Increased co-activation during clinical tests of spasticity is associated with increased co-activation during reactive standing balance control in cerebral palsy

Abstract

Joint hyper-resistance is a common symptom in cerebral palsy (CP). It is assessed by rotating the joint of a relaxed patient. Joint rotations also occur when perturbing functional movements. Therefore, joint hyper-resistance might contribute to reactive balance impairments in CP. Our aim was to investigate relationships between altered muscle responses to isolated joint rotations and perturbations of standing balance in children with CP. Twenty children with CP and twenty typically developing children participated in the study. During an instrumented spasticity assessment, the ankle was rotated as fast as possible from maximal plantarflexion towards maximal dorsiflexion. Standing balance was perturbed by backward support-surface translations and toe-up support-surface rotations. Gastrocnemius, soleus, and tibialis anterior electromyography was measured. We evaluated alterations in reciprocal pathways by plantarflexor-dorsiflexor co-activation and the neural response to stretch by average muscle activity. We evaluated the relation between muscle responses to ankle rotation and balance perturbations using linear mixed models. Co-activation during isolated joint rotations and perturbations of standing balance was correlated in CP but not in typically developing children. The neural response to stretch during isolated joint rotations and balance perturbations was not correlated. Our results suggest that increased co-activation, possibly due to reduced reciprocal inhibition, during isolated joint rotations might be a predictor of altered reactive balance control strategies in CP.

NEW & NOTEWORTHY

It has been challenging to relate altered muscle coordination during functional movements to altered muscle’s responses to isolated joint rotations in children with cerebral palsy. We performed more comprehensive assessments by not only considering mean muscle activity but also agonist-antagonist co-activation. Our results indicate that muscle co-activation during balance control might partly result from altered reciprocal pathways, e.g. reduced reciprocal inhibition, in the spinal cord. These insights could improve clinical assessments of balance impairments.

Graphical Abstract

INTRODUCTION

Joint hyper-resistance is the most common symptom in children with cerebral palsy (CP) (1). Joint hyper-resistance is clinically evaluated by assessing the resistance against an imposed passive muscle stretch (1). Muscle stretches also occur when functional movements are perturbed (e.g., when standing on a departing bus). Therefore, joint hyper-resistance might contribute to balance impairments that are common in children with CP (2). Yet, little is known about the relation between joint hyper-resistance and balance impairments in CP.

Children with CP have higher muscle activity and higher levels of agonist-antagonist co-activation in response to perturbations of standing balance (3–5). Children with CP activate their muscles more than typically developing (TD) children in response to similar center of mass (CoM) disturbances (4, 5) and have higher muscle co-activation in response to both support-surface translations and rotations (3–5).

Increased muscle co-activation observed in children with CP during perturbations of standing balance might be due to neural deficits rather than being a compensation strategy to improve balance control. In an earlier study, we observed plantarflexor-dorsiflexor co-activation in response to both platform translations and rotations during standing in children with CP (5). Increased ankle stiffness due to increased muscle co-activation might help balance control in response to platform translations by resisting body movement with respect to the platform. However, increased joint stiffness does not help balance control during platform rotations as it couples body motion with platform motion resulting in body tilt (6–8). A potential cause of the increased co-activation are altered reciprocal pathways, i.e., a lack of inhibition or increased excitation of the antagonistic muscle upon activation of the agonist (9–11).

