Manual Therapy 20 (2015) 46e55

Contents lists available at ScienceDirect

Manual Therapy
journal homepage: www.elsevier.com/math

Original article

A new description of scapulothoracic motion during arm movements

in healthy subjects
Alexandra Roren a, Marie-Martine Lefevre-Colau a, Serge Poiraudeau a, f, Fouad Fayad b,

s Roby-Brami a, c, d, e, f, *
Viviane Pasqui c, d, e, Agne
a

Department of Physical Medicine and Rehabilitation, Cochin Hospital (AP-HP), Paris Descartes University, Paris, France
Department of Rheumatology, Hotel-Dieu de France Hospital, Saint-Joseph University, Beirut, Lebanon
ISIR (Institute of Intelligent Systems and Robotics), CNRS UMR 7222, Paris, France
d
Sorbonne Universit
es, UPMC University Paris 06, Paris, France
e
ISIR-AGATHE, INSERM U 1150, Paris, France
f
Institut F
ed
eratif de Recherche sur le Handicap, INSERM, Paris, France
b
c

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 2 September 2013
Received in revised form
18 June 2014
Accepted 24 June 2014

The participation of scapula motion in arm movement is clinically well known and recent three
dimensional (3D) analyses using kinematic techniques have conrmed its importance. Scapular motion
relative to the thorax has a theoretical maximum of 6 degrees of freedom (DoF), resulting from rotations
at both clavicular joints (3 rotational DoF each). However, most recent kinematic studies have only
analysed the 3D rotations of the scapula relative to the thorax. In the present study, the 3D translations of
the barycentre of the scapula were considered in order to complete the description of movement at the
shoulder complex. Eight healthy subjects performed arm elevation in the sagittal and frontal planes,
simulated activities of daily living (hair combing and back washing) and maximum voluntary scapula
movement (forward and backward rolling). Measurements were recorded using a 6 DoF electromagnetic
device and the acromial method of analysis was used. The results showed that 3D scapular rotations and
translation of its barycentre were functionally consistent for all tasks. A principal component analysis
(PCA) yielded three factors, explaining 97.6% of the variance. The rst two factors (protraction and shrug,
according to clinical descriptions) combined rotations and translations, consistent with the hypothesis
that the scapula rolls over the curved thoracic surface. The third factor related to lateral-medial rotation,
thus representing rotation in the plane tangential to the thorax. The PCA suggested that scapular motion
can be described using these 3 DoF. This should be studied in a larger group of individuals, including
patients with pathological conditions.
2014 Elsevier Ltd. All rights reserved.

Keywords:
Kinematics
Scapula rotation
Protraction
Clavicle

1. Introduction
A sound understanding of the kinematics of the shoulder
complex is important for the clinical evaluation and treatment of
many pathological conditions (Sahrmann, 2001). It has long been
observed clinically that the scapula over the thorax is nely coordinated with the rotations which occur at the glenohumeral joint
(scapulo-humeral rhythm) (Codman, 1934). During scapulothoracic motion, there are simultaneous rotations of the clavicle

* Corresponding author. ISIR, Institute of Intelligent Systems and Robotics, University Pierre et Marie Curie, CNRS UMR 7222, 4 place Jussieu, 75005 Paris, France.
Tel.: 33 1 44 27 62 15; fax: 33 1 44 27 51 45.
E-mail address: [email protected] (A. Roby-Brami).
http://dx.doi.org/10.1016/j.math.2014.06.006
1356-689X/ 2014 Elsevier Ltd. All rights reserved.

at the sternoclavicular (SC) and acromioclavicular (AC) joints

(Inman and Saunders, 1946), causing the scapula to both rotate and
translate over the thorax. Recent three dimensional (3D) kinematic
techniques have conrmed these clinical observations and have
demonstrated the importance of scapular motion for full, functional mobility of the arm (Pronk et al., 1993; van der Helm and
Pronk, 1995; Ludewig et al., 1996; McClure et al., 2001; Ludewig
et al., 2004; Magermans et al., 2005; Ludewig et al., 2009; Roren
et al., 2012). In addition, studies have demonstrated that scapular
kinematics are altered in most pathological conditions that affect
the shoulder (rotator cuff pathologies, shoulder instability and
shoulder stiffness); see review in Ludewig and Reynolds (2009).
The description of scapular motion relative to the thorax remains, however, a scientic and clinical challenge. The International Society of Biomechanics (ISB) has proposed a protocol based

