Clinical Biomechanics 21 (2006) 692–700

www.elsevier.com/locate/clinbiomech

In vivo measurements of subacromial impingement:

Substantial compression develops in abduction
with large internal rotation
a,b,*
Toshimasa Yanai , Franz K. Fuss c, Tetsuo Fukunaga d

a
School of Life System Science and Technology, Chukyo University, 101 Tokodachi, Kaizu-cho, Toyota 470-0393, Japan
b
School of Physical Education, University of Otago, Dunedin, New Zealand
c
Division of BioEngineering, College of Engineering, Nanyang Technological University, Singapore
d
School of Sport Science, Waseda University, Tokorozawa 359-1192, Japan

Received 20 September 2005; accepted 1 March 2006

Abstract

Background. Subacromial impingement is a widely recognized mechanism of chronic shoulder pain. The magnitudes of the compres-
sive forces that impinge the subacromial structures were often measured from cadaveric specimens, but it is questionable to use this data
as a sole basis to determine the shoulder motions and/or shoulder configurations that induce impingement in live subjects performing
active motion. The purpose of the present study was to determine in vivo the magnitude of the compressive force at selected shoulder
configurations.
Methods. The subacromial structures may be impinged by the downward-directed forces exerted by the coraco-acromial ligament. The
reactions of these forces push the ligament upwards and deform it into a curved shape. A single resultant of these reaction forces was
determined with an inverse approach to quantify the magnitude of the impingement force. An ultrasound unit was used to visualize the
deformed shape of the coraco-acromial ligament for thirteen subjects with no symptomatic shoulder problem actively holding their
shoulders in five configurations.
Findings. The impingement force in 90° abduction + maximum internal rotation (mean = 21.3 N) and that in the Hawkins test posi-
tion (mean = 18.3 N) were significantly greater than those in 90° abduction + neutral and external rotation (means 63 N).
Interpretation. For young asymptomatic shoulders, the motions that induce impingement are not any arm abduction, but the arm
abduction with a large internal rotation. Further study is indicated to examine the impingement force among various age groups.
Ó 2006 Elsevier Ltd. All rights reserved.

Keywords: Compression; Injury; Range of motion; Rotator cuff; Shoulder

1. Introduction impingement, or compression, of the subacromial struc-

After introduced by Neer (1972), subacromial impinge-
ment has been the most widely recognized mechanism of
chronic shoulder pain. With this mechanism, the rotator
cuff pathology is explained as the consequence of repeated
*
Corresponding author. Address: School of Life System Science and
Technology, Chukyo University, 101 Tokodachi, Kaizu-cho, Toyota 470-
0393, Japan.
E-mail address: tyanai@life.chukyo-u.ac.jp (T. Yanai).
tures between the greater tubercle of the humerus and the
coraco-acromial arch. The tendons of the supraspinatus,
the infraspinatus and the long head of the biceps brachii
are often damaged by the repeated impingement, leading
to degenerative tendonitis and ultimately tendon rupture.
Neer (1972) coined the term ‘‘impingement syndrome’’
for such shoulder problem.
The magnitude of the compressive force that impinges
the subacromial structures under the coraco-acromial arch
has been measured on cadaveric specimens (Payne et al.,

0268-0033/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.clinbiomech.2006.03.001
 T. Yanai et al. / Clinical Biomechanics 21 (2006) 692–700 693

