Speed-Dependent Reductions of Force Output in People With Poststroke Hemiparesis.

Key Words: Exercise, Hemiparesis, Motor activity, Muscle spasticity.

Decreased speed of locomotion is one of the major characteristics that occur as a result of poststroke hemiparesis,[1-4] Walking speed is an effective indicator of the degree of abnormality in gait quality, overall functional status, and clinical progress in people with hemiparesis.[5,6] Furthermore, gait speed has been found to correlate with ability to balance on either one or both lower extremities, degree of lower-extremity force recovery, Barthel Index score, degree of ambulatory independence, cadence of gait, and rating of overall gait appearance.[7-9]

Increased speed is thought by some authors to result in further exacerbation of unwanted or abnormal muscle activity; therefore, people with hemiparesis are advised to avoid moving at faster speeds.[10] This hypothesis stems from studies that demonstrate that speed-dependent stretch reflexes in antagonist muscle groups contribute to the inability to increase movement speed and that increased effort results in greater-than-normal muscle activity that will interfere with movement,[11-13] When a person moves slowly over a long period of time, this contributes to a further loss of ability to generate powerful contractions and maintain endurance at fast speeds because of muscle atrophy and increased muscle fatigability (see McComas[14] for a review). In individuals who have not had strokes, training is known to improve muscle force and endurance at the training speed as well as at slower speeds, although improvements during training at slower speeds do not carry over to training at faster speeds (see McArdle et al[15] for a review). Therefore, the very interventions aimed at improving power at faster walking speeds in people who have not had strokes are thought to be contraindicated in people with hemiparesis because of the exacerbation of unwanted or abnormal muscle activity during movement.

Although these previous studies have focused on reflexes and their contributions to slow movement speeds, it is increasingly recognized that the mechanics of a given movement change as the speed of the movement increases[16] and thus task mechanics may be another contributor to differences in movement ability at faster speeds. In the example of cycling, previous studies[17-19] have shown that increased pedaling speeds required earlier onsets of muscle activity to reach peak force at appropriate points in the cycle. In addition, at faster speeds, speed-dependent interaction forces (eg, inertial forces such as Coriolis forces) increased in magnitude.[16] Given these mechanical alterations at higher speeds, we believe the nervous system must develop strategies to deal with altered task mechanics.

Pedaling allows exploration of the role of task mechanics (ie, mechanical constraints present during a task) during locomotion at different speeds. It is possible to study a wide range of speeds because the nonparetic lower extremity can assist the progression of the crank, via the coupled crank spindle, and therefore overcome impaired control of the paretic lower extremity. Such studies are important because we believe that observations can be made about the behavior of the paretic limb during fast movements, and we can determine whether that behavior is assistive or resistive to the forward progression of the crank. In addition, mechanical measures of pedaling performance have been used to characterize impairments in people with hemiparesis.[20-23] These studies have shown that the mechanics of pedaling are impaired so that, compared with normal pedaling behavior (ie, force generation and muscle activity during a pedaling task), a reduced amount of net mechanical work is done by the paretic lower extremity. This reduced work is a result of a combination of reduced positive work done during the downstroke of the cycle and an exaggerated amount of negative work done during the upstroke.[20] The timing of electromyographic (EMG) activity in individual paretic limb muscles exhibited 2 distinct types of abnormalities that were correlated with this lesser work production: prolonged excitation in the vastus medialis muscle (VM) and phase-advanced excitation (both early initiation and early termination) in the rectus femoris muscle (RF) and the semimembranosus muscle (SM).[20]

The purpose of our study was to quantify the effects of increased speed on motor performance during pedaling exercises in people with poststroke hemiparesis. If pedaling at higher speeds results in less work done by the paretic limb, then we believe at least 3 possible mechanisms may contribute to the effect. One mechanism--increased inappropriate activity--would involve the exacerbation of EMG timing abnormalities so that greater prolonged activity (eg, in VM) may occur. A second possible mechanism would involve a lack of speed-appropriate EMG timing alterations so that peak force generation would be delayed. In the case of pedaling, an inability to generate earlier onset of muscle activity at faster speeds will result in peak pedal forces being generated at later, less appropriate regions of the crank cycle. Third, it might also be possible that EMG timing changes in people with poststroke hemiparesis parallel those of individuals who have not had strokes, but are superimposed on the existing pathological timing of muscle activity. For example, even if the pathologically prolonged VM were to turn off (deactivate) at the same "late" point in the crank cycle, the residual negative work that would be generated by the deactivating muscle would be greater at faster speeds.

