Journal of Sport Rehabilitation, 2009, 18, 502-520
© 2009 Human Kinetics, Inc.
Establishing Normative Data
on Scapulothoracic Musculature
Using Handheld Dynamometry
Nichole Turner, Kristen Ferguson, Britney W. Mobley,
Bryan Riemann, and George Davies
Context: Scapular strength deficits have been linked to shoulder dysfunction. Objec-
tive: To establish normative data on the scapulothoracic musculature in normal sub-
jects using a handheld dynamometer. Design: Descriptive normative data study.
Setting: Field research. Subjects: 172 subjects with varying levels of overhead activ-
ity. Methods: A handheld dynamometer was used to test the upper, middle, and lower
trapezius; rhomboids; and serratus anterior. Main Outcome Measures: A 2-factor
ANOVA was performed for each of the muscles by activity level and unilateral ratio
by activity-level analyses. Post hoc analysis included multiple pairwise comparisons,
using the Dunn-Bonferroni correction method. Results: Activity level did not signifi-
cantly affect the unilateral ratios: Elevation:depression was 2.5:1, upward:downward
rotation was 1.5:1, and protraction:retraction was 1.25:1. A rank order from strongest
to weakest was established through significant comparisons. Conclusion: The unilat-
eral ratios along with the rank order should be considered when discussing scapular
Keywords: handheld dynamometer, scapulothoracic strength, shoulder
Weakness, abnormal positioning, and abnormal timing of the scapular mus-
cles are all contributing factors to scapular dyskinesia. Impairments in scapular
motion can lead to problems such as abnormal stresses on the anterior capsular
structures of the shoulder, increased risk of rotator-cuff compression, and
decreased muscle performance.1,2 Also, dyskinesis influences the position and
decreases the size of the subacromial space, leading to subacromial impinge-
ments.3,4 Changes in scapular motion, such as decreased protraction or imbal-
ances between the upper and lower trapezius, have been reported in patients with
impingement.3–8 Inadequate scapular stabilization has been shown to contribute to
altered biomechanics of the shoulder complex and to increase the risk of musculo-
skeletal problems such as instability and impingement.2–5,7–9
Turner, Ferguson, and Mobley are practicing physical therapists who graduated from Armstrong
Atlantic State University. Riemann is with the Dept of Health Sciences, and Davies, the Dept of
Physical Therapy, at the university.
Normative Data on Scapulothoracic Musculature 503
Research has demonstrated that alterations in scapular kinematics are con-
nected to decreased serratus anterior activity, increased upper trapezius muscle
activity, and imbalances between the upper and lower trapezius.2,7,10,11 Research
has shown that scapular upward rotation, posterior tilt, and external rotation were
decreased in patients with impingement syndrome when compared with healthy
subjects.7 Ludewig and Cook7 also demonstrated that subjects with symptoms of
impingement had significantly more upper trapezius muscle activity and less ser-
ratus anterior activity than a control group. Cools et al5 found that overhead ath-
letes with impingement demonstrated decreased protraction. These studies
underline the importance of scapular kinematics and strengthening exercises in
shoulder rehabilitation protocols.
One way to objectively measure strength in the clinical setting is through the
use of a handheld dynamometer, which is more accurate and less subjective than
manual muscle testing. The intrarater and interrater reliability of handheld dyna-
mometry have been supported in previous studies.12–23 The strength of the rater,
experience, and tester stability all can alter the reliability of the measurement.21,23
However, if these variables are controlled, the handheld dynamometer can be a
valuable assessment tool.
Adequate assessment of the scapulothoracic musculature is essential to
designing better rehabilitation protocols. Research has demonstrated that some
cases of impingement problems have been adequately resolved with rehabilitation
protocols involving scapular muscle reeducation and strengthening exercises.1,24,25
By restoring the normal balance of force couples, physical rehabilitation can
improve the position and motion of the scapula to decrease impingements and also
increase the strength of rotator-cuff muscles.1 Therefore, outcome assessments in
rehabilitation protocols should address these kinematic and muscle-activity altera-
tions to restore normal scapulothoracic and shoulder-complex movements.