1997; Wuelker et al., 1994, 1995). In these studies, the
humerus was moved relative to the stabilized scapula by
means of computer-operated actuators that simulated the
muscle activities of the rotator cuff and the Deltoid, and
the compressive force developed under the coraco-acromial
arch was measured directly with pressure transducers.
These studies demonstrated direct evidence to support that
the compressive force under the coraco-acromial arch
increased as the arm was elevated and that the compressive
forces were distributed over the undersurface of coracoid,
coraco-acromial ligament and acromion. There are, how-
ever, some concerns about the possible overestimation of
the compressive force measured with this approach because
the limited number of muscles running across the glenohu-
meral joint were used to simulate the arm motion and the
artificial activation pattern of muscles might not be an accu-
rate representation of the human movement. In fact, Shar-
key and Marder (1995) demonstrated that the superior
migration of the humeral head was significantly greater in
the cadaver model simulating the humerus movements
mechanically than in live subjects elevating their arms
actively. The increased superior migration of the humeral
head might be due to the absence of the humeral head
depressors such as teres major and latissimus dorsi (Hadler
et al., 2001). A study to examine the compressive force
under the coraco-acromial arch of live human subjects
actively elevating their arms is, therefore, needed to confirm
the evidence demonstrated by the cadaveric studies. The
purpose of this study was to determine in vivo the magni-
tude of the compressive force developed under the coraco-
acromial ligament at selected shoulder configurations.
2. Methods
The subacromial structures being impinged under the
coraco-acromial ligament are compressed by the distrib-
uted downward-directed forces exerted by the undersurface
of the coraco-acromial ligament and the distributed
upward-directed forces exerted by the upper surface of
the humeral head. In reaction, the subacromial structures
push the coraco-acromial ligament upwards by the distrib-
uted upward-directed forces and deform the ligament into
a curved shape (Fig. 1). In the present study, a single resul-
tant of the distributed reaction forces exerted on the cor-
aco-acromial ligament by the subacromial structures was
determined with an inverse approach to quantify the mag-
nitude of the impingement force.
Thirteen male graduate students with no symptomatic
shoulder problem volunteered to participate in this study.
After having provided written informed consent, each sub-
ject was asked to actively maintain his arm in specified
positions for 15–30 s each in the following order: (1) ana-
tomical (neutral) position, (2) 90° abduction + maximum
internal rotation, (3) 90° abduction + neutral rotation, (4)
90° abduction + maximum external rotation, and (5) the
Hawkins impingement test position. Named after the
orthopedic surgeon who introduced this clinical test (Haw-
kins and Kennedy, 1980), the Hawkins test is widely used
to diagnose the pathological condition of subacromial
structures. In this test, the patient’s arm is elevated to 90°
in the sagittal plane and the humerus is internally rotated
passively and forcibly beyond the normal active range so
as to induce compression under the coraco-acromial arch
(Hawkins and Kennedy, 1980). If the patient has a patho-
logical condition of the subacromial structures, localized
pain is elicited at the subacromial region. As the Hawkins
impingement test was reported to be more sensitive to the
impingement of the rotator cuff under the coraco-acromial
ligament than the other clinical test used commonly for
evaluating impingement syndrome (Valadie et al., 2000),
this provocative clinical test was included in the present
study to estimate the impingement force that is large

