Differences in EMG spike shape between individuals with and without non-specific arm pain

Journal of Neuroscience Methods 178 (2009) 148–156

Contents lists available at ScienceDirect

Journal of Neuroscience Methods journal homepage: www.elsevier.com/locate/jneumeth

Differences in EMG spike shape between individuals with and without non-specific arm pain Kristina M. Calder a,∗ , David A. Gabriel b , Linda McLean a a

Motor Performance Laboratory, School of Rehabilitation Therapy, Louise D. Acton Building, 31 George Street, Queen’s University, Kingston, Ontario, Canada K7L 3N6 Electromyographic Kinesiology Laboratory, Brock University, 500 Glenridge Avenue, St. Catharines, Ontario, Canada L2S 3A1

b

a r t i c l e

i n f o

Article history: Received 5 June 2008 Received in revised form 10 November 2008 Accepted 11 November 2008 Keywords: Electromyography Spike shape analysis Extensor carpi radialis brevis Motor unit recruitment Non-specific arm pain Repetitive strain injury

a b s t r a c t The purpose of this study was to determine whether spike shape analysis of surface electromyographic (SEMG) activity is a useful tool to study muscle disorders. This study investigated SEMG spike shape parameters at low levels of contraction and changes in SEMG spike shape across different levels of isometric wrist extension contractions in individuals with non-specific arm pain (NSAP), asymptomatic subjects deemed at-risk for repetitive strain injury, and asymptomatic control subjects. Twenty-two asymptomatic control subjects, 8 at-risk subjects, and 16 subjects with NSAP participated. Bipolar SEMG data were recorded from the ECRB muscle during isometric wrist extension contractions at 10, 20, 30, 40, 50, 60, and 70% of maximum voluntary contraction (MVC) force performed in a randomized order. Five criterion measures: mean spike amplitude (MSA), mean spike duration (MSD), mean spike slope (MSS), mean spike frequency (MSF), and mean number of peaks per spike (MNPPS) were computed from each SEMG signal. A one-way analysis of covariance (ANCOVA) of the spike shape parameters computed from the 10% MVC data, with group as a main effect and age as a covariate, revealed a significant group by age interaction for MSA, and significant group main effects for MSS and MNPPS, where the NSAP group had lower MSS and lower MNPPS than the control subjects. An ANCOVA including group as a main effect and contraction level and age as covariates revealed that all three groups showed predictable changes in the spike shape analysis criterion measures over increasing contraction force levels, where motor unit recruitment and rate coding appear to be the primary mechanisms for increasing force output of the muscle. Significant interactions between group and contraction level were observed for MSD, MSA, MSS, and MNPPS. The NSAP group presented with differences in how the spike shape measures change with increasing contraction level that may be indicative of myogenic changes, a result that is consistent with previous quantitative EMG findings. This work provides evidence that NSAP involves myogenic changes in the ECRB muscle and that spike shape analysis may be a valuable non-invasive tool in the evaluation of neuromuscular disorders. © 2008 Elsevier B.V. All rights reserved.

1. Introduction Non-specific arm pain (NSAP) is a condition whereby individuals have progressive forearm symptoms but the pain and disability cannot be readily attributed to a specific local lesion. NSAP appears to develop as a result of chronic performance of repetitive movements of the arms or hands such as those used during data entry tasks. The diagnosis of NSAP is made through ruling out any other pathology such as tenosynovitis, epicondylitis, or de Quervain’s tendonitis or cervical radiculopathy (Harrington et al., 1998).

∗ Corresponding author. Tel.: +1 905 630 4975; fax: +1 613 533 6103. E-mail address: [email protected] (K.M. Calder). 0165-0270/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jneumeth.2008.11.015

Recently our group used quantitative motor unit potential (MUP) analysis techniques to investigate potential differences in the electrophysiological characteristics of motor units in healthy control subjects, subjects deemed at-risk of developing NSAP, individuals with lateral epicondylitis (LE) and individuals with NSAP. We found evidence that the NSAP group had MUP changes that were consistent with myopathy (Calder et al., 2008). Although the needle electromyographic (EMG) data in the previous study provided valuable information, the approach used was invasive and is not readily available in many laboratories. As an alternative, surface electromyography (SEMG) data acquisition is less invasive, and SEMG systems are commonly available in many laboratories. As such, spike shape analysis of SEMG data (Gabriel et al., 2007) may be a valuable evaluation tool if it is deemed to be useful in differentiating neurogenic from myogenic disorders.

K.M. Calder et al. / Journal of Neuroscience Methods 178 (2009) 148–156

149

Table 1 Predicted changes in the surface electromyography (SEMG) spike shape parameters across increasing force output. The parameters include mean spike amplitude (MSA), mean spike frequency (MSF), mean spike duration (MSD), mean spike slope (MSS), and mean number of peaks per spike (MNPSS). The results from the current study in the extensor carpi radialis brevis (ECRB) muscle in the asymptomatic controls, at-risk subjects, and individuals with non-specific arm pain (NSAP) are presented. Five SEMG spike measures

Fig. 1. An illustration of SEMG spike shape analysis. A spike is defined by an upward and downward deflection in the IP where both deflections cross zero and are at least 100 ␮V in amplitude. In this illustration the apex of each spike is denoted by a circle whereas squares define their base. Note that the deflection that occurs between 1 and 2 is not defined as a spike because it does not cross zero and would be assumed to be background noise.

