Neuromuscular adaptations to eccentric strength training in children and adolescents with cerebral palsy


Dr Siobhán Reid at the School of Sport Science, Exercise, and Health, University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia. E-mail:


Aim  To determine the neuromuscular outcomes of an eccentric strength-training programme for children and adolescents with cerebral palsy (CP).

Method  In this randomised, parallel-group trial with waiting control, 14 participants with CP (six males, eight females; mean age 11y, SD 2y range 9–15y), diagnosed with upper-limb spasticity were compared with 14 age- and sex-matched typically developing participants. Participants with CP completed a 6-week progressive resistance-strengthening programme, performing eccentric lengthening contractions of their upper limb three times a week. Data from dynamometer and surface electromyography (EMG) assessments included peak torque normalised to body mass (T/Bm), work normalised to body mass (W/Bm), angle at peak torque, curve width, and EMG activation.

Results  After training, children with CP had improved eccentric T/Bm (p=0.009) and W/Bm (p=0.009) to a level similar to that of the typically developing children. No change in angle of peak torque occurred, although curve width increased both concentrically (p=0.018) and eccentrically (p=0.015). EMG activity was elevated before training in children with CP but decreased with training to levels similar to those of the typically developing children.

Interpretation  With eccentric strength training, children with CP increased torque throughout range of motion. Results suggest that eccentric exercises may decrease co-contraction, improving net torque development. Eccentric actions may be important in the maintenance of the torque–angle relationship. These results have significant implications for the prescription of strength-training programmes for people with CP.


Peak torque normalised to body mass


Work normalised to body mass

Strength-training programmes are providing encouraging results for children with cerebral palsy (CP), particularly in addressing muscle weakness and improving functional capabilities.1 Muscle weakness disrupts an individual’s ability to accrue sufficient force for the performance of activities and is a major factor in dysfunctional movement, particularly in those with CP.2,3

Muscle weakness may also be associated with muscle shortening.4 Alterations in muscle length potentially affect the force-generating capacity of muscle.5,6 Muscles have an optimal length for the generation of maximal force, with the force–length relationship producing a inverted-U-shaped curve, with the most effective force being generated at mid-range.5 As most children with CP experience altered muscle lengths to some degree, it has been argued that muscle-strengthening programmes should be focused on maintaining muscular length as well as increasing the force-generating capacity of the muscles.4

The potential to increase both force generation and active muscle range may be achieved with the inclusion of eccentric activity in strength-training programmes. In weightlifting terms, eccentric actions are generally thought of as the lowering phase. More precisely, eccentric actions are defined as muscular contractions whereby the load force is greater than the resisting muscle force, and as a consequence the active muscle is lengthened.5

Around a joint axis, the force–length relationship of a muscle is manifest as a torque–angle relationship, which can be determined using dynamometry.7 It is well documented that eccentric training results in shifts in the torque–angle relationship of muscle to that of longer lengths.8,9 Brockett et al.7 provided evidence that a single bout of eccentric exercise induced a sustained shift in the angle at peak torque and width of the torque–angle relationship. Morgan and Proske8 postulate that the mechanism for the shift is related to increased numbers of sarcomeres in series to maximise muscle efficiency. Therefore, eccentric exercise programmes may benefit children with CP by shifting the torque–angle relationship, potentially enhancing biomechanical performance of the muscle.

Reports of the physiological and neurological differences between eccentric and concentric muscle actions paint an encouraging picture for their application to clinical rehabilitation programmes. Muscles are able to produce greater force, at greater velocities, and with less neural drive eccentrically than concentrically.10,11 Data also indicate that the greatest gains in muscle hypertrophy occur in response to eccentric muscle actions.12,13 Although there are still some unknowns, the available evidence appears to support the assertion that eccentric training may be an effective method of increasing muscle-force production.10,14 This premise may be particularly important for people with profound weakness and a very low capacity to generate muscle force.15 Simultaneously, eccentric training may redress the consequences of muscle shortening by altering the length–tension characteristics of muscle to operate at longer lengths.7,16 Eccentric training may also provide a neurological advantage for those with deficient motor control, as it is less demanding of the neurological system.8,9,11,15,17 Despite this strong rationale, there is a distinct lack of research into the applicability of eccentric training in exercise programmes for children and adolescents with CP.