If inappropriate muscle activity is increased with speed, then these faster speeds should be avoided during exercise. If speed-dependent reductions in paretic work are the result of the already-present pathological timings at slower speeds, however, then we contend that rehabilitation efforts should be focused on correcting the original timing abnormality to enable appropriate speed-dependent mechanisms to produce functional adaptations.

Method

Subjects

This study was a subset of a larger study to investigate the pathological mechanisms associated with pedaling and the effects of workload on pedaling behavior, and the methods have been described in detail elsewhere.[20] Twelve elderly subjects (7 male, 5 female) without known neurological impairments and 15 subjects (12 male, 3 female) with poststroke hemiparesis of greater than 6 months' duration (mean poststroke period=42.3 months, SD=43.1, range=7-132) were tested (Table). The subjects without hemiparesis had a mean age of 69.5 years (SD=8.4, range=65-82), and the subjects with hemiparesis had a mean age of 65.3 years (SD=5.3, range=57-77). Ten subjects had left-sided hemiparesis, and 5 subjects had right-sided hemiparesis. All of these subjects had sustained a single, unilateral cerebrovascular accident with residual lower-limb paresis; had no observable perceptual, cognitive, or sensory deficits; exhibited no evidence of lower-limb contracture; had no history of cardiovascular impairments that would make pedaling contraindicated; and could tolerate sitting on a bicycle seat for approximately 1 hour. All subjects gave informed consent prior to participation in the study.

Table. Subject Characteristics

                              Age (y)

Subject                [bar]X   SD    Range   Sex

With hemiparesis         65.3   5.8   57-77   12 M
(n=15)                                         3 F

Without hemiparesis      69.5   8.4   62-82    5 F
(n=12)

                            No. of Months
                             Poststroke
                                                 Synergy
Subject                [bar]X    SD    Range   Performance(a)
                                               (Maximum=22)

With hemiparesis         42.3   43.1   7-132       0-14, n=3
(n=15)                                            15-18, n=6
                                                  19-22, n=6
Without hemiparesis
(n=12)

                            Total Modified
                              Fugl-Meyer
                              Assessment
                              Score(%)
                                               Paretic
Subject                [bar]X   SD    Range    Side

With hemiparesis         86.7   8.7   70-100     10 L
(n=15)                                            5 R

Without hemiparesis
(n=12)

(a) Synergy performance reflects the ability to move within (0-14), combine (15-18), or move outside of (19-22) extensor/flexor synergy patterns, as measured in a portion of the modified Fugl-Meyer Assessment.[24]

Procedure

All subjects with hemiparesis underwent the lower-limb portion of the Fugl-Meyer Assessment[24] a reliable and valid measure of motor recovery poststroke,[25] for assessment of global motor function. The Fugl-Meyer Assessment allowed us to identify 3 subjects who were unable to move their paretic lower limb out of mass flexor or extensor synergy, 6 subjects who were able to move their paretic lower limb with combined flexor and extensor synergy only, and 6 subjects who were able to generate isolated movements outside of mass synergy patterns. The subjects with hemiparesis ranged in walking ability from mildly impaired to nonambulatory. The subjects without hemiparesis showed no signs or symptoms of neurological disease or lower-limb orthopedic impairment.

We modified a standard ergometer with a frictionally loaded flywheel by including a backboard seating mechanism with shoulder and lap harnesses to stabilize the subject and remove the need to control balance[26] (Fig. 1). The subject sat on a standard cushioned bicycle seat while leaning, with an upright trunk posture, against the backboard. This posture allowed a subject with hemiparesis to pedal without the need to hold on to handlebars and, we believe, provided greater trunk stability. Reaction forces oriented normal and fore-aft to the pedal surfaces were measured by instrumented pedals with well-documented measurement reliability.[27] Reliability of the linear response to loads has been documented at r=.99. Accuracy checks confirmed the dynanometer's ability to measure forces with an absolute error of [+ or -] 5 N. Other tests demonstrated both mechanical and electrical decoupling between the normal and shear forces and a relatively flat dynamic response to sudden impacts. The subject's feet were firmly attached to footplates on the pedal surface, which allowed the subject to create shear and vertical forces. Angular rotation of the crank and pedals were measured by optical encoders that were fixed to the centers of rotation of the crank and pedals.

