609

Acromial
Angle
of the Shoulder:

Correlation
with the
Syndrome
and Rotator Cuff

Impingement
Tears

Michael
J. Tuite1
David A. Toivonen2
John F. Orwin2
Douglas
H. Wright1

on Radiographs

OBJECTIVE.
This study was undertaken
to determine
the reproducibility
of measurements
of an acromial
angle on radiographs
and to correlate
those measurements
with the presence
of the impingement
syndrome
and rotator cuff tears.
MATERIALS
AND METHODS.
Ninety-nine
shoulders
in 95 patients who had an arch
radiograph
and had undergone
shoulder arthroscopy
were included in this retrospective
study. The acromial angle was measured
on the arch view of the shoulder independently
by three observers
who were blinded to the name, history, and arthroscopic
results. The
angle was measured
at the intersection
of lines drawn along the inferior cortex of the
anterior and posterior portions of the acromion.
Interobserver
variability
was determined
by the

intercorrelation

coefficient

(a test

of reproducibility

of quantitative

measurements).

The average of the three measurements

for each patient was correlated
with the preoperative diagnosis
and the arthroscopic
findings.
RESULTS.
A correlation
was found between
increasing
severity
of cuff disease as
determined
on arthroscopy
and increasing
acromial
angle (p < .01). In 67 patients
(70
shoulders)
with impingement,
patients with a full-thickness
tear (29%) accounted
for
43% of those with an angle of 30#{176}
or greater. The average
acromial
angle for patients
with impingement
was greater than that for either patients
with instability
or patients
with trauma (p < .05 for both).
An angle
of 25#{176}
or greater
was measured
in 63%
of
patients with impingement
but in only 1 8% of those with instability.
The average
acromial angle in patients
with impingement
and an intact rotator cuff was also greater
than the average
angle in patients
with instability
(p = .001). The interobserver
vanability

had

an intercorrelation

coefficient

of 0.90.

CONCLUSION.
The acromial
angle is an objective
and fairly reproducible
measure
of anterior
acromial
shape. The angle is useful in identifying
patients
with a greater
likelihood
of having a rotator cuff tear and in distinguishing
patients
with primary
impingement

AJR

from

those

27, 1 995: accepted

after re-

Presented
at the annual meeting of the American
Roentgen
Ray Society, Washington,
DC, May 1995.
1 Department
of Radiology,
University
of Wisconsin Clinical Science Center, 600 Highland Ave.,
Madison,
WI 53792-3252.
Address
correspondence to M. J. Tuite.
2Department
of Orthopedic
of Wisconsin
Clinical Science
53792-3252.
0361-803X/95/1653-609
American
Roentgen

Surgery,
University
Center, Madison,
WI

Ray Society

instability.

1995;i65:609-6i3

The morphology
[1], who used the
types: flat, curved,
server variability
is
Received
February
vision April 7, 1995.

with

of anterior

of the anterior acromion

was first categorized
by Bigliani et ab.
arch view to classify
anterior
acromion
processes
into three
and hooked.
Some authors have found, however,
that intemobconsiderable
when attempts
are made to classify the spectrum

acromial

posed modifications
found in the original
We undertook

this

shapes

into

one

of three

types

of the three types to achieve

series of Bigliani et ab. [3, 4].
study

to develop

a more

objective

[2].

Other

authors

a distribution
and

reproducible

have

similar

pro-

to that

method

for

describing
anterior acromial
shape. Three important
features
of the anterior
acromion have been described
in previous articles. First, Mabbon et ab. [5] identified
antenon and posterior
acmomiab axes in a series of cadavemic shoulders
and therefore
were

able

to measure

the

angle

of the

acromion.

Second,

several

articles

have

reported
that most patients
have an apex along the undersurface
of the acromion
(Fig. 1) [3, Zuckemman
JD et al., presented
at the American
Shoulder
and Elbow
Surgeons
meeting,
November
1993]. Third, when we used this apex and drew our

TUITE

610

ET AL.

