American Journal of Sports

Medicine
http://ajs.sagepub.com

Relative Balance of Serratus Anterior and Upper Trapezius Muscle Activity During Push-Up Exercises
Paula M. Ludewig, Molly S. Hoff, Erin E. Osowski, Shane A. Meschke and Peter J. Rundquist
Am. J. Sports Med. 2004; 32; 484
DOI: 10.1177/0363546503258911

The online version of this article can be found at:

http://ajs.sagepub.com/cgi/content/abstract/32/2/484

Published by:

http://www.sagepublications.com

On behalf of:

American Orthopaedic Society for Sports Medicine

Additional services and information for American Journal of Sports Medicine can be found at:

Email Alerts: http://ajs.sagepub.com/cgi/alerts

Subscriptions: http://ajs.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations (this article cites 20 articles hosted on the

SAGE Journals Online and HighWire Press platforms):
http://ajs.sagepub.com/cgi/content/abstract/32/2/484#BIBL

Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008

© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
10.1177/0363546503258911

Relative Balance of Serratus Anterior

and Upper Trapezius Muscle Activity
During Push-Up Exercises
Paula M. Ludewig,* PhD, PT, Molly S. Hoff, MS, PT, Erin E. Osowski, MS, PT,
Shane A. Meschke, MS, MA, PT, ATC, and Peter J. Rundquist, PhD, PT
From the Program in Physical Therapy, The University of Minnesota,
Minneapolis, Minnesota

Background: Serratus anterior strengthening is used in prevention and treatment programs for poor scapular control. In certain
clinical cases, exercises substantially activating the serratus with minimal upper trapezius activation are preferred.
Hypothesis: The standard push-up plus would show both the highest serratus anterior activation and lowest upper trapezius/
serratus ratios for both groups and all phases.
Study Design: Controlled laboratory study.
Methods: Thirty subjects, grouped as healthy or with mild shoulder dysfunction, were evaluated performing standard push-up
plus exercises and modifications on elbows, knees, and against a wall. Surface electromyography of the serratus anterior and
upper trapezius was compared between exercises.
Results: Both groups responded similarly across exercises. The standard push-up plus demonstrated the highest activation of
the serratus (to 123%) and lowest trapezius/serratus ratios (<0.2) during plus phases. The wall push-up plus and phases of other
exercises demonstrated higher upper trapezius/serratus ratios (to 2.0).
Conclusions: In clinical cases where excess upper trapezius activation or imbalance of serratus and trapezius activation occurs,
the push-up plus is an optimal exercise. Other cases may benefit from a progression of modified push-up exercises.
Clinical Relevance: Clinical selection of exercises for improving scapular control should consider both maximum serratus acti-
vation and upper trapezius/serratus anterior ratios.
Keywords: shoulder; rehabilitation; electromyography; scapula; shoulder exercise

The lower and middle divisions of the serratus anterior
(SA) muscle are key contributors to normal and abnormal
scapular motion and control.5,9 The SA’s insertion into the
scapular vertebral border and inferior angle results in
larger moment arms for production of scapular upward
rotation and posterior tipping than any of the other mus-
cles linking the scapula and thorax.5 Thus, the SA has been
described as the prime mover of the scapula.5 The SA is
also unique among the scapulothoracic muscles in that it
has the ability to contribute to all components of the nor-
* Address correspondence to Paula M. Ludewig, Program in Physical
Therapy, 420 Delaware St., MMC 388, The University of Minnesota,
Minneapolis, MN 55455 (e-mail: ludew001@umn.edu).
This work was previously presented as a poster presentation at the
Combined Sections Meeting of the American Physical Therapy
Association, Boston, MA, February 2002.
The American Journal of Sports Medicine, Vol. 32, No. 2
DOI: 10.1177/0363546503258911
© 2004 American Orthopaedic Society for Sports Medicine
mal three-dimensional (3-D) movement of the scapula on
the thorax during elevation of the arm.16 Specifically, this
muscle can produce scapular upward rotation, posterior
tipping, and external rotation (Fig. 1) while stabilizing the
vertebral border and inferior angle of the scapula to the
thorax, preventing scapular “winging” (Fig. 2A). Normal
3-D motion of the scapula during humeral elevation is crit-
ical to elevation of the acromion for preservation of the sub-
acromial space, maintenance of normal glenohumeral
rhythm, maximizing the length tension relationship of the
deltoid, and maximizing available humeral motion relative
to the thorax.9,15,16
The importance of the SA is further evidenced by the
presence of abnormal muscle activation in various shoul-
der pathologies.6,15,22,27 Reduced SA electromyographic
(EMG) activity has been demonstrated in throwers with
glenohumeral instability,6 construction workers with
shoulder impingement,15 and swimmers with shoulder
pain.27 This reduction in muscle activity has been related
to abnormal scapular motion.15 Therefore, the SA is

Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008

484
© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
Vol. 32, No. 2, 2004 Muscle Activity During Push-Up Exercises 485

Figure 1. Scapular motions. Modified and reprinted with per-

mission from Ludewig PM, Functional shoulder anatomy and
biomechanics. Solutions to Shoulder Disorders, HSC 11.1.1.
La Crosse, WI, Orthopaedic Section, APTA, Inc, 2001.

