CLINICAL CONSIDERATIONS IN THE USE OF SURFACE

ELECTROMYOGRAPHY: THREE EXPERIMENTAL STUDIES

Gregory J. Lehman, MSca

ABSTRACT

Background: The validity and applicability of erector spinae electromyogram (EMG) use in clinical
practice is questionable. Differences in the amplitude of the EMG signal between populations with low
back pain (LBP) and controls have been hypothesized but not sufficiently documented.
Objective: To examine issues of EMG asymmetry and repeatability in populations with LBP and
populations without LBP in 3 separate studies.
Study Design: Three separate experimental research studies.
Methods: Study 1 examined the intraclass correlation of the EMG signal of the paraspinal musculature
at L3 on 3 separate days during quiet stance with use of 3 different normalization techniques: percent
maximum voluntary contraction, percent submaximal voluntary contraction, and percent averaged
submaximal contractions. Study 2 used a population with chronic LBP to compare the bilateral
asymmetry of paraspinal musculature in segments exhibiting pain with segments not exhibiting pain
during quiet stance. A 1-way analysis of variance was used to compare differences between asymmetry
scores in the pain and nonpain groups. In study 3, persons with LBP and persons in the control group had
the dynamic asymmetry of paraspinal musculature compared during forward bending. A cross-correlation
coefficient assessing bilateral muscle activity was calculated for each subject and differences between
groups were evaluated using a 1-way analysis of variance.
Results: Study 1: Subjects showed excellent repeatability (ICC ⬎ 0.75) regardless of the normalization
technique. Study 2: During quiet stance no differences were found in bilateral asymmetry between painful
and nonpainful motion segments. Study 3: No differences existed in bilateral dynamic symmetry for the
upper erector spinae; however, differences between groups did exist for the lower erector spinae.
Conclusion: Repeatability of the EMG signal during quiet stance is acceptable; however, bilateral
asymmetry may not be a definitive indicator of dysfunction. (J Manipulative Physiol Ther 2002;25:293-9)
Key Indexing Terms: Electromyography; Erector Spinae Muscles; Low Back Pain

INTRODUCTION
he use of the erector spinae electromyogram (EMG) ers, a general consensus on the use of surface EMG in clinical

T signal has been researched in an attempt to dis-

cern differences between those with low back
injury and those without injury. Despite the work of research-
a
Research Associate, Graduate Studies and Research Department,
Canadian Memorial Chiropractic College, Toronto, Ontario, Canada.
Funded by the Natural Sciences and Engineering Research
Council (NSERC).
Submit reprint requests to: Gregory Lehman, DC, Canadian
Memorial Chiropractic College, 1900 Bayview Ave, Toronto,
Ontario, M4G 3E6, Canada (e-mail: gregorylehman@hotmail.
com).
Paper submitted February 13, 2001; in revised form March 15,
2001.
Copyright © 2002 by JMPT.
0161-4754/2002/$35.00 ⫹ 0 76/1/124423
doi:10.1067/mmt.2002.121410
practice is lacking. Questions regarding its validity and useful-
ness exist, yet its use is often embraced by many without
reserve. In contrast to this willful acceptance is an attitude
that quickly shuns its use without fully exhausting the EMG
signals’ potential for clinical decision making. The purpose
of this article is to add to the EMG knowledge base by
addressing the 3 following questions:
1. Can normalization tasks improve the repeatability of
EMG values during quiet standing?
2. Is the paraspinal EMG activity at a painful motion
segment more asymmetrical than at a nonpainful mo-
tion segment in the same person?
3. Is the bilateral paraspinal dynamic EMG activation
profile more asymmetric in a population with low
back pain (LBP) than in an asymptomatic population?