Fig. 1. The ultrasound images of the coraco-acromial ligament recorded from the same subject. The occurrence of impingement is observable from the
shape of the coraco-acromial ligament. When the subacromial structures are not impinged under the coraco-acromial ligament, no force develops between
the upper surface of the subacromial structures and the undersurface of the ligament, and thus the ligament remains flat (as illustrated in the left figure).
When impinged, the subacromial structures are compressed by the distributed downward-directed forces exerted by the undersurface of the coraco-
acromial ligament and the distributed upward-directed forces exerted by the upper surface of the humeral head (as illustrated in the right figure). In
reaction, the subacromial structures push the coraco-acromial ligament upwards by the distributed upward-directed forces (indicated by the arrows in the
figure on the right) and deform the ligament into a curved shape.
694 T. Yanai et al. / Clinical Biomechanics 21 (2006) 692–700
enough for impingement patients to elicit pain. An ade-
quate time interval was taken between the successive mea-
surements to minimize the influences that the viscoelastic
effect and the muscle fatigue might make on the measure-
ments. In addition, the order of the measurements was
arranged to maximize the time interval between the mea-
surements of the positions in which large ligamentus defor-
mations were expected. An unpublished pilot study showed
that the deformation of the ligament was maximal in the
Hawkins test position and the abduction + internally rota-
tion, and was minimal (near zero) in the other positions.
The order of the measurements was, therefore, arranged
to separate the abduction + internal rotation from the
Hawkins test position, and to place between them two posi-
tions in which the deformations were minimal.
While the subject was maintaining each arm position,
the shape of the coraco-acromial ligament was visualized
and recorded with an ultrasound unit. An ultrasound unit
(EUB-6500, Hitachi Medico, Tokyo, Japan) with an elec-
tronic linear array probe of 10 MHz wave frequency was
used for the first five subjects and another ultrasound unit
(ProSound 5500, ALOCA, Tokyo, Japan) with an elec-
tronic linear array probe of 13 MHz wave frequency was
used for the rest of the subjects to visualize and record
the shape of the coraco-acromial ligament at each shoulder
configuration. The ultrasound unit with a higher frequency
was used to improve the clarity of the recorded images of
the coraco-acromial ligament. The increase in the resolu-
tion, however, did not result in a substantial improvement
on the clarity of the recorded images of the approximately
30 mm long and 1 mm thick ligament. The measurements
from all subjects were, therefore, included in the present
study.
The probe was positioned approximately perpendicular
to the plane of the ligament, so that the technique of 2D
videography could be used to determine the shape of the
ligament at rest and in the positions described above. To
standardize the positioning of the probe, the following
steps were taken: (1) The probe was moved over the acro-
mial area to visualize the coracoid and the acromion simul-
taneously on the monitor, and identified the coraco-
acromial ligament that spanned between the two bony
prominences. (2) The probe was translated laterally and
rotated adequately to visualize the lateral edge of the cor-
aco-acromial ligament on the monitor. (3) The probe was
translated medially by a few millimeters to visualize the
axial deformation of the coraco-acromial ligament at the
middle of the ligament’s width. (4) The probe was slowly
tilted in the medio-lateral direction to find the optimum
angle at which the image of the upper and lower surfaces
of the coraco-acromial ligament was most intense and dis-
tinct. The visualized image was then recorded. The step (4)
ensured that the probe was positioned approximately per-
pendicular to the plane of the ligament, because the inten-
sity and distinctness of the ultrasound image had positive
relation to the amplitudes of the echoed ultrasound waves
received by the probe, which, in turn, attained the largest
values when the ultrasound waves were echoed perpendi-
cularly from the reflecting surface, i.e., the ligament’s sur-
faces (Ito and Mochizuki, 2002).
The recorded ultrasound images were manually digitized
with a motion analysis software (Frame-DIASII, DKH,
Tokyo, Japan). The insertions of the ligament were visually
identified, and thirty approximately equally spaced points
were digitized along each contour of the upper and lower
surfaces of the coraco-acromial ligament. The digitized
data were scaled from digitizing unit to the real world
dimension [mm] and transformed into a coordinate system
that had the origin at the lowest left corner of the digitized
landmarks and the abscissa directed from the origin to the
lowest right corner of the digitized landmarks. The multiple
regression was applied to fitting and testing higher-order
polynomial models, and the order and the coefficients of
the best-fit polynomial equation were determined to
express the shape of the upper and lower surfaces of the lig-
ament mathematically. The shape of the ligament as a
whole was determined by averaging the two best-fit polyno-
mial equations for each shoulder configuration. Each coef-
ficient of the determined average polynomial equation was
computed as the mean value of the two coefficients of the
corresponding order determined for the two surfaces. The
length of the ligament for the given shoulder configuration
was determined from the averaged best-fit equation, as fol-
lows (Anton, 1988):
Z xa qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
L¼ 1 þ ½f 0 ðxÞ dx
xc
L is the length of the ligament at given shoulder configura-
tion; f 0 (x), the first derivative of the averaged best-fit
polynomial equation f(x); xa, the domain of the best-fit
polynomial at the acromial attachment and xc is the do-
main of the best-fit polynomial at the coracoid attachment.