increasingly a focus of therapeutic exercise protocols for

prevention and rehabilitation of shoulder dysfunction.
Other authors have suggested shoulder pathology and
A
abnormal scapular motion may be linked to imbalances of
muscle activity rather than global weakness of scapulotho-
racic muscles.25 In particular, excess activation of the upper
trapezius (UT) has been proposed as contributing to abnor-
mal scapular motion.15,25 In some clinical patients, excess
UT activity may attempt to compensate for a weak SA and
is believed to contribute to impingement through abnormal
rotation of the scapula.15 Evidence of increased UT activa-
tion combined with reduced SA activation as well as evi-
dence of increased superior translation of the scapula has
been demonstrated in persons with shoulder pain.15,17,22
Imbalances of force production of the SA and UT can result
in a shoulder-shrugging motion (Fig. 2B) causing excess
superior translation of the scapula with less efficient
upward rotation and reduced posterior tipping. Clinical
consequences of these alterations can include subacromial
impingement, associated subacromial bursitis, and rotator
cuff or biceps tendinitis that can progress to rotator cuff
tears, as well as altered acromioclavicular joint forces and B
possible predisposition to degenerative changes. Sub-
sequently, in patients with a UT/SA imbalance, rather than Figure 2. Subjects demonstrating excessive scapular wing-
exercises that globally activate several scapulothoracic ing and poor scapular control on the thorax (A) and shoulder
muscles, selective activation of the SA while minimizing shrug during attempted humeral elevation leading to exces-
activation of the UT (low ratio of UT/SA activation) may be sive superior scapular translation and abnormal scapular
advantageous, allowing selective SA strengthening to rotation contributing to impingement (B).
reduce the imbalance.
Previous investigators have explored the activation of
the SA as well as other shoulder muscles during a variety found the interpretation of EMG including changes in mus-
of shoulder exercises.3,4,8,12,18,21 The majority of these past cle length, velocity, or type of contraction across exercise
investigators have advocated exercises that produce sub- conditions.7 Changes in muscle length, velocity, or type of
stantial activation of a number of muscles surrounding the contraction have been shown to alter the magnitude of
shoulder girdle, resulting in global strengthening pro- EMG signal even when force is held constant.7
grams for scapulothoracic muscles, including the trapezius. Subsequently, if these factors are not controlled for,
These past studies have investigated only healthy subjects, changes in EMG amplitude across exercises may be due to
and many have not controlled for all factors that can con- these confounding factors rather than improved force pro-

Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008

© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
486 Ludewig et al
duction for a particular exercise, and clinical interpretation
of exercise comparisons may not be possible.
Several past authors have explored EMG activation
specifically during push-up exercises, although with differ-
ing rationale and procedures for their investigations.3,4,12,21
Moseley et al included the standard push-up plus (SPP) as
a core exercise in their shoulder rehabilitation program
based on the high EMG activity of several shoulder mus-
cles, including the SA.21 Decker et al reported the highest
average SA activation during the plus phase of an SPP
as compared to a number of rehabilitation exercises.4
Modifications to the SPP are commonly used clinically,
including a knee push-up plus (KPP), elbow push-up plus
(EPP), and wall push-up plus (WPP). These modified push-
ups are believed to be less demanding alternatives for clin-
ical use, generally advocated earlier in a rehabilitation pro-
gram when many patients may not be able to repetitively
perform the SPP. Only Decker et al investigated any of
these push-up modifications in their study.4 These authors
reported reduced but still high SA activation during the
KPP as compared to the SPP exercise.4 We are unaware of
any past investigations of SA activation during EPP or
WPP exercises. Several of these past authors also captured
UT data during their investigations of push-up exercises,
although they did not report phase-specific magnitudes of
UT activation or UT/SA ratios.3,4,12,21
The purpose of this study was to compare push-up exer-
cises for their ability to maximally activate the SA and to
maximally activate the SA while minimally activating the
UT. In addition, we further intended to expand on previous
research by (1) considering overall SA activation as well as
the ratio of UT to SA activation when comparing exercises;
(2) controlling for confounding variables of length, type,
and velocity of contraction; and (3) including subjects with
shoulder dysfunction in the data analysis. Our hypothesis
was that for both the healthy and shoulder dysfunction
groups, the SPP exercise during the plus phase would show
both the highest SA activation and the lowest UT/SA ratio.
METHODS
Subjects
Thirty subjects were evaluated from each of two groups
(Table 1). All were between 18 and 50 years of age and were
recruited through verbal and posted announcements.
Group 1 consisted of subjects with normal, healthy shoul-
ders without a history of shoulder pain, trauma, fracture,
dislocation, or surgical procedure (n = 19). Group 2 consist-
ed of subjects with shoulder pain or dysfunction for which
SA-strengthening exercises would be advocated clinically
(n = 11). Diagnoses included shoulder impingement or rota-
tor cuff tendinitis, scapular winging, muscle sprain/strain,
past acromioclavicular separation, and general shoulder
laxity. These subjects presented with mild shoulder pain or
dysfunction. The original onset of symptoms averaged 4.5
years previous, with a range of 1 to 11 years. Subjects
reported chronic persistent symptoms with repetitive over-
head or athletic activities involving the upper extremities.
Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008
© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
The American Journal of Sports Medicine
TABLE 1
Average Subject Demographic Dataa
Group 1 Group 2
Age 25.2 (3.7) 26.8 (6.1)
Height (m) 1.72 (0.08) 1.78 (0.10)
Weight (kg) 71.5 (13.1) 74.6 (12.9)
Males 7 6
Females 12 5
a
Standard deviations are in parentheses.
Patients did not report pain at rest, but symptoms could be
reproduced during clinical provocation tests (impingement
tests, resisted shoulder motions, laxity tests). Most sub-
jects’ (7 of 11) onset of symptoms was due to cumulative
rather than acute trauma. SA-strengthening exercises are
advocated in these patient populations due to documented
alterations in scapular motion and position, demonstrated
muscle weakness, and theorized muscle imbalances.4,15,17
Subjects with history or clinical exam revealing pain or
dysfunction that substantially limited shoulder motion or
led to gross instability of the shoulder with daily activities
or joint laxity testing were excluded. Additional exclusion
criteria were signs and symptoms of cervical pain, adhesive
capsulitis, thoracic outlet syndrome, or a current complaint
of numbness or tingling in the upper extremity. Inclusion/
exclusion criteria were assessed with a clinical screening
examination. Prior to participation, all subjects read and
signed the university-approved human subjects consent
form. The investigation was approved by the University
Institutional Review Board.
Instrumentation
EMG data were collected using silver-silver chloride pre-
amplified (×35) surface electrodes with a 20-mm interelec-
trode distance and an 8-mm diameter. The EMG amplifier
had an input impedance of 15 megohms at 100 Hz, a com-
mon mode rejection ratio of 87 dB at 60 Hz, and a band-
width of 20 to 4000 Hz (Therapeutics Unlimited, Iowa City,
Iowa). The root mean square (rms) time constant was set at
25 milliseconds. Processed signals were collected online to
a microcomputer at a sampling rate of 300 Hz. An oscillo-
scope (Tektronix, Wilsonville, Oregon) was used to verify
the quality of raw signals.
The Polhemus FASTRAK (Polhemus Inc., Colchester,
Vermont) 3-D electromagnetic motion-capturing system
was used to capture shoulder motion and position data. It
consists of a transmitter, digitizing stylus, small electro-
magnetic sensors, and a systems electronics unit. The sty-
lus was used to manually digitize anatomical coordinates.
The systems electronics unit converts induced voltage data
from the sensors to a 3-D position and orientation at a 40-Hz
sampling rate. Computer software synchronized the collec-
tion of EMG and kinematic data. Within a source-to-sensor
separation of 76 cm, reported accuracy of the system is 0.3
to 0.8 mm (rms) to less than 1 mm (absolute) for position
and 0.15° (rms) to 0.5° (absolute) for orientation.23
Vol. 32, No. 2, 2004 Muscle Activity During Push-Up Exercises 487