293
294 Lehman
Clinical Use of PMG
Study 1 addresses issues of between-day repeatability of
raw and normalized EMG signals during quiet standing.
Although well-researched, conflicting results have been
shown. Dynamic tasks have shown moderate to high repeat-
ability,1,2,17 whereas evaluation of static tasks have demon-
strated poor repeatability4 or excellent repeatability,5 de-
pending on the study. These inconsistencies in static tasks
motivated study 1; various normalization techniques were
employed in an attempt to improve the reproducibility of the
EMG signal during quiet stance across days.
The remaining 2 studies address the issue of asymmetry
between bilateral erector spinae muscle groups during sym-
metrical tasks. EMG asymmetry is often considered to be a
sign of dysfunction, although support for this belief is not
fully supported by the literature. Table 1 presents a brief
overview of the studies investigating EMG amplitude and
asymmetry differences between populations with LBP and
pain-free controls. The studies appear to suggest that dif-
ferences exist between the groups during dynamic flexion
tasks at peak flexion (ie, those with LBP typically lack the
flexion relaxation phenomenon) and between the ratio of
activity during forward flexion and re-extension. However,
EMG activity during other dynamic tasks and static tasks
does not appear to be statistically different between the 2
groups. However, a general theme seems to be a trend for
increased activation in the group with LBP. The concept of
EMG amplitude symmetry between left and right erector
spinae has also been investigated. Those who promote the
use of scanning EMG suggest that left/right differences in
EMG activation level are indicative of pathology and there-
fore are suggestive of a biomechanical lesion. The research
to lend credence to this idea is mixed and sparse. Four large
studies4,6-8 have compared left/right EMG activity between
populations with LBP and populations without LBP during
static and dynamic tasks and demonstrated no differences
between the 2 groups. Conversely, 2 smaller studies have
shown contrary results. Aleksiev9 found differences in sym-
metry between groups with LBP and control groups during
various levels of static contraction but no differences were
found during rest. Similar results were found by Leach et
al10 in the lower erector spinae. Although the research on
asymmetry appears mixed, the more robust studies appear to
indicate that amplitude asymmetry is not a distinguishing
feature of low-back injured patients. However, none of
these studies specifically directed the location of electrode
placement to the level of a spinal lesion (indicated by pain)
to adequately assess the idea that problematic segments are
inherently more asymmetrical than asymptomatic spinal
levels in controls or within the subject’s own spinal column.
The reason the majority of studies to date have not
demonstrated differences in symmetry between populations
may be a result of many factors that modulate measured
EMG activity level that are not related to the level of neural
drive. Electrode placement, skin temperature, moisture, cu-
taneous fat distribution, and muscle-fiber type and size can
Journal of Manipulative and Physiological Therapeutics
June 2002
all influence measured EMG activity level. Nonhomogene-
ity in these factors between sides of the body may relegate
asymmetry in measured EMG activity to be the norm even
though it is possible that bilateral muscles are contracting at
equal intensities. Because of nonhomogeneity in the preced-
ing factors we may be unable to adequately measure it.
Timing or activation profiles (linear envelope) of EMG
activity is not influenced by these factors. One can compare
the shape of the EMG profile between bilateral muscle
groups. In a small study (n ⫽ 6), Grabiner11 demonstrated
differences in bilateral coupling of the EMG waveform in
persons with LBP and control groups during isometric ex-
ertions at different loads. However, the statistical test used
to assess asymmetry compared the phasic properties of the
EMG signal with the amplitude characteristics; this may
detract from the conclusion of the study based on the
limitations in comparing amplitudes because of factors dis-
cussed earlier. It is possible to assess the phasic qualities of
the EMG signal by using a simple cross-correlation func-
tion.
In an attempt to explore all concepts of asymmetry,
studies 2 and 3 address 2 asymmetry concepts. Study 2
explores whether a spinal segment that exhibits pain and
tenderness has more amplitude asymmetry than a segment
within the same person who is free of pain and tenderness.
This study was motivated by the use of scanning EMG
equipment that is often used to direct treatment of perceived
problematic segments because of their EMG asymmetry.
Study 3 evaluates the bilateral asymmetry of the erector
spinae linear envelope during a dynamic forward flexion
movement using a cross-correlation function. This function
only addresses issues of phasic imbalance but not amplitude
differences between the signals. The methods and results
will be presented together for each study followed by a
general discussion involving all 3 studies.
METHODS
Three separate studies with different study protocols and
different subject populations were conducted. The follow-
ing sections will present the 3 different methods for each of
the separate experiments.
Study 1: Investigating Normalization
Patient characteristics. Eight subjects were recruited with no
history of LBP within the past 6 months (5 men, 3 women;
age ⫽ 24.6 ⫾ 2.3 years, height ⫽ 175.5 ⫾ 10.3 cm, and
weight 73.2 kg ⫾ 12.15 kg). All read and signed informa-
tion and consent forms approved by the University of Wa-
terloo Office of Research Ethics.
Study protocol. The between-day repeatability of the average
EMG activity during quiet standing of the erector spinae
muscle at the L3 vertebrae was assessed on 3 different days.
Day 1 and day 2 were 1 day apart, whereas day 3 was 6 days
after day 2. During a single test session, bilateral EMG
Journal of Manipulative and Physiological Therapeutics Lehman 295
Volume 25, Number 5 Clinical Use of PMG