A. Roren et al. / Manual Therapy 20 (2015) 46e55

on Euler angle formalism for the geometrical description of the

kinematic chain (van der Helm, 1997; Wu et al., 2005). Scapulothoracic motion is described by two distinct sequences of three
Euler angles: one sequence for the rotations of the clavicle relative
to the thorax (pro-retraction, elevation-depression, and axial
rotation at the SC joint), and the other for the rotations of the
scapula relative to the clavicle (pro-retraction, medial-lateral
rotation and anterior-posterior tilt at the AC joint). Scapular motion can also be described relative to the trunk (thus neglecting
clavicular motion) using the same sequence. There are, however
three problems related to the ISB protocol.
First, the clinical usability of the ISB model is limited by the fact
it is based on the use of data obtained either by invasive recording
methods or by external, but imprecise and incomplete methods.
The ISB protocol is based on the measurement of the 3D position of
two bony landmarks which dene the clavicle (most ventral point
on the SC joint and most dorsal point on the AC joint) and three
bony landmarks which dene the scapula (angle of the acromion,
root of the scapular spine and inferior angle of the scapula). The
protocol was validated using sensors inserted directly into the
bones (Karduna et al., 2001; Mc Clure, 2001; Ludewig et al., 2009).
Using this technique, Ludewig et al. (2009) showed that substantial
rotations occur in all the DoF of the SC and AC joints. However, it is
not possible to insert sensors into the bones in clinical practice.
Biplane uoroscopy (Giphart et al., 2013) is another highly reliable
technique but cannot be used in clinical practice either.
Motion capture of the scapula using external sensors xed over
bony landmarks is subject to artefacts caused by skin sliding and
muscle bulging (van Andel et al., 2009). A well accepted method for
the capture of scapular motion consists of using one 6 DoF sensor
xed over the acromion process to dene the position and orientation (i.e. 6 DoF) of an acromial reference frame (acromial
method) (Karduna et al., 2001). Bony landmarks of the scapula are
then digitized and computed in the acromial reference frame
before being projected in the global 3D space (Meskers et al., 1998).
This method allows real continuous dynamic tracking of scapular
motion during natural movements (Karduna et al., 2001; Meskers
et al., 2007). Several types of 6 DoF sensors have been used in
studies: electromagnetic sensors (Meskers et al., 1998; Fayad et al.,
2006), inertial devices (Parel et al., 2012) and optical marker clusters (van Andel et al., 2009; Lempereur et al., 2010; Senk and Cheze,
2010). The level of error associated with acromial methods is
acceptable for ranges of arm elevation below 120 (Karduna et al.,
2001; Ludewig et al., 2009). These methods can be used to assess
3D shoulder kinematics in a variety of conditions in healthy subjects (Dayanidhi et al., 2005; Fayad et al., 2006; Yano et al., 2010)
and in patients with different shoulder pathologies (Ludewig and
Cook, 2000; Lin et al., 2006; Rundquist, 2007; Amasay and
Karduna, 2009). The analysis of clavicular function is usually
limited to the 2D static measurement of protraction and elevation
(2 DOF) using imaging or goniometric methods (Conway, 1961;
Inman et al., 1996; Hallaceli and Gunal, 2002). Motion capture of
the clavicle with external markers is very new. Two studies in the
literature have used clavicle motion capture techniques. One
technical study proposed the 2D dynamic measurement of clavicular rotations using the acromial method with the digitization of
two additional landmarks (AC and SC) (Helgadottir et al., 2010). The
other study presented SC and AC joint rotations in healthy subjects
(Teece et al., 2008). The quantication of axial rotations of the
clavicle is particularly challenging due to skin artefacts and the
crank-shape of the clavicle (Ludewig et al., 2004; Helgadottir et al.,
2011).
In the present study, we propose to extend the acromial method
in order to capture, in addition to scapular rotations, the linear
translations (displacements) of the centre of the scapula relative to

47

the thorax. This technique yields 6 DoF, thus providing a direct and
complete measurement of the kinematics of the shoulder complex,
including the clavicle.
The second problem with the ISB protocol relates to the inconsistent use of clinical and biomechanical terminology. In anatomical
and clinical descriptions, scapula motion is reported to consist of
three rotations and two translations (Peat, 1986; Culham and Peat,
1993; Sahrmann, 2001; Tytherleigh-Strong, 2008). The rotation of
the scapula in its main plane is often called upward/downward
rotation (sometimes lateral and medial rotation), the other rotations involve winging around the vertical axis: internal and
external rotation (sometimes named protraction/retraction) and
tilting around a horizontal axis (anterior/posterior tilt or tilting or,
less frequently, tipping). The translations are reported as elevation/
depression (sometimes named upward/downward translation) and
abduction-adduction (toward/away from the vertebral spine). According to some authors, protraction is not a pure rotation but a
combination of rotation and translation involving both scapular
and clavicular motion. It is dened as a forward movement of the
scapula around the thoracic wall [which] combines linear translation
away from the vertebral column, rotation of the scapula around the
end of the clavicle (winging) and anterior movement of the lateral end
of the clavicle (Culham and Peat, 1993) see also (Solem-Bertoft
et al., 1993). Clinical observations thus incite the inclusion of
translations in the kinematic description of global scapula motion.
The last problem of the ISB protocol relates to the dimensions
(i.e. the minimum number of DoF) needed to describe scapular
motion relative to the trunk. Theoretically, the motion of a solid in
space is dened by 6 parameters: 3 for position and 3 for orientation. The ISB model using 6 DoF within an open kinematic chain
thus provides a sufcient description of scapula motion relative to
the trunk (Bao and Willems, 1999). However, the shoulder complex
is a closed chain due to the contact between the scapula and the
thorax. The anatomy of the bony and musculo-ligamentous structures is likely to constrain the motion of the scapula in space and to
create couplings, limiting the effective dimensions of scapulothoracic motion. Several mechanical models of these constraints
have been proposed. However, there is no agreement on the minimum number of independent DoF needed for the description of
scapulothoracic motion; 4 according to van der Helm (1994) and
Seth et al. (2010, 2012); 5 according to Bao and Willems (1999) and
6 according to Lenarcic and Stanisic (2003). To our knowledge, this
question has not been investigated experimentally using kinematic
methods.
The present study had a double purpose. The rst aim was to
completely describe and quantify the motion of the scapula relative
to the trunk, including both scapular rotations and translations
(linear displacements of the centre of the scapula (CS)).
The second aim was to specify the functional dimension of
scapular motion during a large variety of motor tasks in healthy
subjects. For this, movements involving analytical arm elevation in
the sagittal and frontal planes, simulations of activities of daily
living (hair combing (HC) and back washing (BW)) and the
maximum voluntary scapula-thoracic movement in different directions (forward and backward shoulder rolling: F-Roll and B-Roll)
were recorded and analysed. The hypothesis was that 6 DoF might
not be necessary for the description of scapulothoracic motion, due
to functional coupling between rotations and translations.
2. Method
2.1. Population
A convenience sample of eight healthy volunteers (age
31.14 9.30 years, range 24e50; height 1.72 0.82 m, range

