| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Reports |
T.M. Stoeckmann, PT, DSc, is Clinical Assistant Professor and Neurologic Residency Program Coordinator, Department of Physical Therapy, Marquette University, Schroeder Health Complex, PO Box 1881, Milwaukee, WI 53201-1881 (USA). At the time of the study, she was a student in the Graduate Program in Neurology, Rocky Mountain University of Health Professions, Provo, Utah.
K.J. Sullivan, PT, PhD, is Associate Professor, Department of Biokinesiology and Physical Therapy, and Director, Professional Doctorate in Physical Therapy Program, University of Southern California, Los Angeles, California.
R.A. Scheidt, PhD, is Associate Professor, Department of Biomedical Engineering, Marquette University, Milwaukee, Wisconsin, and Department of Physical Medicine and Rehabilitation, Feinberg School of Medicine, Northwestern University, Chicago, Illinois.
Address all correspondence to Dr Stoeckmann at: tina.stoeckmann{at}mu.edu
Submitted April 30, 2008;
Accepted March 18, 2009
 | Abstract |
|---|
Objective: The purpose of this study was to determine the effect of load type (elastic, viscous, or mass) on muscle activation and co-contraction during resisted forward reaching in the paretic and nonparetic arms after stroke.
Design: This investigation was a single-session, mixed repeated-measures pilot study.
Methods: Twenty participants (10 with hemiplegia and 10 without neurologic involvement) reached forward with each arm against equivalent elastic, viscous, and mass loads. Normalized shoulder and elbow electromyography impulses were analyzed to determine agonist muscle recruitment and agonist-antagonist muscle co-contraction.
Results: Muscle activation and co-contraction levels were significantly higher on virtually all outcome measures for the paretic and nonparetic arms of the participants with stroke than for the matched control participants. Only the nonparetic shoulder responded to load type with similar activation levels but variable co-contraction responses relative to those of the control shoulder. Elastic and viscous loads were associated with strong activation; mass and viscous loads were associated with minimal co-contraction.
Limitations: A reasonable, but limited, range of loads was available.
Conclusions: Motor control deficits were evident in both the paretic and the nonparetic arms after stroke when forward reaching was resisted with viscous, elastic, or mass loads. The paretic arm responded with higher muscle activation and co-contraction levels across all load conditions than the matched control arm. Smaller increases in muscle activation and co-contraction levels that varied with load type were observed in the nonparetic arm. On the basis of the response of the nonparetic arm, this study provides preliminary evidence suggesting that viscous loads elicited strong muscle activation with minimal co-contraction. Further intervention studies are needed to determine whether viscous loads are preferable for poststroke resistive exercise programs.
| Introduction |
|---|
 TopAbstract Introduction MethodResultsDiscussionConclusionReferences |
|---|
The effect of impaired coordination associated with co-contraction after stroke remains controversial, in part because of the variety of tasks and analytical approaches used as well as the operational definitions of co-contraction. For example, abnormal co-contraction in hemiparetic muscles has been described as "markedly altered timing"8 and as a "delay in initiation and termination."9 Inappropriate co-contraction has been reported during dynamic reaching after stroke9,10 or in association with certain stages of recovery.11,12 In contrast, others have reported normal co-contraction levels during isometric contractions13 and appropriate sequential activation during reaching after stroke.14
Clinical studies have demonstrated that resistive exercise programs after stroke can increase strength (force-generating capacity)15–19 and improve the performance of functional tasks20–22 with no increase in spasticity (velocity-dependent hypertonicity).17,23 In addition, improvements in strength have been associated with increased accuracy and timing (ie, coordination) in dynamic upper-extremity tasks.20,24 Our research hypotheses were formulated, in part, on the basis of the potential of resistive exercise to improve the strength and coordination of agonist and antagonist muscle groups impaired after stroke.
Although physical therapists commonly use weights, elastic bands, pneumatic or hydraulic exercise machines, or pools for resistive exercise, no systematic studies have been conducted in people after stroke to investigate the effects of various resistive load types during strength training. The kinetic properties of resistive loads place unique demands on muscles during movement: the force required to elongate an elastic load increases with the distance the material is stretched; the force required to move against a viscous load (such as water) increases with movement speed; and for weights (mass loads), balanced and appropriately timed acceleration and deceleration forces are required to move and stop the load. Studies comparing these types of resistance have investigated adaptations to load type,25 muscle responses to unexpected loading,26 and variations in load magnitude27—but only in people who were neurologically intact. These subjects did not demonstrate co-contraction with any resistive load.8,27 It is not clear how these load types would affect muscle activation during resistive exercise in people after stroke.