Table 1. Ability to discriminate populations based on EMG amplitude during static and dynamic tasks

Authors No. of Subjects Methods Major findings

Kravitz, Moore, and LBP ⫽ 22 Evaluated EMG activity during No differences between groups. Trends for
Glavos15 Control ⫽ 17 prone lying increased elevation in LBP

Sherman8 Control ⫽ 15 Tasks included sitting, standing, No left/right asymmetry differences between
History of pain ⫽ 28 bending and rising. groups, no single aspect of EMG had
Experiencing pain ⫽ 83 predictive value.
Large variability in raw values.
Trends existed suggesting pain sufferers had
higher EMG levels.

Miller16 LBP ⫽ 11 Performed sitting and standing No differences in EMG amplitude between
Control ⫽ 11 trials groups

Nouwen7 LBP ⫽ 20 Performed flexion, extension, Loss of FRP in LBP group

LBP ⫽ 20 lateral bend and twist Both groups showed bilateral asymmetry but no
difference between groups.

Ahern et al.4 LBP ⫽ 20 Performed flexion/extension, Static stance correctly classified 59% of subjects
Control ⫽ 20 rotation and stance with no difference found in EMG amplitude
or bilateral symmetry between groups.
Using lumbar ROM, EMG activity during peak
flexion, EMG activity during flexion correctly
classified 89% of subjects.

Triano et al3 Loss of FRP in 39/40 subjects

Peach and McGill 17 Loss of FRP

Arena et al 19 Control ⫽ 29 Evaluated EMG levels during Discriminant function analysis correctly
Spondyloarthritis ⫽ 19 sitting, standing, bending and classified 52.4% of subjects.
Disk ⫽ 52 rising in various populations ANOVA analysis found no differences in
Nonspecific ⫽ 66 of LBP patients bilateral asymmetry between groups, controls
Combined ⫽ 17 had less activity during stance than all other
Other ⫽ 24 groups.
Different groups showed differences in EMG
activity levels depending on movement task.

Sihvonen et al2 Controls ⫽ 25 Performed forward bending, Did not quantify differences between groups
LBP ⫽ 87 (52 had no extension and stance during stance.
pain during testing) 41% of all patients lacked FRP (31 total)
Only 10% of pain free showed no FRP.
Patients differed from controls in ratio of EMG
activity found during flexion and extension.

Alexiev9 LBP ⫽ 40 EMG collected during static No significant bilateral differences found at 0%
Control ⫽ 40 exertions of 50%, 30% and 0% in either group.
of maximum voluntary Significant differences in bilateral EMG at 30%,
contraction. 50% & 100% MVC in LBP group but not in
controls.

Leach, Owens, and LBP ⫽ 10 Collected paraspinal EMG Significant differences between groups in ratio
Giesen10 non-LBP ⫽ 6 during, stance, forward of thoracic ES activity and lumbar ES activity
flexion-extension and left/right asymmetry at lower ES.
Loss of FRP in LBP group

Watson18 Control ⫽ 2 Performed forward flexion and
LBP ⫽ 70 extension. Developed a
flexion/relaxation ratio
comparing peak activity during
flexion with that found during
maximally flexed position.
No differences between groups during stance.
Patients had a significantly higher activity
during the peak flexion posture and
significantly lower flexion/relaxation ratio.
Discriminant analysis revealed sensitivity of
75% and sensitivity of 93%.