Strictly speaking, the shape and the length of the coraco-
acromial ligament determined with the present method did
not measure those of the ligamentus portion of the coraco-
acromial ligament alone, but they are rather the shape and
the length of the entire structure of the coraco-acromial lig-
ament including the bone-ligament junctions at the two
attachment sites. The material and structural properties
of the coraco-acromial ligament inclusive of the two
attachment sites, therefore, were adopted for modeling
the ligament for the present study. Such properties were
found in the study of Soslowsky et al. (1994) in which they
measured the material and structural properties of the
specimens of the coraco-acromial ligament as a whole
(inclusive of the two attachments) and in another study
of Soslowsky et al. (1996) in which they measured the stress
and strain naturally imposed on the ligament in situ. The
properties adopted for the present study are summarized
in Table 1.
A two-dimensional static analysis was conducted to
determine the forces acting on the coraco-acromial liga-
ment. First, a single resultant of the distributed tensile
forces acting across the cross-sectional area of the ligament
 T. Yanai et al. / Clinical Biomechanics 21 (2006) 692–700
Table 1
Model ligament
Elastic Cross-sectional Resting Resting stress
modulus (E) area (A) length (L0) (r0)
290 MPa 0.000008 m2 D/1.049a 1.9 MPa
The adopted elastic modulus was determined as the overall property of the
specimens of coraco-acromial ligaments which included the two attach-
ments (bones and bone-ligament junctions) at coracoid and acromion.
a
D = distance between the two insertions of the ligament.
was determined. In their 1994 study, Soslowsky et al. pre-
sented a stress–strain curve for the coraco-acromial liga-
ment which illustrated the existence of the ‘‘toe region’’
in the range of strain from 0% to approximately 2.4%.
The in situ strain of 4.9% was assumed to be in the linear
region of the stress–strain relationship, the stress was deter-
mined using the following formula:
ExtraStrain ¼ ðL  L0 Þ=L0  1:049
Stress ¼ E  ExtraStrain þ r0
Tensile force ¼ Stress  A
The single resultant of the distributed tensile forces acting
across a given cross-sectional area of the coraco-acromial
ligament was considered to be directed normal to the
cross-sectional area, and the magnitude was considered to
be uniform at every cross-sectional area along the entire
length of the ligament. At the insertions of the ligament,
the tensile force was exerted by the coracoid and the acro-
mion (Fig. 2, FCoracoid and FAcromion), which acted as the
constraint forces to maintain the constant distance between
the attachment sites. At a given shoulder configuration, the
coraco-acromial ligament deformed into a certain shape
due to the distributed external forces acting over the upper
and lower surfaces of the ligament. The sum of these con-
straint forces and the distributed external forces acting on
the coraco-acromial ligament (Fig. 2) equaled zero because
the shoulder structures were not in motion in the present
experiments. The equation of the motion could, therefore,
be expressed as follows:
Fig. 2. Free-body diagram of the coraco-acromial ligament.
695
F ¼ F Deltoid þ F Impinge þ F Coracoid þ F Acromion ¼ 0
F is the resultant external force acting on the ligament;
FDeltoid, the single resultant of the distributed forces acting
on the upper surface of the ligament; FImpinge, the reaction
force to the single resultant of the distributed forces that
impinge the subacromial structures; FCoracoid, the resultant
force exerted on the ligament by the coracoid and FAcromion
is the resultant force exerted on the ligament by the
acromion.
The distributed forces acting on the upper surface of the
ligament (FDeltoid) were assumed zero in the present study.
This assumption is believed to be reasonable because the
ligament was flat in the anatomical position for all subjects
who participated in the present study although the evidence
demonstrated that the subacromial structure was not in
contact with the undersurface of the ligament to push it
upwards in this position (Burns and Whipple, 1993).
The directions of the constraint forces (FCoracoid and
FAcromion) were computed as the first derivatives of the
best-fit polynomial equation at the respective insertions,
and the magnitudes of these constraint forces equaled the
magnitude of the tensile force. With the modeling informa-
tion, the single resultant of the distributed reaction forces
that impinge the subacromial structures (FImpinge) was
determined.
A repeated-measures single-factor ANOVA was used to
test the effect of different shoulder configurations on the
magnitudes of the impingement force (FImpinge) at a signif-
icant level of 0.05. Post hoc tests were conducted to com-
pare the mean magnitudes of the impingement force
recorded for the five shoulder configurations.
3. Results
The ultrasound images showed that the coraco-acromial
ligament that was flat initially at the anatomical position
was pushed upward by the subacromial structures and
deflected to form a curved shape when the arm was ele-
vated and internally rotated (Fig. 3). The extent of the
deformation exhibited at the maximally internally rotated
position was similar to that provoked at the Hawkins test
position. These observations reflected well in the deter-
mined impingement forces (Fig. 4). The mean magnitude
of the impingement force recorded in the abduction + max-
imum internal rotation (21.3 N, SD 13.2 N) and that in the
Hawkins test position (18.3 N, SD 12.0 N) were not signif-
icantly different from each other, but they were signifi-
cantly greater (p < 0.01) than those determined in other
positions (53 N).
4. Discussion
The present study demonstrated that (a) the impinge-
ment force developed under the coraco-acromial liga-
ment was negligibly small when the abducted humerus
was held in neutral or externally-rotated position, (b) the
696 T. Yanai et al. / Clinical Biomechanics 21 (2006) 692–700

