscle activation pattern is abnormal, it may not be improved by strength training and it may even be reinforced. Farmer⁶ suggests gait in subjects with CP is immature rather than abnormal, as the muscle firing sequences are similar to those of infants developing typically, but not similar to those of children who are maturing. Weakness and contractures induce compensatory strategies, eventually producing a gait that is both altered and slower.^64,69 The wide range of gait results at self-selected and fast velocities indicate subjects showed various abnormalities and compensations, so general conclusions cannot be drawn.

Gait Deterioration

The maintenance of walking into adulthood is a concern often voiced by parents, this may be influenced by many factors.⁶⁶ Stride length reduces with age and double support increases.⁶ Joint deterioration is a significant factor in loss of walking, and the increased patellofemoral force induced by the crouched posture can cause chronic pain.^5,56 The effort required for walking increases with increasing height and weight, which causes premature fatigue.^70,71 Furukawa et al⁷² found reduced coordination and stability because of deformity, spasticity and weakness, which explained the deterioration in walking ability of 18 children with CP aged 9 to 14 years. General fitness was not a significant factor. Loss of confidence also affects young people’s walking ability.⁷⁰ Strength training could positively influence many of these factors by increasing stride length^{29–31,34,35} and joint excursion,^30,31,35 and by decreasing double support³² and crouch.^30,31,35,41 Psychological benefits were not examined by this review, but some evidence exists for increased self-confidence and self- esteem which may encourage young people to realize greater physical potential.⁷³ Nevertheless, literature concerning gait in young people with CP lacks specific evidence about who is most susceptible to deterioration and why.⁷⁴ Future research should examine factors predisposing to gait deterioration and the use of strength training in conjunction with other interventions to maximize retention of functional and gait abilities.

Limitations of this Review

Unpublished or nonindexed foreign-language studies may have been missed by the search strategy. These could be more likely to have negative results which would affect the review’s findings.^21,75 Study comparisons were limited as not all measured the same variables; some data were insufficient and missing data were not requested from the authors. Some eligible subjects in other studies spanning the age-limit of 20 years were excluded, as authors could not be contacted for individual data within the parameters of this review. The MAL tool has face and content validity and interrater reliability, but no other established psychometric properties.⁷⁶ This is in common with many quality assessment tools used in rehabilitation research. Although it was developed from 2 tools with well-established validity and reliability, the Jadad Scale²⁴ and the Delphi List²⁵, their psychometric credentials cannot be directly transferred to the new tool.⁷⁶ The use of Operational Guidelines for this systematic review resulted in substantial interrater reliability of the MAL.²⁷ Within-group effect size calculations enabled comparisons between all studies, but between-group calculations were not performed for the 5 RCTs. Incomplete age and diagnosis subgroup data limited inferences.

CONCLUSIONS

Isotonic strength training was associated with moderate-to- large strength and function gains which were maintained after detraining, and small-to-moderate gait improvements that partially deteriorated during detraining. Two studies found isokinetic training enhanced strength with less significant effects on gait and function and some loss after detraining. Strength training may enhance motor development in younger children; it may counteract deterioration in youth, where statistically nonsignificant results may be clinically meaningful. Strength training may ameliorate some aspects of gait in children and youth with CP in the absence of more limiting factors. Secondary impairments, particularly in postadolescent youth, may limit carryover of strength gains to gait and function. Subjects experienced no increased spasticity or other serious side-effects.

The 3 previous reviews of English publications^18–20 concluded strength training in CP improves muscle strength but were inconclusive regarding gait and function. They included more heterogeneous populations than this review, but presented no subgroup results. Detraining effects were not discussed. This review is therefore in accordance with previous findings, following a wider language-inclusive search and analysis of more recent literature. This review concerns a narrower population, adds findings for subgroups and detraining effects, and contributes to knowledge regarding gait and function outcomes.