Grabiner, Koh, and LBP ⫽ 7 EMG activity recorded during Significant differences were found between
Ghazawi11 Control ⫽ 6 isometric exertion at 2 trunk groups in bilateral coupling.
angles

FRP, Flexion relaxation phenomenon; ES, erector spinae; ROM, range of motion.
296 Lehman
Clinical Use of PMG
activity was conducted during 5 trials of quiet standing.
Each quiet-standing trial lasted 5 seconds and each trial was
preceded by 5 seconds of stepping forward and backward.
Disposable bipolar Ag-AgCl disc surface electrodes with
a diameter of 1 cm were adhered bilaterally over the erector
spinae musculature. Raw EMG was band-pass filtered (10
and 500 Hz) and analogue-to-digital (A/D)– converted at
1024 Hz. The signal was then full-wave rectified, and low-
pass filtered (Butterworth filter cut-off frequency of 2.5Hz)
to produce a linear envelope. The signals were then nor-
malized to the amplitude obtained during the submaximal
reference task and the peak amplitude obtained during the
maximum voluntary contraction (MVC) tasks.
Normalization tasks procedure. The different normalization pro-
cedures occurred on the same task session as the quiet-
standing trials. The MVC tasks occurred before the stance
trials, and the 4 trials of the submaximal task occurred after
the tasks. The MVC required the subject to lie prone on a
bench with his or her trunk over the ledge while the legs
were strapped to the table. The subject was then required to
perform a maximum extension effort against the resistance
of 1 or 2 experimenters. Two MVCs were performed with 3
minutes of rest between each exertion. The maximum ac-
tivity found was used as the reference amount. The sub-
maximal task required the subject to hold a 5-kg weight
suspended from 45-cm chain and bar 1 cm off the floor
while maintaining a lordotic back with a slight bend at the
knee and a greater bend at the hips in a standing position.
The EMG signals were processed in the same manner as
those from the standing trials.
Normalization techniques. The raw data during stance was nor-
malized using 3 different techniques. The data was first
expressed as a percent of the maximum activity found
during the maximum exertion (% MVC). The second nor-
malization technique expressed the data as a percent of the
peak activity found during one submaximal reference task.
The third normalization technique expressed the raw stance
data as a percent of average EMG activity during the four
submaximal tasks. It should be stressed that the same-day
stance data was normalized to the same-day normalization
task data.
Data analysis. A 1-way random-effects model calculated the
single measure intraclass correlation coefficient (ICC) for
the quiet-standing EMG values for the 4 normalization
conditions: (1) raw, (2) percent MVC, (3) percent submaxi-
mal, and (4) percent average of the EMG average during
submaximal tasks. In addition, the repeatability of the sym-
metry between bilateral muscle groups was also assessed
between days. The symmetry between bilateral muscle
groups was quantified by first using a symmetry transform.
The transform allows the comparison of ratios between left
and right muscle groups across persons when ratio values of
⬎1 or ⬍1 are produced. A transformed score of 0 indicates
perfect symmetry. A transformed score of ⫹1 indicates the
right muscular activity was twice as great as the left,
Journal of Manipulative and Physiological Therapeutics
June 2002
whereas a transformed score of –1 indicates the left mus-
cular activity was twice as great as the right. After all
transformed scores were calculated, the absolute value of
each score was used to find the average ratio for the group.
To find the transformed score, the right erector spinae
activity was first divided by the left erector spinae activity.
If the ratio was ⬎1, then 1 was subtracted from the ratio to
give the transformed score. If the ratio was ⬍1, then the
following formula was used: (1/ratio – 1) ⫻ (–1).
Study 2: Asymmetry in Paraspinal EMG Amplitude During Quiet Stance
Patients. Seventeen subjects with back pain (14 with LBP,
