Reproducibility of lumbar paraspinal surface electromyogram power spectra

Clinical

Copyright

Biomechunics

0 1996 Elsevicr

Vol. 11. No. 6. pp. 317-371. 1996 Science Limited. All rights rcservcd Printed in Great Britain 026X-0033/90

$15.00

+ 0.00

ELSEVIER PII: SO268-0033(96)00020-4

Reproducibility electromyogram

of lumbar paraspinal power spectra

C W Oliver FRCS~, K M Tillotson C G Greenough MD’

CStat’,

surface

A P C Jones mD3, R A Royal

‘Middlesbrough General Hospital, Middlesbrough; ‘Spinal Research Unit, University Huddersfield; and 3Northern Regional Medical Physics Department, South Cleveland Hospital, Middlesbrough, UK

BSC~,

of

Abstract Objective. To establish the reproducibility of surface recordings of electromyograms from lumbar erector spinae muscles. Design. Repeated measurements in 10 male volunteers under controlled conditions. Methods. While isometric load was held constant at 213 and l/3 of maximum voluntary contraction, surface electromyograms were recorded from the 4th/5th lumbar interspace on three occasions separated by at least a day. Fresh surface electrodes were applied on each test occasion. The raw electromyographic signal was filtered and digitized and the signal processed by fast Fourier transformation to give median frequency and total amplitude spectra. Linear regression lines of median frequency and amplitude against time were calculated. The ‘spectral halfwidth’ was defined as the full width at half maximum of the composite amplitude spectrum. Reproducibility was assessed by calculation of the intraclass correlation coefficient. Results. Reproducibility of initial median frequency and spectral halfwidth were found to be excellent at both 2/3 (ICCC 0.91 and 0.88) and l/3 (ICCC 0.84 and 0.83) maximum voluntary contraction. Initial power and regression slopes of power and median frequency were not reproducible. Conclusions. The reproducibility of two electromyographic variables, the initial median frequency and the spectral halfwidth, were found to be satisfactory. Relevance The surface electromyogram provides objective evidence of the functioning of erector spinae muscles and may be valuable in further research in low back pain. Before applying any new technique, however, it is necessary to prove that it is reproducible. Copyright @ 1996 Elsevier Science Ltd. Key words: Surface electromvonraohv, ._. spectral halfwidth

C/in. Biomech.

soectral

Vol. 11, No. 6, 317-321,

analysis.

reoroducibility,

1996

Introduction Over the last few years a greater understanding has evolved concerning the relation between disability, deconditioning and lumbar trunk strengthrm4. Since paraspinal muscle function is thought to be essential for providing extrinsic spinal support, sub-optimal or impaired muscle control abnormalities could permit spinal instability and predispose to further spinal damage. The currently used techniques and devices for Received: 30 November 1993; Accepfed: 13 February 1996 Correspondence and reprint requessts to: Charles G Greenough FRCS, Consultant Orthopaedic Surgeon, Middlesbrough General Hospital, Ayresome Green Lane, Middlesbrough, Cleveland TS5 5AZ, UK

MD Mchlr

low back pain, fitness testing,

evaluating the performance of back muscles measure mechanical variables associated with force, velocity or displacement of the trunk. These approaches all share a common flaw in that the measured kinematics and force variables are cognitively perceived by the central nervous system and thus can be voluntarily regulated. A highly motivated individual interested in knowing the upper limits of their strength would perform to the full extent of their capability, whereas an individual with less motivation would perform at a lesser level, thus not revealing their full potentia14,‘. Dynamometric methods of testing lumbar spine function have gained popularity but have not developed enough to gain acceptance in British medico-legal practice”. It has also been shown that a stronger back is not protective against future back injury’.

31 ii

Clin. Biomech.

Vol.

11, No. 6, 1996

A more objective method of assessment has been sought by using electromyographic (EMG) techniques to monitor shifts of the power spectrum during fatiguing isometric contractions’.s ’ ’ Surface techniques are attractive because they are non-invasive and provide dynamic internal information of muscular function that may more directly reflect physiological changes. Electromyographic recording and analysis has limitations such as the great inter-individual variation of the amplitude of the myoelectric signals. differences in the electric conductivity of tissue, electrode geometry about muscles, muscle length, fibre size, posture and isometric load”. “. The control of as many as possible of these factors within and between subjects is important in long-term studies. There have only been scant reports of reproducibility and reliability of median frequency (MF) recording techniques. Roy et al.” ascertained the error induced in the value of the EMG spectral variable by repeating a contraction within 15 min under similar conditions; the error was found to be 2% for the initial MF value and 6% for the slope of the MF regression line. Performing similar evaluations on test-retest measurements, Biedermann et al.“-‘” found that the error in the MF slope measurement increased to approximately 10% when electrodes were replaced at 5-day intervals. The aim of this study was to examine the reproducibility of surface EMG measurements in healthy volunteers, over a period of time with replacement of electrodes. Figure 1. isometric

reference

frame.