Experimental Procedure

The EMG recording electrodes were placed on the subject’s

dominant arm for the healthy subjects or involved arm for
the subjects with shoulder dysfunction. Electrode prepara-
tion included shaving hair from the electrode placement
area if present, an alcohol wipe over the skin surface, and
use of conductive gel on the electrode’s active surface. No
skin abrasion was necessary due to the high input imped-
ance of the amplifier.28 One surface electrode was applied
to the SA parallel to the muscle fibers, below the axilla,
anterior to the latissimus dorsi, and posterior to the pec-
toralis major (Fig. 3).4,12,15,16 Correct electrode placement
was verified through observation of the oscilloscope during
resisted scapular protraction.11 In addition, resisted shoul-
der extension was observed to limit potential cross talk
from the latissimus dorsi.11 A second surface electrode was
applied to the UT parallel to the muscle fibers, two thirds
of the way between the spinous process of the seventh cer-
vical vertebrae and the acromion process (Fig. 3).10 The ref-
erence electrode was applied to the opposite ankle on the
lateral malleolus.
To normalize EMG data, subjects performed maximal
voluntary contractions (MVC) of the SA and UT against
manual resistance using previously documented proce-
dures.11,15,26 The SA MVC was performed with the subject
standing and the shoulder and elbow flexed to 90°.
Maximal resistance was applied by the investigator
through the elbow against scapular protraction.11,15 The
UT MVC was performed with the subject seated, arm
flexed to 90°, and elbow fully extended. Maximal resistance
was applied to shoulder flexion superior to the elbow.26 We
normalized the UT and SA in shoulder flexion positions of
approximately 90° because the push-up exercises included
in our study required shoulder flexion in this range.28
Furthermore, a previous investigation demonstrated that a
resisted shoulder flexion normalization produces a higher
UT activation than a shoulder shrug.26 A single examiner
performed all normalization procedures. Figure 3. EMG electrode and FASTRAK sensor placement.
A FASTRAK position sensor was applied to the sternum EMG surface electrodes were placed over the upper trapez-
inferior to the sternal notch. A plastic cuff containing ius and serratus anterior muscles. Electromagnetic sensors
another sensor was secured to the distal humerus with were attached to the sternum and a cuff on the distal
Velcro straps (Fig. 3). Bony landmarks on the thorax and humerus.
humerus were digitized to allow creation of anatomically
based local coordinate systems for both the trunk and
humeral sensors.15 upper extremity weight borne on the elbows, and lower
All exercises were performed with a target angle based extremity weight borne on the feet in the standard manner.
on visual estimate of 90° of shoulder flexion during the plus The exercise then includes the full shoulder protraction or
phase of the exercises. The exercise velocity was controlled plus phase only of the push-up. The WPP is performed in a
by use of a metronome set at 30 beats per minute, with sub- standing position with the hands in contact with the wall.
jects completing one repetition of each exercise in 4 sec- Positions of the extremities for all exercises were based on
onds. The exercises were performed in a random order, creating a resulting shoulder flexion angle of 90° for the
including SPP, KPP, EPP, and WPP (Fig. 4). plus phase. Subjects were uniformly instructed on per-
The SPP is a standard push-up with the addition of full formance of each exercise by a single examiner and were
shoulder protraction (the “plus”) after obtaining full elbow allowed to practice a few repetitions until the proper
extension at the end of the usual push-up. The KPP was motion and timing were achieved based on visual assess-
performed in the same way as the SPP except that the ment. Each exercise was completed as a set of five repeti-
knees are the distal point of contact with the ground rather tions. Subjects were allowed to rest up to 2 minutes between
than the feet. The EPP begins with the elbows flexed to 90°, exercise sets.

Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008

© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
488 Ludewig et al The American Journal of Sports Medicine

Figure 4. Push-up exercises including standard push-up plus (A), knee push-up plus (B), elbow push-up plus (C), and wall push-
up plus (D).

Data Reduction The estimated humeral head center translates anterior-

The local coordinate systems developed from the digitized
anatomical landmarks for the trunk and humerus allow
clinically relevant motion descriptions.15 The humeral
coordinate system was mathematically transformed to
describe humeral motion relative to the trunk including
plane of elevation, elevation angle, and long axis rotation
(z, y′, z″ Euler angles).1 The estimated humeral center for
each subject was found by passively moving the humerus
to multiple angular positions less than 45° from the neu-
tral rest position and using a least squares algorithm to
identify the humeral pivot point.2 Subsequently, transla-
tions of the estimated humeral head center relative to the
trunk could be tracked throughout each exercise.
ly during the concentric phases of each exercise, particu-
larly during the plus phase (Fig. 5). These translation data
were used to identify phases of the exercises. These phases
included concentric (beginning of each repetition until the
beginning of the plus phase), concentric plus (beginning of
the plus phase until the peak of the plus phase), eccentric
plus (peak of the plus phase until the end of the plus phase)
and eccentric (end of the plus phase until the beginning of
the next repetition) (Fig. 5). High reliability (ICC > 0.86) of
tracking humeral anterior/posterior translations with a
surface sensor has been previously demonstrated.13
The EMG data for each muscle and subject were aver-
aged for each phase across the three intermediate repeti-
tions of the five repetitions completed. A moving window

Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008

© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
Vol. 32, No. 2, 2004
Figure 5. Identification of phases of exercise from humeral
anterior translation data for a push-up. 1, beginning of con-
centric phase; 2, beginning of concentric plus phase; 3,
beginning of eccentric plus phase; 4, beginning of eccentric
phase; 5, end of push-up.
averaging technique was used with the MVC data to iden-
tify the highest value averaged over a 500-millisecond peri-
od as the reference 100% value for each subject. The aver-
aged EMG data for each phase were then expressed as a
percent maximum value (phase EMG voltage value divided
by the MVC EMG voltage value and then multiplied by
100) for each exercise.28 These EMG data collection and
normalization procedures have previously demonstrated
good trial-to-trial reliability (ICC > 0.73) during active
shoulder motions.15
Using the humeral position data, the average humeral
elevation angle relative to the trunk was also determined
for each phase, exercise, and subject. High trial-to-trial
reliability (ICC > 0.93) and validity (rms error < 3°) of these
angular calculations have been previously demonstrat-
ed.14,15 These procedures allowed for experimental and sta-
tistical control of the confounding variables of length,
velocity, and type of contraction.7,28 With humeral angles
controlled, muscle length was maintained consistently. Use
of the metronome controlled velocity of motion and was
verifiable based on the time per cycle data. Separation into
phases ensured that comparisons were made within the
same type of contraction. Examiners were trained and their
ability to obtain consistent data and procedures verified
during pilot testing.
Statistical Analysis
Means and standard deviations were determined across
subjects within each group for the normalized SA EMG
values, UT EMG values, and the UT/SA ratio from each
phase of the four exercises. Statistical analysis was com-
pleted using the NCSS Version 2000 Statistical Program
(Kaysville, Utah). The independent variables were group,
exercise, and phase. The dependent variables of primary
interest were the normalized SA EMG values and the
UT/SA ratio. The sample distributions were tested for nor-
Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008
© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
Muscle Activity During Push-Up Exercises 489
mality, and due to nonnormal distributions, a logarithmic
transformation was performed to achieve normality prior
to further statistical testing.24
The plus portions (concentric plus and eccentric plus
phases) of the exercises were of primary interest, and the
data for each dependent variable were analyzed using a
three-way mixed model analysis of variance with one
between (group) and two within (exercise and phase) fac-
tors and an overall alpha level of .05.24 In the presence of
significant interactions with phase (exercise or group
effects were dependent on the phase), two-way group by
exercise analyses of variance were calculated for each
phase. If significant exercise effects were present, the
hypothesis regarding specific exercise differences was eval-
uated using Bonferroni-corrected planned comparisons
between exercises, comparing each exercise to the SPP.24
For all exercises except the EPP, the concentric and eccen-
tric nonplus phases of the exercises (concentric, eccentric)
were also compared across exercises following the same
statistical approach.
As EMG values can be influenced by changes in muscle
length, the target humeral angle for all exercises was 90° of
humeral flexion during the plus phase of the exercise. To
test for any influence of variation in the actual humeral
flexion angle, Pearson product moment correlations were
assessed between humeral angle and EMG dependent
variables for each phase and exercise. In the presence of
significant confounding of humeral angle on EMG values,
analysis of covariance adjusting for humeral angle effects
was planned. However, the humeral flexion angle did not
account for a significant amount of the EMG variability for
any exercises or phases. The highest correlation was –0.18
between average humeral angle and either dependent vari-
able. As such, it was determined that the experimental pro-
cedures adequately controlled confounding of length
change on EMG values, and humeral flexion angle was not
included as a covariate in the remainder of the analyses.
RESULTS
Differences between exercises for normalized SA EMG and
the UT/SA ratio were dependent on the phase of the exer-
cise (significant phase by exercise interactions, P ≤ .01).
Subsequently, exercise differences were analyzed separate-
ly for each phase. Both groups responded similarly across
exercises, with no significant differences between groups or
interactions with group identified (P > .05, Fig. 6). For the
concentric plus phase, the subjects with shoulder dysfunc-
tion tended to have higher SA EMG values as compared to
the healthy group and equal or lower values for all other
phases, but differences did not reach statistical significance.
Analyses completed for both groups and all phases yield-
ed significant differences between exercises (significant
exercise main effects, P < .01). As hypothesized, normalized
SA EMG activity progressively decreased across push-up
exercises with the highest values for the SPP (Fig. 6A and
B). SPP normalized SA EMG was significantly greater (P <
.017) than all other exercises for all phases except the con-
centric plus phase, where differences between the SPP and
490 Ludewig et al The American Journal of Sports Medicine