Fig. 3. Ultrasound images of the coraco-acromial ligament recorded from the same subject positioning his shoulder in the five configurations. The
ligament appears as a white line connecting the coracoid and acromion. The ligament is flat in the anatomical position and in the abducted position with
neutral rotation and the maximum external rotation. The flat shaped ligament indicates that the subacromial structures are not impinged under the coraco-
acromial ligament. The ligament was pushed upward by the subacromial structures and deflected to form a curved shape when the abducted arm was
rotated internally to the maximum, indicating that the subacromial structures are impinged under the coraco-acromial ligament. The sharpness of the
curvature in this position was similar to that exhibited in the Hawkins test position, indicating that the intensity of impingement is similar in these two
shoulder configurations.

60 tures at an intensity similar to that provoked with the Haw-

kins impingement test.
50
The major limitation of the present approach is that the
modeling parameters for the structural property (dimen-
Impingement force [N]

40
30
20
10
0
Anatomical position External rotation Neutral rotation
Fig. 4. The impingement forces recorded from all 13 subjects positioned in
the five shoulder configurations.
impingement force developed under the coraco-acromial
ligament increased when the abducted humerus was inter-
nally rotated actively, and (c) the magnitudes of the
impingement force recorded from the abducted shoulders
with the maximum internal rotation attained similar values
to those recorded with the Hawkins impingement test. The
results indicate that the shoulder motions that induce
impingement are not any arm abduction, but rather the
arm abduction with a large internal rotation. The results
also indicate that the shoulder motions beyond the active
range are not necessary to impinge the subacromial struc-
sion) and material property (elastic modulus) of the
ligament were adopted from the literature, and the magni-
tudes of the compressive force for the given deformation of
the subject’s ligament were determined mathematically.
The validity of the present methodology used for determin-
ing the impingement force was tested with a cadaveric
study. Ten shoulder specimens were dissected from ten
Internal rotation Hawkins test
fresh bodies within 48 h after the death. Each shoulder
specimen was fixed to a wooden frame, which provided a
firm support for the measurements. The known magnitudes
of forces in the range from 10 to 70 N were applied in the
normal direction to the undersurface of the coraco-acro-
mial ligament, and the axial deformation and the shape
of the ligament were recorded using an electromagnetic
digitizer/tracker (Polhemus 3Space, Isotrak M100, Col-
chester, VT, USA). The axial deformation and the shape
of the ligament were used as input for the mathematical
model, and the magnitude of the force applied to the liga-
ment was estimated. For each specimen, there was a signif-
icant correlation between the applied and estimated forces
(mean = 0.96, 95% confidence interval ranged from 0.93 to
0.99). The specimen-to-specimen variability in the slope of
the regression line (Fig. 5) was found high (mean = 1.06
 T. Yanai et al. / Clinical Biomechanics 21 (2006) 692–700
80
60
Estimated Force [N]
40
20
0 aco-acromial ligament (12 mm), the maximum deviation
0 20 40 60
Applied Force [N]
Fig. 5. The relation between the estimated force and the applied force for
the ten fresh cadaver specimens.
and 95% confidence interval ranged from 0.54 to 1.58)
whereas the intersection of the regression line and the ordi-
nate was near zero for most subjects (mean = 0.57 and the
95% confidence interval ranged from 2.0 to 3.1). Since the
slope indicates the ratio of the modeled value of the elastic
modulus (290 MPa) to the true value of the specimen’s
elastic modulus, the true value could be estimated for each
specimen from the slope. The estimated elastic modulus
was found to have the mean value of 318 MPa with the
standard deviation of 154 MPa, which matched well with
the values (290 MPa, SD 154 MPa) reported by Soslowsky
et al. (1994). Although these values are estimated with the
assumption that the specimen-to-specimen variability in
the ligament’s cross-sectional area is negligible, the close
matching with the Soslowski’s data supports that the
observed range of slope could be explained to an extent
by an individual variability in the elastic modulus. These
results indicate that the comparisons of the relative intensi-
ties of the impingement force recorded for the given shoul-
der in different configurations yield accurate results, but the
comparisons of the absolute intensity of the impingement
force between subjects may not account for actual results.
Another limitation was associated with the data collec-
tion procedure. Although a standardized procedure
described earlier was used to align the probe with the longi-
tudinal axis of the ligament and to position the probe per-
pendicular to the plane of the ligament (Fig. 6A), some
degree of misalignment and/or deviation might have caused
an error in the determined impingement force. The maxi-
mum error due to misalignment occurs when the ligament
is scanned diagonally (Fig. 6B). For the 30 mm long and
8 mm wide ligament, the maximum misalignment angle is
approximately 15° [= tan1 (8/30)]. The recorded image of
the ligament with this extent of misalignment is expected
697
to have 1.035 [= 1/cos 15°] times greater length between
the attachments (L) and the same height of bending (H)
when compared with the correctly scanned image of the lig-
ament. This misalignment should, thus, cause an underesti-
mation in the determined impingement force. The
magnitude of the underestimation was estimated for the
data collected in the present study and found that the mean
values across the subjects were <0.1 N for the anatomical
position and the abductions with neutral and external rota-
tions, 1.5 N for the abduction with internal rotation, and
1.1 N for the Hawkins test. The maximum error due to a
deviation of the probe occurs when the ultrasound waves
echoed obliquely from the ligament surfaces are received
only just by the edge of the probe (Fig. 6C). For the given
dimension of the alley transducers of the probe (10 mm
width) and the distance between the probe and the cor-
of the probe was estimated to be 23° [= tan1 (5/12)]. The
80
recorded image of the ligament with this extent of deviation
is expected to have the same length between the attachments
(L) and 1.086 [= 1/cos 23°] times greater height (H) when
compared with the correctly scanned image of the ligament.
This deviation should, thus, cause an overestimation in the
determined impingement force. The magnitude of the over-
estimation was determined and found that the mean values
across the subjects were <1.5 N for the anatomical position
and the abductions with neutral and external rotations,
4.5 N for the abduction with internal rotation, and 3.2 N
for the Hawkins test. These results corresponded well with
the results obtained from the test-retest reliability analysis
in which repeated measurements were taken for eight sub-
jects. The mean value of the absolute difference between
the two repeated measurements for all trials for all subjects
was 3.4 N. The difference in the repeated measures fell
within the estimated range of error due to the misalignment
and deviation of the probe, indicating that (a) the estimated
range of error predict the measurement error well and (b)
the error associated with the present two-dimensional
approach does not jeopardize the main finding of the pres-
ent study.
The main finding of the present study is that the shoul-
der motions inducing impingement are not any abduction,
but rather the abduction plus large internal rotation. This
result is compared with three relevant studies found in the
literature (Table 2). These studies are unique combina-
tions of the study design (in vivo vs in vitro) and the type
of motion used for the analysis (active vs passive motion).
The finding was consistent with the in vivo study of pas-
sive motions (Nobuhara, 2003) and the in vitro study of
simulated active motions (Payne et al., 1997). In the study
of Nobuhara, the impingement force under the coraco-
acromial ligament was measured directly from 260
patients by using a pressure sensor called the ICP catheter
(Nihon-Kohden, Tokyo, Japan) during open surgery.
Nobuhara reported that (a) the subacromial pressure
exhibited ‘‘no marked increase’’ during the arm elevation
and depression in the scapular plane, (b) the subacromial
698 T. Yanai et al. / Clinical Biomechanics 21 (2006) 692–700