48

A. Roren et al. / Manual Therapy 20 (2015) 46e55

1.62e1.87; weight 66.71 9.11 kg, range 54e82; BMI

22.55 2.31 kg/m2, range 19.13e25.91) with no prior history of
upper extremity disorders took part in this study. All the subjects
were right-handed, healthy and active adults; they were all working in the rehabilitation unit. The study protocol was approved by
the local Institutional Review Board and all subjects provided
informed consent.
2.2. Instrumentation
Real-time 3-D position and orientation of the thorax, scapula
and humerus were tracked (30-Hz sampling rate) using the four
sensors of the Polhemus Fastrak electromagnetic device. The reported root mean square (RMS) accuracy of this system is
0.3e0.8 mm for position and 0.15 for orientation when used
within a 0.76 m source to sensor separation (SPACE FASTRAK User's
Manuel, Revision F. Colchester, VT; Polhemus Inc., 1993). The
transmitter, which provided the general frame of reference, was
tted on a rigid plastic frame to an adjustable camera tripod placed
in front of the subject. The reference frame was positioned
approximately at the level of the navel, so that the sensors were
within a range of 0.3 me0.7 m from the transmitter.
2.3. Experimental procedure
2.3.1. Kinematics
The thoracic and scapular sensors were xed on the patient's
skin over the sternum and the at surface of the superior acromion
process respectively. The humeral sensor was strapped to the arm
with Velcro, just below the insertion of the deltoid. The fourth
sensor, mounted on a pointer, was used to manually digitize the
bony landmarks palpated on the thorax (xiphoid process, suprasternal notch, and spinal processes of C7 and T8), scapula (acromial
angle, root of the spine and inferior angle) and humerus (medial
and lateral epicondyle). The local coordinates of the bony landmarks were computed in the reference frame of the corresponding
sensor (Mc Clure, 2001; Fayad et al., 2006). The centre of the glenohumeral joint was computed by a regression method (Biryukova
et al., 2000). Joint rotation matrices between the reference frames
based on the bony landmarks of the rigid bodies were then
computed and the rotations expressed using Euler angle sequences
(van der Helm, 1997). Scapular rotations were: internal/external
rotation1 around the main thoracic axis, then medial/lateral rotation around an axis perpendicular to the scapular plane, followed
by anterior/posterior tilt (Fig. 1). The 3D coordinates of the centre of
the scapula (CS) (i.e. the barycentre of the triangle representing the
scapula) were calculated in the reference frame of the thorax. Thus,
the displacement of CS was expressed by projections in three directions (forward, lateral and upward) relative to the centre of the
thorax (barycentre of the four points: xiphoid process, suprasternal
notch, and spinal processes of C7 and T8) (Hanneton et al., 2011).
Arm elevation relative to the thorax was also computed as the angle
between the main axis of the humerus and that of the trunk irrespective of the plane of elevation. Left side data were projected to
the right side. To that purpose, the symmetrical 3D positions of all
the anatomical landmarks with respect to the sagittal plane were
calculated before further computations.
2.3.2. Recording procedure
Skin preparation and xation of the markers on the skin surface
(which is the longest phase of the whole procedure) were carried
out with subjects seated to avoid fatigue.

According to Ludewig et al. (2009) instead of pro/retraction .

Fig. 1. Schema of the body segments and corresponding reference frames for the
thorax and scapula. Bony landmarks on the scapula: acromial angle (AA: angulus
acromialis), root of the spine (TS: trigonum spinae scapulae) and inferior angle (AI:
angulus inferior). CS is the barycentre of the scapula and CT the barycentre of the
trunk. The arrows show the rotations as Euler angles in the following order: 1: Internal/External rotation, 2: Medial/Lateral/rotation, 3: Antero-Posterior Tilt. HTE :
humero-thoracic elevation.

During the digitization process, the subjects were asked to adopt

a reference position: standing upright with both arms hanging
beside the body. Then, kinematic data were recorded at 30 Hz while
the following movements were carried out in the same order: for
forward rolling (F-Roll), the subjects were instructed to perform the
largest possible movement of the scapula in the forward, upward
and backward directions, keeping their arm by their sides. For
backward rolling (B-Roll) the instruction was the same for the
reverse succession of backward, upward and forward directions
(Sheikhzadeh et al., 2008). The subjects were then instructed to
perform maximal active analytic arm elevation: rst in the sagittal
plane (Flexion) then in the frontal plane (Abduction). Finally, they
were instructed to simulate two activities of daily living: hair
combing (HC) and back washing (BS) (Roren et al., 2012). Before
each recording, each movement was described to the subject who
then performed it twice for a warm-up and to ensure he/she carried
out the movement as instructed: avoid exing the elbow or moving
the trunk, complete the movement in about 10 s and remain in a
xed plane of motion for analytical arm elevation. During the
recording, the observer gave permanent verbal support so that the
subject achieved the movement within the recording time and
remained in the correct plane of motion. Incorrect trials were
repeated in order to obtain two analysable trials. The recording of
the tasks lasted less than 10 min for each side. The whole procedure
was carried out for both shoulders, starting with the right side, and
lasted a maximum of 60 min.
2.3.3. Data analysis
Kinematic data (arm elevation, scapulothoracic rotations and CS
translations) were analysed using custom made interactive

A. Roren et al. / Manual Therapy 20 (2015) 46e55

49

Fig. 2. Time normalized averages of arm elevation relative to the thorax during the different tasks (between t0 and tf). Each trace represents the average of the two successive trials
in the group of subjects, right and left sides are averaged: Flexion: maximum elevation in the sagittal plane, Abduction: maximum elevation in the frontal plane HC: simulated hair
combing: BW: simulated back washing; F-Roll: forward rolling; B-Roll backward rolling. Thick lines indicate the means and thin lines one standard error of the mean (sem). The
arrows indicate the time points analysed: t1 to t3.

software (Labview 8.2). The values of arm elevation were used to

compute the beginning (t0) and end (tf) of each movement with a
threshold equal to 3 times the standard deviation of the resting
value (measured during the rst ten recording samples). Two time
points were determined for Flexion, Abduction, BW and HC. The
rst (t1) corresponded to the initial part of the movement
computed at an arm elevation of 90 (Flexion, Abduction and HC) or
at the rst maximum (BW). The second (t2) corresponded to the
nal posture characteristic of the task. Three time points (t1-3)
were determined for F-Roll and B-Roll, relating to the three phases
of the roll.
For the data analysis, the mean of the values obtained for the
two successive repetitions of each task was used. Then, the means
within the group of subjects were computed for each side and each
task. Statistical analysis (ANOVA and Student's t-test) was carried
out using Statview. The signicance level was xed at alpha 0.05.
The arm elevation angle and the 6 scapulothoracic parameters
were computed at each of the time points (t0 to t2 or t3). Changes in
values were then calculated relative to the initial values at t0. The
absolute amplitude of the 3D translation of the CS was also
calculated.
In addition, the kinematic data corresponding to each movement were time-normalized to 100 time intervals between t0 and tf
in order to remove spontaneous variations in movement duration.
Time normalized signals were averaged within subjects and sides.
Principal Component Analysis (PCA) was used to investigate the

dimensionality of the data. The mathematical procedure of the PCA

transforms a number of possibly correlated variables into a smaller
number of uncorrelated variables called principal components (PC,
PCs in the plural), ranked according to the amount of variance
explained by each. PCA was performed with varimax transformation (Statview software) on the normalized time course of the
6 scapulothoracic parameters cumulating all the tasks (i.e. on a
matrix with 6 columns and 100  6 lines). This analysis yields the
amount of variance explained by each PC and the correlation between PCs and DoF.