[Figure 1 ILLUSTRATION OMITTED]

The experimental protocol, conducted in an hour, consisted of measurement of pedal forces, pedal and crank kinematics, and EMG activity during pedaling at 12 randomly ordered workload and cadence combinations (workloads: 45 J="very low," 90 J="low," 135 J="medium," and 180 J="high"; speeds: 25 rpm="slow," 40 rpm="medium," and 55 rpm="fast"). The crank angular speed was displayed for the subjects, and they were instructed to maintain a steady cadence while pedaling. Once a steady cadence was achieved, 15 seconds of EMG, pedal force, and encoder data were collected (1,200 samples per second).

Surface EMG data were recorded from the RF, VM, SM, and biceps femoris (BF) muscles of the right lower extremity of the control subjects and from the same muscles of both lower extremities of the subjects with hemiparesis. Silver-silver chloride EMG eletrodes(*) were positioned over the distal half of the muscle belly such that contact surfaces were aligned longitudinally to the muscle fibers. Electrode sites were prepared by cleaning the skin with isopropyl alcohol and shaving the hair, when necessary, to ensure good contact.

Data Processing and Analysis

The net mechanical work done by each limb was calculated from the kinematic and kinetic data and used as a measure of motor performance. First, a third-order Butterworth low-pass filter was used to filter pedal forces (20-Hz cutoff) and the crank and pedal angles (8-Hz cutoff). Filtering was used to eliminate background noise due to movement artifact. The pedal force components oriented parallel and tangential to the crank arm were calculated from the normal and shear forces. The tangential component of the pedal force created a torque about the crank center (referred to as the "crank torque"), which is the only component that contributed to the angular acceleration or deceleration of the crank. Crank torque was plotted against crank angle, and calculating the area under the resulting curve yielded the net mechanical work done by the lower extremity. The positive and negative areas were also computed separately as measures of the positive (propulsive) and negative (retarding) work done by the limb.

Because muscle excitation cannot be consistently characterized by a single set of on-off times in many people with hemiparesis, each EMG measurement was quantified in terms of the percentage of activity present in 4 equal quadrants (90 [degrees]) of the pedaling cycle relative to the entire activity over the crank cycle (Fig. 1). The four quadrants were defined by axes parallel and perpendicular to the seat tube. Quadrants I and II coincided with limb extension (foot moving away from pelvis), and quadrants III and IV coincided with limb flexion. For each quadrant, the excitation, quantified by integrating the rectified EMG (IEMG), was expressed as the percentage of IEMG over the entire cycle. The relative IEMG in a quadrant provided a measure to quantitatively test whether the paretic limb's EMG offset was inappropriately prolonged (eg, quadrant III for the VM) or appropriately earlier in onset at higher speeds (eg, quadrant IV for the VM and RF, quadrant I for the BF and SM). Using this analysis method, we were able to calculate any undue speed-dependent increases in prolonged muscle activity in the spastic VM and any appropriate speed-dependent earlier activity in the VM, RF, SM, and BF.

We visually examined individual, nonaveraged crank kinematics, kinetics, and EMG activity for general trends. We then calculated the total positive, total negative, and net mechanical work and percentage of IEMG activity in each quadrant for each revolution and averaged the data to obtain the mean values for the paretic limb in each trial. These mean values represented a consistent work and EMG output from each subject at any given workload and speed condition. Within subjects, the within-trial variability of work measurements had intraclass correlation coefficients (ICC[1,1]) between .91 and .93 for net total, positive, and negative work values. The within-trial variability of EMG quadrant measurements had ICC(1,1) values of .72 for the SM, .87 for the VM, .88 for the RF, and .91 for the BF.

To establish the relationship between dependent variables (ie, net mechanical, positive, and negative work and percentage of quadrant EMG activity) and pedaling speed, the best-fit linear regression was calculated. Due to the large intersubject variability among subjects with hemiparesis, the individual subject data are also presented wherever possible to demonstrate the robustness of findings within this group. These data show that statistical findings are supported by the data from a vast majority of the subjects.

