Relationship between neuromuscular body functions and upper extremity activity in children with cerebral palsy
This study was funded by the Norwegian Directorate of Health. We thank Rannei Sæther and Torarin Lamvik for providing study participants, Tom Ivar Lund Nilsen for statistical advice, and Brian Hopkins for English editing.
Siri Merete Brændvik at Clinical Services, St Olav’s University Hospital, Olav Kyrres gt 17, 7006 Trondheim, Norway. E-mail: email@example.com
Aim Our aim was to investigate the relationship between the dimensions of neuromuscular body function and elbow, forearm, and hand activity in the upper extremities in children/adolescents with spastic cerebral palsy (CP), within the framework of the World Health Organization International Classification of Functioning, Disability and Health.
Method Twenty-three participants (10 males, 13 females, mean age 13y, SD 3y, range 8–18y) with spastic CP (21 with hemiplegia, two with diplegia) at Manual Ability Classification System levels I to III participated in the study. Neuromuscular body function measures were (1) muscle strength in the elbow, forearm, and grip, (2) muscle tone in elbow flexors and forearm supinators, (3) active supination range and elbow extension range, and (4) force control at submaximal level in elbow flexion. Activity measures were actual use of the affected hand in bimanual activities (Assisting Hand Assessment) and instructed use of the affected hand (Melbourne Assessment of Unilateral Upper Limb Function).
Results Nearly all the neuromuscular body function variables were significantly correlated with activity. The combination of active supination range and strength explained 74% of the variance in actual use, and the combination of active supination range and force control explained 74% of the variance in instructed use.
Interpretation In high-functioning children and adolescents with CP, limited active supination range and difficulties in generating and modulating force are strongly related to limitations in hand activity. Further studies are needed to establish cause and effect in this relationship.
List of Abbreviations
Assisting Hand Assessment
Coefficient of variation
Root mean square
Cerebral palsy (CP) is the most common cause of physical disability in childhood1 and may affect the child on several health dimensions, as outlined by the framework of the World Health Organization International Classification of Functioning, Disability and Health (ICF).2 The motor signs include primary neuromuscular deficits, such as spasticity, muscle weakness and decreased selective motor control, and secondary musculoskeletal problems, such as bony malformations and contractures. Although primarily classified as a motor disorder, CP is often accompanied by disturbances of sensation, perception, cognition, communication, and behaviour. Activity limitations are presumed to result from these combined factors.3 Much of the treatment offered is targeted towards the dimension of neuromuscular body function, whereas the ultimate goal is to improve activity. The question arises as to what extent these initiatives alter the motor prognosis or make a clinically significant change at the level of disability.4,5 The ICF has been used increasingly in both the clinical and the research field. It classifies functioning and disability associated with health conditions and explicitly cautions against assuming a direct relationship between factors at the dimension of body function and structure and changes in activity. This relationship has rarely been investigated with respect to the CP population, and the studies so far suggest that it is complex.6–8 It is difficult to draw robust conclusions from these studies because they differ in focus (upper/lower extremities), variables investigated, and assessment tools used.
A few studies have addressed the upper extremity, but these focused mainly on hand function.6,7 However, activity of the upper extremities also includes the use of the elbow and forearm, and the CP population often experiences neuromuscular impairments in these areas.9 To our knowledge, no other study has included the forearm and the elbow when investigating hand activity in the context of the ICF. Such studies could provide additional knowledge to improve further the grounds for optimizing treatment and, eventually, the rehabilitation management of children with CP.
The objective of this study was to examine the relationship between neuromuscular body functions and activity in the upper extremities in children and adolescents with CP. Specifically, we wanted to evaluate how strength, muscle tone, active range of motion (ROM), and ability to modulate force in the affected arm are related to capacity in functional activities.
Twenty-three children and adolescents (10 males, 13 females; mean age 13y, SD 3y, range 8–18y), diagnosed with spastic CP participated in the study. They were recruited from the outpatient records of the neuro-orthopaedic team situated at St Olav’s Hospital in the region of mid-Norway. Twenty-one participants were classified as having hemiplegic CP and two with diplegic. The most affected hand in the participants with diplegia was chosen for investigation. All participants had a functional use of the affected upper extremity corresponding to level III or better on the Manual Ability Classification System and were able to take instructions verbally. Exclusion criteria were treatment with botulinum toxin A in the last 6 months and/or surgery on the upper extremity in the last 2 years. The study was approved by the Regional Committee for Medical Research Ethics; written informed consent was obtained from the parents before participation.