Differences in the response to perturbations of standing balance between children with CP and TD children, i.e., higher levels of muscle activity and co-activation, have striking similarities with differences in the response to isolated joint rotations in a relaxed condition. Muscle excitation in response to passive joint rotations is higher in children with CP than in TD children (12–14) and this increased muscle excitation is often attributed to spasticity or hyper-excitability of the stretch reflex (13). Furthermore, isolated joint rotations elicit co-activation between the stretched muscle and its antagonist (e.g., gastrocnemius and tibialis anterior when rotating the ankle joint towards dorsiflexion) in children with CP, but not in TD children (11, 15, 16). The afferent fibers from the stretch receptors project to an inhibitory interneuron, which in healthy individuals inhibits the motor neurons of the antagonist muscle. In CP, decreased activity of this inhibitory interneuron, called reduced reciprocal inhibition, is thought to cause increased agonist-antagonist co-activation in response to passive stretch (11, 16–18). However, increased excitatory inputs to the antagonistic muscle upon activation of the agonist, called reciprocal facilitation, can also lead to co-activation. It is unclear to what extent reduced reciprocal inhibition versus reciprocal facilitation contribute to co-activation in CP (10, 19). Neither current clinical scales nor instrumented tests of joint hyper-resistance assess muscle co-activation during isolated joint rotations in a relaxed patient (11, 16). Clinical scales are commonly limited to the classification of the subjective feeling of the overall resistance to stretch by the examiner (20). During instrumented spasticity assessments, muscle activity is measured by electromyography (EMG) with the aim to distinguish neural and non-neural contributions to joint hyper-resistance (21, 22). Typically, the activity of the stretched muscle is reported as a measure of hyper-reflexia, while agonist-antagonist co-activation is not reported as a measure of altered reciprocal activation.

The failure to assess alterations in reciprocal activation might explain why no or only limited correlations have been found between the muscle response to isolated joint rotations and functional movement impairments (23–25). Two prior studies investigated the relationship between joint hyper-resistance and reactive standing balance. These studies found no relationship between the Modified Ashworth Scale (MAS) and respectively CoM movement during standing in children with CP (26) and the ability to withstand perturbations without stepping in children with hereditary spastic paraparesis (27). To our knowledge, the relationship between increased muscle co-activation in response to isolated joint rotations and increased muscle co-activation during functional movements has not been studied.

We hypothesized that reduced reciprocal inhibition underlies plantarflexor-dorsiflexor co-activation in response to both isolated passive joint rotations and perturbations of standing in children with CP. Therefore, we expect that muscle co-activation during isolated joint rotations will be correlated to muscle co-activation during perturbations of standing balance. Secondary, we hypothesized that the neural response of the plantarflexors in response to isolated ankle dorsiflexion does not explain increased plantarflexor activity during perturbations of standing balance. Therefore, we do not expect that average muscle activity in response to isolated joint rotations correlates with average muscle activity during perturbations of standing balance.

MATERIALS AND METHODS

Participants

As there was no prior data, sample size was determined to enable detection of medium correlations (Spearman correlation of 0.4) with a power of 95%. The ethical committee of UZ/KU Leuven (S63321) approved this observational study. Twenty-one children with spastic CP and twenty typically developing children participated in the study (table 1) between January 2021 and August 2021. Children with CP were diagnosed by a neuro-pediatrician and met the following inclusion criteria: (1) aged 5 to 17 years; (2) Gross Motor Classification Scale (GMFCS) I-III; (3) able to stand independently for at least 10 minutes; (4) no orthopedic/neurological surgery or botulinum neurotoxin injections in respectively the previous 12 or 6 months. One child with CP was excluded due to a lack of cooperation. Typically developing children were age matched with children with CP.

Table 1:

Demographic data of participants.

		CP			TD
		Mean	SD	Range	Mean	SD	Range
Girls/ Boys		9/11			8/12
Age	(years)	12.3	3.1	7–17	12	3.0	7–17
Length	(cm)	153	16	130–185	156	16	126–184
Weight	(kg)	47	17	27–88	44	13	26–67
Unilateral/Bilateral		14/6
GMFCS I/II		15/5
MAS 0/1/1+/2/3/4		4/10/4/0/2/0

SD = standard deviation; GMFCS = Gross Motor Function Classification Scale (Range 1–5); MAS = Modified Ashworth Scale (Range 0–4). CP = cerebral palsy; TD = typically developing.

Protocol

Children and their legal representative signed respectively informed assent and informed consent before the start of the measurements following the Declaration of Helsinki. All data was collected during a single session. The protocol consisted of a clinical assessment of range of motion and MAS; instrumented spasticity assessment of the plantarflexors, hamstrings and rectus femoris; and reactive balance assessments. Note that not all data was used in this study.

We performed an instrumented spasticity assessment of the plantarflexors by applying isolated joint rotations following a previously developed and reliable method (21, 28). Participants lay supine and were asked to relax. The lower leg was supported by a customized frame that allowed ankle rotation. A researcher (same person across all participants) rotated the ankle joint as fast as possible (± 1s) from a plantar flexed position to the end range of motion towards dorsiflexion (figure 1, column 1). At least 7 seconds of rest were provided between each of the five trials to control for movement history dependence in muscle resistance to stretch (29, 30).