Results

Because many workload and speed combinations were performed either faster or slower than the target speeds, trials were recategorized at 1 of 7 speed levels that fell within [+ or -] 3 rpm of a specified speed category measure (level 1: 23.0-29.0 rpm, level 2: 29.1-35.0 rpm, level 3: 35.1-41.0 rpm, level 4: 41.1-47.0 rpm, level 5: 47.1-53.0 rpm, level 6: 53.1-59.0 rpm, and level 7: 59.1-65.0 rpm) (Fig. 2). This recategorization allowed analysis of all successful pedaling trials at the actual attained speeds and provided a finer gradation of speed categories. The criterion for a successful pedaling trial was the ability to pedal the crank for 30 seconds without stopping.

[Figure 2 ILLUSTRATION OMITTED]

Kinetic Responses to Increased Speed

Subjects without neurological impairments demonstrated typical kinetic responses to increased speed in that the crank torque profiles remained relatively unchanged (Fig. 3). When workload was normalized as a percentage of total work done by both lower extremities, these subjects showed no changes in net mechanical work done by each lower extremity (P [is greater than] .05) (Fig. 4A). On average, however, the subjects with hemiparesis showed decreased net mechanical work done by the paretic lower extremity as speed increased ([r.sup.2]=.89, P [is less than] .001) (Fig. 4A). All 15 subjects showed declining values of net total work as speed increased (Fig. 4B). Although less overall work was done by the paretic lower extremity, the nonparetic lower extremity was capable of generating greater positive work at higher speeds because the coupled cranks allowed it to compensate for reduced work in the paretic lower extremity. Therefore, pedaling at higher speeds resulted in reduced force output by the paretic limb.

[Figures 3-4 ILLUSTRATION OMITTED]

As previously reported with these data, the performance deficit in people with hemiparesis can be characterized as reduced force output during the downstroke phase and increased resistive force output during the upstroke phase.[20-22] At faster pedaling speeds, the decrease in total work done by the paretic lower extremity was primarily accounted for by increases in the resistive component (Fig. 5). The net negative work done by the paretic lower extremity occurred during the upstroke phase of the pedaling cycle and resulted in greater resistance to forward crank acceleration (Fig. 3). Although there was no linear relationship between net negative work and speed in the subjects without hemiparesis (P [is greater than] .05), the subjects with hemiparesis, on average, showed a linear increase in net negative work by the paretic limb ([r.sup.2]=.92, P [is less than] .001) (Fig. 5A). This average trend is supported by the individual trends in 12 subjects with hemiparesis who showed increased net negative work at higher speeds (Fig. 5B). Net positive work (ie, downstroke force) did not decrease in the paretic limb. Therefore, the reduced force output that resulted from higher speeds was primarily due to the greater amount of resistive torque generated during the upstroke phase of pedaling.

[Figure 5 ILLUSTRATION OMITTED]

Muscle Activation Responses to Increased Speed

On average, subjects with hemiparesis did not show increased percentages of prolonged VM activity at faster speeds of pedaling (P [is greater than] .05) (Fig. 6). Seven subjects with hemiparesis showed a trend for increasing percentages of inappropriate activity at higher speeds (Fig. 6). Only 3 of the 7 subjects with increased inappropriate VM activity, however, showed a gain of over 5% from the lowest speed to the fastest speed attained. Therefore, more prolonged VM activity is unlikely to be the main factor in explaining the consistent increase in negative work done by the paretic limb at higher speeds in all subjects.

[Figure 6 ILLUSTRATION OMITTED]

The effect of equal percentages of inappropriate VM activity generated over shorter cycle times (as happens at faster speeds) appeared to result in less total work done by the paretic lower extremity. When normalized by cycle time, an inappropriate percentage of VM activity was shown to be strongly correlated to net total work done by the paretic lower extremity ([r.sup.2]=.85, P [is less than] .0001) (Fig. 7). Therefore, because inappropriate VM activity is strongly correlated with reduced total work[20] and because negative work was done by the VM over a greater portion of the cycle at faster speeds, shorter cycle times resulted in greater net negative work (and, therefore, less net total work) done during the upstroke.