Methods

Ten healthy male subjects were recruited from the local hospital community. All subjects gave informed consent for testing and the study was approved by the loca! ethics committee. These subjects were tested on three separate occasions in a period between 5 days and 4 weeks with at least 1 day’s rest between test occasions. All were employed. with the majority working in clerical or administrative positions. Anthropometric details are given in Tabfc 1. The body mass index was calculated as the body mass in kilograms divided by the height in metres squared Isometric lumbar extension testing was performed on a standard reference frame shown in Figure I. The feet were placed 20 cm apart with the knees straight and the pelvis attached to a bar by a single strap. The pelvic stabilizer height was set 6 cm below the anterior superior iliac $pines with the subject erect. Pelvic stabilizer height and load cell chain length were recorded for future testing. Table 1. Anthropometric Variabie

.-

details .- .__-.

Age Weight Height Body mass index Percentage body fat Lean body mass Max. voluntary contraction

Meat7 . . .. .-.. __-----38.0 years 70.1 kg 1,618 m 27.3 kg/m’ 22.4% 46.1 kg 87.5 kg

..(range 26-50) (SD, 11.6) (SD, 0.119) (SD, 4.9) (SD, 4.6) (SD, 18.6) (SD, 33.2)

Loading was performed by pulling on a bar attached to a load cell fixed to base of the frame. The lumbar spine was positioned at a forward inclination of 30 degrees by positioning the lumbar region parallel to a body contour formulator (BCF) which was fixed at 30 degrees to the vertical. The adjustable BCF rods were pressed down and secured onto the curve of the lumbar spine. Once the subject was positioned on the test frame the BCF arm was swung away, and the rod profile recorded on a piece of card. This template was used to facilitate future repositioning of the test subject on the frame in an identical posture. The level and distance from the midline of the surface electrodes was also recorded for future testing. The position of the head and shoulders was not specifically recorded but during testing the subject observed a computer screen which was always in the same position. Subjects were asked to maintain a maximal contraction for 3 s and the peak load was recorded. The subject was unable to see the load cell during the attempt. This was repeated on three occasions with at least 4 min rest between tests. The maximal, voluntary contraction (MVC) was defined as the average of these three peak loads. A subject was to be rejected if the three trials did not fall within 10% of each other, but this did not occur in this series. The MVC was determined once only for each subject. After at least 4 min rest two tests were made, with

Oliver et al: Reproducibility

4 min rest between. The first test was made at l/3 MVC and the second at 213 MVC, each being held for 30 s whilst surface electromyographic recordings were made. Isometric load levels were maintained at target load f 10% by the subject observing a computer display of the target and actual load level. One low-load l/3 MVC rehearsal was allowed and subjects learned control of the load cell within 10 s. Tests were to be rejected if the isometric load level drifted more than 5% from the target level, but this did not occur. Surface electrode recording sites were carefully identified in every subject. The 4thi5th lumbar interspace was palpated in the midline guided by the line between the posterior superior iliac spines of the iliac crests. The central earth electrode was applied at this point. Two surface electrodes were applied on each side over the greatest convexity of the erector spinae muscles at the LdPs level with a vertical inter electrode distance (between adjacent edges) of 40 mm. Typical sites of application of surface electrodes are shown in Figure 2. The surface electrodes were Ag/AgCl (Biotrace*, Medicotest Neonatal ECG electrodes) and measured 18 by 38 mm (total contact area 684 sq. mm.). The skin surface was prepared with alcohol and fine clinical abrasive (Cardiopreps*). Contact resistances were checked and only accepted if surface impedance was below 8 kQ. An assessment,using hidden reference marks, of the accuracy of positioning of the four electrodes on the sites defined by the above protocol was carried out in four

of lumbar

surface electrom yograms

Change in Median Frequency

319

over Time

-TRW, 0

to

a

20

30

R-a=o.yIz

The (S)

Figure 3a. Median

frequency

plot overtime

for three trials in one subject.

Change in Power over Time

.

lmloo~ 0 b

Figure 3b. Total amplitude subjects.

10

20

--

Time (S)

plot over time for the initial trial in two

subjects. Three applications of the electrodes were carried out and 88% of electrode placements had a more than 50% overlap with a target rectangle defined as f 10 mm horizontally and + 20 mm vertically. The signal was preamplified (X loo), filtered, amplified, digitized and then stored on floppy discs. Filter bandwidth was set at 3-200 Hz and the signal was sampled at 1000 Hz. The EMG signals were monitored throughout the test on an oscilloscope screen. Fast Fourier transformation (FIT) was performed in l-s epochs on the digitized signal and the total amplitude and median frequency of each of the 30 spectra were calculated against time (Figure 3). The initial amplitude and initial MF were defined as the constant term of the appropriate linear regression. A composite spectrum was produced by averaging the 30 individual l-s spectra. From this composite spectrum, a new variable, the ‘spectral halfwidth’, was defined as the full width at half amplitude (Figure 4). Statistical analysis

Figure 2. Typical

positioning

of surface electrodes.