Equipment and procedure
Measurements at the dimension of body function included (1) active ROM in elbow extension and forearm supination, (2) muscle tone in the elbow flexors and forearm pronators, (3) maximum muscle strength in elbow flexion/extension, forearm supination/pronation, and grip, and (4) force control at submaximal levels in elbow flexion.
Active ROM was assessed using a manual goniometer. The measurements were carried out by two experienced therapists: one stabilizing the arm to prevent compensations, the other performing the measurements.
A stationary dynamometer Biodex System 3 Pro (Biodex Medical Systems, Shirley, NY, USA) was used to evaluate muscle tone, strength, and force control. Torque (N.m), angle, and angular velocity were detected by the Biodex, and the analogue Biodex signals were digitized (sample frequency of 1000Hz) through a National Instruments card (DAQCard-6036E; EMG Works® 3.1, Delsys Inc., Boston, MA, USA). During measurements in the elbow, the forearm was in a neutral position, and during measurements in the forearm the elbow was in 90° flexion. The shoulder was slightly abducted in all the measurements.
Muscle tone was evaluated as resistance to passive movement in elbow extension and forearm supination. Three trials were performed at two velocities each (10°/s and 180°/s) with the instruction to relax and just let the machine move the arm.
To evaluate muscle strength in the elbow and forearm, three dynamic maximal voluntary contractions at a velocity of 60°/s were performed. This was achieved using the passive mode in the Biodex; the machine moved the arm while the participant was instructed to push as hard as possible against the lever arm. This allows any force exerted to be recorded, enabling the recording of strength even in weak individuals.10 One test trial was conducted to ensure that the task was performed correctly. There was a 1-minute break between the trials. The procedure was carried out first on the forearm, then on the elbow.
Force control capacity was evaluated through a force modulation task. The participants were instructed to follow a target force by isometrically flexing the elbow. The target signal consisted of a linear increase followed by a linear decrease between 0 and 40% maximal voluntary contractions in 30 seconds. Online visual feedback of both the target force and the force actually generated was provided on a monitor (full screen) placed at a convenient distance in front of the participant. The elbow angle was set to 60° (from full extension) and the force level was normalized as the percentage of the best of three isometric maximal voluntary contractions performed before the modulation task. The task was performed twice with a 10-second break between each trial. Two test trials were allowed before measurements were taken to ensure that the task was understood.
Grip strength was measured using Grippit® (AB Detektor, Gothenburg, Sweden) in the standardized position recommended by the American Society of Hand Therapists for hand-grip dynamometry.11 The participants were instructed to grip the handle and squeeze with maximum force for 10 seconds. The procedure was repeated three times, with a 2-minute rest period between repetitions.
The Assisting Hand Assessment (AHA, version 4.3) and the Melbourne Assessment of Unilateral Limb Function (Melbourne Assessment) were used as activity measures. The AHA measures actual use of the affected hand in bimanual activity performance, which is achieved through a semistructured game session addressing everyday task performance, such as opening boxes, unscrewing bottle tops, or drawing. Movement details were not instructed. Although the AHA is validated for children up to 12 years, the participants in our study were up to 18 years old. However, Krumlinde-Sundholm and co-workers12 found low association between the AHA measures and age, suggesting that age per se may be of minor importance concerning the use of the assisting hand in bimanual activities. The Melbourne Assessment assesses instructed unilateral activities of the affected hand that involve reach, grasp, release, and manipulation. Both tests address the capacity domain. They were performed and videotaped according to standardized procedures.13,14
We estimated the reliability of our measurements by correlating two test sessions with an interval of 5 months for all the variables except force control, for which the interval was 8 weeks. Interclass correlation coefficients were between 0.70 and 0.82 for strength measurements, between 0.81 and 0.90 for force control measurements, between 0.89 and 0.94 for active ROM, and were 0.55 for muscle tone and 0.99 for the AHA and the Melbourne Assessment.