[Figure 7 ILLUSTRATION OMITTED]

Because subjects without neurological impairments have shown progressively earlier onsets of muscle activity and relatively equal offsets of muscle activity with faster speeds,[17-19] we were interested in determining whether subjects with hemiparesis demonstrated the same control strategy. To do this, we measured the percentage of activity present during the quadrant of onset of muscle activity and during the quadrant of offset of muscle activity. The onset of muscle activity occurred during the first 90 degrees of the upstroke (quadrant III) with the RF, occurred during the last 90 degrees of the upstroke (quadrant IV) with the VM, and during the initial 90 degrees of the downstroke (quadrant I) with the SM. When EMG profiles from all subjects without hemiparesis were averaged, muscle activity was observed to advance on the order of 20 degrees in the crank cycle for all 3 muscles (Figs. 8B, 9B, and 10B). Similar effects were observed in the SM, RF, and VM in individuals with hemiparesis (Figs. 8C, 9C, and 10C). On average, subjects with hemiparesis had linear trends indicating increasing percentages of activity in the quadrant of onset for the VM and SM, but not for the RF (VM: [r.sup.2]=.89, P [is less than] .001; SM: [r.sup.2]=.851, P [is less than] .001; RF: P [is greater than] .05). Fourteen subjects with hemiparesis showed linearly increasing trends in the VM (Fig. 8A), 10 subjects with hemiparesis showed linearly increasing trends in the RF (Fig. 9A), and 13 subjects with hemiparesis showed linearly increasing trends in the SM (Fig. 10A). With the offset of muscle activity, however, the RF, VM, and SM activity remained relatively unchanged (P [is greater than] .05 for each muscle). Therefore, individuals with hemiparesis exhibit (as do individuals without hemiparesis) earlier onsets of muscle activity in the VM and SM and constant offsets in the VM, RF, and SM when pedaling at faster speeds.

[Figures 8-10 ILLUSTRATION OMITTED]

Discussion and Conclusion

The main finding from our study was that the already reduced force output of the paretic lower extremity was further reduced at higher speeds of pedaling. Reductions occurred mostly because of increased negative work done during the upstroke phase in the pedaling cycle. That is, greater resistive torque was generated at faster speeds. This reduction in pedal force output was consistent among all subjects with hemiparesis, regardless of the initial level of force output in the paretic lower extremity. The nonparetic limb was able to compensate for this decrement in pedal force by increasing its force output, and we therefore were able to observe the behavior of the paretic lower extremity despite its resistive torque increasing during faster speed pedaling.

Even though the force output of the paretic lower extremity was reduced at faster pedaling speeds, there was little indication of exacerbation of inappropriate muscle activity. Despite the fact that the agonist burst was prolonged, we observed that the duration of the burst in absolute time was actually lessened at faster speeds because the offset of muscle activity in the VM, RF, and SM occurred at similar points in the crank cycle at progressively faster speeds. Other researchers[12,13,28,29] have attempted to show a strong relationship between speed-dependent hyperactive stretch reflex activity and speed-dependent exacerbation of movement dysfunction in people with hemiparesis. Therefore, because our results did not show an increased amount of EMG activity at progressively faster speeds during the prolonged activation in the VM, a speed-dependent reflex effect could not be identified. In addition, Sahrmann and Norton[30] demonstrated that prolonged activation, not necessarily due to hyperactive stretch reflexes, was a major determinant of movement speed and could result in slowing of movement, especially during reversal of movement. Other studies that have quantified what is often called "spasticity" during both walking and pedaling have focused on the abnormal timing of muscle activity rather than the occurrence of triggered stretch reflex responses.[23,31] The results of our study and other reports[20-22] appear to support the finding that mistiming of muscle activity rather than stretch-induced abnormal muscle activity is responsible for producing counterproductive decelerating forces in the limbs during cyclical locomotive tasks.

Our results also demonstrate an intact strategy for generating pedal forces at faster pedaling speeds.[17-19] Subjects without neurological impairments were shown to generate (in this study and in previous studies[17-19]) EMG activity earlier in the cycle, supposedly to produce peak level of forces during equivalent regions of the crank cycle. We believe this strategy is used to compensate for the decreased time available to reach peak force. A similar strategy was observed in the majority of subjects with hemiparesis in our study, even though their timing of EMG activity was impaired at slower speeds.

Even though the abnormally prolonged activity in the VM was not increased at higher speeds, unchanged prolonged activity may still contribute to greater negative work because the force production during deactivation of the VM will affect a greater portion of the cycle when the crank moves at a faster speed. Therefore, the occurrence of prolonged activity that is present with paretic muscle at slower speeds will generate lessened force output simply because of the greater mechanical demands associated with faster pedaling speeds.