Michener et al22 examined the construct validity, reliability, and error of
handheld dynamometry testing of scapular muscles in subjects with shoulder pain
and functional loss. They stated, “At present, the error value of a single-time hand-
held dynamometry scapular muscle measurement is unclear without the establish-
ment of normative values of scapular muscle measurements.”22(p1135) To our
knowledge, no studies have been performed to develop normative strength data on
the scapular muscles using a handheld dynamometer. Therefore, the purpose of
this study was to establish normative data for strength of the scapulothoracic mus-
cles in healthy individuals, as well as examine the effects of overhead-activity
level on these measurements. We hypothesized that those who participate in activ-
ities that require increased scapular stabilization, such as overhead athletes, would
have higher strength values than those who do not.
One hundred ninety subjects were recruited from local high school, college, and
adult sports teams. A cross-section of sports and activities was sampled to examine
the differences in subjects with bilateral, unilateral, or no overhead activity. A
sample of convenience was used based on the availability, accessibility, and cha-
racteristics of the surrounding community. Because this was a unique study, we did
not have reasonable estimates of effect sizes to perform an actual power analysis
504 Turner et al
for the group by muscle comparisons. Thus, based on the central-limit theorem, a
goal of 50 subjects per group was set to ensure a reasonable chance of obtaining
normally distributed data for the statistical analyses. The subjects each fit into only
1 category, with little crossover training between unilateral and bilateral activities.
For example, there were no baseball, tennis, or volleyball subjects who regularly
participated in swimming. Eighteen subjects were excluded based on the following
criteria: history of shoulder surgery or macrotrauma; current shoulder, neck, or
thoracic pain; scoliosis; or observable winging of the scapulae. Institutional-
review-board approval was obtained, and the remaining 172 subjects (and parents
when applicable) gave informed consent or assent. All testing was carried out in
keeping with the spirit of the Helsinki declaration. After testing, an exercise pro-
gram for the glenohumeral and the scapulothoracic musculature was provided to
the subjects as an incentive to participate in the study.25,26
Subjects were asked to fill out a questionnaire regarding their activity level, height,
weight, history of previous injury to the upper extremities, and basic demographic
information. Body-mass index (BMI) was then calculated based on height and
mass. Total arm length and upper-arm circumference were recorded for later com-
parisons with strength data. Honesty in reporting height and weight was assumed
in some cases because of equipment limitations; most of the testing took place at
team practices or meetings. Total arm length was defined as the distance from the
lateral edge of the acromion to the end of the middle finger. We also measured
shoulder-girdle length, defined as the distance from the mastoid process to the
lateral edge of the acromion. The halfway point of this measurement was marked
for dynamometer placement over the muscle bulk of the upper trapezius. In addi-
tion, upper-arm length was measured, defined as the length from the lateral acro-
mion to the lateral epicondyle, and the halfway point was marked for dynamometer
placement during rhomboid testing. Arm circumference was also measured at this
halfway point of the upper arm.
The manual-muscle-testing positions described by Hislop and Montgomery28
were adapted for use with the Baseline hydraulic push–pull handheld dynamom-
eter with analog gauge (see Figures 1 and 2). Muscles rarely work in an isolated
manner, so these muscle tests can be considered “biased” toward each muscle. For
example, several studies have shown increased middle-trapezius electromyo-
graphic activity during the position described in Figure 3 for the lower trape-
zius.11,22,27 However, isolating the middle trapezius’ action of scapular retraction
is commonly done in clinical practice, as shown in Figure 4.28 Michener et al22
established construct validity for the lower- and upper-trapezius muscle tests
using the positions described in Figures 3 and 5, respectively. The testing proce-
dure for the serratus anterior targets both functions of the muscle: upward rotation
and protraction (see Figure 6).11,27,29 The testing procedure for the rhomboids is
illustrated in Figure 7. Smith et al30 demonstrated that the rhomboid manual
muscle test described by Hislop et al was not significantly different than the rhom-
boid manual muscle test described by Kendall et al when considering the percent
maximum voluntary contraction of the rhomboids.
Figure 1 — Rectangular end piece used
for testing the middle and lower trapezius.