The aims of the present study, involving adolescents with CP, were, twofold: (1) to determine the effect of an upper-limb eccentric-only training programme on neuromuscular function, specifically peak torque, work, variables of the torque–angle relationship, and electromyographic (EMG) activation; and (2) to compare these data with results from typically developing children and adolescents. We predicted that, before training, children with CP would have decreased peak torque and work across the strength assessments, compared with their typically developing peers, and that this would improve after eccentric training. Second, we predicted that the torque–angle relationship would be biassed to that of shorter muscle lengths in the children with CP and that this would show improvement after eccentric training. Finally, we predicted that children with CP would have depressed EMG activation across all isokinetic tasks compared with typically developing children and that this would again improve after eccentric strength training.


We used a randomised, parallel-group trial design, with a waiting control group and secondary comparisons with a matched group of typically developing children and adolescents. Participants with CP were randomised into two groups and completed the same eccentric training programme: group I (n=7) trained in the first 6 weeks of the study, while group II (n=7) acted as a control group, maintaining their normal activity for 6 weeks, and then completed the 6-week eccentric training programme. Data from the non-training period of group II were analysed to ensure no change as a result of exposure to the testing environment. Data from the combined cohort (n=14) were analysed to determine the effect of eccentric strength training on neuromuscular function in adolescents with CP. Secondary comparisons were made with data from 14 typically developing participants. The study received ethical approval from the human research ethics committee of the University of Western Australia, and written informed consent was obtained from each participating family.


Participants were recruited through advertisements at therapy centres across the Perth Metropolitan area. Fourteen of a possible 15 participants met the selection criteria of elbow flexor spasticity, the ability to follow two-step instructions, no previous upper-limb surgery, and no upper-limb strength training or pharmacological treatment for spasticity (botulinum toxin A) in the past 12 months.

The cohort included six males and eight females, with a mean age of 11 years (SD 2y, range 9–15y). Thirteen of the participants had a diagnosis of spastic hemiplegia, and one had spastic triplegia. Three participants had right-side involvement, and 11 had left-side involvement. Four children were classified at level I, eight at level II, and two at level III on the Manual Ability Classification System,18 with an even distribution between the two groups. An age- and sex-matched sample of 14 typically developing participants was recruited for comparative purposes.

Eccentric intervention and assessment

All of the children with CP underwent a 6-week, home-based, upper-limb eccentric training programme targeting the elbow flexors. Each child was provided with an individually adapted eccentric training rig that permitted eccentric extension of the elbow flexors only. Each child exercised seated at the training rig, with his or her shoulder flexed to 90°, and eccentrically extended the elbow through a standardised range of motion from 110° to 40°, with 0° equating to full elbow extension. The training rig provided loaded assistance to draw the participants into elbow extension in a gravity-eliminated position. Participants were required to actively resist the forced extension by recruiting their elbow flexors eccentrically to control the movement. Participants were instructed to count to 10 through the eccentric lowering phase, working at approximately 7°/second. With the help of a training partner, the loaded arm of the rig was returned to 110° elbow flexion ready for the next repetition. Participants did not perform a concentric muscle action during this recovery period.

Participants trained three times weekly, completing three sets of 10 repetitions at each session. The training loads were individually calculated on the basis of participants’ maximum eccentric performance during dynamometry assessments. Training progressed incrementally by 5% each week, commencing at 50% maximum eccentric torque, and, after reassessment of strength at week 3, loads continued to advance in 5% increments up to 70% by week 6.

Strength assessments were performed on a Biodex dynamometer (System 2; Biodex Medical Systems Inc., New York, USA) in a gravity-eliminated position. All children with CP completed five strength assessments during the 12-week study, at weeks 1, 3, 6, 9, and 12. Data from weeks 3 and 9 were not used in the analysis, serving only to enable load progression monitoring during training. Participants without CP completed the strength assessment once only, these data forming reference values for comparison. Strength assessments involved a series of three isokinetic concentric trials at each of the following velocities, 30°/s, 60°/s and 90°/s and three eccentric trials at 30°/s. Isokinetic trials replicated the range of motion from the training sequence. Participants were given continual verbal encouragement to achieve maximal results. The best performance of the three trials was then used in the analysis. Simultaneous surface EMG, collected at 1000Hz, recorded the activity of the biceps brachii and brachioradialis during the isokinetic assessments.

