- Email: [email protected]
Measuring myoelectric fatigue of the serratus anterior in healthy subjects and patients with long thoracic nerve palsy
~
ELSEVIER
~~
Journal of Ort hopaedic Research
Journal of Orthopaedic Research 22 (2004) 872-877
www.elsevier.com/locate/ort hres
Measuring myoelectric fatigue of the serratus anterior in healthy subjects and patients with long thoracic nerve palsy
Abstract Proper function of serratus anterior plays a vital role in thc movement and stability of thc scapula, and thus of the glenohumeral joint and the upper limb. The unique anatomy of this muscle makes direct measurements of its fatigue properties impossible. Here we describe for the first time indirect measurements of the myoelectric manifestations of Fatigue in the serratus anterior. Eight healthy volunteers (29-35 years) were tested. four of them on two different occasions, using two exercise protocols (60 s isometric maximum upward force in 120" arm flexion, and 60 s maximum forward force at 90" arm flexion) with simultaneous recording by surface and wire electrodes applied according to established methods. Signals were analysed to obtain the rate of fall of median EMG frequency and the rate of rise of amplitude. Both exercise protocols gave similar results. Frequency-slope measurements (mean rate of Fall 0.6 k O.I'XI initial value per second ('XI s I ) with both surface and wire electrodes) were more precise than those of amplitude (mean rate of rise 2.6 f 0.3% s with surface electrodes, only 1.3 f 0.2% s with wire electrodes). Surface electrodes gave much lower variation than fine wires, the coellicient of variation of slopes for surface electrodes being approximately 20-40'X both between studies in a single subject and between subjects. In 5 patients (aged 22-59 years) recovering from long thoracic nerve palsy studied using surface electrodes the frequency slopes was normal (0.6 k O.I'XI S F ' ) , while the amplitude slope was reduced (0.9 f 0.4% s ', P = 0.01). This shows abnormal fatigue properties of the reinnervated muscle and a dissociation between the frequency and amplitude inanifcstations of fatigue. Crown Copyright 0 2003 Published by Elseiver Ltd. on behalf of Orthopaedic Research Society. All rights reserved.
'
.
ii.ort/~v:EMG:
Isometric contraction: Long thoracic nerve palsy: Median frequency: Muscle fatigue: Serratus anterior
Introduction
Thc movcmcnt and stability of the scapula, and thus of the glenohumeral joint and the upper limb, depend heavily on the scapulothoracic muscles [ 1,161. The role of trapezius and serratus anterior in scapular elevation and upward rotation cannot be assumed by other muscles [8]. The effect of serratus anterior dysfunction is dramatically seen in the scapular winging of long thoracic nerve palsy [l]. Serratus anterior is a broad, flat muscle formed by multiple digitations arising from the upper 8-9 ribs in the midaxillary line and attaching to the vertebral aspect of the ventral surface of the scapula. Three functional components have been identified: a superior cylindrical mass that attaches to the superio-
~~
'
~
*Correspondence author. Tel.: +44-151-7O6-4121: 1.a~: +44- I5 1706-58 15. E-rmri/ rrt/dw\.c.- ~cyinan_ebied(~hotmaii.coiii (A.M. Ebied).
medial border of the scapula and anchors the arm during overhead rotation; a long wide intermediate band that connects the 3rd-5th ribs with the vertebral border of the scapula and draws the scapula forward; and the lower five slips from the 6th-10th ribs that converge on the lower angle of the scapula to rotate its inferior angle upward and laterally [8]. Scapular winging secondary to LTNP demonstrated by asymmetrical rotation of the inferior medial border of the scapula is not uncommon and potentially disabling. Non-operative measures may produce satisfactory recovery [30], otherwise operative intervention is usually warranted [6]. Many patients continue to suffer residual winging and fatigue especially with overhead and endurance activities. However, to our knowledge there have been no measurerncnts of the fatigue properties of this anatomically unique muscle. The obvious way to quantify fatigue is as the timedependent decrease in the ability of a muscle to maintain a specified force or joint torque [29]. This is not feasible for muscles that contract in more than one movement,
0736-0266/S - see rront matter Crown Copyright 0 2003 Published by Elseiver Ltd. on behalf of Orthopaedic Research Society. All rights reserved. doi: 10.1016/i.orthres.2003.12.004
A.M. Ehiecl et ul. I Joirrriol of Orrlropaetlic~Rewrrch 22 (2004) 872 877
on more than one joint, or (as in the case of serratus anterior) whose contribution cannot be separated from that of other muscles. Indirect methods of studying myoelectric fatigue depend on analysing the electromyogram (EMG). Power spectral analysis of EMG signals has been used as a predictor and indicator of muscle fatigue during isometric contractions of many muscles [18,20]. Myoelectric signals have been shown to shift during fatiguing exercises with decline in both median M D F and mean M P F frequency. The decline in both parameters has been correlated to a reduction in the muscle fibre conduction velocity, among other things. The conduction velocity can be influenced by accumulation of metabolites at the sarcolemma and the type of muscle fibres [26]. The second EMG parameter commonly used in fatigue studies is the EMG amplitude, calculated as the integrated EMG or root mean square. EMG amplitude is commonly found to increase during exercise of different muscles while the force remains steady or declines. This increase in EMG amplitude has been attributed to increased frequency of motor unit discharge or synchronisation of motor unit firing rate [33]. Other studies showed an initial increase in amplitude followed by plateau curve [4]. EMG variables have been found to have a good level of repeatability, providing that care is taken in placing electrodes and in the setup of the test [19,25]. Here we describe for the first time measurements of the fatigue properties of serratus anterior using EMG methods. EMG signals can be acquired using surface and finewire electrodes: the latter have advantages of anatomical specificity (low 'cross-talk') and access to deep muscles, while the former are better tolerated and give more reproducible results [28]. Both have been employed in EMG fatigue studies (the former more rarely) [31,34]. For serratus anterior muscle, there is no consensus on the most appropriate electrode technique, and the only reports have concentrated on the pattern of muscle activation rather than fatigability [10,22]. Here we used both methods. The serratus anterior is involved in a complex way in a wide range of arm movements. Classically it is loaded by pushing forward with the arm in 90" forward flexion. Also, patients with serratus anterior deficiency usually find overhead activities difficult due to sensation of fatigue and scapular instability [32] that may be related to an intrinsic problem of the muscle itself or secondary to inhibition of the inferior part of the trapezius and rotator cuff muscles. We therefore tested the serratus anterior muscle in both movements.