The reproducibility’ of the measured variables over the three occasions was assessed using estimates of reliability 17. Based on one way analysis of variance, they give the proportion of the variance of an observation due to true inter subject variability. This proportion is termed the intra-class correlation coefficient (ICCC). A figure of 0.75 or above may be regarded as ‘excellent’17. Each estimate is complemented by its one-sided

UIU.

320

3 z I

Biomech.

Vol. ?1, No. 6, 1996

1

Table 3. Details of values 30)

of initial

median

Mean

SD

lowest

Highest

213 MVC Initial MF (Hz) Halfwidth (Hz)

61.8 57.5

10.1 11.7

45.0 36.0

76.4 87.0

2.14 2.94

l/3 MVC Initial MF (Hz) Halfwidth (Hz)

65.1 70.3

10.1 15.4

49.0 46.0

83.4 105.0

2.90 4.59

Variable

0.5

lt-tl

0 0

100

Halfwidth (full width at half maximum)

200

300

500

400

Frequency Figure 4. Illustration

of spectral

600

(Hz)

halfwidth

95%) contidence lower bound and the percentage of the variance of the observations which can be ascribed to cxperimcntal error (including day to day variability). Where the cxpcrimental error is small a 95% least significant difference (LSD) is indicated. If a variable changes by lesu than this value. the change may reasonably be ascribed to experimental error alone. Results Table 2 summarizes the reproducibility of the five measures at both 2i3 MVC‘ and 113 MVC. Only initial MF; and spectral halfwidth can be classed as reproducible under the conditions of testing. For these measures the 9540 LSD is quoted. Although 7X0/o of regression lines demonstrated statistically significant fatigue, variability due to electrode placement and dayto-day changes renders the other measures statistically unsuitable. 7’abIe 3 describes the sample values obtained for the reproducible measures and so provides an initial basis for future work with these measures.

Discussion 1‘0 date there have been very few reproducibility studies of power spectra and MF using fresh electrodes. Any new technique that is introduced as a method of (:-valuation must have its reproducibility demonstrated. Table 2. Reproducibility experimental error ix)

confidence lone-sided 95% and least significant differences Reproduobility

Experimental error (%I

213 MVC Initial MF initial powe: Regression siope MF Regression slope power Halfwidth

0.96 0.66 0.72 0.43 0.94

10.91) 10.41) (0.50! (0.13! (0.881

4 34 28 57 6

li3 M\/L Initial MF initial power Regression slope MF Regression slope pow9 Halfwidth

0.92 0.60 0.44 0.18 0.92

(0.84j (0.331 (0 24) (ns., (0.83)

8 40 56 82 8

bound),

95% LSD (Hz) 6.3

87 8.5 _. 13.5

frequency

and halfwidth

(n =

Within-person SD

This is particularly true for the surface EMG techniques because there is a need to overcome the bias that has evolved from a history of unfulfilled expectation?. Electromyographic recording and analysis has limitations such as the great interindividual variation of the amplitude of the myoelcctric signals, differences in the electric conductivity of tissue, electrode geometry in relation to the muscles, muscle length and fibre size’.” “,‘Ix. i?\ previous study on reproducibility was made by Stokes’” who used the efficiency of electrical activity to characterisc the integrated EMG-extensor torque relationship. The efficiency index was found to be non-linear in each patient. The coefficient of variability (within subjects) was greater in day to day testing (24%) than with repeated pulls at the same testing session ( 14%). Approximately 25% of the variability in Stokes series was found to be related to anthropometric differences. The efficiency of electrical activity is therefore not a useful parameter in assessing the torque of the lumbar muscles as assessed by the EMG. Komi”” compared the reproducibility of surface and needle electrodes and found that surface techniques were more reliable than needle techniques due to difficulties in accurate reinsertion of needle electrodes. The precautions taken in controlling posture, electrode placement, electrode resistance and isometric force are important to give long-term reproducibility. The results from this study estimate the variation from the mean of initial MF and MF regression slope at 4 and 28% respectively, which can be compared with the 2 and 6% found by Roy8 using similar test equipment but with no change of electrodes and a short interval between tests. The intraclass correlation coefficient is a useful method of assessing reproducibility which has been used in the biceps2’ where the ICCC for the MF was 0.99 for two tests 5 days apart. The reproducibility at l/3 and 2/3 MVC for initial MF and halfwidth can be classed as ‘excellent’. The slopes of the regression for MF and total amplitude were not reproducible. The relatively stable determinant of initial MF and halfwidth can be contrasted with the more unpredictable processes of metabolite production, re-utilisation and vascular flow that are the primary determinants of the MF slope and total power slope’. These different physiological correlates may further help to explain the differences in reproducibility between initial MF and MF slope parameters. Although the inclination of the spine was controlled, the posture of the upper body was not. We would anticipate that