Data analysis

Raw EMG data were high-pass filtered using a fourth-order, zero-lag 30Hz Butterworth filter, full-wave rectified, and low-pass filtered using a fourth-order, zero-lag 5Hz Butterworth filter. EMG data were normalised to individual isometric maximum, and the activation patterns of the biceps brachii and brachioradialis were combined to provide an estimation of overall flexor activity. As changes in the torque–angle relationship were predicted, EMG data were then analysed in subsections according to elbow range of motion of 60 to 69°, 70 to 79°, and 80 to 100°.

The torque and joint-angle data were filtered using a fourth-order, zero-lag Butterworth 5Hz low-pass filter. The variables of peak torque (T/Bm) and work (W/Bm) were normalised to body mass. The torque–angle relationship data were analysed by fitting least-squares Gaussian curves to data greater than 90% maximum torque, similar to the method described by Brockett et al.7 This method identifies the angle at which peak torque is generated and the width of the curve at which muscular torque is sustained above 50% of maximum.

Normal-probability Q–Q plots revealed normal distributions on the variables of interest; therefore parametric statistics were used. Independent t-tests were conducted to establish group equivalence at baseline. Paired t-tests were conducted to determine stability of group II data during the non-training control period and the outcome of training in the combined cohort of children with CP. Finally, data from participants with and without CP were compared using between-groups analyses of variance.


On average, participants completed 16 of a possible 18 eccentric training sessions in the 6-week period. No child reported muscle soreness as a consequence of the eccentric training.

Peak torque and work

No statistically significant difference (p>0.05) was found between groups I and II for T/Bm or W/Bm in any of the baseline assessments, denoting group equivalence. Peak torque improved in the concentric task at 30°/s (F(1,5)=6.641, p=0.042) over the non-training control period for group II, possibly reflecting familiarity with testing; however, no difference was found in any other condition.

Significant improvements from pre- to posttraining were seen in T/Bm concentrically at 90°/s (t(13)=4.825, p=0.047), improving from a mean of 37.30Nm/kg (SD 19.47) to 42.87Nm/kg (SD 18.86; Table I). Similarly, T/Bm improved in the eccentric task (t(13)=9.532, p=0.009), showing a 25% increase from pretraining (mean 59.33Nm/kg, SD 19.93) to posttraining (mean 73.79Nm/kg, SD 26.57). Significant improvements were also seen in eccentric W/Bm (t(13)=9.488, p=0.009), improving by 42% from pretraining (mean 43.3Jj/kg, SD 22.47) to posttraining (mean 61.69J/kg, SD 26.32).

Table I.   Peak torque and work rates, normalised to body mass, across isokinetic concentric and eccentric tasks at various velocities for children with cerebral palsy (CP, n=14) before and after training and compared with typically developing children (n=14)
ConditionChildren with CPTypically developing children
Before trainingAfter training
  1. ap<0.01 vs typically developing children; bp<0.05 vs pretraining scores of participants with CP.

Torque/body mass, Nm/kg, mean (SD)
 Concentric 30°/s42.37a (18.92)46.45a (19.74)67.54 (13.80)
 Concentric 60°/s40.31a (20.69)44.05a (20.71)66.06 (17.52)
 Concentric 90°/s37.30a (19.47)42.87a,b (18.86)61.24 (13.87)
 Eccentric 30°/s59.33a (19.27)73.79b (26.57)84.21 (19.06)
Work/body mass, j/kg, mean (SD)
 Concentric 30°/s38.64a (21.02)41.32a (18.40)69.87 (15.58)
 Concentric 60°/s34.08a (21.85)39.61a (20.28)67.78 (17.34)
 Concentric 90°/s32.01a (21.21)36.90a (18.08)60.75 (14.60)
 Eccentric 30°/s43.37a (22.47)61.69b (26.32)79.96 (23.86)

Comparisons between children with and without CP revealed significant differences in T/Bm and W/Bm across all tests before eccentric training (p<0.01). By the completion of the training, these differences in T/Bm and W/Bm no longer existed for the eccentric condition between children with and without CP (see Table I).

Torque–angle relationship

Analysis of the torque–angle relationship revealed no significant difference in curve width (p>0.05) between the two CP groups at baseline, although the mean angle at peak torque differed significantly between the groups for a single task, the concentric 30°/second condition (t(13)=−2.230, p=0.046). Therefore, as there was minimal discrepancy, group equivalence was assumed. Variability in angle at peak torque in the concentric tasks was demonstrated for group II during the non-training control period. However curve width remained stable, with no significant difference detected across the conditions over the control period.