Material and methods We carried out surface and wire electrode studies on 8 male healthy volunteers aged 29-35 years (the right side was tested in 7 and the left in 1). of whom 4 were studied twice to define repeatability. We also
873
carried out surface electrode studies in 5 patients aged 22 59 years with long thoracic nerve palsy presented with the clinical picture of dynamic winging of the inferior medial border of the scapula and confirmed by EMG and nerve conduction studies: 4 had been treated ( 2 12 months before study) by surgical decompression of the long thoracic nerve [ 171 after failure t o improve with 2 12 months physiotherapy. All patients had achieved some recovery with improvement of their ability to elevate their arms and scapular rotation and sliding along the chest wall. Using Kibler's test [14]. side-to-side difference was < I cm at the 3 different positions. However. this group of patients did not regain their full function and continued to complain of easy fatigue compared to the unaffected shoulders particularly with lengthy overhead or strenuous activities. This difference could be attributed to the altered activation pattern of the other scapular stabilisers and disadvantageous performance of the rotator cuff or to deficiency in reinnervated serratus anterior muscle itself. We hypothesised that in this group of patients reinnervated serratus anterior may be more fatigable and E M G manifestations of fatigue would be greater than normal. Fully informed consent was obtained. and the study was approved by the local ethics committee.
Muscle-testing protocols Two isometric exercise tasks were performed. The first task consisted of 60 s 'upward pushing' at 120" of forward flexion, using a KIN-COM machine (Chattanoga. Inc.. Oxfordshire. UK). With the subject sitting, a cuff round the arm 10 cm distal to the tip of the acromion was linked to a lever with centre of rotation at the level of the humeral head, and the arm in 120" forward flexion from neutral. Maximum voluntary force was determined as the best of three 3 s upward pushes separated by 5 s relaxation. Subjects were then instructed to reproduce this force for 60 s, given visual force feedback. The second task consisted of 60 s maximum 'forward pushing' against a wall with the arm in 90" forward elevation (the classic test for examining the serratus anterior clinically and demonstrating scapular winging). As the recorded torque measures not the force generated by the muscle, but rather the indirect effects of the overall loading of the scapular stabilisers. it was used only to give feedback about the level of performance. As a side observation in this study patients and healthy volunteers have felt fatigue around their shoulders and over the serratus anterior area by the end of their exercise protocol. Indeed the group with abnormal sei-ratus anterior function demonstrated scapular winging with lifting o f the infero-medial angle of their scapula during the test, in addition to feeling of discomfort and cramp like pain over the serratus anterior muscle.
E M G ucyuisition untl trntr1~~si.s E M G data were acquired simultaneously using one set of surface electrodes and two double wire electrodes. Surface electrodes were applied as described by Basmajian [2]:with the arm raised. two electrodes were centred in a vertical oval (around 4 cm long) above the inferior angle of scapula and along the midaxillary line. One wire electrode was inserted as described by Delagi [7]: with the subject prone, arm dangling over the edge of a bed, the wire was inserted just lateral to inferior angle of scapula. A second wire electrode was inserted as described by Goodgold [9]: with the subject prone, the vertebral border of the scapula was identified and lifted manually, and the wire inserted between the vertebral border of the scapula and chest wall, the tip lying close to the thoracic surface of the inferior angle of scapula. A note was taken of the sites of electrode insertion, and an attempt was made in the second test to re-insert the electrodes in same sites. The disposable 50 p n fine wire electrodes, based on the design of Basmajian and Stecko [3], are supplied ready-loaded in a 25 Ga. hypodermic needle (Chalgren Enterprises. Inc. Gilroy California). They are of nickel-chromium alloy. insulated with nylon and sterilised by electron beam: 2 mm of metal are exposed at the tips, the hooked ends being staggered. Surface electrodes were Ag-AgCl with an 8 mm contact area (BIOPAC Systems). The concavity was filled with conductive gel. Disposable Ag AgCl electrodes with I0 mni contact areas were used over the tip of the olecranon as a ground. The electrodes were connected to three channels of a 100 A E M G (BIOPAC Systems, Inc., Santa Barbara. California). Differential input amplifiers were used with a common mode rejection ratio of 100 dB minimum, input
874
A . M . Ehird
('1 rtl.
I Journul
of' Orthoparrlic R C S ~ W22~ I(2004) 872-877
impedance 10 R, bandwidth 104000 Hz and gain 2000. Data were acquired with a Biopac system M P 100 data acquisition unit (BIOPAC systems, Inc. Santa Barbara, CA) with A/D resolution of 16 bits. Signals were sampled at 2500 Hz, and bandpass filtered at 10- 1000 Hz for fine wires and 10-500 Hz for surface electrodes. Data were analysed using Acqknowledge software (BIOBAC Systems, Inc.). dividing each 60 s task into 20 intervals (data assigned to the midpoint). Median frequency was obtained by fast Fourier transform. Data from the first 5 s. which show a rapid artifactual decrease, were omitted from the analysis. To measure amplitude the signals were smoothed (2000 saniple/window) then rectified (400 samplelwindow), then nornialised as a percentage of the mean value for the whole task. as previously described [21].
A
Surface electrodes: amplitude & frequency
(YOmean) __ Frequency (Hz)
..... Amplitude
._.
=.
..E.
......;.:.: . _. _ _ . _-..:.... .-...... .. ...
-
-
,-, ,
I
.
I
Crilculation.s & sfrrti.sric:c
0 Data were transferred to spreadsheets for calculation of rates of change ('slopes'), expressed relative to the starting values: for frequency this was done by linear regression; for amplitude. where the time course was markedly nonlinear, a two-point slope calculation was performed between I = 3-9 s and I = 42-60 s (these time-points being chosen empirically to yield maximum robustness and precision). Between-study variability was calculated for each subject's pair of separate measurements. expressed as a coefficient of variation (CV = SDI mean'%!). and the root mean square value taken for all subjects. Between-person CV was calculated using the means of the paired data for all subjects. correcting for the duplicate measurements by subtracting from the observed between-person CV' half the between-study CV'. Differences were compared using the paired or unpaired t-test. as appropriate. and P < 0.05 is taken as statistically significant.
6
In heulthy subjects Fig. 1 shows a sample of the individual data collected using a surface electrode (Fig. 1A) and a fine wire electrode (Fig. IB), showing the decreases in median fiequency and increase in amplitude with time. Fig. 2A summarises mean data on the relative changes in both median frequency and amplitude. With the surface electrode, changes in amplitude are larger but more variable, especially after about 30 s. The rate of fall in frequency was similar with surface and both wire electrodes: the mean rate of fall was 0.6 & 0.1'%1initial value per second (%I S K I ) with both electrodes. The rate of rise of amplitude with both fine wire electrodes was significantly smaller than that recorded by surface electrodes (1.3 f0.2'%1s-' vs 2.6&0.3'%1s-I, P = 0.02), and more variable. With the wire electrodes, frequency measurements were less variable than amplitude, but still more variable than with surface electrodes. Fig. 2C shows that the use of surface electrode gives acceptable agreement between measurements for both amplitude and frequency slopes, and Fig. 2D shows that both slopes are similar for the two exercise protocols. For the surface electrodes, overall mean between-study CV was 29% for frequency slope and 23'Yn also for amplitude slope (both exercise protocols); for comparison, the corresponding overall figures for the wire methods were very poor, 100'%,and SO'%). The open symbols in Fig. 2E summarise the slopes estimated by the surface and wire electrode methods; it
20
30
40
50
60
Goodgold wire electrodes: frequency (Hz)
! : : :
Results
10
50 00
10
20
30
40
50
60
Time (s) Fig. 1. Some individual data from single trials as a function of time (s) in healthy subjects 'forward pushing' (1 line= I measurement on I subject). (A) Median frequency and amplitude acquired using the surface electrode. (B) Median frequency data acquired with fine wires inserted using the Goodgold technique (similar to data acquired tising the Delagi method): amplitude data (not shown) are more variable.