Oliver

et al: Reproducibility

further improvements in reproducibility might therefore be possible, however our results should be achievable in the less controlled environment of clinical testing. Conclusion Lumbar muscle function is considered to be an important component of chronic lower back pain. It has been demonstrated in this study that at l/3 and 2/3 MVC the initial MF and halfwidth are reproducible in subjects with no history of back pain. Spectral analysis may be of practical importance for the assessment of muscle deficits associated with low back pain and evidence presented here indicates that the technique is reliable and gives encouragement to its use in future studies.

surface

electromyograms

321

7 Battie MC, Bigos S, Fisher LD et al. Isometric lifting

strength as a predictor of industrial back pain reports. Spine 1989 14; 8: 851-6 8 Roy SH, DeLuca CJ, Casavant DA. Lumbar muscle fatigue and chronic lower back pain. Spine 1989; 14: 992-1001 9 Roy SH, DeLuca CJ, Snyder-Mackler

10

11 12

13

Acknowledgements The authors wish to acknowledge technical advice from Dr M Adams and Dr P Dolan, Comparative Anatomy Department, University of Bristol.

14

References

15

1 Oliver CW, Greenough CG. The role of lumbar paraspinal surface electromyography in low back pain. J Bone Joint Surg 1994; 76[B] Suppl 1: 44 2 Mayer TG, Gatchel R, Kishino N. Objective assessment of spine function following industrial injury. A prospective study with comparison group and one year follow up. lY85 Volvo Award in Clinical Sciences. Spine 1985; 10: 482-93 3 Mayer TG, Kondraske G, Mooney V et al. Lumbar myoelectric spectral analysis for endurance assessment, a comparison of normals with deconditioned subjects. Spine 1989; Y: 986-991 4 Stokes M, Cooper RG. Morris G, Jayson MIV. Selective changes in multifidus dimensions in subjects with chronic low back pain. Eur Spine J 1992; 1: 38-42 5 DeLuca CJ. Myoelectrical manifestations of localised muscular fatigue in humans. BCrit Rev Biomed Eng 1984 11;4:251-79 6 Jayson MIV. Trauma. back pain, malingering and compensation. Br Med J 1992; 305: 7-8

of lumbar

16 17 18

L et al. Fatigue. recovery and low back pain in varsity rowers. Med Sci Sports Exert 1990;22: 463-9 Klein AB, Snyder-Mackler L, Roy SH, Deluca CJ. Comparison of spinal mobility and isometric trunk extensor forces with electromyographic spectral analysis in identifying low back pain. Phys Ther 1991; 71: 445-54 Dolan P, Adams MA. Back muscle fatigue in static and dynamic activities. Paper presented to Society of Back Pain Research (Bristol) 1992 Rosenburg S, Siedel H. Electromyography of lumbar erector spinae muscles- influence of posture. inter electrode distance, strength, and fatigue. Eur J Appl Physioll989; 59: 104-14 Basmajian JV, DeLuca CJ. Muscles Alive -- Their Functions Revealed by Electromyography. 5th edn. Williams and Wilkins, 1985 DeLuca CJ. Limits on the Use of Surface Electromyography in Biomechanics. Wartenweiler Memorial Lecture. International Society of Biomechanics. Paris 1993 Biedermann HJ, Shanks GL, Forrest WJ, Inglis J Power spectrum analyses of electromyographic activity discriminators in the differential assessment of subjects with chronic low back pain..Spine 1991; 16: 1179-84 Biedermann HJ, Shanks GL, Inglis J. Median frequency estimates of paraspinal muscles: reliability analysis. Electromyogr Clin Neurophysioll990; 30: 83-8 Fleiss JL. The Design and Analysis of Clinical Experiments. John Wiley and Sons, 1986, Canada DeLuca CJ. Use of the surface EMG signal for performance evaluation of back muscles. Muscle Nerve 1993;16:210-16

19 Stokes IAF, Rush S, Moffroid Met al. Trunk extensor-EMG torque relationship. Spine 1987; 12: 770-6 20 Komi PV, Buskirk ER. Reproducibility

of electromagnetic measurements with inserted wire electrodes and surface electrodes. Electromyography

1970;10:357 21 Daanen HAM,

Mazure M, Holewitjn M, Van der Velde G. Reproducibility of mean power frequency of the surface EMG. Eur JAppl Physioll990; 61: 274-7