No significant change in mean angle at peak torque was observed across any isokinetic task from pre- to posteccentric training for the entire cohort of children with CP. Figure 1 shows changes in the torque–angle relationship for the concentric 60°/second task for a child with CP before and after training compared with the mean of the typically developing participants, the angle at peak torque is marked with the arrow head. In the entire CP cohort, significant improvements were seen from pre- to posttraining in the mean curve width concentrically at 30°/second (t(13)=−2.356, p=0.035) and 90°/s (t(13)=−2.699, p=0.018). Similarly, the mean curve width improved from pre- to posttraining eccentrically (t(13)=−2.803, p=0.015) from 23.25° (SD 6.49) to 37.03° (SD 15.95).

Figure 1.

 Torque–angle relationship during the concentric 60°/s task for a child with cerebral palsy compared with the mean of the typically developing participants.

When the pre- and posttraining angles at peak torque of the children with CP were compared with those of typically developing children, no significant difference was found for any isokinetic condition (p>0.05). Before the eccentric training programme, significant differences in curve width were found in each isokinetic condition between participants with and without CP (p<0.05). In each case the range over which torque could be sustained above 50% of maximum (curve width) was greater in typically developing participants. By the completion of the eccentric training programme, these differences in curve width no longer existed across any isokinetic task (p>0.05).


EMG data could not be collected for two participants (one in each CP group). Therefore, data from these participants have been removed from the analysis. No significant difference was found in EMG activation between the two groups of children with CP at baseline on any isokinetic task, except for the 60 to 69° phase of the concentric 60°/second task (t(11)=−2.303, p=0.044). As the discrepancy between the groups was minimal, group equivalence was assumed. No difference (p>0.05) was observed in EMG activation over the non-training control period for participants in group II (n=6).

EMG activation decreased relative to peak activation from pretraining (mean 56.18%, SD 8.73) to posttraining (mean 41.20%, SD 17.04) in the 80 to 100° phase of the eccentric task (t(11)=9.334, p=0.011; Fig. 2). After training, a consistent trend was found for decreased neural activation relative to peak activation across the concentric tasks; however, this failed to reach statistical significance (see Fig. 2a).

Figure 2.

 Electromyographic EMG activation during (a) concentric (90deg/second) and (b) eccentric tasks.

When the pretraining EMG activation scores of the children with CP were compared with those of typically developing children, significant differences were found in EMG activation during the 80 to 100° phase in the three concentric tasks at 30°/s (F(1,23)=4.507, p=0.045), 60°/s (F(1,23)=5.594, p=0.027), and 90°/s (F(1,23)=6.894, p=0.015; Fig. 2a). In each case the children with CP displayed greater EMG activation than typically developing participants. By the completion of eccentric training the differences in EMG activation in the 80 to 100° phase of the concentric tasks ceased to exist between those with and without CP (p≥0.05).

Before eccentric training, no difference was found in eccentric EMG activation between participants with and without CP. However, by the completion of the eccentric training, a significant difference in EMG activation was evident in the 80 to 100° phase (F(1,23)=8.274, p=0.009). In a reversal of the pretraining findings, by completion of the training children with CP performed the eccentric task with less EMG activation (mean 41.70%, SD 17.04) than their typically developing peers (mean 56.58%, SD 5.52; Fig. 2b).


The aims of this research were to determine whether an eccentric training programme altered torque and work production, the torque–angle relationship of muscle, and EMG activation for children and adolescents with CP relatively to their typically developing peers. The limitations associated with a randomised parallel-group design were reduced by the inclusion of a waiting control group (group II), which showed little change across the non-training control period, with the exception of improved concentric T/Bm. This result may be due to increased familiarity with the strength testing scenario and presents as a learning effect. However, as the improvement occurred concentrically while training targeted eccentric outcomes, this may be deemed as of little consequence to the generalisability of the results.

Impact of training on peak torque (T/Bm and angle of peak torque)

Before eccentric training, T/Bm was significantly different between participants with and without CP across all tasks. In every case, those with CP produced significantly less torque than their typically developing peers. These results are in agreement with previous research findings for the lower limb, in which children with CP displayed approximately half the strength capability of their able-bodied peers.19,20 However, contrary to our predictions, the angle at which peak torque occurred was not different between those with and without CP. However, differences in the torque–angle relationship may be more apparent in children with greater involvement than the current sample (Manual Ability Classification System levels I–III) or in the presence of muscle contracture, and this relationship warrants further investigation.