Figure 1:

Experimental set-up (a) and exemplar responses from three children with CP (b-d) and one typically developing child (e) for the co-contraction index (CCI) during isolated joint rotations (IJR, column 1), toe-up rotations level 2 (TO, column 2), and backward translations level 2 (BW, column 3). The child with CP represented in blue and the TD child represented in green have low muscle co-activation. The child in orange has moderate co-activation. The child in red has high co-activation. Starting position of the experimental set-up in grey, perturbed position in black. The co-contraction index was calculated as the average of the minimum EMG signal of lateral gastrocnemius (LG, dark grey) and tibialis anterior (TA, light grey).

Muscle responses to perturbations of standing balance were measured on a Caren platform (Motek, Netherlands). Participants stood barefoot on the platform (starting position was marked and consistent between trials) and were secured using a safety harness. Instructions were to stand upright and to maintain balance without stepping unless necessary to avoid falling. The perturbation protocol consisted of (1) backward translations (figure 1, column 3), followed by (2) toe-up rotations (figure 1, column 2). The plantarflexors are stretched by both backward translations and toe-up rotations but since backward translations cause a forward rotation of the body and toe-up rotations cause a backward rotation of the body, plantarflexor activity elicited by muscle stretch will only aid in maintaining an upright posture in response to translational perturbations. In addition, increased ankle stiffness due to muscle co-activation will only aid to stay upright in response to translational perturbations but not in response to rotational perturbations. We applied ten series of eight identical perturbation trials starting with six increasingly difficult translational perturbation levels (increasing platform displacement, velocity, and/or acceleration) followed by four increasingly difficult rotational perturbation levels (details on platform movement are described in Supplementary material S1, figure S1, https://doi.org/10.6084/m9.figshare.c.7805309.v1). When the participant stepped in more than three trials within one level, we did not continue to the next level (4, 5). If needed, rest was given between levels.

Muscle activity of the lateral gastrocnemius (LG), medial gastrocnemius (MG), soleus (SOL), and tibialis anterior (TA) was measured using surface electromyography (EMG) at 1000Hz (ZeroWire EMG Aurion, Cometa, Italy). Electrodes (Ambu Blue Sensor, Ballerup, Denmark) were placed according to SENIAM guidelines (31).

Data processing & analysis

EMG data was filtered using a fourth order Butterworth band-pass filter with 10 and 450Hz cut-offs, rectified, and low-pass filtered with a fourth order Butterworth filter with 40Hz cut-off. The filtered EMG signal was scaled to the maximum value observed across all movements performed during the protocol (i.e., maximum voluntary contractions for plantarflexors and TA, isolated joint rotations for ankle and knee, perturbations of standing, squats, and jumps; all movements were performed by all participants). For the isolated joint rotations, average scaled muscle activity was calculated across all five trials for each participant. For the perturbations of standing, average scaled muscle activity was calculated across all non-stepping trials within one level for each participant.

Outcome parameters

We computed the co-contraction index (CCI), a measure of muscle co-activation, as the minimum of TA and respectively LG, MG, and SOL filtered and scaled EMG averaged over the time interval of interest, which is proportional to the common area under the EMG trajectories (32) (equation 1) (figure 1).

$$\mathrm{C}\mathrm{C}\mathrm{I}=\frac{\sum _{i=start}^{end}\mathrm{m}\mathrm{i}\mathrm{n}(EM{G}_{PF}\left(i\right),EM{G}_{TA}(i)}{\#frames}$$ (Equation 1)

With PF referring to LG, MG, or SOL; $i$ referring to the different time frames, and # frames the length of the analyzed time period with start the first frame and stop the last frame. For the isolated joint rotations, we analyzed the EMG signal over 1s following rotation onset. For the perturbations of standing balance, we analyzed the EMG signal from 0.5s before until 1.5s after onset.