140 140

120 120
Normalized SA EMG (%)

Normalized SA EMG (%)

100 100

80 80

60 d 60

40 40

20 20

0 0
SPP KPP EPP WPP SPP KPP WPP
A Exercise B Exercise
0.6 2.5

0.5 2.0

0.4 UT/SA Ratio

UT/SA Ratio

1.5

0.3
1.0
0.2

0.5
0.1

0.0 0.0
SPP KPP EPP WPP SPP KPP WPP

C Exercise D Exercise
Figure 6. Serratus anterior (SA) activation across exercises for concentric (solid lines) and eccentric (dashed lines) plus phases
(A), nonplus phases (B), and upper rapezius/serratus anterior (UT/SA) ratios across exercises for concentric and eccentric plus
phases (C) and nonplus phases (D). Solid symbols are the healthy group data, and clear symbols represent the group of subjects
with mild shoulder dysfunction.

KPP did not reach statistical significance. The magnitude both groups. This exercise also had a low UT/SA ratio for
of differences ranged from 13% to 45%. most phases of the exercises, although many of the other
UT/SA ratios were generally low (<0.3, Figs. 6C and D) tested exercises demonstrated similar low ratios. A lower
for all phases with the exception of the eccentric nonplus UT/SA ratio signified that the SA was highly activated,
phase. UT/SA ratios tended to be higher in the nonplus proportionately to the UT being minimally activated. The
phases due to both reduced SA EMG activity during these SPP exercise appears to be optimal in comparison to the
phases and increases in UT EMG activity (Table 2), tested modified push-up exercises if both maximum SA
although the greatest average UT activity for any phase or activation and a low UT/SA ratio are desired for a SA-
exercise reached only 25%. The hypothesis of lower UT/SA strengthening program.
ratios for the SPP was supported fully (across all exercises) For patients with an imbalance of UT to SA activation,
only for the eccentric nonplus phase (P < .017). In addition, an exercise that demonstrates a low UT/SA ratio would be
the SPP had significantly lower ratios than the WPP for all an important component of rehabilitation to allow selective
phases except the concentric plus phase. During the con- SA strengthening and reduction of the imbalance. All push-
centric plus phase, the SPP UT/SA ratio was significantly up exercises tested except the WPP demonstrated low
greater (P < .017) than for the KPP, although both ratios UT/SA ratios during the plus phases. If low ratios are
were low (below 0.1, Fig. 6C). desired in an exercise program, the WPP would be least
indicated, particularly during the eccentric nonplus phase,
where the UT activation was nearly twice that of the SA.
DISCUSSION Patients with excess UT activation may present clinically
with a shoulder-shrug motion when attempting to reach
Our results were supportive of the hypothesis that the SPP overhead (Fig. 2B). Patients with an UT/SA imbalance
during the plus phase would maximally activate the SA for would also tend to present with less scapular control and

Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008

© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
Vol. 32, No. 2, 2004
TABLE 2
Upper Trapezius Normalized EMG Data
Across Exercises and Phases
Upper
Trapezius
Exercise Phasea EMG (%)
Push-up plus 1 14.5
2 9.2
3 8.0
4 19.2
Knee push-up plus 1 11.2
2 4.6
3 4.2
4 20.1
Elbow push-up plus 1 NA
2 7.6
3 6.2
4 NA
Wall push-up plus 1 11.5
2 5.7
3 12.2
4 25.3
a
Phase 1, concentric; phase 2, concentric plus; phase 3, eccentric
plus; phase 4, eccentric. Note that the elbow push-up plus exercise
does not include nonplus phases.
more scapular winging during shoulder flexion as com-
pared to abduction because the trapezius is less able to
compensate for a weak serratus in flexion.9
Higher UT/SA ratios were present for all exercises in the
nonplus phases, particularly the nonplus eccentric phase
where ratios for all exercises approached or exceeded one.
A ratio of one indicates an equal training stimulus for the
UT and SA. As values approach or exceed this equal stimu-
lus level, they are indicative of exercises that would not
selectively activate or train the SA. If higher UT/SA ratios
are of concern when choosing a particular patient’s exer-
cise program, instructing them in performing only the plus
phases of any of the push-up exercises might be indicated.
However, interpretation of any concern for higher ratios
should be considered in light of the relatively low UT activ-
ity that was present. With the exception of the WPP, the UT
activity during even the nonplus phases remained at 20%
MVC or below. The higher ratios were due in part to the
expected reductions in SA activity in the nonplus phases.
Although EMG activation and load are not directly
interchangeable, past authors have indicated in untrained
subjects that loads less than 66% of maximum do not in-
crease strength even with 150 contractions per day, where-
as with loads greater than 66%, increases in strength can
be obtained each training session, and higher loads can
increase strength with fewer repetitions.19 In general, our
results were supportive of past studies investigating push-
up exercises and further substantiate that the SPP exer-
cise is highly effective for SA activation with levels above
66% MVC except for the eccentric nonplus phase. As
expected, our study and others that separated the plus
phase from the remainder of the push-up or modified push-
up rather than averaging EMG throughout the full push-
Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008
© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
Muscle Activity During Push-Up Exercises 491
up demonstrated higher values for the SA during the plus
phase.4,21 These data support the use of the plus movement
for all push-up exercises in addition to the standard push-
up without a plus as levels of activation during the concen-
tric plus phase were above 80% MVC for all exercises.
Standard In addition to supporting past evidence of high SA acti-
Error
vation during a SPP, our investigation added data for sev-
2.71 eral modified push-up exercises and the UT/SA ratio and
2.24 added data from subjects with mild shoulder dysfunction.
1.56 Although SA activation was reduced as compared to SPP,
2.88 the KPP and EPP were both exercises that demonstrated a
1.85 low UT/SA ratio and relatively high SA activation. These
0.78 modified push-ups may be important alternatives for clini-
0.52 cal use. To perform the SPP, loads greater than 35% of body
2.16 weight can be required on a single upper extremity.29
NA
Higher level athletes with chronic symptoms could likely
1.72
1.35
complete SPP exercises fairly early in a rehabilitation pro-
NA gram; however, many patients may not be able to repeti-
1.53 tively perform the SPP as early in the rehabilitation
1.31 process. The modified push-up exercises could be used pro-
1.37 gressively, beginning with the plus phase of the WPP early
2.48 in a rehabilitation program and progressing to the EPP,
KPP, and eventually SPP. All of these exercises involve
weightbearing on the upper extremities, which has been
advocated to improve proprioception and joint stability as
well as strength.29 Progression of an exercise program
should be based on the ability to perform the exercises
without increased pain or loss of scapular control on the
thorax. We did not quantify scapular position during our
investigation, but visual observation for both groups did
suggest adequate scapular control on the thorax during the
performance of the exercises.
Our data for UT activation are generally consistent with
the findings of both Moseley et al and Decker et al who
reported that UT activation did not meet their threshold
for substantial activation during push-up plus exercises.4,21
Their criteria for substantial activation were reported at
40% and 20% MVC, respectively, indicating that UT activi-
ty was below these levels.4,21 Lear and Gross reported high-
er UT activation (approximately 50% MVC) during a push-
up plus.12 Their UT activation levels may have been higher
due to the method used to determine the maximal normal-
ization contraction.12 Lear and Gross used a manually
resisted seated shoulder shrug that does not produce levels
of activation as high as our resisted shoulder flexion nor-
malization.12,26 Therefore, Lear and Gross’s UT EMG data
are likely referenced to a lower normalization contraction
value, which would result in a higher percentage MVC.12
None of these past authors reported UT/SA ratios or phase-
specific UT activation levels.3,4,12,21
All past investigations of SA and UT EMG activation
during push-up exercises have been completed in healthy
subjects.3,4,12,21 Our investigation added data from subjects
with mild shoulder dysfunction. Statistical analysis
revealed no significant differences in how the groups
responded across exercises. A number of possible factors
may have contributed to this lack of group differences. Our