Figure 2 — Curved end piece used for
testing the upper trapezius, serratus ante-
rior, and rhomboids.
Figure 3 — Lower-trapezius bias. The subject was prone with the arm in 150° of scaption
with the thumb up. The dynamometer was placed on the posterolateral corner of the acro-
mion, and the rectangular end piece was used. The clinician stabilized the contralateral hip
while matching the force of the subject.
Figure 4 — Middle-trapezius bias. The subject was prone with the arm in 90° of abduction
and the elbow in 90° of flexion. The dynamometer was placed on the posterolateral corner
of the acromion, and the rectangular end piece was used. The subject was instructed to
“squeeze your shoulder blades together” to retract his or her scapula. The clinician stabi-
lized the contralateral scapula while matching the force exerted by the subject.
Figure 5 — Upper-trapezius bias. The sub-
ject was seated with his or her arms at the
sides. The dynamometer was placed halfway
between the mastoid and lateral acromion
over the muscle bulk, and the curved end piece
was used. The examiner was allowed to use
both hands, so no stabilization was necessary.
Figure 6 — Serratus anterior bias. The sub-
ject was seated with the arm in 130° of flexion.
The dynamometer was placed on the humerus
distally to the deltoid attachment, and the
curved end piece was used. The clinician stabi-
lized the inferior angle of the scapula being
tested while matching the force of the subject.
Normative Data on Scapulothoracic Musculature 507
For each test, the subject was asked to perform the motion through his or her
full range of motion, back off into midrange, and hold the position. A “make”
muscle contraction was used rather than a “break” muscle contraction. A make
test was used to avoid overpowering the subjects in an effort to measure their
force-producing capabilities. Make tests are used almost exclusively with hand-
held dynamometry.13–21,23 Subjects were asked to build their force gradually to a
maximum voluntary effort over a 2-second period. They maintained a maximum
voluntary effort for a 5-second period. The examiner kept the dynamometer in
place by matching the force exerted by the subject, and the peak force was
recorded. If the subject “broke” against resistance, the data were not recorded, and
the muscle test was repeated. Strength measurements were collected for each sub-
ject’s dominant upper extremity. One trial of each muscle test was used, which has
been established in the literature as adequate for measuring muscle strength in
healthy subjects.20 To avoid any possible fatigue factor, all the testing occurred
before practices, competitions, or heavy exercise.
The order for testing the muscles was semirandomized to facilitate the
speed of testing. Because the end piece of the dynamometer had to be changed,
muscle tests requiring the curved end piece were tested together. These muscles
included the upper trapezius, serratus anterior, and rhomboids. The middle- and
lower-trapezius muscle tests required the rectangular end piece and were there-
fore tested last. For the next subject, the rectangular end-piece configuration was
used first. Within these limitations, the order of testing was randomized. The
handheld dynamometer with both end pieces used in testing is pictured in
Figures 1 and 2.
Figure 7 — Rhomboids bias. The subject was prone with the hand on the small of the
back. The dynamometer was placed on the humerus halfway between the acromion and
lateral epicondyle, and the curved end piece was used. The clinician stabilized the contra-
508 Turner et al
Before actual data collection, the 2 testers involved in the study performed exten-
sive practice with subjects not included in the study analysis to develop consistent
techniques, adopt stable positions for resisting subjects’ force, and improve reli-
ability. After sufficient practice, a pilot study was performed to determine the intra-
rater and interrater reliability of the 2 examiners. Sixteen subjects were recruited,
all of whom met inclusion criteria for the study. Both examiners were blinded to the
measurements, and the procedure used was identical to that of the larger study.
None of the subjects were included in the larger investigation. Time between testers
(intertester) was 15 minutes, and intratester-reliability interval was 72 hours. Again,
the subjects were similar in anthropometric and demographic characteristics to the
subjects in the larger study. Rationale for sample size used was based on a reliabil-
ity power analysis.31 Data were analyzed using Statistical Package for the Social
Sciences (version 15.0) to calculate interclass coefficients (ICCs) for intrarater and
interrater reliability. The (3, 1) formula was used to calculate these ICCs. After ICC
calculation, the standard errors of measurement were calculated. See Table 1.