The SEMG signal represents the summation of motor unit (MU) activity occurring under the recording electrodes during a voluntary or involuntary muscular contraction. Changes observed in the time and frequency domain parameters can be linked to changes in MU firing patterns, however there is no way to distinguish between rate coding and motor unit recruitment (Zhou and Rymer, 2004a,b), or the occurrence of synchronization (Fuglevand et al., 1993). An alternative method of SEMG analysis was recently developed by Gabriel (2000) to investigate differences in MU firing patterns across increasing levels of contraction. This non-invasive technique, called ‘spike shape analysis’, uses changes in the shape of individual spikes within the SEMG interference pattern (IP) to make inferences about changes in MU firing patterns with increasing motor output or with fatigue. Spike morphological features such as size and shape are the only characteristics of a motor unit potential that are still preserved in a full IP, since the peaks are still detectable (Magora and Gonen, 1970). Spike shape analysis uses two major components of the SEMG signal: spikes and peaks (Fig. 1). A spike is composed of a single upward and downward deflection that is at least 100 ␮V in amplitude (Hirose and Sobue, 1972). A peak is defined as a pair of upward and downward deflections within a spike that do not together constitute a discrete spike (Beach et al., 1982). The variables computed to perform a spike shape analysis include mean spike amplitude (MSA), mean spike frequency (MSF), mean spike slope (MSS), mean spike duration (MSD) and mean number of peaks per spike (MNPPS). The specific algorithms used to compute these measures are described in detail in Gabriel et al. (2007), and are briefly presented in the methods section of this paper. The main advantage of spike shape analysis is that the simultaneous evaluation of all five measures presents a unique characterization of the underlying MU activity across a muscle contraction. For example, the changes in spike shape associated with increased recruitment (increase in MSA, MSF, MSS, MNPPS; decrease in MSD) are different from those associated with synchronization (increase in MSA, MSS, MSD; decrease in MSF and MNPPS), a distinction that is not possible with traditional time and frequency-based analyses of SEMG (see Table 1). The reliability of SEMG spike shape parameters has been previously established to be good (ICCs ranged from 0.76 to 0.93) (Gabriel, 2000), and spike shape measures have recently been shown to change in a predictable manner with increases in isometric force production in the biceps brachii (BB) muscle (Gabriel et al., 2007). This method has

MSA

MSF

MSS

MSD

MNPPS

Motor unit firing pattern Increased firing frequency Increased recruitment Increased synchronization

– ↑ ↑

↑ ↑ ↓

– ↑ ↑

↓ ↓ ↑

– ↑ ↓

Muscle ECRB; asymptomatic subjects ECRB; at-risk subjects ECRB; NSAP subjects

↑* ↑* ↑*

↑* ↑* ↑*

↑* ↑* ↑*

↓ ↓* ↓*

↑* ↑* ↑*

*

Denotes that changes were significant at the alpha = 0.05 level.

not been used to identify neuromuscular changes seen in clinical populations. In myopathy, there is a need to recruit more MUs with higher firing rates and faster conduction velocities earlier in the force gradation process than in healthy individuals. This is because the lower number of muscle fibers per MU decreases the twitch force of each MU (Beardwell, 1967). Because of the higher number of MU discharges (firing rate) at a given effort level required to compensate for the loss of muscle force generated from each MU firing, it can be hypothesized that MNPPS and MSF, both indicators of firing frequency, will increase earlier in myopathic muscle than in healthy muscle when force output is increased (Magora and Gonen, 1970; Muro et al., 1982; Latash, 1988a,b). The MSA and MSS may also increase sooner in myopathic muscle because there is earlier recruitment of larger MUs (reflected in MSA) with faster conduction velocities (reflected in MSS) to compensate for lower MU density (Viitasalo and Komi, 1975; Komi and Vitasalo, 1976). Recruitment of additional MUs also results in a greater amount of asynchronous activity (Muro et al., 1982). The superposition of asynchronous MUPs may result in a more complex IP with a greater number of peaks per spike (Magora and Gonen, 1970). As such, if MSA, MSS, MSF and MNPPS all increase earlier in a graded contraction with increasing force output, one can deduce that recruitment is occurring sooner in that population. If an increase in the MSF is observed without a substantial change in the other spike shape parameters, one can deduce that rate coding is likely dominating. In neuropathic conditions, motor neuron loss may lead to a higher degree of motor unit potential summation and MU synchronization since collateral sprouting increases the fiber density within the existing MUs (Erminio et al., 1959; McComas et al., 1971; Stalberg, 1982). Using the spike shape analysis method, a neuropathic condition, for a given force level the muscle might exhibit higher MSA and MSD values, because of the larger MU size, and lower MNPPS and MSF values as patients would generate more force per MU compared to healthy control subjects (Magora and Gonen, 1970; Muro et al., 1982; Latash, 1988a,b). An increased occurrence of MU synchronization can increase the amplitude and slope of the resulting spikes in the IP, which might be observed as a higher MSS for a given contraction level, or as an earlier increase in MSS with increasing contraction levels. The first purpose of this study was therefore to determine whether spike shape analysis would result in findings during lowlevel muscle contractions that are consistent with those previously found using quantitative EMG analysis in the same sample of individuals with NSAP, individuals deemed at-risk of developing NSAP and healthy control subjects (Calder et al., 2008). The second purpose of this study was to determine if the rate of change

150

K.M. Calder et al. / Journal of Neuroscience Methods 178 (2009) 148–156

in the spike shape measures is different among the three study groups. 2. Methods 2.1. Subjects Subjects were recruited from the Kingston community through advertisements posted in local newspapers, and through posters placed in physiotherapy clinics, physicians’ offices, and workplaces. The study was approved by the Queen’s University Health Sciences Research Ethics Board and all subjects provided informed consent prior to participation. Twenty-two asymptomatic control subjects (8 males, 14 females; 27 ± 5 years old), 8 at-risk subjects (2 males, 6 females; 45 ± 13 years old), and 16 subjects with NSAP (7 males, 9 females; 50 ± 9 years old) participated in the study. 2.2. Experimental design Subjects were excluded if they presented with clinical signs or symptoms of cervical radiculopathy, radial nerve palsy, and/or carpal tunnel syndrome. A screening examination of the upper extremities, including myotome testing (C5–C8), dermatome (light touch, pin prick) testing (C5–C8), and assessment of the deep tendon reflexes at the C5–C8 levels was performed, and those with muscle weakness, hypo or hyper-aesthesias, or hypo- or hyperreflexive deep tendon reflexes were excluded from participation. Cervical spine range of motion was assessed in sitting to ensure that cervical movements did not reproduce the forearm signs and symptoms. The movements tested included lower cervical spine flexion, extension, lateral flexion, rotation, and combined extension and lateral flexion. These positions were held at the end of the available range of motion for 10 s and if the subject reported symptoms during these tests, he or she was excluded from the study. Subjects were assigned to the NSAP group if they experienced pain on palpation of the extensor carpi radialis brevis (ECRB) muscle belly, and if they had forearm pain when performing repeated wrist extension contractions during their daily activities, but on clinical examination passive wrist flexion with their elbow extended did not reproduce their signs and symptoms. Because our goal was to characterize individuals with NSAP, we did not include any subjects who had clinical signs that could be attributed to LE or to both LE and NSAP. Potential participants were excluded if resisted wrist flexion, and palpation of the medial epicondyle reproduced symptoms as these would indicate the presence of wrist flexor pathology. Control and at-risk subjects had no pain on resisted wrist extension, passive wrist flexion, or palpation of the lateral epicondyle or the ECRB muscle. Subjects who were assigned to the asymptomatic at-risk group had no history of arm injury or pain and worked in jobs that demanded frequent repetition of wrist extension contractions, whereas the subjects in the control group did not perform activities that required repetitive wrist motions at work or during their leisure time. Subjects were seated in a straight-back chair with the elbow of their most affected limb (NSAP subjects) or dominant arm (control subjects and at-risk subjects) flexed at 90◦ and their forearm pronated and resting on a custom-built table. This table was designed to stabilize flexion and extension contractions of the wrist. Isometric force of the wrist extensor group was measured in Newtons (N) by a strain gauge (OmegadyneTM LC101-500 lb “S” beam load cell) secured to the bottom of the testing table with adjustable straps passing through an opening and secured around the dorsum of the hand. Force signals were sampled (1000 Hz) using a 16-bit National Instrument analog to digital converter (PCI-MCIAEIC) with custom LabviewTM software (v.6.1; National Instruments,