(A) Correct positioning of probe

e
ob ce
rs
Pr f tr
an
sd
u Recorded ultrasound image

le yo
Al

Coraco-acromial
H
ligament
L
H

(B) Misalignment of probe (C) Deviation of probe from

perpendicular plane

H H/cos
L/cos L

Fig. 6. Possible sources of error in two-dimensional scanning for the measurement of axial deformation of the coraco-acromial ligament. (A) The axial
deformation of the coraco-acromial ligament can be scanned correctly by aligning the probe with the long-axis of the ligament and directing the
ultrasound waves straight toward the ligament so that the waves reflect on the ligament surfaces at 90°. (B) The misalignment of the probe might occur
within the range of ±15° for the given width (approx. 8 mm) and the length (approx. 30 mm) of the coraco-acromial ligament, which was estimated to
cause an underestimation error of <1.5 N for the 13 subjects. (C) The deviation of the probe from the perpendicular plane might occur within the range of
±23° for the given width of the probe transducer (10 mm) and the given distance between the probe and the ligament (approx. 12 mm), which was
estimated to cause an overestimation error of <4.5 N. The V-shaped lines indicate the ultrasound waves emitted from the probe and the echoed (reflected)
waves traveling back to the probe. If the angle / exceeds ±23°, the echoed waves do not travel back to the probe and the ligament cannot be visualized in
the recorded image.