investigation had adequate statistical power to detect dif-
ferences of 13% to 15%. If differences between groups of a
smaller magnitude were of interest, a larger sample size
492 Ludewig et al
would be required. Also, a lack of homogeneity of diagnoses
in the group with shoulder dysfunction might reduce the
ability to detect differences between groups. However, the
within-group standard deviations were no higher in the
dysfunction group than in the healthy group, indicating
this factor likely did not substantially influence the results.
None of the subjects with shoulder dysfunction was cur-
rently experiencing an acute exacerbation of symptoms, and
all were able to complete the push-up exercises without sig-
nificant pain (<2 on a 10-point scale). This mild dysfunction
status was likely the factor most influencing the lack of
group differences. Generalization of our results to more
acute patients is not recommended without further study.
Despite no statistical differences, there was a tendency
for the subjects with shoulder dysfunction to have higher
values of activation than the healthy group during the con-
centric plus phase of the exercises and equal or lower val-
ues during other phases. Reduced ability to produce force
during the MVC in the group with dysfunction could result
in exercise activation during the more demanding concen-
tric plus phases, being a higher percentage of their MVC.
The less demanding phases would be less likely to show
such differences. However, the same patterns of progres-
sively increasing activation from the WPP to the SPP were
seen in both groups. This supports the potentially effective
use of similar exercise progressions not only for preventive
programs in healthy subjects but also for treatment of
patient populations with mild shoulder dysfunction or in
the later stages of rehabilitation. When planning a rehabil-
itation program, it should always be considered that even
with the same diagnosis, clinical patients do not have iden-
tical pathological presentations, and therefore one exercise
may be better suited for a particular patient even though
the means across subjects might show two exercises to be
similar.
EMG is not a direct measure of muscle force production.
Varying length, velocity, or type of muscle contraction dur-
ing an exercise can confound the EMG force relationship.7
If these confounding factors are controlled as they were in
our study, an indirect measure of force can be obtained,
with higher EMG during one exercise as compared to
another representing higher force production and subse-
quently the potential for greater strength gains.19,28 Past
studies have compared exercises with the arm at very dif-
ferent humeral angles ranging from extension to full over-
head elevation.4,8,18 During elevation of the arm, the high-
est gravitational force demands are known to occur at 90°
of abduction, yet higher EMG for some muscles can be seen
with full abduction due to the confounding influence of
muscle length change on EMG activation.9,16 For studies
where humeral angles vary widely, EMG cannot be consid-
ered representative of higher force production for one exer-
cise as compared to another. With a target angle of 90° of
humeral flexion during the plus phase of the push-up exer-
cises, our data suggest that further precise control of mus-
cle length is not needed to reduce EMG variability for these
push-up activities.
Limitations of our investigation should be noted. Use of
surface electrodes for the SA and UT muscles allowed for
representation of general whole-muscle activity in a reli-
Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008
© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
The American Journal of Sports Medicine
able and noninvasive manner.4 However, all areas (specifi-
cally the upper portions) of the SA are not represented by
the placement of this electrode where the muscle is super-
ficially located.20 Surface SA EMG is also difficult to obtain
in subjects with a high body mass index. Our subjects were
generally slim with no subjects presenting with an exces-
sively high body mass index. Fine wire electrodes provide a
more localized description of muscle activation and are less
susceptible to cross talk but are less reliable and may pro-
duce pain during exercises.28 Another limitation of our
study is that other muscles beyond the SA and UT that can
contribute to scapular control were not considered in this
investigation. This was because the UT and SA are fre-
quently targeted in rehabilitation and are primary con-
trollers of the scapula.4,5,15 In addition, the restricted age
range of our subject population (19 to 35) affects generaliz-
ability outside of this age group. Future investigations may
wish to consider additional muscle groups, larger subject
samples with a broader age range, and a broader sample of
clinical patients.
SUMMARY AND CONCLUSIONS
We investigated the activation of the SA and UT/SA ratios
during SPP exercises and modifications on elbows, knees,
and against the wall. Tested subjects included healthy indi-
viduals and a group with mild shoulder dysfunction.
Subjects with shoulder dysfunction responded similarly to
healthy subjects across exercise conditions. The SPP con-
sistently demonstrated the highest activation of the SA as
well as low UT/SA ratios during all phases of the exercise
except the eccentric nonplus phase. EPP and KPP exercis-
es also demonstrated relatively high SA activity with low
UT/SA ratios. The WPP and eccentric nonplus phase of all
push-up exercises demonstrated higher UT/SA ratios.
In clinical cases of shoulder impingement or scapular
winging where maximum activation of the SA with mini-
mal activation of the UT is desired, the SPP is an optimal
exercise. Other patients not immediately able to perform
the SPP without exacerbating their symptoms may benefit