A growing body of literature has consistently demonstrated that force production and coordination (ie, speed and accuracy) are impaired in both the paretic and the nonparetic arms after stroke.28–31 The purpose of this pilot study was to investigate the effects of commonly used resistive load types (mass, elastic, and viscous) on muscle activation and timing in both arms after stroke. The significance of this study is that it contributes important clinical insights related to the effects and, potentially, the effectiveness of specific physical rehabilitation interventions for people after stroke. Thus far, there has been insufficient evidence to guide therapists in selecting the resistive load type that can result in both an increase in strength and an improvement in muscle coordination.
We hypothesized that the level of muscle activation during reaching would be higher against the viscous load because the peak force requirements of viscous loads coincide with the peak velocity profile of the movement, arguably the "weakest" part of the reach, on the basis of the force-velocity relationship of muscle contraction. In contrast, we hypothesized that muscle timing for the mass load would be associated with the largest amount of abnormal co-contraction in both the nonparetic and the paretic arms after stroke because the mass load requires appropriately synchronized agonist and antagonist muscles to successfully move (accelerate) and stop (decelerate) the load.
| Method |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
|
) mounted in series with its respective load. An OPTOTRAK 3020 motion analysis system
recorded force, kinematic, and electromyography (EMG) data for this study at a rate of 1,000 samples per second (sample data from one trial are shown in Fig. 2A).
|
|
Participants were required to perform 10 successful reaches for all 3 loads with each arm. They were instructed to wait for the "go" signal and then push the cart forward until the cart marker touched the target post and remained at the target briefly before returning to the start position. A successful reach involved moving the cart forward to the spatial target (criterion: 16±1 cm) in about 0.5 second (criterion: 700±100 milliseconds). Data collection for each trial lasted 4 seconds from the time of the auditory "go" signal. Participants were informed of the criteria for a successful reach, and feedback about reach time and distance was provided after each trial. Participants were allowed to practice until they were successful but generally needed only a few trials. Each participant reached against one type of resistance until 10 successful trials were achieved. This process was repeated for each of the remaining load types, resulting in 30 trials of data. Load order was randomized across subjects. Participants were allowed to rest as often as needed, although no subject requested a rest. Participants with stroke started with the nonparetic limb so that they could learn the task before performing it with the less-coordinated arm32; control participants started with the dominant (right) arm.
Before and after the reaching trials, maximum-effort force data were collected from 3 isometric pushes against a cart locked in the midreach position. The pretrial pushes were used to determine how much resistance to use for the subsequent experimental reaches. Pre- and postexperimental maximum-effort pushes also were compared to assess for fatigue; all participants exceeded their initial efforts by a small amount (<7%) in postexperimental testing, suggesting a modest learning effect and no fatigue.
Data Collection and Analysis
Load equivalency.
Because the different load types placed substantially different kinetic demands on the subjects during reaching, we developed a method to determine resistive loads that would require approximately equivalent efforts across the 3 load types. On the basis of previous protocols of resistive movement after stroke,12,23 our goal was for each participant to reach against loads requiring approximately 30% of the preexperimental maximum isometric push. Resistive loads for each arm were matched to create "triplets" of equivalent force impulses, so that Imass = Iviscous = Ielasticwith I being calculated as follows:
|
|
EMG data collection and analysis.
Surface EMG data were collected from the anterior and posterior deltoid muscles, both heads of the biceps muscle, and the lateral and long heads of the triceps muscle (Fig. 2A). The prime movers for our task were identified as the anterior deltoid and triceps muscles in a pilot study of adults who were healthy; the antagonists were identified as the posterior deltoid and biceps muscles.
The EMG signals were preamplified with a gain of 1,000 and band limited to frequencies between 10 and 500 Hz before sampling with a Noraxon MyoSystem 1200.
Postprocessing within MATLAB|| included calculating root-mean-square amplitudes of the full-wave-rectified EMG signals with a 25-millisecond sliding-window root-mean-square filter and a 50-millisecond sliding-window low-pass filter. To capture the onset of contraction that precedes movement, we analyzed the EMG data beginning 300 milliseconds before the onset of movement. Movement onset was identified as the point at which the cart velocity exceeded 0.02 m/s, and onset ended when the velocity returned to less than 0.02 m/s (Fig. 2A, red cursors).