Data analysis was carried out using Matlab (The Math Works Inc., Natick, MA, USA), version 7.6. Before further analyses, torque signals were low-pass filtered with a cut-off frequency of 6Hz. The torque recorded during the passive movements at 10°/s was used to correct the other torque measures for arm weight. Peak torque during 180°/s, in the trial with the least resistance, was used to reflect muscle tone in the elbow flexors and forearm pronators. The best of three dynamic maximal voluntary contractions, represented by peak torque value in the isokinetic period, was used to estimate strength in the elbow and the forearm. The highest peak force value across three trials in the grip force dynamometer was used for further analysis.
In the force modulation task, the first and last second of both the increasing and decreasing contractions were omitted from further analyses. For the remaining period, the root mean square (rms) and the coefficient of variation (CV) of the difference between the applied force and the target force were calculated. The trials with the lowest rms during increase of force and during decrease were selected for further analysis. Three of the participants did not complete the force modulation task, leaving a sample of 20 for these data.
Both the AHA and the Melbourne Assessment were scored by an experienced occupational therapist (certified for AHA), and functional scores were converted to percentage scores.
Statistical analysis was performed in SPSS, version 15.0 (SPSS Inc., Chicago, IL, USA). Mean and SD of all the variables were calculated. Before further statistical analyses, data were normalized, giving a mean of 0 and SD of 1 for all variables. Linear correlations were used to test for a relationship between body function and activity. Depending on the distribution of the variable, parametric (Pearson’s) or non-parametric (Spearman’s rank) correlations were used. To test for a possible predictive value of the body function variables, forward stepwise multiple linear regression analysis was performed. The criterion for entry into the model was p<0.05 and for removal p>0.10. To reduce the amount of variables, the five significantly intercorrelated strength variables (elbow flexion and extension torque, forearm supination and pronation torque, and grip force) were averaged and presented as one strength variable (total strength). The same procedure was applied to the two significantly intercorrelated force modulation variables (0–40% and 40–0% of maximal voluntary contractions) for both rms and CV. Finally, the total rms and total CV were averaged into one force control variable (total force control). Thus, six independent variables were tested for prediction. The cumulative coefficient of determination (R2) and level of β are reported for goodness-of-fit and the individual predictive value, respectively. To avoid overestimation, the adjusted value of R2 is reported. Significance level was set to 0.05.
The mean and SD of all variables are listed in Table I. Active ROM was reduced more in the forearm than in the elbow. Twelve out of 23 participants had full extension range in the elbow, whereas only two had full supination range.
Table I. Mean (SD) of non-normalized variables
|Flexion torque elbow (N.m)||7.83||4.96|
|Extension torque elbow (N.m)||10.83||6.78|
|Supination torque forearm (N.m)||0.69||0.65|
|Pronation torque forearm (N.m)||1.17||0.84|
|Grip force (N)||104.83||66.50|
|Resistance torque elbow extension (N.m)||2.33||1.16|
|Resistance torque forearm supination (N.m)||0.35||0.20|
|AROM elbow extension (°)||−9.34||13.43|
|AROM forearm supination (°)||−55.21||39.21|
|Force control 0–40 rms (%)||4.75||4.36|
|Force control 40–0 rms (%)||6.11||5.09|
|Force control 0–40 CV||0.19||0.19|
|Force control 40–0 CV||0.53||0.99|
|Melbourne Assessment (%)||80.78||15.23|
Correlations showed that nearly all the variables at the dimension of neuromuscular body function were associated with activity, as measured by the AHA and the Melbourne Assessment (Table II). Total strength and active ROM forearm supination showed the strongest correlations with both activity measures. Extension resistance was negatively correlated with both the AHA and the Melbourne Assessment, whereas no significant correlation was found for supination resistance. Total force control was negatively correlated with the Melbourne Assessment, whereas there was no significant relationship with the AHA. Active ROM elbow extension was not significantly correlated with any of the activity variables. Excluding the participants without limitation in extension range (n=11) gave a positive correlation between range and activity for both the AHA (Spearman’s rank correlation; rS=0.87) and the Melbourne Assessment (rS=0.88), suggesting that a larger sample with more variation in range may have given a significant correlation.