3 with thoracic pain below T4) were referred from the
University of Waterloo-Canadian Memorial Chiropractic
College (UW-CMCC) Research Clinic. Nine subjects were
male (height ⫽ 171 ⫾ 5 cm. weight ⫽ 73.5 ⫾ 8.04 kg, age
⫽ 26.6 ⫾ 13 years) and 8 subjects were female (height ⫽
164.25 ⫾ 7.5 cm, weight ⫽ 57.6 ⫾ 6.4 kg, age ⫽ 20.1 ⫾
1.5 years). All read and signed information and consent
forms approved by the University of Waterloo Office of
Research Ethics.
Task protocol. Subjects were examined by the clinic chiro-
practor. A segment exhibiting pain and a therapist-per-
ceived motion restriction was identified by manual palpa-
tion. We recognized the concerns regarding the reliability of
motion palpation; therefore pain and local tenderness were
the deciding factor in determining the site for electrode
placement. Disposable bipolar Ag-AgCl disc surface elec-
trodes with a diameter of 1 cm were adhered bilaterally over
the erector spinae musculature at the painful segment level
and at a nonpainful and nonrestricted segment, 5 segments
superior or inferior to the painful segment. Typically, the
nonpainful segment was superior to the painful segment.
Only when the painful and restricted segment was above T7
was a nonpainful segment located inferiorly (3 occurrenc-
es). EMG signals were collected for 20 seconds during quiet
standing 2 times and 4 minutes apart.
EMG processing. Raw EMG was band-pass filtered (10 and
500 Hz) and A/D converted at 1024 Hz. The amplifier had
a CMRR of 10,000:1. The signal was then full-wave recti-
fied, and low-pass filtered (filter cut-off frequency of 2.5Hz)
to produce a linear envelope.
Statistical analysis. After the average activity of the linear
envelope during quiet stance was calculated, a symmetry
ratio could be calculated. The method for calculating the
symmetry appears in the “methods section” for study 1.
After all transformed scores were calculated, the absolute
value of each score was used to find the average ratio for the
group. The absolute values of the symmetry ratios for the
painful segment and for the nonpainful segment were then
compared using a 1-way analysis of variance.
Study 3: Asymmetry of Paraspinal EMG During Dynamic Tasks
Patient characteristics. The data analysed in this study had been
previously collected in 2 separate previous studies. The first
Journal of Manipulative and Physiological Therapeutics
Volume 25, Number 5
study7 investigated lumbar kinematics and associated trunk
muscle EMG in a healthy population. Data from 14 random
subjects from the study’s 24 subjects (17 men [age ⫽ 22 ⫾
3.4 years, height ⫽ 178.1 ⫾ 5.1 cm, and weight ⫽ 77 ⫾ 6.7
kg] and 7 women [age ⫽ 20.4 ⫾ 0.5 years, height ⫽ 166.1
⫾ 4.4 cm, and weight ⫽ 71 ⫾ 9.7 kg]) was assessed in the
present study. The second study population was drawn from
a previous study12 that examined trunk kinematics and
associated EMG in a convenience sample of persons with
chronic LBP drawn from the UW-CMCC Research Clinic.
Data from 12 subjects (7 women and 5 men [age ⫽ 29.4 ⫾
3.05 years, height ⫽ 167.7 ⫾ 11.1 cm, and weight ⫽ 70.6
⫾ 13.6 kg]) was assessed. All read and signed information
and consent forms approved by the University of Waterloo
Office of Research Ethics.
Task procedure. With feet shoulder width apart and knees
slightly bent, the subject flexed forward in the sagittal plane
until full flexion was reached with the arms and head
hanging freely. The subjects were instructed to pause for a
moment at full flexion, then rise to an upright posture. The
subjects performed this task 3 times.
EMG collection and processing. Raw EMG signals were collected
during the flexion extension tasks using disposable bipolar
Ag-AgCl disc surface electrodes with a diameter of 1 cm
that were placed parallel to the muscle fibers with a center-
to-center spacing of 3 cm over the following muscles bilat-
erally: thoracic erector spinae (approximately 3-cm lateral
to T9 spinous process); lumbar erector spinae (approximate-
ly 3-cm lateral to L3 spinous process). EMG signals were