Further, the eccentric training programme did not alter the torque–angle relationship. In contrast to our expectations, the peak of the torque–angle curve did not shift to longer muscle lengths as described in current literature.7,9 This finding implies that the eccentric training programme was not of sufficient intensity to induce changes in the torque–angle relationship, commonly inferring increased number of sarcomeres in series. In spite of this, children with CP did show significant increases in T/Bm after training. Specifically, changes were noted in the eccentric condition and concentrically at 90°/s. The significant differences in eccentric T/Bm no longer existed between those with and without CP after training. Other researchers have also reported this phenomenon after lower-limb strength training,9,21 leading to speculation that eccentric actions may be an important inclusion in any strengthening routine, particularly for people with CP.

Impact of training on torque production through range of motion (W/Bm and curve width)

Significant differences were revealed in W/Bm and curve width between participants with and without CP across all tasks before training. In each case, those with CP produced less W/Bm, and the width of the curve was narrower, than their typically developing peers. Improvements were demonstrated in both W/Bm and curve width after the eccentric training programme for those with CP. Unexpectedly, the improvements in W/Bm were such that the statistical differences in eccentric capabilities no longer existed between those with and without CP.

After the eccentric strength-training programme, significant improvements in the curve width were demonstrated during both concentric and eccentric tasks. Once again, these improvements in curve width resulted in no statistical difference between those with and without CP, for any isokinetic task. Damiano et al.21 have reported that, after a concentric/eccentric training protocol, participants with CP could approximate more closely the torque–angle relationship of typically developing children. This suggests that a ‘normal’ width of the torque–angle curve may be maintained with regular eccentric loading of muscle. Again, this finding has important implications for the prescription of strength-training programmes for people with CP. These results suggest that the components of the torque–angle relationship (namely angle at peak torque, curve width, and W/Bm) are robust, sensitive estimates of the integrity of muscle torque-generating capacity in this population.

The question remains whether the resultant changes from eccentric exercise are neural or mechanical in origin. Previous researchers have proposed that the mechanism relates to the addition of sarcomeres in series, leading to a muscle fibre that generates tension over a greater range of motion.8,9 No change in the angle at peak torque makes it unlikely that the training resulted in the addition of sarcomeres in series. However, it is conceivable that the change in curve width could be attributed to an increase in the number of functional sarcomeres in series,7 leading to greater torque generation throughout range of motion. Alternatively, neurological adaptations may be the source of improved force generation.

Impact of training on EMG activation through range of movement

In contrast to our expectations, children with CP displayed increased EMG activation across the concentric tasks in comparison with their typically developing peers, particularly in the phase corresponding to angle at peak torque (80–100°). After the training programme a consistent trend occurred whereby EMG activation decreased for those with CP. This trend reached significance in the eccentric task, possibly indicating increased specificity of neural recruitment patterns after eccentric training. Generally, these decreases lead to activation scores more closely representing those of the typically developing reference population. The downregulation of EMG activation to that of normal levels may be linked to the evidence that eccentric muscle actions require less active neural control than do concentric actions.13,16,17 A decrease in motor unit recruitment in eccentric muscle actions may suggest better use of elastic energy or a more favourable length–tension relationship.22 Exercising eccentrically may increase neural efficiency, while simultaneously increasing torque production, a benefit that is of particular relevance to people with spasticity.

Overall, the eccentric training programme resulted in increases in eccentric T/Bm and W/Bm to levels no different from the typically developing population, increases in the width of the torque–angle relationship across both concentric and eccentric tasks, to a range no different from the typically developing population, and no change in angle of peak torque, although no difference from the normative population existed at baseline. These changes all occurred while EMG activation tended to decrease, particularly over the 80 to 100° phase, the area of peak torque production. This increased output (torque) with decreased input (EMG activation) suggests a decrease in co-contraction after eccentric training. It is well known that children with CP have diminished net force production and excessive co-contraction during functional activities compared with typically developing children.21 However, the conclusions of the current study are limited by the non-inclusion of the antagonist EMG activation (triceps). It is conceivable that a decrease in co-contraction potentially allows the agonist to increase peak torque (T/Bm), work output (W/Bm), and the capacity to sustain torque throughout range of motion (curve width), thereby improving task efficiency. These data add credence to the body of evidence suggesting that co-contraction hampers the force-production capabilities of the agonist muscles during voluntary activations in children with CP,23 and this warrants further investigation. The suggestion that a decrease in co-contraction follows eccentric strength training has significant implications for the prescription of strength training for people with CP.