Austin). Subjects were trained to perform isometric contractions of the wrist extensor muscles by pulling on the strap with the dorsum of their hand. During wrist extension contractions, subjects were instructed to keep their elbow and forearm on the testing table, and to only extend at the wrist. Once subjects became familiar with the testing situation, they performed three brief (3–5 s) maximal wrist extension contractions. The highest force value computed using a 200 ms moving window (root mean square; 199 ms overlap) across the three data files represented the maximal voluntary contraction (MVC) effort for that subject (contraction level = 100%). In subsequent contractions the trajectory of the smoothed (25 ms moving window over which the root mean square (RMS) value was computed) force signal was displayed relative to the target force on a PC monitor placed in front of the subjects to provide visual feedback. The order of the target forces of 10, 20, 30, 40, 50, 60, and 70% of MVC was randomly selected prior to each testing session. For each contraction, subjects were instructed to “pull up with the back of your hand by extending at the wrist until your force tracing falls between the two bars on the computer screen, and maintain your force tracing between the two bars as stable as possible until I tell you to relax”. The two bars represented ±2 S.D. of the target force level (which is a % of the MVC). Once the subjects had held their required force level stable (within the tolerance limits, the two bars) for approximately 5 s while EMG data were recorded, they were told to relax. Subjects were provided with a 3-min rest before each contraction. 2.3. Force and SEMG recordings Prior to electrode placement, the skin on the upper forearm and dorsum of the hand was cleaned with rubbing alcohol to reduce impedance at the electrode skin interface. The motor point of the ECRB was then determined. The motor point was defined as the region of the muscle where the lowest possible electrical stimulus produced a visible muscle twitch. The cathode portion of a stimulating probe was placed in the estimated motor point region, approximately two finger widths from the cubital crease. With the train rate on the stimulator set at 10 pulses per second (pps), and the stimulation duration set at 1 ms (Pierrot-Deseilligny and Mazevet, 2000), the cathode was moved around the muscle belly. Once the motor point was determined, two Ag/AgCl electrodes (Kendall-LTP, Chicopee, Massachusetts cut in half to measure 10 mm × 30 mm) were placed distal to the motor point in a traditional bipolar SEMG configuration with a center to center inter-electrode distance of 20 mm. A reference electrode (20 mm × 30 mm) was placed on the posterior aspect of the hand. The bipolar SEMG data were amplified (1000×) and bandpass filtered from 10 to 500 Hz using Bortec AMT 8 SEMG amplifiers (Bortec, Calgary, Canada), digitized using a 16-bit National Instruments Analog to Digital Conversion card (PCI-MCIAEIC) and sampled at 1 kHz with an anti-aliasing filter of 500 Hz using custom LabviewTM software (v.6.1; National Instruments, Austin). The data were stored on a PC for off-line processing. 2.4. Data reduction and analysis Starting from the beginning of each data file, a computer algorithm identified a 2 s window of force data during which the force signal had the lowest RMS error relative to the target force level. The 2 s of SEMG data recorded simultaneously with this force window were then used to calculate the five spike shape measures: MSA, MSF, MSS, MSD, and MNPPS (Gabriel et al., 2007). The algorithm for detection of peaks and definition of spike amplitude (SA) was taken from Beach et al. (1982) and Hirose and Sobue (1972), respectively, and implemented using MATLAB (the MathWorks Inc., Natick, MA). Refer to the first spike in Fig. 1 in reference to the following calculations. Note that the highest peak in a spike is taken as the apex

K.M. Calder et al. / Journal of Neuroscience Methods 178 (2009) 148–156

(B) for all calculations, and the x and y subscripts are the x- and ycoordinates of points A, B and C on the spike. The MSA calculation is shown as SAy =

(By − Ay ) + (BY − CY ) , 2

(1)

and MSA =

NS  SA

NS

i=1

(2)

NS , TD

(3)

where NS is the number of spikes and TD is the total duration of a given sample of SEMG. Individual spike slopes are calculated from the beginning of each spike (point A) to its peak (point B) simply based on the x- and ycoordinates of points A and B on the spike. Once the slope for each spike is calculated, the mean spike slope across the data window is determined as the sum of the spike slopes divided by the number of spikes (NS) seen within that data window. The MSS algorithm is shown in as SS =

BY − AY , BX − AX

(4)

and MSS =

NS  SS

,

NS

i=1

(5)

where SS is the spike slope and NS is the number of spikes of a given sample of SEMG. MNPPS is calculated by determining the number of peaks (P) in the recording window and dividing it by the NS in the same window. The MNPPS algorithm is shown as MNPPS =

P . NS

(6)

MSD is calculated by determining the duration of each individual spike. The duration of all of the spikes within the data window are then summed and divided by the NS. The MSD algorithm is shown as MSD =

NS  CX − AX i=1

NS

interactions (group × contraction level, group × age, and contraction level × age) in our model. The ␣-level was set at 0.05 for all tests. Post hoc analyses were performed using Tukey’s pair-wise comparisons where appropriate. Data in the text are reported as group means ± standard deviations, and were analyzed using MINITAB® Statistical Software (v.14). 3. Results

,

where SAy is spike amplitude and NS is the number of spikes in a given sample of SEMG. MSF is calculated by taking the NS seen in the data window and dividing by the total duration (TD) of that data analysis window. The MSF algorithm is shown as MSF =