shows how the frequency slopes are similar but the amplitude slope smaller in wire electrodes. For the surface electrodes, the estimated true between-person CV was 33Yn for frequency slope and 42% for amplitude slope; the corresponding figures for the wire methods were 45% and 62% In patients with long thorucic nerve palsy Fig. 2B show the mean response in the group of patients in a similar way to the control data in Fig. 2A. The median frequency falls in a similar way to controls, while the amplitude response is very different, showing a reduced and delayed rise: This can be seen from the mean slope data shown as the filled symbol in Fig. 2E: the amplitude slope is lower than in controls (0.9 t 0.4% s-', P = 0.01), the frequency slope not significantly different (0.6 k O.I%I s-I).
Discussion
The protocol This is the first report of fatigue measurements of the serratus anterior, a muscle which combines an essential role in the movements of the upper limb with an
875 A
Iielative change in amplitude & frequency
C
Slope in second test
0
0Amplitude .Frequency
i
0.4
0
@
2-
Slope in first test 0
1''
0 D
I-
0.0
-0.4
0
8
4-
0.8
4
2
6
'Forward' slope
-t- Surface: amplitude
J +All +All
wires: amplitude wires: frequency
tSurface: frequency
-0.8
20
Time (s)
40
0
. - -..
0
60
...... Controls +Amplitude (surface): patients &Frequency: patients
!, ,
E
6-
_ _.. -. . _.._ _
0
Time (s) 40
60
6
4
+
...
0 ' 20
2
Frequency slope (fall)
4 0.0 -0.3 -0.6
'Upward' slope
0
.-
I
0 Controls: wire 0 Controls: wire
0Controls: surface Patients: surface
Amplitude slope (rise) 1
2
3
Fig. 2. Amplitude and frequency changes during exercise in healthy subjects and controls. (A) Shows the time course of the relative change in amplitude and frequency measured by wire and surfice methods (see key) during the contraction. For concision the results of forward and upward pushing exercises. the results of the first and second tests. and the results of the two wire electrodes are combined. ( B ) Shows the corresponding data from patients obtained using surface electrodes. with the control mean data (seen in (A)) shown as dashed lines for comparison. (C) and (D) Shows the consistency and reproducibility of surface-electrode: (C) shows mean frequency and amplitude slopes (see key) in the second test as a function of those in the first (results from forward and upward pushing are combined): (D) shows mean frequency and amplitude slopes (see key) in forward pushing as a function of those in upward pushing (results from first and second tests combined). (E) Shows the mean amplitude slope ('% s - ' ) as a function of mean frequency slope ( % s ' ) for control subjects studied using surface electrodes and wire electrodes, and patients studied by surface electrodes (results combined from both protocols, and from both wire electrodes). The amplitude slopes direr between patients and controls ( P = 0.01) and wire and surface electrodes ( P = 0.02). but the frequency slopes do not. Errors bars show SEM.
anatomy which makes direct force measurements impossible. We find, as expected, a measurable decrease in median frequency and an increase in amplitude, the former being the more precise measurement. The two exercise protocols (forward and upward pushing) give very similar results. In general surface electrodes are usually used to study superficial muscles, while wires are reserved for small, deep muscles (such as the rotator cuff muscles [21]) or when localised fibre or motor unit action potentials are being characterised. As in studies of other muscles [12,13], reproducibility was better with surface electrodes than with inserted wires. One factor in the variability with wire electrodes could be changing
position of the wires in relation to the active motor units (although the hooked wires we used are designed to engage in the muscle fibres). However, surface electrodes are liable to similar errors due to skin movements. A second factor is the relation between the wire electrodes and the active muscle bulk. Serratus anterior is composed of many digitations that combine before inserting on the anteromedial surface of the inferior angle of the scapula. The muscle is functionally divided into upper and lower parts, with different levels of activity in various tasks and arm positions. Both the Delagi and Goodgold techniques insert wires into the musculo-tendinous part of the muscle which may not always represent the overall
876
A . M . Ehicvl ri ul. I Jourtiul of Orlhopuc&
activity. Surface electrodes were inserted within an elliptical 4 cm area, perhaps giving a more representative measure of activity. Another factor might be a heterogeneous pattern of intramuscular distribution of muscle fibre types (muscle fibre regionalisation) [34]. Other reported reasons for the poor reproducibility of wire electrodes are pain (not a problem here), intramuscular bleeding and inhomogeneous structure of the musculature [ll]. Also, the difficulty in re-inserting wires exactly into the same place no doubt explains poor reproducibility between test days. A well-known possible problem with surface electrodes is cross-talk (noise from other muscles or cardiorespiratory activities). As this was negligible in the calf except in the presence of a thick layer of subcutaneous fat [28], there is no reason to think crosstalk was a problem in our study. As in other surface-electrode studies [ 15,251, there is considerable variation in absolute values of median frequency (see Fig. 1A). However, the absolute slopes correlated significantly with the starting values (not shown), so that the relative slope used here is the appropriate quantity, and shows only modest variability.
Reseurcli 22 (2004) 872-877
tigue, it is affected by various local reflexes originating from the muscle itself. The denervation-reinnervation process has been shown to influence the distribution of muscle fibre type, size of sprouting motor neurones and recovery of sensory receptors and afferents [24]. These changes may interact to change the muscle characteristics resulting in more fatigable muscles, as seen in rat medial gastrocnemius and human soleus [27]. However, the conversion in muscle characteristics was found to be correlated to the paralysis period as well as the severity of nerve injury. In our case it may be that the period of denervation or severity of nerve compression were not severe enough to lead to full conversion of the muscle fatigue characteristics despite the changes seen in EMG amplitude that may represent lack of muscle ability to recruit large or fast contracting motor units or the lack of fast contracting fatigable units in the reinnervated muscles. Although this study could not demonstrate definitive difference in fatigability of the serratus anterior in this group of patients with abnormal serratus anterior function we have established a method that can be expanded to study other groups of serratus anterior dysfunction in future experiments.