Table II. Pearson’s (r) and Spearman’s rank correlation (rS) between body function variables and activity
| Flexion elbow||r=0.73a||rS=0.69a|
| Extension elbow||rS=0.55a||rS=0.42a|
| Supination forearm||r=0.66a||rS=0.60a|
| Pronation forearm||r=0.64a||rS=0.63a|
|Total force control||rS=−0.30||rS=−0.45a|
| Force control rms||rS=−0.38a||rS=−0.45a|
| Force control CV||rS=−0.29||rS=−0.46a|
|AROM elbow extension||rS=−0.09||rS=−0.15|
|AROM forearm supination||r=0.80a||rS=0.81a|
The results of the regression model (Table III) showed that both active ROM forearm supination and total strength significantly contributed to the outcome of the AHA, accounting for 74% of the variance observed. The remaining variables, extension and supination resistance, total force control, and active ROM elbow extension, did not play a significant role. The β-value of the predictors indicated an individual contribution of 0.52 for active ROM forearm supination and 0.45 for total force (both significant). Running separate regressions on the individual strength variables identified grip strength as the only significant contributor to the variance in AHA, explaining nearly 63% of the variance. It was also the strength variable with the highest correlation with the AHA (Pearson’s correlation; r=0.80).
Table III. Forward stepwise multiple regressions of neuromuscular body function variables against activity
| AROM forearm supination||0.63||0.52 (p=0.003)||0.207; 0.835||0.67||0.64 (p<0.001)||0.346; 0.937|
| Total strength||0.74 (p=0.007)||0.45 (p=0.007)||0.138; 0.767|| || || |
| Total force control|| || || ||0.74 (p=0.027)||−0.34 (p=0.027)||−0.635; −0.044|
| AROM elbow extension|| ||0.16 (p=0.263)|| || ||0.11 (p=0.434)|| |
| Total strength|| || || || ||0.14 (p=0.390)|| |
| Total force control|| ||0.14 (p=0.365)|| || || || |
| Extension resistance|| ||−0.20 (p=0.146)|| || ||−0.18 (p=0.178)|| |
| Supination resistance|| ||−0.24 (p=0.050)|| || ||−0.08 (p=0.523)|| |
With respect to the Melbourne Assessment, outcomes were slightly different. Here, a combination of active ROM forearm supination and total force control explained approximately 74% of the total variance. The β-values for active ROM forearm supination and total force control indicated an individual contribution of 0.64 and −0.34, respectively.
The possible influence of age on the outcome of the activity measures was tested, showing a non-significant correlation with both the AHA (rS=0.25) and the Melbourne Assessment (rS=0.40).
In this study, the relationship between neuromuscular body function and upper extremity activity was examined. Moderate to high correlations were found for nearly all the neuromuscular body function variables, except for active extension range in the elbow and forearm supination resistance, for which the correlations were weak. However, the combination of active supination range of the forearm and strength was the only significant predictor of actual use of the affected arm, and the combination of active supination range and the ability to modulate force was the only significant predictor of instructed use.
Active supination range, strength, and force modulation were all identified as significant predictors for activity, and they explained a considerable amount of the variance observed in actual and instructed use (74% for both). As both active movement and force modulation contain elements of force, these results suggest that force per se is important for activity, and that decreased force may give activity limitations. Muscle weakness, defined as ‘the inability to generate normal voluntary force in a muscle or normal voluntary torque about a joint’ (see reference 15, p 2161), is acknowledged as a sign of CP, but its contribution to activity limitations has been investigated only rarely. Arnould et al.6 found a moderate correlation (r=0.56) between grip strength and activity, and identified gross manual dexterity of the dominant hand and grip strength of the affected hand as the best predictors of manual ability. However, this combination explained only 58% of the variance at the activity dimension, which is lower than in our study. The differences in correlation and explanatory strength may be a result of the inclusion of the dominant hand and/or use of an indirect measure of activity (e.g. parental perception of a child’s difficulty in performing manual activities). This interpretation is supported by the findings of Van Meeteren et al.,7 who found a weak relationship (r=0.35) between grip strength and activity when measured by parental perception, but a stronger relationship using a direct measure. This suggests that the relationship between strength and activity may actually be stronger than concluded in these studies, a suggestion supported by the recent identification of a strong relationship between strength and activity in the lower extremities.16 This supports the use of strength training as an intervention, but documentation on the effect on activity is, so far, ambiguous.17,18
Active supination range was reduced with a mean of approximately −55°, being the variable with the strongest correlation with both actual (r=0.80) and instructed (rS=0.81) use of the affected arm. It was also a significant predictor of both actual (β=0.52) and instructed (β=0.64) use. The potential impact of elbow extension range remains unclear, as several participants in our study had full range. More variation in extension range may have yielded a significant impact on activity. To our knowledge, no other study has investigated the relationship between range of motion and activity in the upper extremities of this population. Ostensjø et al.8 explored the relationship between body function and activity in the lower extremities. They found a moderate relationship between passive range of motion and activity, but range of motion was not a significant predictor. The stronger and predictive relationship between range of motion and activity in our study may indicate that active range is related more to activity than passive range. Furthermore, fine motor abilities in the upper extremity may be more sensitive to limitations in movement range than gross motor functions in the lower extremities. In addition, our results corroborate the clinical assumption that range of motion is one of the most important determinants for the acquisition of motor abilities.19 Substantial effort is put into increasing or maintaining range of motion in the rehabilitation management of CP and various stretching techniques are commonly used. However, the literature on the effectiveness of stretching is inconclusive.20,21 More research is thus needed, both about the impact of limited ROM on activity and about designing effective interventions aimed at maintaining or improving active ROM.