151

,

(7)

where NS is the number of spikes in a given window of SEMG. To test for group differences in the spike shape parameters computed from low-level contraction data, and thus to address the first objective, a one-way analysis of covariance (ANCOVA) was used to compare MSA, MSF, MSS, MSD and MNPPS across the three groups at 10% MVC. Age was a potential confounding factor and was therefore included as a covariate. The F-test for group used the mean square error term of the interaction between group and age as the denominator in order to determine whether or not a group effect was present regardless of any age effects (Zar, 1984). To determine if the trends in the spike shape analysis features with increasing force output were similar across the three groups, a backward stepwise ANCOVA model including group as a main effect and contraction level and age as covariates was used. For the backward stepwise ANCOVA model, we included the threeway interaction (group × contraction level × age) and all two way

Age was found to be significantly different among the three groups, where the control subjects were significantly younger than the at-risk and NSAP subjects (p < 0.05), but no difference in mean age was found between the at-risk and NSAP groups (p > 0.05). To address this, as noted above, we included age as a covariate in our statistical models. 3.1. Comparison of the three groups at 10% of MVC The ANCOVA revealed a significant group by age interaction for MSA (p < 0.05), but not for any of the other variables. To test for a group difference in this case, therefore, the F-test was performed by dividing the mean square value for group by the mean square value for the interaction between group and age. The result indicated that there was no significant group effect for MSA at 10% MVC. Significant group differences were found for MSS and MNPPS at the 10% MVC level (p < 0.05). Post hoc analysis revealed that the NSAP group had significantly lower MSS (NSAP: 0.03 mV/ms and control: 0.07 mV/ms; p < 0.05) and MNPPS (NSAP: 0.90 and control: 1.10; p < 0.05) values than the control group. The at-risk group was not found to have significantly different MSS or MNPPS values than the other two groups. 3.2. Within-group changes in spike shape measures with increasing force 3.2.1. Asymptomatic healthy controls The means ± standard deviations and slope values for the five spike shape measures across the seven levels of % of MVC are presented in Table 2. MSA showed a significant increase of 1.1 mV (550%) from 10 to 70% of MVC (p < 0.05). MSF significantly increased from 10 to 70% of MVC (22.1 Hz; 27.7%; p < 0.05). MSD decreased from 10 to 70%, however the change was not found to be significant (0.3 ms 3.8%; p > 0.05). As force increased from 10 to 70% of MVC, MSS exhibited a significant increase of 0.35 mV/ms (500%, p < 0.05). MNPPS showed a 0.05 increase in number of peaks per spike from 10 to 70% of MVE, accounting for a significant increase of 4.6% (p < 0.05). 3.2.2. At-risk asymptomatic subjects The means ± standard deviations and slope values for the five spike shape measures across the seven levels of % of MVC are presented in Table 3. MSA showed a significant increase of 0.43 mV (430%) from 10 to 70% of MVC (p < 0.05). MSF significantly increased from 10 to 70% of MVC (24.4 Hz; 26.9%; p < 0.05). MSD significantly decreased from 10 to 70% of MVC (1.53 ms; 18.2%; p < 0.05). As force increased from 10 to 70% of MVC, MSS exhibited a significant increase of 0.15 mV/ms (375%, p < 0.05). MNPPS showed a significant increase of 0.11 peaks per spike from 10 to 70% of MVC, accounting for an 11.6% increase (p < 0.05). 3.2.3. Subjects with non-specific arm pain The means ± standard deviations and slope values for the five spike shape measures across the seven levels of % of MVC are presented in Table 4. MSA significantly increased by 0.39 mV across the contraction levels, showing a 325.0% increase (p < 0.05). MSF significantly increased from 10 to 70% of MVC (36.0 Hz; 49.1%; p < 0.05).

152

K.M. Calder et al. / Journal of Neuroscience Methods 178 (2009) 148–156

Table 2 Spike shape measures across increasing contraction levels in the control group. Data from the asymptomatic control subjects (n = 22) (means (M) ± standard deviations (S.D.), and slope values) are presented for the spike shape analysis measures including mean spike amplitude (MSA), mean spike frequency (MSF), mean spike duration (MSD), mean spike slope (MSS), and mean number of peaks per spike (MNPPS) across the seven contraction levels (%MVC). MSA (mV) M ± S.D. %MVC 10 20 30 40 50 60 70 Change Percent Slope values *

0.2 ± 0.2 0.4 ± 0.3 0.6 ± 0.5 0.7 ± 0.5 0.9 ± 0.6 1.1 ± 0.8 1.3 ± 0.9 +1.1 550% +0.19*

MSF (Hz) M ± S.D. 81.6 ± 17.6 90.0 ± 17.8 94.0 ± 19.5 103.2 ± 12.6 103.6 ± 15.9 102.6 ± 15.4 103.7 ± 14.1 +22.1 27.1% +0.36*

MSD (ms) M ± S.D. 7.9 ± 0.8 7.6 ± 0.8 7.5 ± 1.0 7.2 ± 0.9 7.4 ± 1.2 7.6 ± 1.3 7.6 ± 1.4 −0.3 3.8% −0.004

MSS (mV/ms) M ± S.D. 0.07 ± 0.05 0.11 ± 0.08 0.17 ± 0.12 0.23 ± 0.16 0.30 ± 0.21 0.36 ± 0.26 0.42 ± 0.31 +0.35 500% +0.006*

MNPPS M ± S.D. 1.09 ± 0.09 1.10 ± 0.06 1.13 ± 0.08 1.11 ± 0.07 1.13 ± 0.07 1.14 ± 0.08 1.14 ± 0.08 +0.05 4.6% +0.0001*

Denotes a significant slope at the alpha = 0.05 level.