3. Results
The 6 DoF describing the initial posture of the scapula relative to
the thorax were analysed using a two factors ANOVA (side and task)
with participants as repeated measures. For scapula internal rotation, there was a signicant effect of task (F72, 5 5.2, p < 0.001),
but not side (F72,1 0.8). Post-Hoc analysis showed that the
scapula was less internally rotated at the beginning of the B-Roll
than for the other tasks (4.6 0.5 , non signicant, ns). For the
other DoF, there was no effect of task (F72,1 < 1.6, ns) or side
(F72,1 < 0.9, ns). In addition, there were no signicant differences
between sides, for movement at t1 and t2, whatever the task
(Student's t test). Thus data from the two sides were averaged and
this average was used for the rest of the analysis.

50

A. Roren et al. / Manual Therapy 20 (2015) 46e55

The time course of humerothoracic elevation and the selected

time points are shown in Fig. 2. For the Flexion, Abduction and HC
tasks there was a raising and a lowering of the humerus relative to
the thorax. For the BW task, arm elevation increased then
decreased to complete the task, then followed the reverse pattern
for the return movement. The B-Roll and F-Roll tasks both followed

a tri-phasic pattern, as expected from the instructions given. The

time points (t1 to t3) were placed in the middle of the plateau (or at
the local maxima, if any) for each phase.
An example of scapula kinematics during the different tasks is
shown for a representative subject in Fig. 3. The normalised time
courses of humerothoracic elevation (Fig. 2) and scapula rotation

Fig. 3. Examples of scapular kinematics during the six tasks in a representative subject. The (blue) polygon on the side view represents the initial posture of the thorax. Each
triangle represents the scapula at a given time (from t0 to t3 for Flexion, Abduction, HC, BW and from t0 to t4 for F-Roll and B-Roll). CS is represented by a (red) cross. The thick
triangle indicates the initial posture. For the sake of clarity, only every third sample has been represented (~0.1 s). (For interpretation of the references to colour in this gure legend,
the reader is referred to the web version of this article.)

A. Roren et al. / Manual Therapy 20 (2015) 46e55

angles during the different tasks are illustrated in Figs. 4e6. The
amount of rotation between the initial posture and t1-t3 are
summarized in Tables 1 and 2.
During Abduction (Fig. 4, black lines), there was a large increase
in scapular lateral rotation (29.64 1.60 ), accompanied by a small
amount of external rotation (4.06 1.07 ) and posterior tilting
(9.57 0.79 ). The small translation of the CS (10.0 1.1 mm in 3D)
was mainly medial (4.2 1.2 mm). During Flexion (Fig. 4, grey
lines), there was also a large amount of lateral rotation
(27.06 1.24 ) accompanied by internal rotation (16.86 0.78 )
and posterior tilting (5.02 1.36 ). The CS moved laterally and
forward (24.5 1.8 mm, 16.1 2.1 mm, respectively), corresponding to a total 3D amplitude of 31.8 1.2 mm.
The results obtained for simulated activities of daily living are
illustrated in Fig. 5. During simulated hair combing (Fig. 5, HC, grey
lines), the movement consisted mainly of a large amount of lateral
rotation (32.03 1.05 ) and posterior tilt (6.36 0.81 ). The CS
moved mainly laterally (12.6 1.8 mm). During the initial elevation,
the scapula rotated medially (6.11 0.47 ) and anterior tilt
increased (8.06 0.74 ). The translation of CS was medial
(8.7 1.2 mm) and mainly upward (27.6 1.9 mm).
Shoulder rolling involved maximal voluntary movements of the
scapula in three successive phases (Fig. 6 and Table 2). During F-Roll
(Fig. 6, grey lines), the rst phase was characterized at t1 by scapular internal rotation (22.36 1.68 ) and scapular lateral rotation
(21.34 1.49 ), and was accompanied by a large translation of the
CS in all directions (77.8 6.8 mm in 3D). The second phase was
characterized at t2 by a further increase in lateral rotation

51

(39.67 1.76 ) accompanied by anterior tilting (25.62 1.92 )

and upward CS translation (96.8 6.8 mm) while internal rotation
decreased (21.10 2.72 ). The maximum scapular translation
occurred at this time (108.9 7.9 mm in 3D). The last phase was
characterized at t3 by a large scapular external rotation
(43.51 1.18 ), a decrease in lateral rotation (15.59 1.06 ) and a
posterior tilt (3.73 0.67 ). The CS moved medially
(84.4 3.3 mm) and backwards (33.1 2.3 mm) (92.6 3.3 mm
in 3D). During B-Roll (Fig. 6, black lines), scapular kinematics were
similar, in the reverse order.
Coupling between the 6 DoF of scapulothoracic motion was
investigated during the different movements using PCA. PCA extracts the common factors of the time course of the 6 DoF during all
the tasks. Three PCs were identied, explaining respectively 47.4%
(PC1), 31.5% (PC2) and 18.7% (PC3) of the variance for a total of
97.6%. PC1 was correlated with internal rotation (r 0.866), lateral
and forward CS translation (r 0.951 and r 0.687, respectively)
but not the other DoF (r < 0.042). PC2 was mainly correlated with
upward translation (r 0.970) and anterior tilt (r 0.889) and
more weakly with forward and medial translation, (r 0.456 and
r 0.501 respectively). PC3 was mainly correlated with medial/
lateral rotation (r 0.984).
4. Discussion
The current study showed, in a variety of tasks, that scapulothoracic motion involves both 3D rotations of the scapula and
signicant 3D translations. The direction of the 3D linear

Fig. 4. Time normalized averages of the coordinates of the centre of the scapula (CS) and scapular rotations during Flexion (grey lines) and Abduction (black lines). Same legend as
Fig. 2.