result.^ in patients with long tlzonnc‘ic nerve palsy
Having concluded that surface electrodes and ‘upward pushing’ were a valid method for studying the altered fatigue properties of this important but technically difficult muscle, we also studied 5 patients with scapula winging due to long thoracic nerve palsy (4 following partially effective treatment). We established first that these patients could perform the protocol. We wished to test the hypothesis that manifestations of fatigue would be increased. This was not so. The frequency slope was normal, while the amplitude slope was reduced by about half. This is unlikely to be explained by differences in the muscle loading (impossible to assess, because of the possible compensation by other muscles), because in biceps brachii the frequency slope, not the amplitude slope, increases linearly with isometric contraction force [25]. Why this contradiction? A well-described phenomenon in normal muscles is the order of recruitment of motor unites. Slowly contracting, fatigue resistant motor units with low rate of frequency discharge are usually recruited early during the exercises while fast contracting, strong units are usually reserved until the end, perhaps until the muscle starts to demonstrate some evidence of fatigue [23]. As contraction continues and fatigue develops, motor units start to decrease their firing frequency, but even a reduced firing rates can produce the same level of muscle activation, a phenomenon known as “muscle wisdom” [ 5 ] . This recruitment order is influenced by the characters of the motor units and the central drive or motoneurone impulse rate and frequency. Although it can be a central mechanism of fa-
Acknowledgements The authors wish to thank Mr Alfred Morris for his assistance in the early stages of this project. This research project was supported by a grant from Menoufiya University, Egypt. References Barnett ND, Mander M, Peacock JC. Bushby K, Gardner Medwin D, Johnson G R . Winging of the scapula: the underlying biomechanics and an orthotic solution. Proc Inst Mech Eng ( H ) 1995;209:215-23. Basmajian J. In: Desmedt J, editor. New concepts of the motor unit in neuromuscular disorders: electromyographic kinesiology. Basel: S Karger; 1973. Basmajian J, Stecko G . A new bipolar electrode for electromyography. Appl Physiol 1962;I7:849. Bazzy AR, Donnelly DF. Diaphragmatic Failure during loaded breathing: role of neuromuscular transmission. J Appl Physiol: Respir Environ Exercise Physiol 1993:74: 1679-83. Bigland Ritchie B, Cafarelli E, Vollestad NK. Fatigue of submaximal static contractions. Acta Physiol Scand Suppl 1986: 556: 1 3 7 4 . Connor PM, Yamaguchi K, Manifold SG, Pollock RG. Flatow EL, Bigliani LU. Split pectoralis major transfer for serratus anterior palsy. Clin Orthop 1997341:13442. Delagi E, Perotto A. Iazzetti J, Morrison D. Anatomic guide for the electromyographer. New York: Springfield & Illinois: 1975. Duralde XA. Evaluation and treatment of the winged scapula. J South Orthop Assoc 1995;4:38-52. Goodgold J . Anatomical correlates of clinical electromyography. Second ed. Baltimore: Williams and Wilkins; 1984.
[lo] Jobe FW. Moynes DR, Tibone JE, Perry J. An E M G analysis of the shoulder in pitching. A second report. Am J Sport Med 1984;12:218 20. [ I I ] Jonsson B. Bagge U. Displacement deformity and fracture of wire electrodes for electromyography. Electromyography l968:8:32947. [I21 Jonsson B, Komi P. In: Desmedt J. editor. New Concepts of the Motor Unit Neuromuscular Disorders Electromyographic Kinesiology. Basel: S Karger AG, National-Zeitung AG: 1973. [I31 Kadaba MP, Wootten ME. Gainey J. Cochran GV. Repeatability of phasic muscle activity: performance of surface and intramuscular wire electrodes in gait analysis. J Orthop Res 1985:3:350-9. [I41 Kibler WB. The role of the scapula in athletic shoulder function. Am J Sport Med 1998;26:325-37. [I51 Krivickas LS. Taylor A. Maniar RM. Mascha E. Reisman SS. Is spectral analysis of the surface electromyographic signal a clinically useful tool for evaluation of skeletal muscle Fatigue? J Clin Neurophysiol: Off Pub1 Am Electroencephalogr Soc l998;15:13845. [I61 Ludewig PM, Cook TM, Nawoczenski DA. Three-dimensional scapular orientation and muscle activity at selected positions of hiimeral elevation. J Orthop Sport Phys Ther 1996:24:5765. [I71 Manning P, Frostick S, Wallace W. Winging of the scapula, a fresh look at the long thoracic nerve. In: 9th Congress of the European Society for Surgery of the Shoulder and Elbow, Nottingham, 19th-21st September, 1996. [I81 Merletti R, Farina D. Gazzoni M, Schieroni MP. Effect of age on muscle functions investigated with surface electromyography. Muscle Nerve 2002;25:65-76. [I91 Merletti R, Fiorito A, LoConte LR, Cisari C. Repeatability of electrically evoked EMG signals in the human vastus medialis muscle. Muscle Nerve 1998;21:184 93. [20] Merletti R. Knaflitz M, De Luca C. Myoelectric manifestation of fatigue in voluntary and electrically elicited contractions. J Appl Physiol l99O:69: 1810-9. [21] Morris AD. Kemp GJ, Lees A. Frostick SP. A study of the reproducibility of three different normalisation methods in intra-
muscular dual fine wire electromyography of the shoulder, J Electromyogr Kinesiol 1998:8:3 17-22, [22] Nuber GW, Jobe FW, Perry J. Moynes DR, Antonelli D. Fine wire electromyography analysis of muscles of the shoulder during swimming. Am J Sport Med 1986;14:7-1 I . [23] Olson CB, Carpenter DO. Henneman E. Orderly recruitment of muscle action potentials. Arch Neurol 1968:19:591-7. [24] Rafuse VF, Gordon T. J Physiol 1998;509(Pt 3):909-26. [25] Rainoldi A, Galardi G, Maderna L. Comi G , Lo Conte L. Merletti R . Repeatability of surface E M G variables during voluntary isometric contractions of the biceps brachii muscle. J Electromyogr Kinesiol 1999:9:105-19. [26] Sadoyama T, Masuda T. Miyata H. Katsuta S. Fibre conduction velocity and fibre composition in human vastus lateralis. Eur J Appl Physiol Occup Physiol 1988;57:767-71. [27] Sheilds RK. Fatigability. relaxation properties, and electromyographic responses of the human paralyzed soleus muscle. J Neurophysiol 1995:73:2195--206. [28] Solomonow M. Baratta R. Bernardi M. Zhou B. Lu Y, Zhii M. et al. Surface and wire E M G crosstalk in neighbouring muscles. J Electromyogr Kinesiol 1994;4: 131 42. [29] Vollestad NK. Measurement of human muscle fatigue. J Netirosci Meth 1997:74:219-27. [30] Warner JJ. Micheli LJ. Arslanian LE, Kennedy J, Kennedy R. Scapulothoracic motion in normal shoulders and shoulders with glenohumeral instability and impingement syndrome. A study using Moire topographic analysis. Clin Orthop 1992:285:191-9. [31] Westgaard RH, de Luca CJ. Motor unit substitution in longduration contractions of the human trapezius muscle. J Neuropliysiol 1999:82:5014. [32] White SM. Witten CM. Long thoracic nerve palsy in a professional ballet dancer. Am J Sport Med 1993;21:626--8. [33] Yao W, Fuglevand RJ. Enoka RM. Motor-unit synchronisation increases EMG amplitude and decreases force steadiness of stimulated contractions. J Neurophysiol 2000:83:441--52. [34] Zijdewind I, Zwarts MJ. Kernell D. Fatigue-associated changes in the electromyograni of the human first doi-sal interosseous muscle. Muscle Nerve 1999:22:1432 6.