Table 3 Spike shape measures across increasing contraction levels in the at-risk group. Data from the asymptomatic at-risk subjects (n = 8) (means (M) ± standard deviations (S.D.), and slope values) are presented for the spike shape analysis measures including mean spike amplitude (MSA), mean spike frequency (MSF), mean spike duration (MSD), mean spike slope (MSS), and mean number of peaks per spike (MNPPS) across the seven contraction levels (%MVC). MSA (mV) M ± S.D. %MVE 10 20 30 40 50 60 70 Change Percent Slope values *

0.10 ± 0.09 0.18 ± 0.18 0.24 ± 0.23 0.28 ± 0.33 0.41 ± 0.44 0.46 ± 0.49 0.53 ± 0.49 +0.43 430% +0.007*

MSF (Hz) M ± S.D. 90.6 ± 29.2 99.1 ± 32.0 110.4 ± 20.2 112.4 ± 14.8 114.0 ± 16.4 115.1 ± 13.0 115.0 ± 13.4 +24.4 26.9% +0.4*

MSD (ms) M ± S.D. 8.40 ± 1.62 7.27 ± 1.01 7.04 ± 1.00 6.88 ± 0.61 6.93 ± 0.66 6.83 ± 0.65 6.87 ± 0.77 −1.53 18.2% −0.02*

MSS (mV/ms) M ± S.D. 0.04 ± 0.03 0.07 ± 0.07 0.08 ± 0.08 0.10 ± 0.12 0.14 ± 0.14 0.16 ± 0.17 0.19 ± 0.18 +0.15 375% +0.003*

MNPPS M ± S.D. 0.95 ± 0.15 1.02 ± 0.08 1.05 ± 0.07 1.02 ± 0.08 1.07 ± 0.10 1.07 ± 0.08 1.06 ± 0.07 +0.11 11.6% +0.002*

Denotes significant slopes at the alpha = 0.05 level.

MSD significantly decreased from 10 to 70% of MVC (1.79 ms; 19.7%; p < 0.05). As force increased from 10 to 70% of MVC, MSS exhibited a significant increase of 0.14 mV/ms (466.7%, p < 0.05). MNPPS displayed an increase from 10 to 70% of MVC (0.17 peaks per spike; 18.9%), which was significant (p < 0.05). 3.3. Group differences in changes in spike shape measures with increasing force The repeated measures ANCOVA revealed no significant threeway interactions between group, level and age in any of the spike shape parameters (p > 0.05), suggesting that age did not affect differences in the slopes among the groups as contraction level increased. Age also did not significantly interact with contraction

level (p > 0.05) for any of the spike shape parameters, again suggesting that age does not affect the slope of the regressions of each variable with contraction level. A significant group by age interaction was found for MSA and MSS (Fig. 2a and b) (p < 0.05) indicating that for these variables, the age difference between the groups caused a difference in the intercepts in the MSA and MSS regressions with contraction levels. There were significant group by contraction level interactions for the MSA, MSS (Fig. 3a and b), MNPPS and MSD (Fig. 4a and b) (p < 0.05) features, indicating that the groups exhibited different slopes across contraction level. The interactions revealed that the control subjects had significantly higher slopes for MSA, and MSS (0.0185 mV/%MVC and 0.006 mV/ms/%MVC, respectively) than individuals at-risk (0.007 mV/%MVC and 0.003 mV/ms/%MVC,

Table 4 Spike shape measures across increasing contraction levels in the NSAP group. Data from the subjects with NSAP (n = 16) (means (M) ± standard deviations (S.D.), and slope values) are presented for the spike shape analysis measures including mean spike amplitude (MSA), mean spike frequency (MSF), mean spike duration (MSD), mean spike slope (MSS), and mean number of peaks per spike (MNPPS) across the seven contraction levels (%MVC). MSA (mV) M ± S.D. %MVE 10 20 30 40 50 60 70 Change Percent Slope values *

0.12 ± 0.14 0.17 ± 0.16 0.22 ± 0.16 0.29 ± 0.22 0.32 ± 0.21 0.42 ± 0.31 0.51 ± 0.42 +0.39 325.0% +0.006*

Denotes significant slopes at the alpha = 0.05 level.

MSF (Hz) M ± S.D. 73.3 ± 29.7 89.1 ± 21.0 98.3 ± 23.2 100.9 ± 19.3 98.6 ± 27.0 111.9 ± 11.0 109.3 ± 12.1 +36.0 49.1% +0.55*

MSD (ms) M ± S.D. 9.07 ± 3.04 8.03 ± 1.82 7.63 ± 1.16 7.49 ± 1.06 7.37 ± 0.86 7.09 ± 0.79 7.28 ± 0.80 −1.79 19.7% −0.03*

MSS (mV/ms) M ± S.D. 0.03 ± 0.03 0.05 ± 0.04 0.07 ± 0.06 0.09 ± 0.07 0.10 ± 0.07 0.14 ± 0.10 0.17 ± 0.14 +0.14 466.7% +0.002*

MNPPS M ± S.D. 0.90 ± 0.17 0.96 ± 0.17 1.01 ± 0.14 1.03 ± 0.15 1.03 ± 0.11 1.05 ± 0.10 1.07 ± 0.12 +0.17 18.9% +0.003*

K.M. Calder et al. / Journal of Neuroscience Methods 178 (2009) 148–156

Fig. 2. (a) Mean spike amplitude (MSA) across the different ages for the control subjects (solid black circles), at-risk subjects (open boxes with a cross inside), and NSAP subjects (open triangles). (b) Mean spike slope (MSS) across the different ages for the control subjects (solid black circles), at-risk subjects (open boxes with a cross inside), and NSAP subjects (open triangles). Regression lines for the groups are; control group, thick black line; at-risk group, thick dashed line; and NSAP group, thin dashed line. The slopes for groups marked with * are significantly different from those of groups marked with ** (p < 0.05).

respectively) and individuals with NSAP (0.006 mV/%MVC and 0.002 mV/ms/%MVC, respectively) (p < 0.05). No significant differences were observed between the at-risk group and individuals with NSAP for the MSA and MSS features. The NSAP group had a significantly higher slope than the at-risk and control subjects for MNPPS (0.003, 0.002, and 0.0001 peaks per spike per %MVC, respectively, p < 0.05). For MNPPS the at-risk subjects had a significantly higher slope than the control subjects (p < 0.05). For MSD, the at-risk and NSAP groups had significantly lower slopes across increasing contraction levels than the control subjects (−0.020, −0.027, and −0.004 ms per %MVC, respectively; p < 0.05). No significant group by contraction level interaction was found for MSF, and MSF was found to have a significant group main effect (p < 0.05). The post hoc analysis revealed that the MSF of the at-risk group was significantly higher (108.02 ± 21.62 Hz) than the control (96.99 ± 17.88 Hz) and NSAP (97.34 ± 24.13 Hz) groups (p < 0.05). 4. Discussion The overall goal of the current work was to determine if a new SEMG methodology could reveal differences in the ECRB muscle activity in individuals with NSAP relative to healthy control subjects deemed at-risk of developing NSAP, and healthy control subjects deemed not at-risk of developing NSAP. We found that, similar to our previous decomposition-based quantitative electromyography findings in the same subjects (Calder et al., 2008), data collected