sampled at 1024 Hz (Watscope A/D Convertor 12 bit,
Northern Digital Inc, Waterloo, Ontario, Canada) and band-
pass filtered (10 and 500 Hz) for a total of 4 channels. The
amplifier had a common mode rejection ratio of 10,000:1.
The signals were then full-wave rectified and low-pass
filtered (Butterworth filter cut-off frequency of 2.5 Hz).
Statistical analysis. A cross-correlation comparing the left and
right erector spinae dynamic EMG profile (upper and lower)
for the 3 flexion trials was used. The correlation numbers
produced for the populations with and without with LBP
were assessed for differences using a 1-way analysis of
variance.
RESULTS
Study 1
Repeatability was acceptable regardless of the normaliza-
tion procedure. Each normalization improved the repeat-
ability of the measured EMG signal relative to unnormal-
ized data. It should also be noted that although repeatability
was deemed statistically acceptable, within- and between-
variability still existed. The average standard deviation of
EMG activity across all days expressed as a percentage of
the mean was 17.3% for raw EMG, 20.9% when normalized
to MVC, 17.8% when normalized to a submaximal refer-
Lehman 297
Clinical Use of PMG
ence task, and 17.2% when normalized to the average value
found during repeated submaximal reference tasks.
Asymmetry in muscle activity was present regardless of
normalization procedure. On average, differences between
the left and right side ranged from 65% to 74% (SD ⫽ 54%
to 60.9%; Tables 2 and 3). The ICCs for symmetry between
the 2 sides showed less repeatability than the other variables
measured (ICC ⫽ 0.43-0.773). Normalization to a maxi-
mum voluntary contraction did not improve the repeatabil-
ity of symmetry scores relative to using raw EMG. How-
ever, the normalization procedures using submaximal ref-
erence tasks improved repeatability relative to raw EMG
values.
Study 2
The mean absolute-transformed ratio score was .51 (SD
⫽ .36) for the segments exhibiting pain and .69 (SD ⫽ .64)
for the segment with no pain. There were no statistically
significant differences between the 2 segments of interest.
Study 3
For the upper erector spinae, the mean cross-correlation
coefficient healthy population was .86 (SD ⫽ .06), whereas
the coefficient for the group with LBP was .85 (SD ⫽ .07).
No significant difference was found between these two
groups for this muscle study. However for the lower erector
spinae, a statistically significant difference in cross-correla-
tion coefficients was found between groups. The population
without LBP had an average score of 0.93 (SD ⫽ 0.02),
whereas the population with LBP averaged a score of 0.86
(SD ⫽ 0.08).
DISCUSSION
Normalization techniques did improve the repeatability
of the EMG signal; however, all trials, normalized and
non-normalized, demonstrated excellent repeatability of the
EMG signal, with ICCs above 0.75. This suggests that
normalization is not necessary to compare the results of 1
person with himself over time for quiet-stance trials. How-
ever, without normalization it is still difficult to compare
across persons and to allow for a biologically relevant
interpretation of the EMG signal as a result of the various
factors, other than muscle activation, that modulate EMG
signal amplitude. A large between-subject variability in
EMG amplitude has been demonstrated in previous studies,8
as has the risk of non-normalization in proper biologic
interpretation of EMG signals.12 An interesting observation
was the variability of the EMG signal about the mean during
within-day trials. This suggests the importance of conduct-
ing multiple trials to average the signal across trials. This
variablility is most likely a result of the relatively low
activity during stance and the way in which this activity can
be easily influenced by small changes in posture and sub-
sequent stability demands on the erector spinae. Subjects
298 Lehman
Clinical Use of PMG
Table 2. ICC scores for erectors spinae activity during quiet
stance
Normalization Technique Right ES Left ES
None raw amplitude 0.78 0.87