Subjects were classified into 1 of the following 3 groups: no overhead activity,
unilateral overhead activity, or bilateral overhead activity. No overhead activity
was defined as not participating in any athletic activity that required the arm to be
elevated above 90°, including runners, soccer players, and nonathletic participants.
Unilateral overhead activity was defined as participating in any athletic activity
that required predominately 1 arm to be elevated above 90°, such as tennis and any
throwing-sport athletes. Bilateral overhead activity was defined as participation in
Table 1 Reliability Analysis for the Pilot Study
Upper trapezius (left)
Upper trapezius (right)
Middle trapezius (left)
Middle trapezius (right)
Lower trapezius (left)
Lower trapezius (right)
Serratus anterior (left)
Serratus anterior (right)
E1, examiner 1; E2, examiner 2; IR, interrater.
Normative Data on Scapulothoracic Musculature 509
athletic activity that required both arms to be elevated above 90°, such as swim-
mers and triathletes. Based on the operational definitions used, to be considered for
either the unilateral or bilateral athlete categories, one had to have actively partici-
pated in an organized sport (ie, high school, college, master’s level) for a minimum
of 1 year. Descriptive statistics for the demographics for each group are provided
in Table 2.
First, the data were correlated with the anthropometric measurements to
attempt to normalize scapulothoracic strength to each individual. The dominant
arm was used for all data analysis. A 2-factor repeated-measures ANOVA was
used to analyze the effects of activity level on scapulothoracic strength measure-
ments. The 5 muscles were entered as a within-subject factor using 5 levels. Activ-
ity level was used as a between-subjects factor.
For each ANOVA, the average force production of each muscle across each
activity level was analyzed to determine a rank order for the strength of the 5
muscle groups tested. Then the unilateral strength ratios were determined. The 3
ratios studied were elevation versus depression (upper vs lower trapezius), pro-
traction versus retraction (serratus anterior vs middle trapezius), and upward
versus downward rotation (serratus anterior vs rhomboids). For simplicity, the
serratus anterior was used to represent upward rotation, instead of the force-
couple concept using upper trapezius, lower trapezius, and serratus anterior. The
middle trapezius was selected because it has the unilateral function of retraction,
as opposed to the rhomboids, which have the dual function of retraction and
downward rotation. As stated before, muscle weaknesses and imbalances can lead
to impingement, so using these unilateral ratios may help establish normal bal-
ance of the scapulothoracic muscles.3–8 A separate 2-factor ANOVA of ratio by
activity level was used to examine the effects of activity level on the strength
ratios. Each ratio served as a within-subject factor with 3 levels, and activity level
constituted a between-subjects factor with 3 levels.
When significant interactions on main effect were revealed, main interactions,
only effects post hoc comparisons were conducted using the Dunn–Bonferroni
procedure. Specifically, for significant interactions, only within-group–between-
muscles and within-muscle–between-groups comparisons were considered. Statis-
tical significance was considered P < .05.
Table 2 Subjects Classified by Activity Level
Age range, y13–60
Mean age24.9 ± 9.4
17.9 ± 4.8
28.6 ± 13.2
510 Turner et al
Correlations With Anthropometric Data
Correlations are shown in Table 3. Analysis demonstrated very weak relationships
between muscle force and these anthropometric relationships.
Muscle by Overhead Activity-Level Analysis
The 2-factor ANOVA demonstrated a significant interaction between muscle
strength and activity level (F5.2,438 = 4.364, P = .001). A significant main effect
was observed for muscle (F2.6,438 = 713.971, P < .001), as well as activity level
(F2,169 = 10.889, P < .001). The means, SDs, and 95% confidence intervals are
presented in Table 4.
Generally, the overhead athletes (both unilateral and bilateral) had signifi-
cantly higher muscle strength than the group with no overhead activity (Figure 8).
This pattern was true for every muscle except the lower trapezius, which was the
weakest across all 3 groups. There were no significant differences between the
unilateral and bilateral overhead-activity groups. There were no statistically sig-
nificant differences found among any of the groups with respect to the lower tra-
pezius (Table 4).