We assessed mean muscle activity in response to stretch as the time-averaged processed and scaled EMG for all muscles individually using the same time periods as for the CCI. During isolated joint rotations, the mean muscle activity is a measure of the neural response to stretch consisting of both short-latency stretch reflexes and prolonged muscle activity in response to stretch that is often observed in CP (12). During reactive balance, the mean muscle activity captures the neural response to stretch as well as balance correcting responses.

The most affected leg (based on MAS) was analyzed.

Statistical analysis

We evaluated correlations between CCI and mean muscle activity during isolated joint rotations and perturbations of standing and reported descriptive statistics (mean and range of correlations across all levels).

We assessed the relation between the muscle response to isolated joint rotations and to translational or rotational perturbations across perturbations levels using mixed linear models (equation 2). Independent variables were perturbation level and the muscle response (CCI or mean activity) during isolated joint rotations and the dependent variable was the muscle response during the reactive balance conditions. A participant factor (i.e., subject in equation 2) was included as random factor nested within group.

$$CC{I}_{RB}~Perturbationlevel\mathrm{*}CC{I}_{IJR}+1|subject$$ (Equation 2)

With $CC{I}_{RB}$ the co-contraction index for toe-up rotational perturbations or backward translational perturbations, and $CC{I}_{IJR}$the co-contraction index for the isolated joint rotations.

We created separate models for the different outcomes (CCI and mean muscle activity), different muscle pairs/muscles, different standing balance perturbations (translations or rotations), and different participant groups (CP and TD). Regression coefficients (ẞ) and p-values are reported.

All statistical analysis were performed using Matlab (2018, Mathworks, United States) with differences considered significant at p<0.05.

RESULTS

Due to empty EMG batteries, we had to exclude LG data for one child (TD), SOL data for two children (1 CP and 1 TD), and TA data for one child (CP).

Standing balance performance

For toe-up rotations, all children performed level 1 but respectively four (CP) and one (CP) children did not perform levels 2–4 and levels 3–4. For backward translations, all children performed level 1 but respectively two (CP), one (CP), three (2 CP, 1 TD), and five (CP) children did not perform levels 2–6, levels 3–6, levels 4–6, and levels 5–6 (supplementary material S2, table S1, https://doi.org/10.6084/m9.figshare.c.7805309.v1).

Muscle co-activation & mean muscle activity

We observed large inter-subject variability in co-contraction index values and mean muscle activity values for the isolated joint rotations, toe-up rotations, and backward translations in children with CP. Yet, our sample covered the complete range without clear outliers (figure 2, Supplementary material S3, table S2–5, figure S2-S3, https://doi.org/10.6084/m9.figshare.c.7805309.v1). TD children have little mean muscle activity and muscle co-activation in response to isolated joint rotations and translational and rotational perturbations of standing balance.

Figure 2:

Co-contraction index for LG-TA (a) and mean muscle activity for LG (b) across all participants for isolated joint rotations (left column), toe-up rotations (middle column), and backward translations (right column). Every dot represents one child. Children with cerebral palsy in color; typically developing children in gray scale. L1-L6 = Levels 1 to 6.

Relation between muscle co-activation during isolated joint rotations and perturbed standing

Children with cerebral palsy:

Linear mixed models revealed a relation between co-activation during perturbations of standing balance and during isolated joint rotations for toe-up rotations for all muscle pairs and backward translations for LG-TA and MG-TA (figure 3, Supplementary material S5, table S8, https://doi.org/10.6084/m9.figshare.c.7805309.v1) in children with CP.

Figure 3:

Spearman correlations between lateral gastrocnemius (LG) and tibialis anterior (TA) co-activation (CCI) during isolated joint rotations (IJR) and during perturbations of standing balance (toe-up rotations (TO) for children with cerebral palsy: upper part; backward translations (BW): lower part) for each level. Each dot represents one child.

We found a significant main effect of LG-TA CCI during isolated joint rotations on LG-TA CCI during toe-up rotations (p-value (p)=0.006, regression coefficient (ẞ)=2.17, mean correlation coefficient across levels (r)=0.49, range of correlation coefficients = [0.26 – 0.61] ) and backward translations (p=0.004, ẞ=3.26, r=0.43 [0 – 0.62]) (supplementary material S5, table S8, https://doi.org/10.6084/m9.figshare.c.7805309.v1). No main effect of level was found. No interaction effect between level and CCI during isolated joint rotations was found for toe-up rotations, whereas there was an interaction effect for backward translations (p=0.025) (supplementary material S5, table S8, https://doi.org/10.6084/m9.figshare.c.7805309.v1), suggesting a different relation for level 6.