153

Fig. 3. (a) Mean spike amplitude (MSA) across the different contraction levels (%MVC) plotted for control, at-risk, and NSAP subjects. (b) Mean spike slope (MSS) across the different contraction levels plotted for control, at-risk and NSAP subjects groups. The slopes for groups marked with * are significantly different from those of groups marked with ** (p < 0.05).

from low-level static contractions in the NSAP group showed evidence of myogenic changes. Specifically, at 10% MVC the NSAP group had lower MSS and MNPPS values compared to the control group. This novel technique also allowed us to investigate differences in muscle recruitment across increasing contraction levels, where again we found evidence of myogenic changes in the NSAP group, specifically that subjects with NSAP had lower slopes for MSA, MSS, and MSD across contraction levels compared to the control subjects, and a higher slope for MNPPS compared to the at-risk and control subjects. In this study, we focused on data recorded from the ECRB muscle. This muscle originates from the common extensor tendon, and has been implicated in LE, a condition that might be associated with NSAP. As such, our previous study on NSAP had a fourth comparison group, those with LE, since we wanted to determine whether NSAP and LE might present similarly in terms of any neuromuscular changes noted. We showed in that study that, in fact, subjects with LE had larger MUPs which suggested a neurogenic process which was distinct from the findings in our NSAP group. In order to enable a comparison between the current study and the previous study (Calder et al., 2008) we located our surface electrodes in this study over the ECRB muscle through palpation while subjects performed resisted third digit extension. However, since the extensor carpi radialis longus (ECRL) muscle runs very close to the ECRB muscle, we cannot guarantee that our SEMG recordings in the current study do not contain crosstalk from the ECRL. That said, our findings in the current study are consistent with those found in our previous study on the same sample

154

K.M. Calder et al. / Journal of Neuroscience Methods 178 (2009) 148–156

for MSA and MSS, which suggest that age caused different effects in the MSA and MSS of different groups. The effects of increased age in the at-risk and NSAP group were lower mean MSA and MSS values compared to the control group (Fig. 2a and b). MU loss is a gradual process, where significant declines occur only after the six to seventh decade of life (Campbell et al., 1973; Doherty, 2003), and the subjects in the at-risk and NSAP groups in the current study were on average in the fifth decade of life, therefore the effects of MU loss was not expected to affect our results. Our findings of lower MSA and MSS slopes across contraction levels in the NSAP group relative to the control group, since the control group was significantly younger, are strongly suggestive that NSAP causes myogenic changes as opposed to MU loss (i.e. neuropathic changes) in the muscles in individuals with NSAP. 4.1. Spike shape differences among the groups during low-level contractions

Fig. 4. (a) Mean spike duration (MSD) across the different contraction levels (%MVC) plotted for control, at-risk, and NSAP subjects. (b) Mean number of peaks per spike (MNPPS) across the different contraction levels plotted for control, at-risk and NSAP subjects groups. The slopes of groups marked with * are significantly different from those of groups marked with ** (p < 0.05). The slopes of groups marked with + are significantly different from those of groups marked with ++ (p < 0.05).

where we had a needle electrode inserted directly in the ECRB muscle. An important factor to consider when interpreting the results of this study is the age difference among our sample groups. The control group was significantly younger than the at-risk and NSAP group. This is related to the main criteria for fitting into the asymptomatic control group, where individuals were to be healthy and not be performing regular repetitive wrist activities in their occupation or leisure time, which resulted in our control sample being primarily composed of undergraduate or graduate students who did not yet have a full-time occupation. With normal aging, muscle atrophy occurs as a result of the total number of fibers within a given muscle being reduced, which is observed to a significant degree around the age of 60 years (McComas et al., 1993). Our NSAP and at-risk subjects were close to 60 years of age (50 ± 9 and 45 ± 13, respectively); therefore the effects of aging on their ECRB muscles must be considered. As such, we included age as a covariate in the analyses and found that, for the most part, age did not have a significant effect on our outcomes. The backward stepwise ANCOVA model revealed no significant three-way interaction between group, contraction level and age so we could conclude that age did not cause any differences in slope values across contraction level. No significant two-way interactions were found between age and contraction level for any of the spike shape parameters, suggesting that age did not have any significant effect on the relationship between the outcome measure and contraction level for any of the spike shape features. We did however find significant group by age interactions

During a 10% MVC of the ECRB muscle, the NSAP group was found to have significantly lower MSS and MNPPS values than the control subjects. These findings are consistent with the findings of our previous needle EMG investigation using DQEMG; where MUPs from the NSAP group were smaller when compared to the control subjects (Calder et al., 2008). The results would have been stronger if MSA had also been significantly lower in the NSAP group, however this was not the case in the present study. Chronic pain conditions like NSAP and trapezius myalgia are believed to be associated with preferential damage to slow-twitch muscle fibers (Larsson et al., 1988, 1990, 1992), which would be evident by reduced MUP size observed in EMG recordings at low-level contractions (Calder et al., 2008). Patients with chronic myalgia related to static muscle loading or repetitive work have been the focus of several studies where biopsies of the descending portion of the trapezius muscle showed an increase in the number of type I fibers in the muscles of myalgic subjects and that some of the type I muscle fibers had signs of mitochondrial damage (‘ragged red’ fibers) (Larsson et al., 1988, 1990, 1992). Very small numbers of damaged fibers were identified in the small samples involved in these studies, so the functional impact of such biopsy results is not clear. Interestingly Dennet and Fry (Dennett and Fry, 1988) took specimens from the unaffected first dorsal interosseous muscle in patients with chronic overuse syndromes of that limb and found an increase in the number of type I fibers and mitochondrial changes consistent with the findings at the trapezius (Larsson et al., 1988, 1992). This suggests that muscle changes such as those seen in the current study may not necessarily be pathologic in nature, but may reflect training effects within muscles that perform lowlevel activity for long periods of time. To account for this, in the current study we included a group of subjects deemed at-risk for developing NSAP based on their performance of work activities that have been shown to lead to NSAP or other repetitive strain injuries. The lower MSS and MNPPS found among patients with NSAP in our study might be related to habitual use (i.e. a training effect) since the at-risk subjects, without having any pathology, were not significantly different than the NSAP group. That said, the data from the at-risk subjects and NSAP subjects with increasing contraction levels do not show similar trends across all parameters, and as such there might be some pathology in the NSAP group that is not simply related to a training effect. The lower MSS and MNPPS values found in the NSAP group compared to the control subjects may be a reflection of damage to the slow-twitch muscle fibers, whereby; (i) the slow-twitch muscle fibers that are recruited early in a contraction are damaged, and therefore have lower twitch properties, and (ii) the reduced twitch properties result in more synchronized behaviour of the muscle fibers, resulting in lower MNPPS.