The EMG data were recorded during preexperimental maximum voluntary isometric contractions (MVICs) for shoulder and elbow flexion and extension and quiet resting baseline trials.28,33 The lowest average EMG value for 3 resting baseline trials was subsequently subtracted from all experimental EMG values (including MVIC EMG values). The highest of the 3 resulting MVIC EMG values for each muscle group was used for normalization during postprocessing data analysis.34,35
All dependent outcome variables were calculated from the average EMG impulse data (V·ms)—expressed as a percentage of the MVIC—and included the average impulse for each agonist muscle and the co-contraction impulse between agonist and antagonist muscles. For each resistance type, the EMG signals from its 10 trials were averaged for each muscle, and the area under the curves was calculated (Fig. 2B). Our co-contraction measure estimated the amount of EMG overlap for agonist and antagonist pairs, calculated as the area under the curve created by the lower of the 2 EMG values at each moment in time (Fig. 2C, red trace).36 This co-contraction impulse value reflected how much the least-active muscle was firing throughout the movement. In this way, agonist and antagonist pairs could be active at different times without any co-contraction, a distinction that was critical for the acceleration and deceleration of mass loads (Fig. 2B, right column).
Statistical Testing
The EMG recordings from one participant with hemiplegia were corrupted by a faulty ground electrode. The EMG data for this participant and the corresponding matched control participant were excluded from further analysis.
Given the multiple comparisons of this study, a multivariate analysis of variance (MANOVA) was used to assess for an effect of arm type or load type on task performance. Separate analyses were completed for the paretic and nonparetic arms, each paired with the respective right or left arm of the age-matched control participant. To determine the effect of specific load types (mass, viscous, and elastic) on our dependent measures of agonist muscle activation and co-contraction, we used separate post hoc 2 (arm type) x 3 (load type) mixed-design, repeated-measures analyses of variance (ANOVAs). In all cases, effects were considered statistically significant at P=.05, as determined with Tukey t tests, when appropriate. Statistical tests were performed with the Minitab# statistics package.
| Results |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
.001 for each). There were no significant strength differences between the right and the left arms of the control participants (P values for grip and push were 1.0 and .62, respectively) (Tab. 2). Although the lower mean values of both strength measures for the nonparetic arm than for the control arm were not significantly different (P values for grip and push were .11 and .24, respectively), these differences reflected medium to large effect sizes37 (.79 and .55 for grip and push, respectively), consistent with previous reports indicating mild impairment of the nonparetic arm.28,30,31,38
|
|
|
|
Both agonist muscle recruitment and co-contraction were affected by load type at the shoulder but not at the elbow (Tab. 3, Fig. 4). Elastic and viscous loads resulted in significantly higher levels of muscle recruitment than the mass load for the anterior deltoid muscle in both the nonparetic and the control arms (Tab. 3 [Tukey post hoc analysis], Fig. 4). Although the elastic load elicited higher levels of shoulder co-contraction than the mass load in the nonparetic arm, the viscous load did not (Tab. 3 [Tukey post hoc analysis], Fig. 4). A marginal interaction between arm type and load type for anterior deltoid muscle recruitment was also demonstrated (P=.06). Despite the different magnitudes of the responses, both arm types demonstrated a sensitivity to load type at the shoulder, reflecting a blend of both normal and impaired responses in the nonparetic arm.
| Discussion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
In contrast, reaching with a hemiparetic (paretic) arm resulted in a high percentage of muscle activation and a higher level of co-contraction across all load types relative to those in people without neurologic involvement.
We also found evidence of impairment during reaching in the nonparetic arm. With the exception of shoulder agonist muscle activation, nonparetic arms also demonstrated higher percentages of muscle activation and co-contraction, which differed in response to load type. Specifically, elastic and viscous loads were associated with a higher level of agonist muscle activation (anterior deltoid muscle), and elastic loads were associated with the most co-contraction at the shoulder.
In light of our findings of differences in muscle activation patterns in the paretic and nonparetic arms of people after stroke and people who were neurologically intact (control participants), future research and clinical applications should include consideration of the types of loads being used to improve strength, and a load type should be chosen on the basis of its effect on muscle activation.