We also found a significant main effect of MG-TA CCI during isolated joint rotations on MG-TA CCI during toe-up rotations (p=0.003, ẞ=2.20, r=0.57 [0.41 – 0.78]) and backward translations (p=0.024, ẞ=2.64, r=0.31 [0.07 – 0.46]) (supplementary material S4, figure S4, supplementary material S5, table S8, https://doi.org/10.6084/m9.figshare.c.7805309.v1). No main effect of level or interaction effect was found for toe-up rotations. For backward translations, there was a main effect for level (p=0.002), suggesting higher levels of co-activation with increasing perturbation level, and an interaction effect for level and co-activation during isolated joint rotations (p=0.001) (supplementary material S5, table S8, https://doi.org/10.6084/m9.figshare.c.7805309.v1).

We found a significant main effect of SOL-TA CCI during isolated joint rotations on SOL-TA CCI during toe-up rotations (p=0.047, ẞ=1.71, r=0.50 [0.43 – 0.59]) but not for backward translations (p=0.272, ẞ=1.11, r=0.26 [−0.32 – 0.46]) (supplementary material S4, figure S5, supplementary material S5, table S8, https://doi.org/10.6084/m9.figshare.c.7805309.v1). No significant main effects for level or interaction effects were found (supplementary material S5, table S8, https://doi.org/10.6084/m9.figshare.c.7805309.v1).

Typically developing children:

Linear mixed models revealed no relations between CCI during perturbations of standing balance and during isolated joint rotations for all muscle pairs, except for the LG-TA CCI during toe-up rotations (p=0.010, ẞ=1.0, r=0.36 [0.26 – 0.50], figure 4). No effects were found for other muscle pairs, nor for backward translations (supplementary material S5, table S9, https://doi.org/10.6084/m9.figshare.c.7805309.v1). No main effect of level was found. No interaction effect between level and CCI was found.

Figure 4:

Spearman correlations between lateral gastrocnemius (LG) and tibialis anterior (TA) co-activation (CCI) during isolated joint rotations (IJR) and during perturbations of standing balance (toe-up rotations (TO) for typically developing children: upper part; backward translations (BW): lower part) for each level. Each dot represents one child.

Relation between muscle activity during isolated joint rotations and perturbed standing

Children with cerebral palsy:

Linear mixed models revealed no relations between mean muscle activity during perturbations of standing balance and during isolated joint rotations for all muscle pairs, except for the soleus during toe-up rotations (Supplementary material S6, Figure S8-S11, Supplementary material S7, table S12, https://doi.org/10.6084/m9.figshare.c.7805309.v1).

We found a significant main effect of mean SOL activity during isolated joint rotations on mean SOL activity during toe-up rotations (p=0.034, ẞ=0.673, r=0.63 [0.47 – 0.74]) (supplementary material S7, table S12, https://doi.org/10.6084/m9.figshare.c.7805309.v1). Further, we found an interaction effect for level and mean activity during isolated joint rotations for LG during toe-up rotations (p=0.033) and backward translations (p=0.005) and for the SOL during toe-up rotations (p=0.005) (supplementary material S7, table S12, https://doi.org/10.6084/m9.figshare.c.7805309.v1).

Typically developing children:

Linear mixed models revealed no relations between mean muscle activity during perturbations of standing balance and during isolated joint rotations for all muscle pairs, except for the soleus during toe-up rotations (p = 0.006, ẞ=1.516, r=0.65, [0.62 – 0.72] and backward translations (p = 0.03, ẞ=1.726, r=0.55, [0.47 – 0.72]) (Supplementary material S6, Figure S12-S15, supplementary material S7, table S13, https://doi.org/10.6084/m9.figshare.c.7805309.v1).