52

A. Roren et al. / Manual Therapy 20 (2015) 46e55

Fig. 5. Time normalized averages of the coordinates of the centre of the scapula (CS) and scapular rotations during simulated hair combing (HC, grey lines) and back washing (BW,
black lines). Same legend as Fig. 2.

translation of the CS was specic to each task and functionally

consistent.
The patterns of 3D scapular rotation recorded in this study
corroborate the ndings of previous 3D studies which also evaluated scapular kinematics during planar arm elevation and simulated tasks such as hair-combing and back-washing (Triftt, 1998;
Magermans et al., 2005; Fayad et al., 2006, 2008; Rundquist et al.,
2009; Roren et al., 2012). Similarly to the results of a previous
study (Sheikhzadeh et al., 2008), the present study showed that
shoulder rolling provoked much larger scapular internal/external
rotation and anterior/posterior tilt than other arm movements. In
addition, the present study provides original data regarding 3D
translations of the CS during analytical arm movements and ADL.
Translations of the CS remained below 30 mm when accompanying
arm movements. This is in accordance with a previous study that
assessed 3D scapular rotations and 2D translations using an electromechanical digitizer at three different degrees of arm elevation
(Wang et al., 1999). In contrast, during shoulder rolling, when the
instruction was to voluntarily perform the largest possible scapular
movements, there were large translations of the CS which could
reach ~100 mm.
The description of scapula motion used in the present study
(three rotations and the translation of its barycentre) is geometrically correct and does not involve any hypothesis regarding the
position of the instantaneous axis of rotation, which is known to
vary during arm movements (Culham and Peat, 1993), or any
possible translation along the instantaneous axis of rotation
(Woltring et al., 1985).

The current study used an acromial method with electromagnetic sensors. Karduna et al. (2001) found that errors due to skin
artefacts were relatively small for arm elevation up to 120 .
Moreover, reliability studies of scapular motion during analytical
arm elevation have shown fair to excellent repeatability in healthy
subjects (McQuade and Smidt, 1998; Tsai et al., 2003; Fayad et al.,
2006; Roren et al., 2013). Most of the variables analysed in the
present study were recorded below 120 of elevation, except for
maximum elevation during Flexion and Abduction (time t2).
However, the good correlations obtained with the PCA showed that
the scapular kinematics obeyed similar coupling laws whatever the
task. The greatest scapular translations and rotations occurred
during the shoulder rolling tasks, as specied by the instruction. It
must be emphasised that these tasks were performed well below
the level of arm elevation which might create skin artefacts.
The current study conrms the empirical descriptions of well
coordinated scapular rotations and translations (Culham and Peat,
1993) and quanties the remarkable range of motion and exibility of the shoulder complex. The combined movements of the
scapula result from the mobility of the clavicle which is linked to
the scapula via the SC and AC joints (Inman and Saunders, 1946;
Teece et al., 2008; Ludewig et al., 2009). However, the relationship between motion of the clavicle (at the AC and SC joints) and
motion of the scapula relative to the thorax is not simple, particularly because of the crank-shape of the clavicle. Axial rotation
around the SC joint may cause signicant displacement of the position of the AC joint and thus of the scapula as a whole (rotation
and translation of CS). For the rst time, the present method affords

A. Roren et al. / Manual Therapy 20 (2015) 46e55

53

Fig. 6. Time normalized averages of the coordinates of the centre of the scapula (CS) and scapular rotations during forward (F-Roll, grey lines) and backward rolling (B-Roll, black
lines). Same legend as Fig. 2.

which occur during activities of daily living, combined with

anatomical factors such as the narrowness and incongruence of the
articular surfaces (Grant, 1965; Corteen and Teitge, 2005; Docimo
et al., 2008; Colegate-Stone et al., 2010). The kinematic protocol
proposed here provides an indirect assessment of 3D clavicular
motion which could be useful to improve understanding of loading
and pathologies of the AC and SC joints.
The PCA showed that the functional dimension of shoulder
girdle movements may be limited to 3 principal components. The
rst factor, PC1 combined internal/external rotation and forward/
backward and medial/lateral translation, consistent with the classical anatomical and clinical denitions of protraction-retraction.
PC2 combined anterior/posterior tilting with upward/downward

a compact expression of 6D scapulothoracic kinematics: the measurement of the linear translation of the CS, in addition to 3D
scapular rotations. It also provides an indirect, non invasive and
global assessment of clavicular motion. This computation can be
easily implemented in any biomechanical method which provides
the 3D position of the scapular landmarks according to the ISB
protocol.
The fact that CS translations occurred during all the tasks suggests that clavicle motion occurs extremely frequently in daily life,
possibly leading to local constraints at the AC and SC joints. The
prevalence of AC and SC arthritis in asymptomatic subjects is high
(Kier et al., 1986; Needell et al., 1996; Stein et al., 2001). This
prevalence seems to be explained by the high and repetitive loads

Table 1
Amount of scapula rotation and displacement of its barycentre (CS) for planar arm elevation and simulated activities.
Arm motion task

Abduction

Time-point

t1

Scapular rotations ( )
Internal/External rotation
4.06 1.07
Medial/Lateral rotation
29.64 1.6
Anterior/posterior tilt
9.57 0.79
Translations of the barycentre CS (mm)
Medial/Lateral
4.2 1.2
Forward/backward
ns
Upward/downward
ns
3D Amplitude
10.0 1.1

Flexion

HC

BW

t2

t1

t2

t1

t2

t1

t2

ns
41.50 1.9
14.87 0.98

16.86 0.78
27.06 1.24
5.02 1.36

17.23 0.83
39.68 1.33
8.59 1.68

4.92 1
32.03 1.05
6.36 0.81

ns
28.34 3.64
7.80 1.21

2.8 0.52
6.11 0.47
8.06 0.74

ns
4.58 0.76
6.6 1.07

ns
ns
ns
13.1 1.6

24.5 1.8
16.1 2.1
ns
31.8 11.2

16.8 2.0
14.0 2.3
ns
25.2 1.8

12.6 1.8
ns
ns
30.2 1.8

7.5 1.7
5.9 1.2
ns
20.3 2.2

8.7
5.8
27.6
16.4

1.2
1.3
1.9
1.6

ns
ns
18.1 2.4
12.6 1.6

Values are mean differences sem of scapular rotation (in degrees) and displacement of its barycentre (CS) (in millimeters) measured between the initial posture (at t0) and
times t1 and t2 during the following tasks : Abduction, Flexion, simulated back washing (BW) and hair combing (HC). The time-point t1 corresponds to the rst part of
movement (elevation 90 ) for Abduction, Flexion and HC and to the rst local elevation maxima for BW. The time-point t2 to the time of target posture (maximal arm
elevation for Abduction, Flexion and HC or to the middle plateau for BW). Bold characters indicate p < .001, non-bold indicate p < .05. ns: non signicant difference. Italics
indicate values obtained for arm elevation >120 .