ELSEVIER
~~
Journal of Ort hopaedic Research
Journal of Orthopaedic Research 22 (2004) 872-877
www.elsevier.com/locate/ort hres
Measuring myoelectric fatigue of the serratus anterior in healthy subjects and patients with long thoracic nerve palsy
Abstract Proper function of serratus anterior plays a vital role in thc movement and stability of thc scapula, and thus of the glenohumeral joint and the upper limb. The unique anatomy of this muscle makes direct measurements of its fatigue properties impossible. Here we describe for the first time indirect measurements of the myoelectric manifestations of Fatigue in the serratus anterior. Eight healthy volunteers (29-35 years) were tested. four of them on two different occasions, using two exercise protocols (60 s isometric maximum upward force in 120" arm flexion, and 60 s maximum forward force at 90" arm flexion) with simultaneous recording by surface and wire electrodes applied according to established methods. Signals were analysed to obtain the rate of fall of median EMG frequency and the rate of rise of amplitude. Both exercise protocols gave similar results. Frequency-slope measurements (mean rate of Fall 0.6 k O.I'XI initial value per second ('XI s I ) with both surface and wire electrodes) were more precise than those of amplitude (mean rate of rise 2.6 f 0.3% s with surface electrodes, only 1.3 f 0.2% s with wire electrodes). Surface electrodes gave much lower variation than fine wires, the coellicient of variation of slopes for surface electrodes being approximately 20-40'X both between studies in a single subject and between subjects. In 5 patients (aged 22-59 years) recovering from long thoracic nerve palsy studied using surface electrodes the frequency slopes was normal (0.6 k O.I'XI S F ' ) , while the amplitude slope was reduced (0.9 f 0.4% s ', P = 0.01). This shows abnormal fatigue properties of the reinnervated muscle and a dissociation between the frequency and amplitude inanifcstations of fatigue. Crown Copyright 0 2003 Published by Elseiver Ltd. on behalf of Orthopaedic Research Society. All rights reserved.
'
.
ii.ort/~v:EMG:
Isometric contraction: Long thoracic nerve palsy: Median frequency: Muscle fatigue: Serratus anterior
Introduction
Thc movcmcnt and stability of the scapula, and thus of the glenohumeral joint and the upper limb, depend heavily on the scapulothoracic muscles [ 1,161. The role of trapezius and serratus anterior in scapular elevation and upward rotation cannot be assumed by other muscles [8]. The effect of serratus anterior dysfunction is dramatically seen in the scapular winging of long thoracic nerve palsy [l]. Serratus anterior is a broad, flat muscle formed by multiple digitations arising from the upper 8-9 ribs in the midaxillary line and attaching to the vertebral aspect of the ventral surface of the scapula. Three functional components have been identified: a superior cylindrical mass that attaches to the superio-
~~
'
~
*Correspondence author. Tel.: +44-151-7O6-4121: 1.a~: +44- I5 1706-58 15. E-rmri/ rrt/dw\.c.- ~cyinan_ebied(~hotmaii.coiii (A.M. Ebied).
medial border of the scapula and anchors the arm during overhead rotation; a long wide intermediate band that connects the 3rd-5th ribs with the vertebral border of the scapula and draws the scapula forward; and the lower five slips from the 6th-10th ribs that converge on the lower angle of the scapula to rotate its inferior angle upward and laterally [8]. Scapular winging secondary to LTNP demonstrated by asymmetrical rotation of the inferior medial border of the scapula is not uncommon and potentially disabling. Non-operative measures may produce satisfactory recovery [30], otherwise operative intervention is usually warranted [6]. Many patients continue to suffer residual winging and fatigue especially with overhead and endurance activities. However, to our knowledge there have been no measurerncnts of the fatigue properties of this anatomically unique muscle. The obvious way to quantify fatigue is as the timedependent decrease in the ability of a muscle to maintain a specified force or joint torque [29]. This is not feasible for muscles that contract in more than one movement,
0736-0266/S - see rront matter Crown Copyright 0 2003 Published by Elseiver Ltd. on behalf of Orthopaedic Research Society. All rights reserved. doi: 10.1016/i.orthres.2003.12.004
A.M. Ehiecl et ul. I Joirrriol of Orrlropaetlic~Rewrrch 22 (2004) 872 877
on more than one joint, or (as in the case of serratus anterior) whose contribution cannot be separated from that of other muscles. Indirect methods of studying myoelectric fatigue depend on analysing the electromyogram (EMG). Power spectral analysis of EMG signals has been used as a predictor and indicator of muscle fatigue during isometric contractions of many muscles [18,20]. Myoelectric signals have been shown to shift during fatiguing exercises with decline in both median M D F and mean M P F frequency. The decline in both parameters has been correlated to a reduction in the muscle fibre conduction velocity, among other things. The conduction velocity can be influenced by accumulation of metabolites at the sarcolemma and the type of muscle fibres [26]. The second EMG parameter commonly used in fatigue studies is the EMG amplitude, calculated as the integrated EMG or root mean square. EMG amplitude is commonly found to increase during exercise of different muscles while the force remains steady or declines. This increase in EMG amplitude has been attributed to increased frequency of motor unit discharge or synchronisation of motor unit firing rate [33]. Other studies showed an initial increase in amplitude followed by plateau curve [4]. EMG variables have been found to have a good level of repeatability, providing that care is taken in placing electrodes and in the setup of the test [19,25]. Here we describe for the first time measurements of the fatigue properties of serratus anterior using EMG methods. EMG signals can be acquired using surface and finewire electrodes: the latter have advantages of anatomical specificity (low 'cross-talk') and access to deep muscles, while the former are better tolerated and give more reproducible results [28]. Both have been employed in EMG fatigue studies (the former more rarely) [31,34]. For serratus anterior muscle, there is no consensus on the most appropriate electrode technique, and the only reports have concentrated on the pattern of muscle activation rather than fatigability [10,22]. Here we used both methods. The serratus anterior is involved in a complex way in a wide range of arm movements. Classically it is loaded by pushing forward with the arm in 90" forward flexion. Also, patients with serratus anterior deficiency usually find overhead activities difficult due to sensation of fatigue and scapular instability [32] that may be related to an intrinsic problem of the muscle itself or secondary to inhibition of the inferior part of the trapezius and rotator cuff muscles. We therefore tested the serratus anterior muscle in both movements.