DISCUSSION

Our results suggest that altered reciprocal activation contributes to altered reactive balance control in children with CP. We found that muscle co-activation in response to isolated joint rotations was related to increased muscle co-activation in response to perturbations of standing balance in children with CP, whereas in TD children, no relations were found except for LG-TA co-activation. Muscle co-activation in response to isolated joint rotations in relaxed patients with CP has been attributed to reduced reciprocal inhibition in the spinal cord given the absence of other control processes in this condition (18). The observed correlations thus suggest that the increased muscle co-activation during standing balance control might at least partially rely in spinal processes. In contrast, we found very few relations between the mean muscle activity during isolated joint rotations and perturbed standing. Current assessment of joint hyper-resistance focuses on reflex hyper-excitability and alterations in passive tissue properties (21) but our results indicate that clinical assessment of altered reciprocal activation during isolated joint rotations might provide information about balance impairments.

In contrast to many previous studies, we found a relation between the response to muscle stretch at rest and during functional movements. Previous research has mainly focused on the relation between spasticity or reflex hyper-excitability and muscle activity during walking (23). Only two prior studies specifically investigated reactive standing balance and found no relation between joint hyper-resistance as measured by MAS, and CoM movement (26) or the ability to withstand perturbations without stepping (27). In contrast to prior studies, we did not only assess general resistance or the neural response to muscle stretch but also muscle co-activation as a sign of altered reciprocal activation. We specifically related those outcomes to muscle coordination during perturbed standing. This allowed us to relate observations at rest to observations during a functional task. An exploratory analysis showed that the MAS score was not correlated to muscle co-activation during reactive standing (supplementary material S9, table S16-S17, figures S16-S18, https://doi.org/10.6084/m9.figshare.c.7805309.v1), confirming prior work and stressing the importance of using more specific outcomes (20). Alterations in passive tissue properties such as contractures might also contribute to functional impairments (23) but an exploratory analysis showed that co-activation during balance perturbations was not related to passive joint stiffness, suggesting that mechanical tissue properties do not explain variability in muscle coordination underlying balance control (supplementary material S10, table S18, https://doi.org/10.6084/m9.figshare.c.7805309.v1).

The associations between muscle co-activation during isolated joint rotations and during reactive standing suggest a common neural deficit. Both spinal and supraspinal pathways that are involved in reactive standing balance, are impaired in children with CP. The exaggerated response to isolated joint rotations is driven by impaired spinal pathways (13), as no supra-spinal contributions are expected. Here, we investigated how alterations in spinal pathways (tested through the isolated joint rotations) contribute to alterations in reactive balance. Associations in co-activation between these two conditions suggest that reactive balance impairments in children with CP partially originate from deficits in spinal pathways. Given that co-contraction during isolated joint rotations explain only up to 50% of the variance (based on the significant correlations), impairments in supraspinal pathways might also contribute. For example, damage to the corticospinal tracts we cannot exclude contributions from reduced selective control commonly leads to reduced selective muscle control and common drive to agonists and antagonists in CP.

We found associations in muscle co-activation between isolated joint rotations and perturbed standing notwithstanding striking differences in baseline muscle activity, body position, and stretch velocity. In both conditions, the ankle was dorsiflexed by an external force. However, during isolated joint rotations, the dorsiflexion movement is performed as fast as possible by the examiner, resulting in a stretch velocity that is a factor 10 higher than during the dorsiflexion movement that is imposed by the Caren platform. In addition, the patient is at rest during the isolated joint rotations, and therefore baseline muscle activity is low. During perturbed standing, the participant should maintain an upright posture resulting in higher baseline muscle activation. These differences may explain the large difference in amplitude of muscle activity observed between movement conditions, which reflects differences in the number of recruited motor units. Notwithstanding these striking differences, we found associations in muscle co-activation between isolated joint rotations and perturbed standing. Hence, altered reciprocal activation might be consistent across stretch velocities and tasks. In contrast, the stretch reflex is known to be velocity- and task-dependent (7, 13, 25), which might explain the limited associations between responsive muscle activity between conditions. It would be interesting to investigate whether associations between muscle activity in response to isolated joint rotations and perturbations of standing balance are present for balance perturbations with higher accelerations, velocities, or displacements. Similarly, it would be interesting to study the behavior of individual motor units (e.g., through high density EMG) to evaluate how recruitment affects reciprocal inhibition.