% MVC 0.88 0.926
% submaximal MVC 0.82 0.925
% average submax MVC 0.84 .93
ES, Erector spinae; MVC, maximum voluntary contraction.
also showed a high asymmetry score, indicating that, on
average, one muscle was at least 60% more active than the
other in an a population without LBP.
This study is limited to clinical applicability because the
relevance of these measures to dysfunction is still unknown.
However, the results imply that abiding to a stringent test
protocol can result in reproducible measurements across
days; therefore it is possible to track a patient through
rehabilitation program using these measures.
Study 2 demonstrates that in terms of amplitude asym-
metry, problematic spinal segments are no more asymmet-
rical than pain-free segments in a population with LBP. An
argument may be made that the 5 segments between muscle
sites is too close to allow for a problematic/non-problematic
comparison. This is a just argument, considering the muscle
evaluated by the EMG electrodes, longissimus thoracis,
crosses more than 5 segments. It is important to note that
smaller segmental muscles are not measured by the elec-
trodes since the electrode pick-up depth is 1 cm.13 With this
anatomical consideration, it may be impossible to detect
segmental dysfunction (ie, discriminating L4 from L5) with
the use of surface or scanning EMG. This study questions
the validity of using a simple static posture (stance) to
identify muscle dysfunction. However, clinicians should be
cautious in concluding that surface EMG provides no in-
sight into abnormal muscle function. The simplicity of the
test protocol rather than the EMG tool may be at fault. To
properly exploit surface EMG measurements and EMG
amplitude ratios between muscle groups, additional exper-
imental rigor and more challenging static exertions may be
necessary. Edgerton and colleagues14 have acquired a data-
base of muscle amplitude ratios during 7 “static end range
positions” and proposed a theoretic model that may be
useful in detecting hyperactive, hypoactive, and compensa-
tory muscle function by assessing 14 separate muscle
groups. This protocol is exceedingly more complex than the
protocols evaluated in this study and is typically used with
scanning EMG equipment. However, it may be this com-
plexity that is essential to adequately assess muscle dys-
function with EMG protocols.
Study 3 showed significant differences between pain and
pain-free populations in the phasic asymmetry of the EMG
signal for the lower erector spinae but not for the upper
erector spinae. The reason for the difference may be a lack
Journal of Manipulative and Physiological Therapeutics
June 2002
Table 3. Average Symmetry score across the population on all
three different experiment days
Normalization Technique Asymmetry Score and ICC
None raw amplitude .65 (.609) ICC ⫽ .59
% MVC .66 (.54) ICC ⫽ .43
% submaximal VC .74 (.56) ICC ⫽ .67
% average submaximal VC .67 (.54) ICC ⫽ .773
ICC, Intraclass correlation coefficient; MVC, maximum voluntary con-
traction; VC, voluntary contraction.
of dysfunction present at the upper erector spinae, consid-
ering the pain population was presenting with LBP. With no
fewer than 7 segments between the muscle sites, different
fascicles would be recruited to perform the same bending
task. Study 2 only had 5 segments between possible muscle
sites, resulting in the recruitment of the same muscle fasci-
cles during stance. Nonetheless, the differences between
populations are noteworthy, but are they clinically signifi-
cant? What does a cross-correlation of 0.86 mean in the
symptomatic population compared with a cross-correlation
coefficient of 0.93 in the control group? How dysfunctional
is this difference? Once again, the clinical relevance for
rehabilitation is unknown.
CONCLUSION
EMG signals during quiet stance show excellent repeat-
ability; segmental differences in asymmetry between dys-
functional and nonproblematic segments are not evident,
which suggests that the diagnostic validity of EMG evalu-
ations during simple quiet-stance tasks is highly suspect.