54

A. Roren et al. / Manual Therapy 20 (2015) 46e55

Table 2
Amount of scapula rotation and displacement of its barycentre (CS) for forward and backward rolling.
Arm motion task

F-Roll

Time-point

t1

B-Roll
t2

t3

t1

t2

t3

21.10 2.72
39.67 1.76
25.62 1.92

43.51 1.18
15.59 1.06
3.73 0.67

36.91 1.74
11.03 1.23
5 0.39

8.27 3.01
40.99 1.04
17.19 2.08

39.36 2.59
25.91 1.88
19.48 1.79

Scapular rotations ( )
Internal/External rotation
22.36 1.68
Medial/Lateral rotation
21.34 1.49
Anterior/posterior tilt
7.02 1.69
Translations of the barycentre CS (mm)
Medial/Lateral
11.1 1.9
Forward/backward
58.7 4.7
Upward/downward
49.2 5.1
3D Amplitude
77.8 6.8

37.9
27.4
96.8
108.9

3.8
5.2
6.8
7.9

84.4
33.1
13.3
92.6

3.3
2.3
3.2
3.3

65.5
24.4
1.8
71.2

3.7
3.0
2.0
3.8

41.7
17.0
87.2
100.4

3.4
4.6
6.7
6.8

8.5
86.3
86.3
123.8

4.2
6.0
6.9
8.7

Values are mean differences sem of scapular rotation (in degrees) and translation of its barycentre (CS) (in millimeters) measured between the initial posture (at t0) and
time-points t1, t2 and t3 during forward rolling (F-Roll) and backward rolling (B-Roll). The time-points t1 to t3 indicate the three main phases of the rolling movement. Bold
characters indicate p < .001, non-bold indicate p < .05.

translation (clinically termed shoulder shrugging). PC3 represented

medial/lateral rotation. This therefore suggests that internal/
external rotation and medial/lateral translation are strongly
coupled and that only 3 DoF are needed to describe scapular
movement in healthy subjects.
The coupling evidenced by the PCA could either be the result of
anatomical factors constraining scapular motion around the
thoracic wall and/or the synergic organisation of movement by the
central nervous system (Latash, 2008); two hypotheses which are
not incompatible. Indeed, it is known that both central neural
networks and local biomechanical constraints shape the control of
movement. This is the case, for example for synergies of the hand
and ngers (Santello et al., 2013) and control of the hand/arm
trajectory (Schaal and Sternad, 2001).

The ne coordination between clavicular and scapular motion

sub serving this coupling might be disrupted in pathological conditions. In the future, the analysis of CS translation should be used
as a complement to acromial methods for the evaluation of
shoulder kinematics in patients with different pathologies of the
shoulder complex (Ludewig and Cook, 2000; Vermeulen et al.,
2002; Lin et al., 2006; Fayad et al., 2008).

5. Limitations

Disclosures

The small size of the convenience sample included must be

considered in the interpretation of the results. Further studies are
needed in a larger group of healthy subjects, taking into account
their size or other anthropomorphic factors. The lack of control of
movement velocity during the tasks could be a limitation; however,
the humerothoracic and scapular recordings were time normalized.
Moreover, it has been shown that the velocity of arm elevation does
not modify 3D scapular kinematics (Fayad et al., 2006). The order of
the tasks and sides (left-right) were not randomized, however, the
recording procedure was not likely to generate fatigue in these
healthy subjects and no signicant differences for scapular rotations or translations were found between sides.

The co-authors do not have a direct nancial relation that might

lead to a conict of interest.

6. Conclusion
The functional decomposition of DoF evidenced by the PCA
provides an innovative model of scapular motion. It showed that
scapular motion involves simultaneous translations and rotations,
constrained by the curved surface of the thorax and likely also by
muscles and ligaments: the rst two PCs summarized motion on
this surface and the third PC related to rotation tangential to this
surface. This mechanical hypothesis is consistent with the model
proposed by Seth et al. (2010, 2012) who used 4 DoF to simulate
scapular movement over the thorax, represented by an ellipsoid,
and ts the data by Ludewig et al. (2009). The present study suggests that 3 DoF may be sufcient in order to fully describe scapular
motion in healthy subjects. Further studies with precise mechanical
models are needed to better understand the contribution of the
different elements of the shoulder complex to the kinematics of the
arm.

Funding
Agnes Roby-Brami is supported by INSERM and Institut
de
ratif de Recherche sur le Handicap. This work was supported
Fe
by the French National Agency for Research (ANR, Brahma project,
PSIROB-ROBO-003).

Acknowledgements
This work was supported by the French National Agency for
Research (Agence Nationale de la Recherche, Brahma project,
 de
ratif de Recherche sur le
PSIROB-ROBO-003) and by the Institut Fe
Handicap (National Institute of Health and Medical Research). The
authors thank Johanna Robertson for editing of the text.
References
Amasay T, Karduna AR. Scapular kinematics in constrained and functional upper
extremity movements. J Orthop Sports Phys Ther 2009;39(8):618e27.
Bao H, Willems PY. On the kinematic modelling and the parameter estimation of
the human shoulder. J Biomech 1999;32(9):943e50.
Biryukova EV, Roby-Brami A, Frolov AA, Mokhtari M. Kinematics of human arm
reconstructed from spatial tracking system recordings. J Biomech 2000;33(8):
985e95.
Codman M. The shoulder. Boston: Thomas Todd Co; 1934. pp. 1e513.
Colegate-Stone T, Allom R, Singh R, Elias DA, Standring S, Sinha J. Classication of
the morphology of the acromioclavicular joint using cadaveric and radiological
analysis. J Bone Jt Surg Br 2010;92(5):743e6.
Conway AM. Movements at the sternoclavicular and acromioclavicular joints. Phys
Ther Rev 1961;41:421e32.
Corteen DP, Teitge RA. Stabilization of the clavicle after distal resection: a biomechanical study. Am J Sports Med 2005;33(1):61e7.
Culham E, Peat M. Functional anatomy of the shoulder complex. J Orthop Sports
Phys Ther 1993;18(1):342e50.
Dayanidhi S, Orlin M, Kozin S, Duff S, Karduna A. Scapular kinematics during humeral elevation in adults and children. Clin Biomech (Bristol, Avon) 2005;20(6):
600e6.
Docimo Jr S, Kornitsky D, Futterman B, Elkowitz DE. Surgical treatment for acromioclavicular joint osteoarthritis: patient selection, surgical options, complications, and outcome. Curr Rev Musculoskelet Med 2008;1(2):154e60.