Agonist Muscle Recruitment
Because the peak force requirement of a viscous load coincides with the peak of the velocity profile (theoretically the weakest part of the reach, on the basis of the force-velocity relationship of the muscles), we hypothesized that the viscous load would induce a higher level of agonist muscle recruitment. This hypothesis was only partially supported by our data. The elastic and viscous loads were equally effective in eliciting significantly higher levels of agonist muscle activation than the mass load—and only at the shoulder for the control and nonparetic arms. As expected for the control group, the elastic and viscous loads elicited only agonist muscle activation, whereas the mass load elicited a brief agonist burst and then a brief antagonist burst, which coincided with the acceleration and deceleration profiles for the load, respectively. This biphasic muscle activation profile is consistent with those in other single-joint studies of subjects who were healthy and who responded to changes in these load types.19,27,39 Gottlieb et al27 reported synchronization of biphasic muscle torque and EMG values at both the elbow and the shoulder in response to resisted reaching, whereas our participants demonstrated this pattern only at the shoulder. This difference likely was attributable to the arm configuration for the reaching task. In the study by Gottlieb et al,27 the reach occurred with 90 degrees of shoulder abduction, whereas the reach occurred with 0 degrees of abduction in the present study. Such posture-dependent and task-specific effects also have been described by other authors.40–43
In contrast, the paretic arm consistently used a higher percentage of maximum voluntary effort and co-contraction across all load types. This finding is consistent with those of other stroke studies reporting that participants with more motor impairment after stroke lacked the ability to individually coordinate joints within a limb and were unable to adapt their motor responses to various upper-extremity tasks.5,44–46 Lum et al associated the presence of these abnormal synergies with strength imbalances,47 suggesting an association between strength and coordination in this population.
Importantly, both paretic and nonparetic muscles used a higher percentage of their voluntary capacity to reach against various types of resistance, as evidenced by increased agonist muscle recruitment and co-contraction in most conditions. Thus, although our results are consistent with the earlier findings that paretic muscles are much less efficient in producing force, requiring more EMG activity to effectively move a load,4,6,7,47 our results also implicate excessive co-contraction of the antagonist muscle10–12,48 as potentially contributing to the clinical presentation of weakness during our dynamic task.
Coactivity
Some investigators have suggested that abnormal co-contraction patterns represent a reduction in the number of muscle combinations or possible synergies available in a paretic limb after stroke,18 reflecting impairments in both agonist muscle recruitment and antagonist muscle inhibition.49 Our second hypothesis proposed that the levels of agonist-antagonist muscle co-contraction would be elevated in the paretic and nonparetic arms in response to the mass load, the only load that inherently required appropriately timed agonist and antagonist muscle activation.39 Our dynamic task, combined with the different kinetic demands for the loads tested, required a measurement of co-contraction that was sensitive to the timing of agonist and antagonist muscle activation. Simply measuring antagonist muscle activation levels may contribute to erroneous conclusions about the presence or absence of co-contraction; it is not just whether antagonist muscles are on but when they are on that is critical. Thus, a unique contribution of this pilot study is the introduction of a temporally sensitive method of quantifying co-contraction. On the basis of this analysis, we were able to demonstrate and support our hypothesis that both the paretic and the nonparetic arms would exhibit a significant amount of co-contraction not observed in the control arms.
Specifically, we found that the nonparetic shoulder showed a significantly higher level of co-contraction with the elastic load than with the mass load but not the viscous load. A post hoc analysis (t tests) confirmed that the level of co-contraction of the nonparetic shoulder was significantly higher than that of the control shoulders for the elastic load (P=.05), but the levels of co-contraction for the viscous and mass loads were lower and comparable to those of the control shoulders (P=.09 and P=.68, respectively). Therefore, the nonparetic arm showed more flexible motor strategies than the paretic arm but this variable level of co-contraction is not consistent with the consistently minimal co-contraction seen in the control arms. We believe that the higher level of co-contraction for the elastic load may reflect the need for large stabilization forces at the end of reach for the elastic load.13,50
The "Unimpaired" (Nonparetic) Arm Also Is Affected
Consistent with the growing body of evidence that the "unimpaired" limb also shows subtle motor impairments after stroke,30,31 we also found significant differences between the nonparetic and control arms in all EMG outcome measures except anterior deltoid muscle recruitment (Fig. 3). Lower baseline strength is consistent with deficits in isometric torque production reported by other authors.28,29 More sensitive kinematic and kinetic studies have consistently demonstrated motor control deficits in the "less paretic" limb after stroke, such as the impaired muscle timing represented in the present study by high levels of co-contraction.51 Such deficits call into question the use of the nonparetic arm as a matched control for research or clinical practice.
Clinical Applications and Future Studies
Because both muscle weakness and co-contraction correlate significantly with motor impairment and disability,52 the results of the present study may have important implications. There is very little literature suggesting effective treatment strategies for temporal coordination deficits after stroke. With mounting evidence supporting the use of resistive strength training to reduce impairments after stroke,15,16 the logical extension of the present study is to determine whether training with a particular type of resistance can preferentially benefit both muscle recruitment and coordination.
In our study, muscle activation patterns (our indicator of coordination) were specific to our task and the kinetic demands imposed on the limb by different resistive loads for both the nonparetic and the control arms. Viscous loads, in particular, appeared to place demands on the muscle that resulted in higher levels of muscle activation with less co-contraction than did elastic or mass loads. However, the paretic arm responded with high levels of muscle activation and co-contraction across all load types. Future intervention studies could investigate whether strengthening exercises with viscous loads are more effective than those with other loads for developing strength without co-contraction after stroke.