Our study design did not allow us to attribute increased muscle co-activation in response to passive muscle stretch to a specific spinal pathway. Afferent fibers from the stretch receptors in the stretched muscle project to an inhibitory interneuron. Decreased inhibitory activity might lead to increased activation of the antagonist (9, 17, 18). In addition, the afferent fibers from the agonist, may also activate excitatory pathways to the antagonist (10, 19). Both pathways, reduced reciprocal inhibition and reciprocal excitation, might result in muscle co-activation.

It was surprising that the co-activation was correlated between isolated joint rotations and perturbations of standing balance while mean TA activity was not. Co-activation is caused by increased activity of TA, upon stretch of the plantarflexors. Whereas we found that TA activity was indeed correlated to the co-contraction index during isolated stretches (r=0.91, p<0.05), TA activity could not fully explain the co-contraction index (See Supplementary material S11, figure S20, https://doi.org/10.6084/m9.figshare.c.7805309.v1). This is explained by TA activity upon stretch being larger than plantarflexor activity in some children. As a result, the co-contraction index was mainly defined by the activity of the plantarflexors – and not TA – in these children (see Supplementary material S11, figure S19 for exemplar trajectories, https://doi.org/10.6084/m9.figshare.c.7805309.v1). Given the relationship between the co-contraction index and mean muscle activity for TA activity, it is likely that we would also find a relationship between mean TA activity during isolated joint rotations and perturbations of standing balance in a larger dataset.

Although increased muscle co-activation might hinder balance control, we did not observe a relation between muscle co-activation and balance performance. Visual inspection of our data suggests that muscle co-activation is not related to the ability to withstand perturbations without stepping (figure 3, figure S2-S3), suggesting that other factors might be important here. Fear of falling might have induced stepping as stepping is an effective way to increase the base of support. Muscle co-activation was also not related to CoM movement during non-stepping responses (supplementary material S8, table S14-S15, https://doi.org/10.6084/m9.figshare.c.7805309.v1), suggesting that children compensate for the higher antagonistic muscle activity by also increasing agonistic muscle activity (4, 5). Whereas such compensation strategies might be effective for small perturbation magnitudes, they are limited as activation bounds will be reached sooner when co-activating antagonistic muscles. In the future, it would be interesting to investigate whether increased co-activation is related to falling.

It is unlikely that our EMG scaling method affected our primary outcomes. EMG was scaled to the maximal signal across different tasks, and it is possible that not all children activated their muscles to the same extent. A higher scaling factor (due to higher activation during one of the tasks) would result in a smaller scaled signal and thus smaller mean muscle activity and CCI for both isolated joint rotations and perturbed standing balance. The absence of correlations between mean muscle activity during isolated joint rotations and during perturbed standing for most muscles, conditions, and levels thus suggests that the observations of the current study were not caused by inter-subject differences in scaling. In addition, we obtained similar results when scaling the EMG to the maximal signal across perturbation trials only or scaling the EMG to the maximal signal observed during translational and rotational perturbation of level 1 (performed by all subjects). Also, we chose a specific outcome measure for co-activation that is sensitive to both the presence and amplitude of the common activation but did not capture whether activation patterns, including timing of activation, were similar between conditions. Whereas evaluating the timing of muscle activation could provide additional information, such analysis would have been hard during reactive standing balance due to the presence of both balance corrective muscle activity and antagonistic activity in both plantarflexors and dorsiflexors.

Conclusion

We demonstrated that muscle co-activation during isolated joint rotations and perturbations of standing balance is related. This suggests that alterations in reciprocal activation, might contribute to muscle co-activation during functional movements but this should be further investigated in other tasks.

GLOSSARY

CP

Cerebral palsy

TD

Typically developing

MAS

Modified Ashworth Score

CoM

Center of mass

GMFCS

Gross function classification scale

EMG

Electromyography

LG

Lateral gastrocnemius

MG

Medial gastrocnemius

SOL

Soleus

TA

Tibialis anterior

CCI

Co-contraction index

IJR

Isolated joint rotations

TO

Toe-up rotation

BW

Backward translation

DATA AVAILABILITY

The data supporting the conclusions of this article will be made available by the authors upon request, without undue reservation.