A. Roren et al. / Manual Therapy 20 (2015) 46e55

Fayad F, Hoffmann G, Hanneton S, Yazbeck C, Lefevre-Colau MM, Poiraudeau S, et al.
3-D scapular kinematics during arm elevation: effect of motion velocity. Clin
Biomech (Bristol, Avon) 2006;21(9):932e41.
Fayad F, Roby-Brami A, Yazbeck C, Hanneton S, Lefevre-Colau MM, Gautheron V,
et al. Three-dimensional scapular kinematics and scapulohumeral rhythm in
patients with glenohumeral osteoarthritis or frozen shoulder. J Biomech
2008;41(2):326e32.
Giphart J, Brunkhorst J, Horn N, Shelburne K, Torry M, Millett P. Effect of plane of
arm elevation on glenohumeral kinematics: a normative biplane uoroscopy
study. J Bone Jt Surg Am 2013;95(3):238e45.
Grant JCB. Method of anatomy. 7th ed. Baltimore: Williams and Wilkins; 1965.
pp. 1e777.
Hallaceli H, Gunal I. Normal range of scapular elevation and depression in healthy
subjects. Arch Orthop Trauma Surg 2002;122(2):99e101.
Hanneton S, Dedobbeler S, Hoellinger T, Roby-Brami A. Direct kinematic modeling
of the upper limb during trunk-assisted reaching. J Appl Biomech 2011;27(3):
272e7.
Helgadottir H, Kristjansson E, Mottram S, Karduna AR, Jonsson Jr H. Altered scapular
orientation during arm elevation in patients with insidious onset neck pain and
whiplash-associated disorder. J Orthop Sports Phys Ther 2010;40(12):784e91.
Helgadottir H, Kristjansson E, Mottram S, Karduna A, Jonsson Jr H. Altered alignment of the shoulder girdle and cervical spine in patients with insidious onset
neck pain and whiplash-associated disorder. J Appl Biomech 2011;27(3):
181e91.
Inman V, Saunders J. Observations on the function of the clavicle. Calif Med
1946;65(4):158e66.
Inman VT, Saunders JB, Abbott LC. Observations of the function of the shoulder
joint. 1944. Clin Orthop Relat Res 1996;330:3e12.
Karduna AR, McClure PW, Michener LA, Sennett B. Dynamic measurements of
three-dimensional scapular kinematics: a validation study. J Biomech Eng
2001;123(2):184e90.
Kier R, Wain SL, Apple J, Martinez S. Osteoarthritis of the sternoclavicular joint.
Radiographic features and pathologic correlation. Invest Radiol 1986;21(3):
227e33.
Latash ML. Synergy. New York: Oxford University Press; 2008. pp. 1e412.
Lempereur M, Brochard S, Burdin V, Remy-Neris O. Difference between palpation
and optoelectronics recording of scapular motion. Comput Methods Biomech
Biomed Engin 2010;13(1):49e57.
Lenarcic J, Stanisic M. A humanoid shoulder complex and the humeral pointing
kinematics. IEEE Trans Rob Autom 2003;19(3):499e506.
Lin JJ, Hanten WP, Olson SL, Roddey TS, Soto-Quijano DA, Lim HK, et al. Shoulder
dysfunction assessment: self-report and impaired scapular movements. Phys
Ther 2006;86(8):1065e74.
Ludewig PM, Behrens SA, Meyer SM, Spoden SM, Wilson LA. Three-dimensional
clavicular motion during arm elevation: reliability and descriptive data.
J Orthop Sports Phys Ther 2004;34(3):140e9.
Ludewig PM, Cook TM. Alterations in shoulder kinematics and associated muscle
activity in people with symptoms of shoulder impingement. Phys Ther
2000;80(3):276e91.
Ludewig PM, Cook TM, Nawoczenski DA. Three-dimensional scapular orientation
and muscle activity at selected positions of humeral elevation. J Orthop Sports
Phys Ther 1996;24(2):57e65.
Ludewig P, Phadke V, Braman J, Hassett D, Cieminski C, Laprade R. Motion of the
shoulder complex during multiplanar humeral elevation. J Bone Jt Surg Am
2009;91(2):378e89.
Ludewig PM, Reynolds JF. The association of scapular kinematics and glenohumeral
joint pathologies. J Orthop Sports Phys Ther 2009;39(2):90e104.
Magermans DJ, Chadwick EK, Veeger HE, van der Helm FC. Requirements for upper
extremity motions during activities of daily living. Clin Biomech (Bristol, Avon)
2005;20(6):591e9.
McClure PW, Michener LA, Sennett BJ, Karduna AR. Direct 3-dimensional measurement of scapular kinematics during dynamic movements in vivo. J Shoulder
Elbow Surg 2001;10(3):269e77.
McQuade KJ, Smidt GL. Dynamic scapulohumeral rhythm: the effects of external
resistance during elevation of the arm in the scapular plane. J Orthop Sports
Phys Ther 1998;27(2):125e33.
Meskers CG, van de Sande MA, de Groot JH. Comparison between tripod and skinxed recording of scapular motion. J Biomech 2007;40(4):941e6.
Meskers CG, Vermeulen HM, de Groot JH, van Der Helm FC, Rozing PM. 3D shoulder
position measurements using a six-degree-of-freedom electromagnetic
tracking device. Clin Biomech (Bristol, Avon) 1998;13(4e5):280e92.
Needell SD, Zlatkin MB, Sher JS, Murphy BJ, Uribe JW. MR imaging of the rotator
cuff: peri tendinous and bone abnormalities in an asymptomatic population.
Am J Roentgenol 1996;166(4):863e7.
Parel I, Cutti AG, Fiumana G, Porcellini G, Verni G, Accardo AP. Ambulatory measurement of the scapulohumeral rhythm: intra- and inter-operator agreement
of a protocol based on inertial and magnetic sensors. Gait Posture 2012;35(4):
636e40.