The upper trapezius was significantly stronger than any other muscle
(Figure 9). Both the middle trapezius and serratus anterior were significantly stron-
ger than the rhomboids and the lower trapezius. There were no significant differ-
ences between the middle trapezius and the serratus anterior. In addition, there
were no significant differences between the lower trapezius and the rhomboids.
The upper trapezius was significantly stronger than any other muscle (Figure
10). The serratus anterior was significantly stronger than the middle and lower
trapezius, as well as the rhomboids. The middle trapezius was significantly stron-
ger than the lower trapezius and rhomboids. There was no significant difference
between the lower trapezius and the rhomboids.
The upper trapezius was significantly stronger than any other muscle (Figure 11).
The middle trapezius was significantly stronger than the lower trapezius and the
Table 3 Correlations (r Values) Between Anthropometric
Measurements and Scapulothoracic Strength
Upper trapezius .213*.288**
Middle trapezius .250*.260*
Serratus anterior .184*.209*
*P < .05. **P < .001.
Normative Data on Scapulothoracic Musculature 511
Table 4 Mean Force Production for Each Muscle Across Overhead-
Middle trapezius None
Lower trapezius None
rhomboids. The serratus anterior was significantly stronger than the middle and
lower trapezius, as well as the rhomboids. The rhomboids were significantly stron-
ger than the lower trapezius, which is a unique finding to the bilateral overhead-
Therefore, given the data, a rank order cannot be established for every activity-
level group. However, Table 5 demonstrates a general template for rank ordering
the strength of the scapulothoracic muscles based on overhead-activity level.
After a 2-way ANOVA to compare muscle-strength ratios across activity-
level groups, no significant interaction was observed (F2.6,217 = 1.929, P = .135). A
significant main effect for muscle ratio was noted (F1.3,217 = 293.617, P < .001),
but the main effect for activity level was not significant (F2,169 = 0.919, P = .401).
Therefore, it is not necessary to discuss each muscle ratio separately by overhead-
activity level. Post hoc analysis revealed that the elevation:depression ratio was
statistically significantly higher than both of the other ratios. The upward:downward
rotation ratio was significantly higher than the protraction:retraction ratio (Table 6
and Figure 12).
Figure 9 — Strength of the scapulothoracic muscles in the no-overhead-activity group.
*Significantly greater than all other muscles for this activity group. †Significantly greater
than the lower trapezius and rhomboid. Error bars represent the SD.
Figure 8 — Strength of the scapulothoracic muscles between activity-level groups. *Sig-
nificantly greater than the group with no overhead activity. Error bars represent the SD.
Figure 10 — Strength of the scapulothoracic muscles in the unilateral overhead-activity
group. *Significantly greater than all other muscles for this activity group. †Significantly
greater than the lower trapezius and rhomboid. ‡Significantly greater than the middle tra-
pezius, lower trapezius, and rhomboid. Error bars represent the SD.
Figure 11 — Strength of the scapulothoracic muscles in the bilateral overhead-activity
group. *Significantly greater than all other muscles for this activity group. †Significantly
greater than the lower trapezius and rhomboid. ‡Significantly greater than the middle tra-
pezius, lower trapezius, and rhomboid. §Significantly greater than the lower trapezius. Er-
ror bars represent the SD.
Table 5 Rank Order for Scapulothoracic Muscle Strength Based on
Post Hoc Analysis of Overhead-Activity Level
Serratus anterior or middle
Rhomboids or lower
Rhomboids or lower
a Indicates the strength of this muscle was significantly greater than those below.
Table 6 Ratio Comparison as Grouped by Overhead-Activity Level
(upper trapezius to lower
rotation (serratus anterior
to middle trapezius)
(serratus anterior to
Normative Data on Scapulothoracic Musculature 515
This study was the first to examine the relationships between scapulothoracic
strength using a handheld dynamometer and anthropometric measurements such
as BMI or arm length. These measurements are clinically relevant in a myriad of
ways. For instance, these data can be used to establish normative strength values
for use during injury-prevention screenings, initial evaluations, and serial reas-
sessments and when setting discharge criteria, as well as being used as guidelines
for designing strength and conditioning programs to ensure that the glenohumeral
joint has a stable scapulothoracic base. Future research in this area, along with our
current study, could lead to an extensive database of strength values for clinicians.