Bilateral amplitude asymmetry appears not to be indicative
of pathology but rather a natural phenomenon in both symp-
tomatic subjects and controls. This null finding was con-
trasted with the observation that differences in phasic asym-
metry during dynamic flexion were found between symp-
tomatic and control populations in the lower erector spinae
EMG linear envelope but not in the upper erector spinae.
The implications of these findings are unclear but suggest
that patterns of muscle activation rather than mere ampli-
tude comparisons may better identify pathology.
ACKNOWLEDGMENTS
The author gratefully acknowledges the financial support
of the Natural Sciences and Engineering Research Council
(NSERC) under SM McGill’s research grant.
REFERENCES
1. Ahern D, Follick M, Council J, Laser-Wolston N. Reliabil-
ity of lumbar paravertebral EMG assessment in chronic
low-back pain. Arch Phys Med Rehabil 1986;67:762-5.
2. Sihvonen T, Partanen J, Hanninen O, Soimakallio S. Elec-
tric behaviour of low back muscles during lumbar pelvic
Journal of Manipulative and Physiological Therapeutics
Volume 25, Number 5
rhythm in low-back pain patients and healthy controls.
Arch Phys Med Rehabil 1991;72:1080-7.
3. Triano JJ, Schultz AB. Correlation of objective measures of
trunk motion and muscle function with low-back pain dis-
ability ratings. Spine 1987;9:561-5.
4. Ahern D, Follick M, Council J, Laser-Wolston N, Litchman
H. Comparison of lumbar paravertebral EMG patterns in
chronic low-back pain patients and non-patient controls.
Pain 1988;34:153-60.
5. Spector B. Surface electromyography as a model for the
development of standardized procedures and reliability test-
ing. J Manipulative Physiol Ther 1979;2:214-22.
6. Nouwen A, Bush C. The relationship between paraspinal
EMG and chronic low-back pain. Pain 1984;20:109-23.
7. Nouwen A, Akkerveeken P, Versloot J. Patterns of muscu-
lar activity during movement in patients with chronic low-
back pain. Spine 1987;12:777-82.
8. Sherman RA. Relationships between strength of low back
muscle contraction and reported intensity of chronic low-
back pain. Am J Phys Med 1985;64:190-200.
9. Alexiev AR. Some differences of the electromyographic
erector spinae activity between normal subjects and low-
back pain patients during the generation of isometric trunk
torque. Electromyogr Clin Neurophysiol 1994;34:495-9.
10. Leach R, Owens E, Giesen J. Correlates of myoelectric
asymmetry detected in low-back pain patients using hand
held post style surface electromyography. J Manipulative
Physiol Ther 1993;16:140-9.
11. Grabiner MD, Koh TJ, El Ghazawi A. Decoupling of
Lehman 299
Clinical Use of PMG
bilateral paraspinal excitation in subjects with low-back
pain. Spine 1992;17:1219-23.
12. Lehman GJ, McGill SM. The importance of normalization
in the interpretation of surface electromyography: a proof
of principle. J Manipulative Physiol Ther 1999;22:444-6.
13. Fuglevand AJ, Patla AE, Winter DA, Stashuk D. Detection
of motor unit action potentials with surface electrodes:
influence of electrode size and spacing. Biol Cybern 1992;
67:143-53.
14. Edgerton VR, Wolf SL, Levendowski, Roy RR. Theoretical
basis for patterning EMG amplitudes to assess muscle
dysfunction. Med Sci Sport Exerc 1996;28:744-51.
15. Kravitz E, Moore M, Glaros A. Paralumbar muscle activity
in chronic low-back pain. Arch Phys Med Rehab 1981;62:
172-6.
16. Miller D. Comparison of electromyographic activity in the
lumbar paraspinal muscles of subjects with and without
chronic low-back pain. Phys Ther 1985;65:1347-54.
17. Peach JP, McGill SM. 3-dimensional kinematics and trunk
muscle myoelectric activity in the young lumbar spine: a
database. Arch Phys Med Rehabil 1998;79:663-9.
18. Watson PJ, Booker CK, Main CJ, Chen ACN. Surface
electromyography in the identification of chronic low-back
pain patients. Clin Biomechanics 1997;12:165-71.
19. Arena J, Sherman R, Bruno G, Young T. Electromyo-
graphic recordings of 5 types of low-back pain subjects and
non-pain controls in different positions. Pain 1989;37:57-
65.