pressure exhibited ‘‘no noted increase’’ during the internal tion, and (c) the subacromial pressure exhibited a ‘‘marked
and external rotation when the arm was in adducted posi- increase’’ with internal rotation and a ‘‘decrease’’ with
external rotation when the arm was abducted at 90°.
Table 2 The consistency of the present study and the Nobuhara’s
Comparisons of the main finding of the present study with three relevant study indicates that the subacromial structure of live sub-
studies in literature jects is impinged with a notable magnitude of compressive
Design Motion Impingement force only when the arm abducted to 90° was, either
Present In vivo Actively positioned IR only actively or passively, rotated internally. In the study of
study (2005) shoulder configurations Payne et al., the humerus of each cadaveric shoulder
Nobuhara In vivo Passive motions IR only was elevated in various vertical planes with respect to
(2003) maneuvered by surgeon
Payne et al. In vitro Active motions stimulated IR only
the stabilized scapula by means of computer-operated
(1997) by mechanical system actuators that simulated the muscle activities of the rota-
Burns and In vitro Passively positioned IR, N, ER tor cuff and the deltoid, and the compressive force devel-
Whipple (1993) shoulder configurations oped under the coraco-acromial arch was measured
maintained by directly with pressure transducers. The coordination of
mechanical system
the force generation pattern for the simulated muscles
 T. Yanai et al. / Clinical Biomechanics 21 (2006) 692–700
was set on the basis of the electromyogram data reported
by Inman and Saunders (1944) so that it could represent a
realistic activation pattern of a living human. The results
demonstrated that the impingement under the coraco-
acromial ligament of the normal shoulder occurred only
when the abducted arm was internally rotated, but that
the impingement force under the coraco-acromial liga-
ment of the shoulders with the type III acromion was
increased even when the abducted arm was not internally
rotated. The results are generally in a good agreement
with the present study, but the additional result has indi-
cated that the acromial morphology may influence the
magnitude of impingement force developed not only
under the acromion, but also under the coraco-acromial
ligament. This indication leads to a delimitation of the
present study: The main finding is delimited to the rele-
vant age groups. According to the recent morphological
studies of symptomatic and asymptomatic shoulders of
live subjects (MacGillivray et al., 1998; Wang and Shap-
iro, 1997; Worland et al., 2003), the incidence of type
III acromions increases with age. The present finding that
the impingement force was near zero in the arm abduc-
tion with external and neutral rotations might be a unique
result for the present age group (graduate students of
mostly 20s) in which type III acromions are not prevalent.
This speculation is further supported by a study reporting
that only 2% of the 200 asymptomatic shoulders of col-
lege athletes had a type III acromion (Speer et al.,
2001). On the basis of this ratio, the chance of having a
shoulder with type III acromion in the thirteen shoulders
evaluated in the present study is so slim. The different
results found in the present study might, thus, be
accounted for by the difference in the acromial morphol-
ogy associated with the subjects’ age. Further study is
indicated to evaluate this speculation by examining the
acromial morphology and its relation to the impingement
force for a greater number of subjects of various age
groups.
The main finding of the present study, on the other
hand, contradicted the in vitro study of passive motion
(Burns and Whipple, 1993). In this in vitro study, the con-
tact between the coraco-acromial ligament and the supra-
spinatus tendon was reported when the glenohumeral
joint was abducted to 30–60° (equivalent to approximately
45–90° of arm abduction with the scapulothoracic motion)
with the humerus in any degrees of humeral rotations. The
different results obtained in this study might be attributable
to the unrealistic activation pattern of the muscle forces
simulated in their study. In their study, the humerus of a
shoulder specimen was moved passively by the operator
to form a variety of configurations, and for each configura-
tion the humerus position was held by the simulated forces
generated by actuators to maintain the configuration for
the measurements. The same magnitude of the actuator
force was applied for all shoulder configurations. This
unrealistic pattern of force coordination might have
increased superior migration of the humeral head (Sharkey
699
and Marder, 1995). In the preparation for the experiment,