Our results also suggest that the ability to generate force at higher levels is predictive for how the child actually uses the affected arm, whereas ability to modulate force at submaximal levels is more predictive for instructed use. It is worth noting, however, that actual use was evaluated in bimanual activities, whereas instructed use focused on the affected arm only. Actual use in a bimanual task allows the motor system to ignore the affected arm as first choice. Instead, the hand of the affected arm may be used to assist when the stabilizing and force-demanding aspects are dominant. Although the AHA is not intended to explain underlying mechanisms,12 it may provide the means for identifying important predictors for bimanual activities. Conversely, instructed use challenges the capacity of the motor system, which may trigger deficits in the ability to tune the movements. This implies that the choice of assessment tools in the clinic has to be considered carefully, and that it may sometimes be necessary to use multiple tests to assess activity limitations.
Spasticity is a commonly reported characteristic of children with CP and is clinically considered to be one of the most significant contributors to activity limitations.19 Neither the nature of spasticity nor its effect on activity limitations is fully understood, and several authors have recently questioned a direct relationship between spasticity and activity limitations.22,23 In this study, we measured muscle tone as resistance to passive movement, which is one component of spasticity. We found moderate correlations between muscle tone in the elbow flexors and activity (AHA r=−0.59; Melbourne Assessment rS=−0.54), whereas the correlations with muscle tone in the pronators were weak (AHA rS=−0.13; Melbourne Assessment rS=−0.15). A possible explanation for this weak correlation might be that decreased active supination range, a mean of −55° among our participants, actually constrains the ability to build up resistance torque during supination, and thus may constitute a confounding variable. Muscle tone was not identified as a significant contributor to activity, although supination resistance nearly reached significance (p=0.050). This finding is in agreement with other studies performed on the lower extremities.16,24 However, the reliability of the muscle tone measurements was lower than for the rest of the variables, indicating possible measurements error. Thus, the relationship between muscle tone and activity might be stronger than our results indicate. Consequently, no firm conclusion can be drawn about the impact of muscle tone in our study.
There are a number of limitations to this study. The reliability estimates reported are not based on a test–retest situation, but on pre- to post test following intervention. Yet, the interclass correlations were good for all the variables (above 0.70), except for muscle tone (0.55). Thus, it may be considered a rather conservative and acceptable stability estimate. Another question to be addressed concerns the use of multiple regressions. Six independent variables were tested for prediction, with a sample size of 23. This could call into question the use of multiple regressions. However, neither β nor its standard error changed much as the number of variables put into the model increased, suggesting that the unsystematic variance was not influenced by the number of variables. Given the small sample size, further validation of this study is needed.
Our results suggest a moderate to strong relationship between neuromuscular body functions and activity of the upper extremity in high-functioning children and adolescents with CP. This indicates that both the forearm and the upper arm should be considered carefully when making inferences about the affected arm in this population. Although not significant in this study, muscle tone cannot be ruled out as important for activity. Further studies are needed to establish cause and effect in this relationship, for example by randomized control trials or experimental procedures.