Because inappropriate muscle activity did not appear to increase at faster speeds, and for the reasons outlined, we hypothesize that the mechanical demands of faster pedaling speeds were the major contributing factor to the reduction in force output that occurred at faster pedaling speeds. This hypothesis implies that there is no harm to the nervous system when training individuals with spasticity to pedal at faster speeds. This result parallels an earlier published report that increased workloads also do not exacerbate inappropriate muscle activity during pedaling.[21] Together, these 2 studies suggest that regimens of pedaling at faster speeds or high workload might allow the nervous system to further adapt to higher speeds of movement and higher workloads so that functional improvements may occur.

Finally, physiological benefits have been ascribed to aerobic ergometer exercise.[32] Therefore, utilizing higher-speed pedaling to reach target heart rates need not be avoided due to fear of exacerbating inappropriate muscle activity. Appropriate screening and monitoring of vital cardiovascular and respiratory signs in people with cerebrovascular disease, in our view, are essential to assure safety.

The interpretation of the results from our study is limited by several issues. First, it may not be valid to compare the behavior of the hemiparetic nervous system during pedaling with behavior during other lower-limb tasks such as walking or rising from a sitting position to a standing position. Perhaps the security supplied by the pedaling paradigm prevents protective mechanisms, typically applied during functional tasks, from being applied during movement. These protective mechanisms, stimulated at faster speeds, may exacerbate already impaired motor patterns during a less-secure task such as unsupported

walking. Furthermore, by specifying target speeds for the subjects in our study, we were unable to determine the fastest speeds at which they can move. It is possible that at least some of these subjects were pedaling at relatively slow speeds compared with their maximum capability, whereas other subjects may have reached their maximum capability with the slower targeted speeds. That is, the relative effort applied by each subject at the given target speeds was probably highly variable. Finally, caution must be used when applying the results of this study, which was conducted over short bouts of pedaling exercise (less than 1 minute), with longer bouts of exercise that are typically applied in the clinic. It is possible that, over time, the ability to control paretic muscles at faster speeds may degrade with fatigue or heightened sensitivity to stretch. We recommend that similarly designed studies be applied to different movement paradigms, at a full spectrum of speeds and over longer exercise durations.

(*) Therapeutics Unlimited, 2835 Friendship St, Iowa City, IA 52240.

References

[1] Perry J. The mechanics of walking in hemiplegia. Clin Orthop. 1969;63:23-31.

[2] Wagenaar RC, Beek WJ. Hemiplegic gait: a kinematic analysis using walking speed as a basis. J Biomech. 1992;25:1007-1015.

[3] Brandstater ME, de Bruin H, Gowland C, Clark BM. Hemiplegic gait: analysis of temporal variables. Arch Phys Med Rehabil. 1983;64:583-587.

[4] Perry J, Garrett M, Gronley JK, Mulroy SJ. Classification of walking handicap in the stroke population. Stroke. 1995;26:982-989.

[5] Dettmann MA, Linder MT, Sepic SB. Relationships among walking performance, postural stability, and functional assessments of the hemiplegic patient. Am J Phys Med. 1987;66:77-90.

[6] Knutsson E, Richards C. Different types of disturbed motor control in gait of hemiparetic patients. Brain. 1979;102(pt 2):405-430.

[7] Bohannon RW, Walsh S. Nature, reliability, and predictive value of muscle performance measures in patients with hemiparesis following stroke. Arch Phys Med Rehabil. 1992;73:721-725.

[8] Bohannon RW, Andrews AW. Correlation of knee extensor muscle torque and spasticity with gait speed in patients with stroke. Arch Phys Med Rehabil. 1990;71:330-333.

[9] Roth E, Merbitz C, Mroczek K, et al. Hemiplegic gait: relationships between walking speed and other temporal parameters. Am J Phys Med Rehabil. 1997;76:128-133.

[10] Bobath B. Adult Hemiplegia: Evaluation and Treatment. 2nd ed. London, England: William Heinemann Medical Books Ltd; 1978.

[11] Dimitrijevic MR, Nathan PW. Studies of spasticity in man, I: some features of spasticity. Brain. 1967;90:1-30.

[12] Corcos DM, Gottlieb GL, Penn RD, et al. Movement deficits caused by hyperexcitable stretch reflexes in spastic humans. Brain. 1986;109: 1043-1058.

[13] Knutsson E, Martensson A, Gransberg L. Influences of muscle stretch reflexes on voluntary, velocity-controlled movements in spastic paraparesis. Brain. 1997;120(pt 9):1621-1633.