all the muscles were removed from the shoulder specimens
except the rotator cuff muscles and deltoid. The removed
muscles included teres major and latissimus dorsi which
act as humeral head depressors (Hadler et al., 2001). The
removal of these humeral head depressors could result in
an increased superior migration of the humeral head to
narrow the subacromial space, which, in turn, predisposes
the subacromial structure to impingement. The contact
between the supraspinatus tendon and the coraco-acromial
ligament observed in the cadaveric arms elevated below the
painful arc and in those held in neutral and externally-
rotated positions might have affected by this mechanism.
The different results might also be partially due to the dif-
ference in the acromial morphology as mentioned earlier.
As neither the present study nor Burns and Whipple
(1993) reported the type of acromion, however, the latter
speculation could not be evaluated.
5. Conclusion
The present study demonstrated (a) that the shoulder
motions that induce impingement are not any arm abduc-
tion, but rather the arm abduction with a large internal
rotation, and (b) that the shoulder motions beyond the
active range are not necessary to impinge the subacromial
structures at an intensity similar to that provoked with
the Hawkins impingement test. These findings are delim-
ited to young asymptomatic adults (20s and 30s of age).
Further study is indicated to examine the impingement
force and its relation to the acromial morphology for a
greater number of subjects of various age groups.
Acknowledgements
This study was partially funded by the Royal Society
of New Zealand, ISAT Linkages Fund (99-BRAP-09-
YANA).
The authors also thank Mr. T. Kurahashi and Mr. H.
Yoshida of the Hitachi Medico, Japan for providing the
ultrasound device for the part of the data collection.
References
Anton, H., 1988. Calculus with Analytic Geometry, third ed. John-Wiley
& Sons, New York.
Burns II, W.C., Whipple, T.L., 1993. Anatomic relationships in the
shoulder impingement syndrome. Clin. Orthop. Res. 294, 96–102.
Halder, A.M., Zhao, K.D., O’Driscoll, S.W., Morrey, B.F., An, K.N.,
2001. Dynamic contributions to superior shoulder stability. J. Orthop.
Res. 19, 206–212.
Hawkins, R.J., Kennedy, J.C., 1980. Impingement syndrome in athletes.
Am. J. Sports Med. 8, 151–158.
Inman, V.T., Saunders, J.B.D.M., 1944. Observations on the function of
the shoulder joint. J. Bone Joint Surg. 26, 1–30.
Ito, M., Mochizuki, T., 2002. Ultrasound Diagnostic Equipment. Corona
Publishing, Tokyo, Japan (in Japanese).
MacGillivray, J.D., Fealy, S., Potter, H.G., O’Brien, S.J., 1998. Multi-
planar analysis of acromion morphology. Am. J. Sports Med. 26, 836–
840.
700 T. Yanai et al. / Clinical Biomechanics 21 (2006) 692–700
Neer II, C.S., 1972. Anterior acromioplasty for the chronic impingement
syndrome in the shoulder. A preliminary report. J. Bone Joint Surg.
54A, 41–50.
Nobuhara, K., 2003. The Shoulder: Its Function and Clinical Aspects.
World Scientific Publishing.
Payne, L.Z., Deng, X-H., Craig, E.V., Torzilli, P.A., Warren, R.F., 1997.
The combined dynamic and static contributions to subacromial
impingement. A biomechanical analysis. Am. J. Sports Med. 25,
801–808.
Sharkey, N.A., Marder, R.A., 1995. The rotator cuff opposes superior
translation of the humeral head. Am. J. Sports Med. 23, 270–275.
Soslowsky, L.J., An, C.H., Hohnston, S.P., Carpenter, J.E., 1994.
Geometric and mechanical properties of the coracoacromial ligament
and their relationship to rotator cuff disease. Clin. Orthop. Res. 304,
10–17.
Soslowsky, L.J., An, C.H., DeBano, C.M., Carpenter, J.E., 1996.
Coracoacromial ligament: in situ load and viscoelastic properties in
rotator cuff disease. Clin. Orthop. Res. 330, 40–44.
Speer, K.P., Osbahr, D.C., Montella, B.J., Apple, A.S., Mair, S.D., 2001.
Acromial morphotype in the young asymptomatic athletic shoulder. J.
Shoulder Elbow Surg. 10, 434–437.
Valadie, A.L., Jobe, C.M., Pink, M.M., Ekman, E.F., Jobe, F.W., 2000.
Anatomy of provocative tests for impingement syndrome of the
shoulder. J. Shoulder Elbow Surg. 9, 36–46.
Wang, J.C., Shapiro, M.S., 1997. Changes in acromial morphology with
age. J. Shoulder Elbow Surg. 6, 55–59.
Worland, R.L., Lee, D., Orozco, C.G., SozaRex, F., Keenan, 2003.
Correlation of age, acromial morphology, and rotator cuff tear
pathology diagnosed by ultrasound in asymptomatic patients. J. South
Orthop. Assoc. 12, 23–26.
Wuelker, N., Plitz, W., Roetman, B., 1994. Biomechanical data concern-
ing the shoulder impingement syndrome. Clin. Orthop. Res. 303, 242–
249.
Wuelker, N., Roetman, B., Roessing, S., 1995. Coracoacromial
pressure recordings in a cadaveric model. J. Shoulder Elbow Surg. 4,
462–467.