Some investigators have suggested that weakness in the paretic elbow musculature reveals a strong task dependence that is attributed to abnormal synergy between the elbow and the shoulder muscles43 and that can be modified with training.46 On the basis of the specificity of the training literature,40,41,53 one could extrapolate that practice accelerating and decelerating mass loads might facilitate the generation of more appropriately timed agonist and antagonist muscle activation, elastic loads might foster stability, and viscous loads might preferentially facilitate agonist muscle recruitment with minimal co-contraction. Lum et al54 found that participants with stroke showed improvements in agonist muscle EMG amplitude and work output and reductions in force direction errors after guided reaching against robotically simulated viscous loads. However, the support provided during the guided reaching task reduced the need to accommodate the mass of the limb, altering the dynamic requirements of the reaching task. Cirstea et al reported that a single session spent practicing arm pointing movements led to improved elbow and shoulder muscle timing in subjects who had had a stroke and had mild to moderate levels of functioning.42 Given that the arm itself acts predominantly as a mass load during reaching tasks, such findings may provide mechanistic support for randomized controlled trials that have demonstrated the effectiveness of task-specific training in people after stroke.55 Further investigation is needed to determine which type of resistive training might best help subjects acquire more appropriate motor responses after stroke.
Although the optimal treatment has yet to be identified, there is a growing awareness of the need to address both strength and coordination in rehabilitation paradigms.43 For example, Sullivan et al21 used a combination of task-specific training and isotonic strengthening of the legs to improve ambulation; that strategy positively affected both impairment and function, with improvements continuing even 6 months later. Using a different exercise sequence, Patten and colleagues22 evaluated an upper-extremity hybrid intervention comprising both resistance training and functional task practice and reported strength gains with increased EMG activation and marked improvement in all clinical and functional measures. Although further research along these lines is needed, our pilot study is a preliminary step in developing a method for directly comparing different types of resistive loads and evaluating their effects on motor control.
Study Limitations
The present study had a few minor limitations. The side on which the lesion was located was not homogeneous across participants (8 of 10 had right-side cerebrovascular accidents). Despite the fact that our participants’ Fugl-Meyer Motor Assessment scores fell within the moderate impairment category,12 our participants had with a fairly wide range of impairment levels within that category. Given that muscle activation patterns may be related to the level of residual arm function,11,12,56,57 these variables should be more tightly controlled in future work. Finally, although the clinically based loads were specifically chosen for their validity, matching these load triplets to each participant's strength was limited to the weight increments available. Robotic techniques have already shown potential in stroke rehabilitation for the arm19,58 and would provide a means for more precisely matching resistive loads to individual abilities.
| Conclusion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
The significance of this pilot study is that it revealed anticipated differences in muscle activation and co-contraction by load type that were not distinguishable in the hemiparetic arm. However, the nonparetic arm provided a model of mildly impaired motor response that might be more sensitive to investigations of intervention efficacy. Future studies should be conducted to determine intervention effectiveness before the initiation of an intervention trial.
| Footnotes |
|---|
The authors are grateful to Guy Simoneau, PhD, for his insightful comments on the analysis of this project and Supriya Asnani, MS, for her assistance in software programming as well as data collection and analysis. They also are deeply indebted to Priyanka Kanade, MS, for building and validating the cart-on-rail system.
This research was done in partial fulfillment of the requirements for Dr Stoeckmann's Doctor of Science degree at Rocky Mountain University of Health Professions, Provo, Utah.
This study was approved by the institutional review boards of Marquette University and Rocky Mountain University of Health Professions.
This work was supported by a Marquette University College of Health Science Faculty Development Award to Dr Stoeckmann, by the National Science Foundation (grant BES 0238442) awarded to Dr Scheidt, and by the National Institute of Child Health and Human Development, National Institutes of Health (grants NIH R24 HD39627 and NIH R01 HD53727) awarded to Dr Scheidt.
* The Hygenic Corp, 1245 Home Ave, Akron, OH 44310. 
Mark-10 Corp, 11 Dixon Ave, Copiague, NY 11726.
Northern Digital Inc, 5555 Business Park, Ste 100, Bakersfield, CA 93309.
Noraxon USA Inc, 13430 N Scottsdale Rd, Ste 104, Scottsdale, AZ 85254.
|| The MathWorks Inc, 3 Apple Hill Dr, Natick, MA 01760-2098.
# Minitab Inc, Quality Plaza, 1829 Pine Hall Rd, State College, PA 16801-3008.
| References |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