Material and methods We carried out surface and wire electrode studies on 8 male healthy volunteers aged 29-35 years (the right side was tested in 7 and the left in 1). of whom 4 were studied twice to define repeatability. We also
873
carried out surface electrode studies in 5 patients aged 22 59 years with long thoracic nerve palsy presented with the clinical picture of dynamic winging of the inferior medial border of the scapula and confirmed by EMG and nerve conduction studies: 4 had been treated ( 2 12 months before study) by surgical decompression of the long thoracic nerve [ 171 after failure t o improve with 2 12 months physiotherapy. All patients had achieved some recovery with improvement of their ability to elevate their arms and scapular rotation and sliding along the chest wall. Using Kibler's test [14]. side-to-side difference was < I cm at the 3 different positions. However. this group of patients did not regain their full function and continued to complain of easy fatigue compared to the unaffected shoulders particularly with lengthy overhead or strenuous activities. This difference could be attributed to the altered activation pattern of the other scapular stabilisers and disadvantageous performance of the rotator cuff or to deficiency in reinnervated serratus anterior muscle itself. We hypothesised that in this group of patients reinnervated serratus anterior may be more fatigable and E M G manifestations of fatigue would be greater than normal. Fully informed consent was obtained. and the study was approved by the local ethics committee.
Muscle-testing protocols Two isometric exercise tasks were performed. The first task consisted of 60 s 'upward pushing' at 120" of forward flexion, using a KIN-COM machine (Chattanoga. Inc.. Oxfordshire. UK). With the subject sitting, a cuff round the arm 10 cm distal to the tip of the acromion was linked to a lever with centre of rotation at the level of the humeral head, and the arm in 120" forward flexion from neutral. Maximum voluntary force was determined as the best of three 3 s upward pushes separated by 5 s relaxation. Subjects were then instructed to reproduce this force for 60 s, given visual force feedback. The second task consisted of 60 s maximum 'forward pushing' against a wall with the arm in 90" forward elevation (the classic test for examining the serratus anterior clinically and demonstrating scapular winging). As the recorded torque measures not the force generated by the muscle, but rather the indirect effects of the overall loading of the scapular stabilisers. it was used only to give feedback about the level of performance. As a side observation in this study patients and healthy volunteers have felt fatigue around their shoulders and over the serratus anterior area by the end of their exercise protocol. Indeed the group with abnormal sei-ratus anterior function demonstrated scapular winging with lifting o f the infero-medial angle of their scapula during the test, in addition to feeling of discomfort and cramp like pain over the serratus anterior muscle.
E M G ucyuisition untl trntr1~~si.s E M G data were acquired simultaneously using one set of surface electrodes and two double wire electrodes. Surface electrodes were applied as described by Basmajian [2]:with the arm raised. two electrodes were centred in a vertical oval (around 4 cm long) above the inferior angle of scapula and along the midaxillary line. One wire electrode was inserted as described by Delagi [7]: with the subject prone, arm dangling over the edge of a bed, the wire was inserted just lateral to inferior angle of scapula. A second wire electrode was inserted as described by Goodgold [9]: with the subject prone, the vertebral border of the scapula was identified and lifted manually, and the wire inserted between the vertebral border of the scapula and chest wall, the tip lying close to the thoracic surface of the inferior angle of scapula. A note was taken of the sites of electrode insertion, and an attempt was made in the second test to re-insert the electrodes in same sites. The disposable 50 p n fine wire electrodes, based on the design of Basmajian and Stecko [3], are supplied ready-loaded in a 25 Ga. hypodermic needle (Chalgren Enterprises. Inc. Gilroy California). They are of nickel-chromium alloy. insulated with nylon and sterilised by electron beam: 2 mm of metal are exposed at the tips, the hooked ends being staggered. Surface electrodes were Ag-AgCl with an 8 mm contact area (BIOPAC Systems). The concavity was filled with conductive gel. Disposable Ag AgCl electrodes with I0 mni contact areas were used over the tip of the olecranon as a ground. The electrodes were connected to three channels of a 100 A E M G (BIOPAC Systems, Inc., Santa Barbara. California). Differential input amplifiers were used with a common mode rejection ratio of 100 dB minimum, input
874
A . M . Ehird
('1 rtl.
I Journul
of' Orthoparrlic R C S ~ W22~ I(2004) 872-877
impedance 10 R, bandwidth 104000 Hz and gain 2000. Data were acquired with a Biopac system M P 100 data acquisition unit (BIOPAC systems, Inc. Santa Barbara, CA) with A/D resolution of 16 bits. Signals were sampled at 2500 Hz, and bandpass filtered at 10- 1000 Hz for fine wires and 10-500 Hz for surface electrodes. Data were analysed using Acqknowledge software (BIOBAC Systems, Inc.). dividing each 60 s task into 20 intervals (data assigned to the midpoint). Median frequency was obtained by fast Fourier transform. Data from the first 5 s. which show a rapid artifactual decrease, were omitted from the analysis. To measure amplitude the signals were smoothed (2000 saniple/window) then rectified (400 samplelwindow), then nornialised as a percentage of the mean value for the whole task. as previously described [21].
A
Surface electrodes: amplitude & frequency
(YOmean) __ Frequency (Hz)
..... Amplitude
._.
=.
..E.
......;.:.: . _. _ _ . _-..:.... .-...... .. ...
-
-
,-, ,
I
.
I
Crilculation.s & sfrrti.sric:c
0 Data were transferred to spreadsheets for calculation of rates of change ('slopes'), expressed relative to the starting values: for frequency this was done by linear regression; for amplitude. where the time course was markedly nonlinear, a two-point slope calculation was performed between I = 3-9 s and I = 42-60 s (these time-points being chosen empirically to yield maximum robustness and precision). Between-study variability was calculated for each subject's pair of separate measurements. expressed as a coefficient of variation (CV = SDI mean'%!). and the root mean square value taken for all subjects. Between-person CV was calculated using the means of the paired data for all subjects. correcting for the duplicate measurements by subtracting from the observed between-person CV' half the between-study CV'. Differences were compared using the paired or unpaired t-test. as appropriate. and P < 0.05 is taken as statistically significant.
6
In heulthy subjects Fig. 1 shows a sample of the individual data collected using a surface electrode (Fig. 1A) and a fine wire electrode (Fig. IB), showing the decreases in median fiequency and increase in amplitude with time. Fig. 2A summarises mean data on the relative changes in both median frequency and amplitude. With the surface electrode, changes in amplitude are larger but more variable, especially after about 30 s. The rate of fall in frequency was similar with surface and both wire electrodes: the mean rate of fall was 0.6 & 0.1'%1initial value per second (%I S K I ) with both electrodes. The rate of rise of amplitude with both fine wire electrodes was significantly smaller than that recorded by surface electrodes (1.3 f0.2'%1s-' vs 2.6&0.3'%1s-I, P = 0.02), and more variable. With the wire electrodes, frequency measurements were less variable than amplitude, but still more variable than with surface electrodes. Fig. 2C shows that the use of surface electrode gives acceptable agreement between measurements for both amplitude and frequency slopes, and Fig. 2D shows that both slopes are similar for the two exercise protocols. For the surface electrodes, overall mean between-study CV was 29% for frequency slope and 23'Yn also for amplitude slope (both exercise protocols); for comparison, the corresponding overall figures for the wire methods were very poor, 100'%,and SO'%). The open symbols in Fig. 2E summarise the slopes estimated by the surface and wire electrode methods; it
20
30
40
50
60
Goodgold wire electrodes: frequency (Hz)
! : : :
Results
10
50 00
10
20
30
40
50
60
Time (s) Fig. 1. Some individual data from single trials as a function of time (s) in healthy subjects 'forward pushing' (1 line= I measurement on I subject). (A) Median frequency and amplitude acquired using the surface electrode. (B) Median frequency data acquired with fine wires inserted using the Goodgold technique (similar to data acquired tising the Delagi method): amplitude data (not shown) are more variable.
shows how the frequency slopes are similar but the amplitude slope smaller in wire electrodes. For the surface electrodes, the estimated true between-person CV was 33Yn for frequency slope and 42% for amplitude slope; the corresponding figures for the wire methods were 45% and 62% In patients with long thorucic nerve palsy Fig. 2B show the mean response in the group of patients in a similar way to the control data in Fig. 2A. The median frequency falls in a similar way to controls, while the amplitude response is very different, showing a reduced and delayed rise: This can be seen from the mean slope data shown as the filled symbol in Fig. 2E: the amplitude slope is lower than in controls (0.9 t 0.4% s-', P = 0.01), the frequency slope not significantly different (0.6 k O.I%I s-I).