K.M. Calder et al. / Journal of Neuroscience Methods 178 (2009) 148–156

4.2. Changes in spike shape parameters with increasing force production The two basic mechanisms responsible for the increases in force production during voluntary muscular contraction are MU recruitment and increases in the discharge rates of already active MUs (Adrian and Bronk, 1928, 1929). The changes in spike shape measures with increases in muscle force output in all three groups are consistent with this phenomenon (Table 1). That is, there was an increase in MSA, MSF, MSS, and MNPPS while MSD decreased. Previous work by Kukulka and Clamann (1981) identified MU recruitment as the main component causing increases in force production from 0 to 88% of MVC in the bicep brachii (BB) muscle, whereas in the adductor pollicis (ADP) muscle, no new MUs were observed to be recruited at forces greater than 50% MVC. Their findings suggested that rate coding plays a more prominent role in later force modulation in the ADP, while recruitment plays a more important role throughout most of the contractile force range in BB (Kukulka and Clamann, 1981). Gabriel et al. (2007) recently reported predictable changes in SEMG spike shape in the BB muscle with increasing isometric force, whereby, similar to Kukulka and Clamann (1981), the increase in force production was shown to be accomplished primarily through MU recruitment. The ECRB muscle is a long, superficial forearm muscle, and is normally comprised of approximately 65% fast-twitch MUs and of approximately 35% slow-twitch MUs (Romaiguere et al., 1989). This distribution of fast and slow-twitch MUs in the ECRB is similar to that of the BB muscle (Le Bozec and Maton, 1987), therefore we expected the main observation with increases in force production would be spike shape parameters reflective of recruitment, that is increases in MSA, MSF, MSS, MNPPS and a decrease in MSD, which was observed in all three groups. 4.3. Differences in spike shape changes among the groups with increasing contraction level Electrodiagnostic testing includes qualitative evaluation of IP complexity (Fuglsang-Frederiksen, 2006). As the intensity of the muscle contraction gradually increases, patients with myopathic disorders will exhibit a more fully developed IP at lower force levels than normal. This does not necessarily mean that EMG amplitude will be greater. Rather, muscle fiber loss requires that more MUs be recruited earlier during the force gradation process. Each MUP is actually smaller but there are more active MUs firing at higher rates at lower force levels. The larger number of MUs firing at lower force levels results in greater asynchronous activity. The superposition of a larger number of smaller, asynchronous MUPs results in a more complex IP, and this can be quantified by the MNPPS (Beardwell, 1967; Magora and Gonen, 1970; Fusfeld, 1971). Even though the EMG magnitudes were smaller for the NSAP group, the IP exhibited a greater rate of increase in MSF and in the MNPPS across force levels when compared to the other two groups, adding further support to our earlier indwelling findings that NSAP is myogenic in nature (Calder et al., 2008). The main difference between the groups for MNPPS and MSD appeared to occur during the lower levels of contractions (Fig. 4a and b), whereas things normalize at higher contraction levels. It must be noted, however that the at-risk subjects exhibited the same behaviour as the NSAP group, albeit not to the same extent. This suggests that it may be the low-threshold fibers that differ between the groups, supporting Larsson (Larsson et al., 1988, 1990) and Larsson’s (Larsson et al., 1992) work, where damage occurs to the low-threshold MUs (ragged-red fibers). Increases in the frequency content of the EMG signal have been linked with the recruitment of high threshold MUs (Broman et al., 1985; Moritani and Muro, 1987; Solomonow et al., 1990; Kupa et al., 1995; Sbriccoli et al., 2003). In the current study, MSF significantly

155

increased from 10 to 70% of MVC, through observation a steady increase occurred from 10 to 50% MVC and then plateaued between 50 and 70% of MVC in the two asymptomatic groups, whereas in the NSAP group, MSF steadily increased to 60% of MVC and then drop slightly at 70% of MVC. This difference might reflect the necessity of recruiting additional MUs until 60% of MVC because there are fewer fibers per motor unit, a phenomenon that would be consistent with myopathic disorders. Future research investigating fiber type composition and myopathic changes through histological studies in this patient population would provide further insight into the pathophysiology of NSAP. 4.4. Summary and conclusions SEMG spike shape is a non-invasive way to examine motor unit properties at low levels of contractions and has the additional advantage over traditional clinical electromyography approaches in that muscle activity can be studied across a broader range of contraction levels. In the current study, utilizing the SEMG to examine spike shape parameters during a low-level contraction and across contraction levels of increasing intensity revealed significant differences between ECRB muscles in an asymptomatic control group, an at-risk group, and a group with NSAP. The spike shape analysis results from the low-level contraction data in the current study concur with our previous findings on the same sample using MUP morphological analysis from quantitative electromyography techniques (Calder et al., 2008). Both studies suggest that individuals who suffer from NSAP have either myopathic changes in their affected muscles or have changes in motor unit properties related to habitual use. The results from spike shape analysis would have been stronger if MSA demonstrated a significant group difference at the 10% MVC level. With increasing contraction levels, the spike shape analysis approach allowed us to note differences in rate coding and recruitment between the study groups, and again these findings were consistent with myogenic changes in the NSAP group. Since our group of individuals at-risk of developing NSAP showed similar trends in spike shape features to those seen in the NSAP group in many but not all variables, there may be a training effect occurring concurrently with myopathic changes in individuals with NSAP. This work provides evidence that researchers and clinicians may be able to use non-invasive techniques such as spike shape analysis to evaluate clinical conditions. Future research should examine the usefulness of SEMG spike shape analysis in patients with neuropathic and myopathic conditions where there is clear quantitative electromyographic evidence of MU changes. Acknowledgements This research was funded through a Canadian Graduate Scholarship from the Natural Sciences and Engineering Research Council of Canada (K.M. Calder) and by the Workplace Safety and Insurance Board of Ontario. References Adrian ED, Bronk DW. The discharge of impulses in motor nerve fibres. Part I. Impulses in single fibres of the phrenic nerve. J Physiol 1928;66:81–101. Adrian ED, Bronk DW. The discharge of impulses in motor nerve fibres. Part II. The frequency of discharge in reflex and voluntary contractions. J Physiol 1929;67:i3–151. Beach J, Gorniak GC, Gans C. A method for quantifying electromyograms. J Biomech 1982;15:611–7. Beardwell A. The spatial organization of motor units and the origin of different types of potential. Ann Phys Med 1967;9:139–57. Broman H, Bilotto G, De Luca CJ. Myoelectric signal conduction velocity and spectral parameters: influence of force and time. J Appl Physiol 1985;58:1428–37.