When the handheld-dynamometer measurements were compared with mea-
surements such as height, weight, BMI, total arm length, and arm circumference,
weak relationships were found. Examining BMI, in addition to height and mass,
was one of the original intents of the current study because currently there is no
other research comparing handheld-dynamometer strength data and anthropomet-
ric data for any body part. However, because there were very weak relationships
(Table 3) we did not think it was appropriate in the final analysis of the data. For
example, the highest correlation was between BMI and the lower trapezius (r =
.271), which only leads to 7.3% shared variance. Although normalizing to body
weight or BMI would have been useful for general application, the data did not
support this relationship. A few possible explanations for the weak relationships
Figure 12 — Unilateral muscle-strength ratios by overhead-activity level. *Significantly
greater than the protraction:retraction ratio and the upward:downward rotation ratio. †Sig-
nificantly greater than the protraction:retraction ratio. Error bars represent the SD.
516 Turner et al
are that the upper extremity is not weight bearing and therefore may not be
affected by body weight. In addition, the scapular muscles work as synergists and
stabilizers, as opposed to prime movers, so this may affect the relationships with
measurements such as body mass or BMI. Subjects with varying levels of over-
head experience and a wide range of ages were studied, so perhaps a more homo-
geneous sample would have yielded stronger correlations.
The strongest anthropometric correlations were seen with upper-arm circum-
ference measurements. Arm circumference may represent muscle mass of the
upper arm, so a higher measure would indicate increased cross-sectional area and
therefore increased force production. The scapular muscles provide the proximal
stability for the arm, so this increase in strength of the upper arm may translate to
increased strength in the scapular muscles. Increased arm circumference could
also indicate obesity, but in this study the mean BMI was 22.69 ± 3.98. The sub-
jects were also generally physically active, so we feel that arm circumference
better represents muscle mass. Further research is warranted in this area to explore
all relationships between anthropometric measurements and scapulothoracic
When comparing muscle strength and activity level, a common pattern was
found. The overhead-activity-level groups (unilateral and bilateral) had signifi-
cantly greater scapulothoracic muscle strength than the group with no overhead
activity. This observation can most likely be explained by a training effect.
Although direct training of the scapular stabilizers is not commonly seen in train-
ing programs, these muscles are a part of the kinetic chain and were therefore
active during overhead activity for stabilization.
There were significant differences between each muscle’s strength in the
bilateral overhead-activity level, which led to establishing a rank order for this
group. The rank order is as follows: upper trapezius, serratus anterior, middle
trapezius, rhomboids, and lower trapezius. For the remainder of the activity-level
groups, an exact rank order could not be determined given the lack of significant
differences between muscles; however, several trends were observed. These trends
were similar to the order established by the bilateral overhead-activity group.
Generally, the upper trapezius was the strongest, followed by the serratus anterior
and middle trapezius, followed by the rhomboids and lower trapezius.
As previously stated, scapulothoracic muscle weaknesses lead to imbalances
that can result in abnormal stabilization and control of the scapula.1 Identifying
weakness that could potentially lead to shoulder dysfunction could potentially
lead to a decreased rate of subacromial impingement. In one study, the rate of
shoulder impingement was 55.1% of 878 patients with shoulder dysfunction.32 A
rank order can be used to identify weak muscles to modify training and condition-
ing programs with the goal of restoring normal muscle balance. This comparison
would be analogous to considering quadriceps-to-hamstring ratios.
However, with respect to the unilateral ratios, no differences were seen across
activity level. The 172 normal subjects included in this study demonstrated upper-
trapezius strength approximately 2.5 times that of the lower trapezius. The ratio
between the upward rotators (represented by the serratus anterior) and the down-
ward rotators (rhomboids) was approximately 1.5:1. The ratio between scapular
protraction (serratus anterior) and scapular retraction was approximately 1.25:1.
We recognize that the scapulothoracic muscles work as synergists to produce scap-