Discussion
The protocol This is the first report of fatigue measurements of the serratus anterior, a muscle which combines an essential role in the movements of the upper limb with an
875 A
Iielative change in amplitude & frequency
C
Slope in second test
0
0Amplitude .Frequency
i
0.4
0
@
2-
Slope in first test 0
1''
0 D
I-
0.0
-0.4
0
8
4-
0.8
4
2
6
'Forward' slope
-t- Surface: amplitude
J +All +All
wires: amplitude wires: frequency
tSurface: frequency
-0.8
20
Time (s)
40
0
. - -..
0
60
...... Controls +Amplitude (surface): patients &Frequency: patients
!, ,
E
6-
_ _.. -. . _.._ _
0
Time (s) 40
60
6
4
+
...
0 ' 20
2
Frequency slope (fall)
4 0.0 -0.3 -0.6
'Upward' slope
0
.-
I
0 Controls: wire 0 Controls: wire
0Controls: surface Patients: surface
Amplitude slope (rise) 1
2
3
Fig. 2. Amplitude and frequency changes during exercise in healthy subjects and controls. (A) Shows the time course of the relative change in amplitude and frequency measured by wire and surfice methods (see key) during the contraction. For concision the results of forward and upward pushing exercises. the results of the first and second tests. and the results of the two wire electrodes are combined. ( B ) Shows the corresponding data from patients obtained using surface electrodes. with the control mean data (seen in (A)) shown as dashed lines for comparison. (C) and (D) Shows the consistency and reproducibility of surface-electrode: (C) shows mean frequency and amplitude slopes (see key) in the second test as a function of those in the first (results from forward and upward pushing are combined): (D) shows mean frequency and amplitude slopes (see key) in forward pushing as a function of those in upward pushing (results from first and second tests combined). (E) Shows the mean amplitude slope ('% s - ' ) as a function of mean frequency slope ( % s ' ) for control subjects studied using surface electrodes and wire electrodes, and patients studied by surface electrodes (results combined from both protocols, and from both wire electrodes). The amplitude slopes direr between patients and controls ( P = 0.01) and wire and surface electrodes ( P = 0.02). but the frequency slopes do not. Errors bars show SEM.
anatomy which makes direct force measurements impossible. We find, as expected, a measurable decrease in median frequency and an increase in amplitude, the former being the more precise measurement. The two exercise protocols (forward and upward pushing) give very similar results. In general surface electrodes are usually used to study superficial muscles, while wires are reserved for small, deep muscles (such as the rotator cuff muscles [21]) or when localised fibre or motor unit action potentials are being characterised. As in studies of other muscles [12,13], reproducibility was better with surface electrodes than with inserted wires. One factor in the variability with wire electrodes could be changing
position of the wires in relation to the active motor units (although the hooked wires we used are designed to engage in the muscle fibres). However, surface electrodes are liable to similar errors due to skin movements. A second factor is the relation between the wire electrodes and the active muscle bulk. Serratus anterior is composed of many digitations that combine before inserting on the anteromedial surface of the inferior angle of the scapula. The muscle is functionally divided into upper and lower parts, with different levels of activity in various tasks and arm positions. Both the Delagi and Goodgold techniques insert wires into the musculo-tendinous part of the muscle which may not always represent the overall
876
A . M . Ehicvl ri ul. I Jourtiul of Orlhopuc&
activity. Surface electrodes were inserted within an elliptical 4 cm area, perhaps giving a more representative measure of activity. Another factor might be a heterogeneous pattern of intramuscular distribution of muscle fibre types (muscle fibre regionalisation) [34]. Other reported reasons for the poor reproducibility of wire electrodes are pain (not a problem here), intramuscular bleeding and inhomogeneous structure of the musculature [ll]. Also, the difficulty in re-inserting wires exactly into the same place no doubt explains poor reproducibility between test days. A well-known possible problem with surface electrodes is cross-talk (noise from other muscles or cardiorespiratory activities). As this was negligible in the calf except in the presence of a thick layer of subcutaneous fat [28], there is no reason to think crosstalk was a problem in our study. As in other surface-electrode studies [ 15,251, there is considerable variation in absolute values of median frequency (see Fig. 1A). However, the absolute slopes correlated significantly with the starting values (not shown), so that the relative slope used here is the appropriate quantity, and shows only modest variability.
Reseurcli 22 (2004) 872-877
tigue, it is affected by various local reflexes originating from the muscle itself. The denervation-reinnervation process has been shown to influence the distribution of muscle fibre type, size of sprouting motor neurones and recovery of sensory receptors and afferents [24]. These changes may interact to change the muscle characteristics resulting in more fatigable muscles, as seen in rat medial gastrocnemius and human soleus [27]. However, the conversion in muscle characteristics was found to be correlated to the paralysis period as well as the severity of nerve injury. In our case it may be that the period of denervation or severity of nerve compression were not severe enough to lead to full conversion of the muscle fatigue characteristics despite the changes seen in EMG amplitude that may represent lack of muscle ability to recruit large or fast contracting motor units or the lack of fast contracting fatigable units in the reinnervated muscles. Although this study could not demonstrate definitive difference in fatigability of the serratus anterior in this group of patients with abnormal serratus anterior function we have established a method that can be expanded to study other groups of serratus anterior dysfunction in future experiments.