156

K.M. Calder et al. / Journal of Neuroscience Methods 178 (2009) 148–156

Calder KM, Stashuk DW, McLean L. Motor unit potential morphology differences in individuals with non-specific arm pain and lateral epicondylitis. J Neuroeng Rehab 2008;5:34. Campbell MJ, McComas AJ, Petito F. Physiological changes in ageing muscles. J Neurol Neurosurg Psychiatry 1973;36:174–82. Dennett X, Fry HJ. Overuse syndrome: a muscle biopsy study. Lancet 1988;331:905–8. Doherty TJ. Invited review: aging and sarcopenia. J Appl Physiol 2003;95:1717–27. Erminio F, Buchthal F, Rosenfalck P. Motor unit territory and muscle fiber concentration in paresis due to peripheral nerve injury and anterior horn cell involvement. Neurology 1959;9:657–71. Fuglevand AJ, Winter DA, Patla AE. Models of recruitment and rate coding organization in motor-unit pools. J Neurophysiol 1993;70:2470–88. Fuglsang-Frederiksen A. The role of different EMG methods in evaluating myopathy. Clin Neurophysiol 2006;117:1173–89. Fusfeld RD. Analysis of electromyographic signals by measurement of wave duration. Electroencephalogr Clin Neurophysiol 1971;30:337–44. Gabriel DA. Reliability of SEMG spike parameters during concentric contractions. Electromyogr Clin Neurophysiol 2000;40:423–30. Gabriel DA, Lester SM, Lenhardt SA, Cambridge ED. Analysis of surface EMG spike shape across different levels of isometric force. J Neurosci Methods 2007;159:146–52. Harrington JM, Carter JT, Birrell L, Gompertz D. Surveillance case definitions for work related upper limb pain syndromes. Occup Environ Med 1998;55:264–71. Hirose K, Sobue I. Quantitative electromyography—a method by computer analysis. Electromyogr Clin Neurophysiol 1972;12:421–9. Komi PV, Vitasalo JH. Signal characteristics of EMG at different levels of muscle tension. Acta Physiol Scand 1976;96:267–76. Kukulka CG, Clamann HP. Comparison of the recruitment and discharge properties of motor units in human brachial biceps and adductor pollicis during isometric contractions. Brain Res 1981;219:45–55. Kupa EJ, Roy SH, Kandarian SC, De Luca CJ. Effects of muscle fiber type and size on EMG median frequency and conduction velocity. J Appl Physiol 1995;79:23–32. Larsson SE, Bengtsson A, Bodegard L, Henriksson KG, Larsson J. Muscle changes in work-related chronic myalgia. Acta Orthop Scand 1988;59:552–6. Larsson SE, Bodegard L, Henriksson KG, Oberg PA. Chronic trapezius myalgia. Morphology and blood flow studied in 17 patients. Acta Orthop Scand 1990;61:394–8. Larsson B, Libelius R, Ohlsson K. Trapezius muscle changes unrelated to static work load. Chemical and morphologic controlled studies of 22 women with and without neck pain. Acta Orthop Scand 1992;63:203–6.

Latash ML. Spectral analysis of the electromyogram (EMG) in spinal cord trauma patients. I. Different types of the EMG and corresponding spectra. Electromyogr Clin Neurophysiol 1988a;28:319–27. Latash ML. Spectral analysis of the electromyogram (EMG) in spinal cord trauma patients. II. Motor unit and interference EMG power spectra. Electromyogr Clin Neurophysiol 1988b;28:329–34. Le Bozec S, Maton B. Differences between motor unit firing rate, twitch characteristics and fibre type composition in an agonistic muscle group in man. Eur J Appl Physiol Occup Physiol 1987;56:350–5. Magora A, Gonen B. Computer analysis of the shape of spike from electromyographic interference pattern. Electromyography 1970;10:261–71. McComas AJ, Sica RE, Campbell MJ, Upton AR. Functional compensation in partially denervated muscles. J Neurol Neurosurg Psychiatry 1971;34:453–60. McComas AJ, Galea V, de Bruin H. Motor unit populations in healthy and diseased muscles. Phys Ther 1993;73:868–77. Moritani T, Muro M. Motor unit activity and surface electromyogram power spectrum during increasing force of contraction. Eur J Appl Physiol Occup Physiol 1987;56:260–5. Muro M, Nagata A, Murakami K, Moritani T. Surface EMG power spectral analysis of neuromuscular disorders during isometric and isotonic contractions. Am J Phys Med 1982;61:244–54. Pierrot-Deseilligny E, Mazevet D. The monosynaptic reflex: a tool to investigate motor control in humans. Interest and limits. Neurophysiol Clin 2000;30:67–80. Romaiguere P, Vedel JP, Pagni S, Zenatti A. Physiological properties of the motor units of the wrist extensor muscles in man. Exp Brain Res 1989;78:51–61. Sbriccoli P, Bazzucchi I, Rosponi A, Bernardi M, De Vito G, Felici F. Amplitude and spectral characteristics of biceps Brachii sEMG depend upon speed of isometric force generation. J Electromyogr Kinesiol 2003;13:139–47. Solomonow M, Baten C, Smit J, Baratta R, Hermens H, D’Ambrosia R, et al. Electromyogram power spectra frequencies associated with motor unit recruitment strategies. J Appl Physiol 1990;68:1177–85. Stalberg E. Electrophysiological studies of reinnervation in ALS. Adv Neurol 1982;36:47–59. Viitasalo JH, Komi PV. Signal characteristics of EMG with special reference to reproducibility of measurements. Acta Physiol Scand 1975;93:531–9. Zar JH. Biostatistical analysis. 2nd ed New Jersey: Prentice-Hall: Englewoods Cliffs; 1984. Zhou P, Rymer WZ. Can standard surface EMG processing parameters be used to estimate motor unit global firing rate? J Neural Eng 2004a;1:99–110. Zhou P, Rymer WZ. An evaluation of the utility and limitations of counting motor unit action potentials in the surface electromyogram. J Neural Eng 2004b;1:238–45.