[14] McComas AJ. Skeletal Muscle: Form and Function. Champaign, Ill: Human Kinetics Inc; 1996:287-298.

[15] McArdle WD, Katch FI, Katch VL. Exercise Physiology. 4th ed. Baltimore, Md: Williams & Wilkins; 1996:393-415.

[16] Kautz SA, Hull ML. A theoretical basis for interpreting the force applied to the pedal in cycling. J Biomech. 1993;26:155-165.

[17] Suzuki S, Watanabe S, Homma S. EMG activity and kinematics of human cycling movements at different constant velocities. Brain Res. 1982;240:245-258.

[18] Neptune RR, Kautz SA, Hull ML. The effect of pedaling rate on coordination in cycling. J Biomech. 1997;30:1051-1058.

[19] Marsh AP, Martin PE. The relationship between cadence and lower extremity EMG in cyclists and noncyclists. Med Sci Sports Exerc. 1995; 27:217-225.

[20] Kautz SA, Brown DA. Relationships between timing of muscle excitation and impaired motor performance during cyclical lower extremity movement in post-stroke hemiplegia. Brain. 1998;121 (pt 3):515-526.

[21] Brown DA, Kautz SA. Increased workload enhances force output during pedaling exercise in persons with poststroke hemiplegia. Stroke. 1998;29:598-606.

[22] Brown DA, Kautz SA, Dairaghi CA. Muscle activity adapts to anti-gravity posture during pedalling in persons with post-stroke hemiplegia. Brain. 1997;120(pt 5):825-837.

[23] Benecke R, Conrad B, Meinck HM, Hohne J. Electromyographic analysis of bicycling on an ergometer for evaluation of spasticity of lower limbs in man. In: Desmedt JE, ed. Motor Control Mechanisms in Health and Disease. New York, NY: Raven Press; 1983:1035-1046.

[24] Fugl-Meyer AR, Jaasko L, Leyman I, et al. The post-stroke hemiplegic patient, I: a method of evaluation of physical performance. Scand J Rehabil Med. 1975;7:13-31.

[25] Duncan PW, Propst M, Nelson SG. Reliability of the Fugl-Meyer assessment of sensorimotor following cerebrovascular accident. Phys Ther. 1983;63:1606-1610.

[26] Brown DA, Kautz SA, Dairaghi CA. Muscle activity patterns altered during pedaling at different body orientations. J Biomech. 1996;29: 1349-1356.

[27] Newmiller J, Hull ML, Zajac FE. A mechanically decoupled two force component bicycle pedal dynanometer. J Biomech. 1988;21: 375-386.

[28] Nathan PW. Factors affecting spasticity. Int Rehabil Med. 1980;2: 27-30.

[29] Phillips CA, Repperger DW, Chelette TL. The acceleration-velocity relationship: identification of normal and spastic upper extremity movement. Comput Biol Med. 1997;27:309-328.

[30] Sahrmann SA, Norton BJ. The relationship of voluntary movement to spasticity in the upper motor neuron syndrome. Ann Neurol. 1977;2:460-465.

[31] Fung J, Barbeau H. A dynamic EMG profile index to quantify muscular activation disorder in spastic paretic gait. Electroencephalogr Clin Neurophysiol. 1989;73:233-244.

[32] Potempa K, Lopez M, Braun LT, et al. Physiological outcomes of aerobic exercise training in hemiparetic stroke patients. Stroke. 1995;26:101-105.

DA Brown, PhD, PT, is Assistant Professor, Programs in Physical Therapy, Northwestern University Medical School, 645 N Michigan Ave, Suite 1100, Chicago, IL 60611 (USA) ([e-mail]). Address all correspondence to Dr Brown.

SA Kautz, PhD, is Biomedical Engineer, The Rehabilitation Research & Development Center, VA Palo Alto Health Care System, Palo Alto, Calif.

Concept and research design, writing, data collection and analysis, project management, fund procurement, subjects, facilities/equipment, institutional liaisons, clerical/secretarial support, and consultation (including review of manuscript prior to submission) were provided by Brown and Kautz. Christine Dairaghi provided technical assistance with data collection and provision of subjects.

This study was approved by the Stanford University School of Medicine Institutional Review Board.

This work was funded, in part, by the Foundation for Physical Therapy and the Department of Veterans Affairs, Rehabilitation Research and Development Division.

This article was submitted October 15, 1998, and was accepted June 15, 1999.3