result.^ in patients with long tlzonnc‘ic nerve palsy
Having concluded that surface electrodes and ‘upward pushing’ were a valid method for studying the altered fatigue properties of this important but technically difficult muscle, we also studied 5 patients with scapula winging due to long thoracic nerve palsy (4 following partially effective treatment). We established first that these patients could perform the protocol. We wished to test the hypothesis that manifestations of fatigue would be increased. This was not so. The frequency slope was normal, while the amplitude slope was reduced by about half. This is unlikely to be explained by differences in the muscle loading (impossible to assess, because of the possible compensation by other muscles), because in biceps brachii the frequency slope, not the amplitude slope, increases linearly with isometric contraction force [25]. Why this contradiction? A well-described phenomenon in normal muscles is the order of recruitment of motor unites. Slowly contracting, fatigue resistant motor units with low rate of frequency discharge are usually recruited early during the exercises while fast contracting, strong units are usually reserved until the end, perhaps until the muscle starts to demonstrate some evidence of fatigue [23]. As contraction continues and fatigue develops, motor units start to decrease their firing frequency, but even a reduced firing rates can produce the same level of muscle activation, a phenomenon known as “muscle wisdom” [ 5 ] . This recruitment order is influenced by the characters of the motor units and the central drive or motoneurone impulse rate and frequency. Although it can be a central mechanism of fa-
Acknowledgements The authors wish to thank Mr Alfred Morris for his assistance in the early stages of this project. This research project was supported by a grant from Menoufiya University, Egypt. References Barnett ND, Mander M, Peacock JC. Bushby K, Gardner Medwin D, Johnson G R . Winging of the scapula: the underlying biomechanics and an orthotic solution. Proc Inst Mech Eng ( H ) 1995;209:215-23. Basmajian J. In: Desmedt J, editor. New concepts of the motor unit in neuromuscular disorders: electromyographic kinesiology. Basel: S Karger; 1973. Basmajian J, Stecko G . A new bipolar electrode for electromyography. Appl Physiol 1962;I7:849. Bazzy AR, Donnelly DF. Diaphragmatic Failure during loaded breathing: role of neuromuscular transmission. J Appl Physiol: Respir Environ Exercise Physiol 1993:74: 1679-83. Bigland Ritchie B, Cafarelli E, Vollestad NK. Fatigue of submaximal static contractions. Acta Physiol Scand Suppl 1986: 556: 1 3 7 4 . Connor PM, Yamaguchi K, Manifold SG, Pollock RG. Flatow EL, Bigliani LU. Split pectoralis major transfer for serratus anterior palsy. Clin Orthop 1997341:13442. Delagi E, Perotto A. Iazzetti J, Morrison D. Anatomic guide for the electromyographer. New York: Springfield & Illinois: 1975. Duralde XA. Evaluation and treatment of the winged scapula. J South Orthop Assoc 1995;4:38-52. Goodgold J . Anatomical correlates of clinical electromyography. Second ed. Baltimore: Williams and Wilkins; 1984.
[lo] Jobe FW. Moynes DR, Tibone JE, Perry J. An E M G analysis of the shoulder in pitching. A second report. Am J Sport Med 1984;12:218 20. [ I I ] Jonsson B. Bagge U. Displacement deformity and fracture of wire electrodes for electromyography. Electromyography l968:8:32947. [I21 Jonsson B, Komi P. In: Desmedt J. editor. New Concepts of the Motor Unit Neuromuscular Disorders Electromyographic Kinesiology. Basel: S Karger AG, National-Zeitung AG: 1973. [I31 Kadaba MP, Wootten ME. Gainey J. Cochran GV. Repeatability of phasic muscle activity: performance of surface and intramuscular wire electrodes in gait analysis. J Orthop Res 1985:3:350-9. [I41 Kibler WB. The role of the scapula in athletic shoulder function. Am J Sport Med 1998;26:325-37. [I51 Krivickas LS. Taylor A. Maniar RM. Mascha E. Reisman SS. Is spectral analysis of the surface electromyographic signal a clinically useful tool for evaluation of skeletal muscle Fatigue? J Clin Neurophysiol: Off Pub1 Am Electroencephalogr Soc l998;15:13845. [I61 Ludewig PM, Cook TM, Nawoczenski DA. Three-dimensional scapular orientation and muscle activity at selected positions of hiimeral elevation. J Orthop Sport Phys Ther 1996:24:5765. [I71 Manning P, Frostick S, Wallace W. Winging of the scapula, a fresh look at the long thoracic nerve. In: 9th Congress of the European Society for Surgery of the Shoulder and Elbow, Nottingham, 19th-21st September, 1996. [I81 Merletti R, Farina D. Gazzoni M, Schieroni MP. Effect of age on muscle functions investigated with surface electromyography. Muscle Nerve 2002;25:65-76. [I91 Merletti R, Fiorito A, LoConte LR, Cisari C. Repeatability of electrically evoked EMG signals in the human vastus medialis muscle. Muscle Nerve 1998;21:184 93. [20] Merletti R. Knaflitz M, De Luca C. Myoelectric manifestation of fatigue in voluntary and electrically elicited contractions. J Appl Physiol l99O:69: 1810-9. [21] Morris AD. Kemp GJ, Lees A. Frostick SP. A study of the reproducibility of three different normalisation methods in intra-
muscular dual fine wire electromyography of the shoulder, J Electromyogr Kinesiol 1998:8:3 17-22, [22] Nuber GW, Jobe FW, Perry J. Moynes DR, Antonelli D. Fine wire electromyography analysis of muscles of the shoulder during swimming. Am J Sport Med 1986;14:7-1 I . [23] Olson CB, Carpenter DO. Henneman E. Orderly recruitment of muscle action potentials. Arch Neurol 1968:19:591-7. [24] Rafuse VF, Gordon T. J Physiol 1998;509(Pt 3):909-26. [25] Rainoldi A, Galardi G, Maderna L. Comi G , Lo Conte L. Merletti R . Repeatability of surface E M G variables during voluntary isometric contractions of the biceps brachii muscle. J Electromyogr Kinesiol 1999:9:105-19. [26] Sadoyama T, Masuda T. Miyata H. Katsuta S. Fibre conduction velocity and fibre composition in human vastus lateralis. Eur J Appl Physiol Occup Physiol 1988;57:767-71. [27] Sheilds RK. Fatigability. relaxation properties, and electromyographic responses of the human paralyzed soleus muscle. J Neurophysiol 1995:73:2195--206. [28] Solomonow M. Baratta R. Bernardi M. Zhou B. Lu Y, Zhii M. et al. Surface and wire E M G crosstalk in neighbouring muscles. J Electromyogr Kinesiol 1994;4: 131 42. [29] Vollestad NK. Measurement of human muscle fatigue. J Netirosci Meth 1997:74:219-27. [30] Warner JJ. Micheli LJ. Arslanian LE, Kennedy J, Kennedy R. Scapulothoracic motion in normal shoulders and shoulders with glenohumeral instability and impingement syndrome. A study using Moire topographic analysis. Clin Orthop 1992:285:191-9. [31] Westgaard RH, de Luca CJ. Motor unit substitution in longduration contractions of the human trapezius muscle. J Neuropliysiol 1999:82:5014. [32] White SM. Witten CM. Long thoracic nerve palsy in a professional ballet dancer. Am J Sport Med 1993;21:626--8. [33] Yao W, Fuglevand RJ. Enoka RM. Motor-unit synchronisation increases EMG amplitude and decreases force steadiness of stimulated contractions. J Neurophysiol 2000:83:441--52. [34] Zijdewind I, Zwarts MJ. Kernell D. Fatigue-associated changes in the electromyograni of the human first doi-sal interosseous muscle. Muscle Nerve 1999:22:1432 6.