from a progression of exercises beginning with the plus
phase of the WPP, progressing to the EPP, KPP, and finally
SPP. Selective activation of the SA without high activation
of the UT may improve the relative strength of the SA and
improve the balance of activation between these muscles in
patients with shoulder dysfunction. Proper balance of UT
and SA activation is believed to reduce excess superior
translation of the scapula, improve scapular posterior tip-
ping, and maximize the available subacromial space
beneath the acromion, thus reducing rotator cuff impinge-
ment. Considering both maximum SA activation and UT/SA
ratios should affect the clinical selection of exercises for
improving scapular control.
ACKNOWLEDGMENT
The authors would like to thank Ryan Trebil for his contri-
butions during the pilot phase of the project. This investi-
Vol. 32, No. 2, 2004
gation was supported in part by an equipment grant from
the Minnesota Medical Foundation.
REFERENCES
1. An KN, Browne AO, Korinek S, et al: Three-dimensional kinematics of
glenohumeral elevation. J Orthop Res 9: 143–149, 1991
2. An KN, Korinek SL, Kilpela T, et al: Kinematic and kinetic analysis of
push-up exercise. Biomed Sci Instrument 26: 53–57, 1990
3. Belle RM, Hawkins RJ: Dynamic electromyographic analysis of the
shoulder muscles during rotational and scapular strengthening exer-
cises, in Post M, Morrey BF, Hawkins RJ (eds): Surgery of the
Shoulder. St. Louis, Mosby-Year Book, 1990, pp 32–35
4. Decker MJ, Hintermeister RA, Faber KJ, et al: Serratus anterior mus-
cle activity during selected rehabilitation exercises. Am J Sports Med
27: 784–791, 1999
5. Dvir Z, Berme N: The shoulder complex in elevation of the arm: A
mechanism approach. J Biomech 11: 219–225, 1978
6. Glousman R, Jobe F, Tibone J, et al: Dynamic electromyographic
analysis of the throwing shoulder with glenohumeral instability. J Bone
Joint Surg 70-A: 220–226, 1988
7. Heckathorne CW, Childress DS: Relationships of the surface elec-
tromyogram to the force, length, velocity, and conduction rate of the
cineplastic human biceps. Am J Phys Med 60: 1–19, 1981
8. Hintermeister RA, Lange GW, Schultheis JM, et al: Electromyographic
activity and applied load during shoulder rehabilitation exercises
using elastic resistance. Am J Sports Med 26: 210–220, 1998
9. Inman VT, Saunders JB, Abbott LC: Observations on the functions of
the shoulder joint. J Bone Joint Surg 26A: 1–30, 1944
10. Jensen C, Vasseljen O, Westgaard RH: The influence of electrode
position on bipolar surface electromyogram recordings of the upper
trapezius muscle. Europ J Appl Physiol 67: 266–273, 1993
11. Kendall FP, McCreary EK, Provance PG: Muscles Testing and
Function. Fourth edition. Baltimore, Williams & Williams, 1993
12. Lear LJ, Gross MT: An electromyographical analysis of the scapular
stabilizing synergists during a push-up progression. J Orthop Sports
Phys Ther 28: 146–157, 1998
13. Ludewig PM, Cook TM: Translations of the humerus in persons with
shoulder impingement symptoms. J Orthop Sports Phys Ther 32:
248–259, 2002
14. Ludewig PM, Cook TM, Shields RK: Comparison of surface sensor
and bone-fixed measurement of humeral motion. J Appl Biomech 18:
163–170, 2002
Downloaded from http://ajs.sagepub.com at CAPES on July 31, 2008
© 2004 American Orthopaedic Society for Sports Medicine. All rights reserved. Not for commercial use or unauthorized distribution.
Muscle Activity During Push-Up Exercises 493
15. Ludewig PM, Cook TM: Alterations in shoulder kinematics and asso-
ciated muscle activity in people with symptoms of shoulder impinge-
ment. Phys Ther 80(3): 276–291, 2000
16. Ludewig PM, Cook TM, Nawoczenski DA: Three-dimensional scapu-
lar orientation and muscle activity at selected positions of humeral
elevation. J Orthop Sports Phys Ther 24: 57–65, 1996
17. Lukasiewicz AC, McClure P, Michener L, et al: Comparison of 3-
dimensional scapular position and orientation between subjects with
and without shoulder impingement. J Orthop Sports Phys Ther 29:
574–586, 1999
18. McCann PD, Wooten ME, Kadaba MP, et al: A kinematic and elec-
tromyographic study of shoulder rehabilitation exercises. Clin Orthop
288: 179–188, 1993
19. McDonagh MJN, Davies CTM: Adaptive response of mammalian
skeletal muscle to exercise with high loads. Eur J Appl Physiol 52:
139–155, 1984
20. Moore KL, Dalley AF: Clinically Oriented Anatomy. Philadelphia,
Lippincott Williams & Wilkins, 1999
21. Moseley JB, Jobe FW, Pink M, et al: EMG analysis of the scapular
muscles during a shoulder rehabilitation program. Am J Sports Med
20: 128–134, 1992
22. Peat M, Grahame R: Electromyographical analysis of soft tissue
lesions affecting shoulder function. Am J Phys Med 56: 223–240,
1977
23. Polhemus Inc., 3SPACE FASTRAK User’s Manual, Revision F.
Colchester, VT, 1993
24. Portney LG, Watkins MP: Foundations of Clinical Research,
Applications to Practice. Second edition. Upper Saddle River, NJ,
Prentice Hall Health, 2000
25. Sahrmann S: Diagnosis and Treatment of Movement Impairment
Syndromes. St. Louis, Mosby, 2002
26. Schuldt K, Harms Ringdahl K: Activity level during isometric test con-
tractions of the neck and shoulder muscles. Scand J Rehab Med 20:
117–127, 1988
27. Scovazzo ML, Browne A, Pink M, et al: The painful shoulder during
freestyle swimming: An electromyographic cinematographic analysis
of twelve muscles. Am J Sports Med 19: 577–582, 1991
28. Soderberg GL: Selected Topics in Surface Electromyography for Use
in the Occupational Setting: Expert Perspectives. U.S. Department of
Health and Human Services, Public Health Service, Centers for
Disease Control, National Institute of Occupational Safety and Health.
Washington DC: Government Printing Office, 1992 (DHHS/NIOSH
Publication No. 91-100)
29. Uhl TL, Carver TJ, Mattacola CG, et al: Shoulder musculature activa-
tion during upper extremity weight-bearing exercise. J Orthop Sports
Phys Ther 33: 109–117, 2003