55

Peat M. Functional anatomy of the shoulder complex. Phys Ther 1986;66(12):

1855e65.
Pronk GM, van der Helm F, Rozendaal L. Interaction between the joints in the
shoulder mechanism: the function of the costoclavicular, conoid and trapezoid
ligaments. Proc Institution Mech Eng Part J Eng Med 1993;207(4):219e29.
Roren A, Fayad F, Roby-Brami A, Revel M, Fermanian J, Poiraudeau S, et al. Precision
of 3D scapular kinematic measurements for analytic arm movements and activities of daily living. Man Ther 2013;18(6):473e80.
Roren A, Lefevre-Colau MM, Roby-Brami A, Revel M, Fermanian J, Gautheron V, et al.
Modied 3D scapular kinematic patterns for activities of daily living in painful
shoulders with restricted mobility: a comparison with contralateral unaffected
shoulders. J Biomech 2012;45(7):1305e11.
Rundquist PJ. Alterations in scapular Kinematics in subjects with idiopathic loss of
shoulder range of motion. J Orthop Sports Phys Ther 2007;37(1):19e25.
Rundquist PJ, Obrecht C, Woodruff L. Three-dimensional shoulder kinematics to
complete activities of daily living. Am J Phys Med Rehabil 2009;88(8):623e9.
Sahrmann S. Diagnosis and treatment of movement impairment syndromes by
Shirley. Saint Louis, USA: Mosby; 2001.
rntell H. Neural bases of hand synergies. Front Comput
Santello M, Baud-Bovy G, Jo
Neurosci 2013;7(23):1e15.
Schaal S, Sternad D. Origins and violations of the 2/3 power law in rhythmic threedimensional arm movements. Exp Brain Res 2001;136(1):60e72.
Senk M, Cheze L. A new method for motion capture of the scapula using an optoelectronic tracking device: a feasibility study. Comput Methods Biomech Biomed Engin 2010;13(3):397e401.
Seth A, Matias R, Veloso A, Delp S. A scapulothoracic joint model for fast and accurate simulations of upper-extremity motion. Gainesville, Florida: ASB 2012,
American Society of Biomechanics; 2012. http://www.asbweb.org/conferences/
2012/abstracts/12.pdf.
Seth A, Sherman M, Eastman P, Delp S. Minimal formulation of joint motion for
biomechanisms. Nonlinear Dyn 2010;62(1):291e303.
Sheikhzadeh A, Yoon J, Pinto VJ, Kwon YW. Three-dimensional motion of the
scapula and shoulder during activities of daily living. J Shoulder Elbow Surg
2008;17(6):936e42.
Solem-Bertoft E, Thuomas KA, Westerberg CE. The inuence of scapular retraction
and protraction on the width of the subacromial space. MRI Study 1993;296:
99e103.
SPACE FASTRAK. User's manuel, revision F. Colchester, VT: Polhemus Inc; 1993.
Stein BE, Wiater JM, Pfaff HC, Bigliani LU, Levine WN. Detection of acriomioclavicular joint pathology in asymptomatic shoulders with magnetic resonance
imaging. J Shoulder Elbow Surg 2001;10(3):204e8.
Teece RM, Lunden JB, Lloyd AS, Kaiser AP, Cieminski CJ, Ludewig PM. Threedimensional acromioclavicular joint motions during elevation of the arm.
J Orthop Sports Phys Ther 2008;38(4):181e90.
Triftt P. The relationship between motion of the shoulder and the stated ability to
perform activities of daily living. J Bone Joint Surg Am 1998;80(1):41e6.
Tsai NT, McClure PW, Karduna AR. Effects of muscle fatigue on 3-dimensional
scapular kinematics. Arch Phys Med Rehabil 2003;84(7):1000e5.
Tytherleigh-Strong G. Pectoral girdle, shoulder region and axilla. In: Standring S,
Borley NR, Crossman AR, Newell RLM, Johnson D, Collins P, et al., editors. Gray's
anatomy: the anatomical basis of clinical practice. London: Churchill Livingstone Elsevier; 2008. p. 803.
van Andel C, van Hutten K, Eversdijk M, Veeger D, Harlaar J. Recording scapular
motion using an acromion marker cluster. Gait Posture 2009;29(1):123e8.
van der Helm FC. A nite element musculoskeletal model of the shoulder mechanism. J Biomech 1994;27(5):551e69.
van der Helm FC, Pronk GM. Three-dimensional recording and description of motions of the shoulder mechanism. J Biomech Eng 1995;117(1):27e40.
van der Helm FCT. A standardized protocol for motion recordings of the shoulder.
In: Proceedings of the First Conference of the ISG. Delft; 1997. pp. 7e12.
Vermeulen HM, Stokdijk M, Eilers PH, Meskers CG, Rozing PM, Vliet Vlieland TP.
Measurement of three dimensional shoulder movement patterns with an
electromagnetic tracking device in patients with a frozen shoulder. Ann Rheum
Dis 2002;61(2):115e20.
Wang CH, McClure P, Pratt NE, Nobilini R. Stretching and strengthening exercises:
their effect on three-dimensional scapular kinematics. Arch Phys Med Rehabil
1999;80(8):923e9.
Woltring HJ, Huiskes R, de Lange A, Veldpaus FE. Finite centroid and helical axis
estimation from noisy landmark measurements in the study of human joint
kinematics 1985;18(5):379e89.
Wu G, van der Helm FC, Veeger HE, Makhsous M, Van Roy P, Anglin C, et al. ISB
recommendation on denitions of joint coordinate systems of various joints for
the reporting of human joint motionePart II: shoulder, elbow, wrist and hand.
J Biomech 2005;38(5):981e92.
Yano Y, Hamada J, Tamai K, Yoshizaki K, Sahara R, Fujiwara T, et al. Different
scapular kinematics in healthy subjects during arm elevation and lowering:
glenohumeral and scapulothoracic patterns. J Shoulder Elbow Surg 2010;19(2):
209e15.