AJR:165,

as having a full-thickness
rotator

tear [7]. Forty-two

cuff, 33 had a partial

rotator

ness tear. Of the 70 shoulders

and Methods

A total of 101 consecutive

patients who had an arch view of the
shoulder and had undergone shoulder arthroscopy
between Octoben 1990 and April 1992 were initially included in the study. Six
patients were excluded: one with a nondiagnostic
arch view, one
because of a prior acromioplasty,
one with synovial osteochondromatosis, and three because of the presence of an os acromiale.
Therefore, 95 patients (99 shoulders) were included in the study.
The group

included

65 men and 30 women

with a mean

view

with either

and the glenoid

cavity

and humeral

nor

profile

of chronic

dens) had instability,

impingement

syndrome,

and 12 patients

16 patients

(12 shoulders)

and

junction

posterior

with

A nonpaired

was measured
who

and

of

20 had

after signif-

on the arch view indepenwere

blinded

to the patients

lines

that

met to form an angle. Angles were mea-

the

posterior

line.

the average acromial angle

between males and females as well as between patients with instability
and those with impingement
but with an intact rotator cuff. The Spearman rank correlation coefficient was used to correlate the acromial angle
with the arthroscopic
findings for the rotator cuff. Analysis of variance

head was used for

measuring the acromial angle.

The preoperative
diagnoses, which had been assigned by the
senior orthopedic surgeon after reviewing the imaging studies and
clinically examining the patients, were obtained from the patients
charts. Sixty-seven
patients (70 shoulders) had the preoperative
diagnosis

angle

tear,

sured to the nearest 1 #{176}with a goniometer

calibrated
at 1 #{176}intervals.
The apex was typically
near the junction of the anterior and middle
thirds of the acromion
(Fig. 1) but occasionally
was within 5 mm of
the anterior tip (Fig. 2). Some patients had flat posterior
but curved
anterior
inferior acromial
surfaces
(Fig. 3). In these patients,
the
anterior line was drawn from the anteroinfenior
tip to the apex at the

the undersurface
of the acromion. A repeat view was obtained if
there was superimposition
of the acromion on the humeral head, the
glenoid cavity, on the base of the scapular spine. If a repeat view was
available, the arch view that showed the greatest distance between
the acromion

tear. Ofthe 12 patients with impingement

seven had a partial or full-thickness
tear.
of the authors,

a partial

diagnosis

a full-thickness
cant trauma,

by three

28 had

ogy fellow. A mean acromial

angle for each patient was calculated
from the independent
measurements.
In addition,
the senior author
remeasured
the angle 2-6 months later for all the patients while also
blinded to the original measurements.
The acromial
angle was determined
by drawing a line along the
inferior acromial cortex on either side of the apex, resulting in ante-

age of 43

a 5#{176}
or a 15#{176}
tilt to better

with the preoperative

cuff,

names, history, and arthroscopic

results. The three observers were
a radiologist, an orthopedic surgeon, and a musculoskeletal
radiol-

a repeat

had an intact

22 had

The acromial

an intact

shoulders

cuff team, and 24 had a full-thick-

impingement,

dently

years (range, 16-76 years).

Arch radiognaphs of the shoulder were obtained in the lateral projection with the patient positioned 40#{176}
anterior oblique and the X-ray
beam coned to the superior half of the scapula [1 6, 9]. The tubescreen distance was 183 cm, and the X-ray voltage was usually 7075 kVp. A 10#{176}
caudal tilt was usually used, but a few patients
required

1995

ment symptoms
after a significant
traumatic
episode. This last group
was defined as patients who had a previously
asymptomatic
shoulden, who had acute pain after a fall or similar traumatic
event, and
who had impingement
signs on examination.
All patients
underwent
shoulder
arthroscopy
performed
by the
senior orthopedic
surgeon
using a standard
technique
[6, 7]. The
rotator cuff was graded prospectively
as intact, as partially torn, or

lines along the inferior cortex of the acromion

instead of down
the center as reported
by Mabbon et al. [5], our posterior
line
could be used preopenativeby
to determine
the amount of bone
to be resected
during acromioplasty
[6-8].
The purposes
of this study were to determine
whether
our
acromial
angle measurement
was reproducible
and whether
a correlation
exists between
the angle and both the impingement syndrome
and rotator cuff teams.

Materials

September

t test was used to compare

was used to correlate the acromial angle with the preoperative

The Pearson correlation
between the acromial

(17 shoul-

had impinge-

diagnosis.

coefficient was used to evaluate the relationship

angle and age for patients with impingement.

_t

Fig. i.-37-year-old
A, Arch

man with impingement

radiograph
shows anterior
(straight
acromial
surfaces,
which meet at apex (curved
B, Lines drawn on same image to determine

symptoms

and intact rotator cuff at arthroscopy.

white arrows)
and posterior
(black arrows)
Inferior
arrow).
acromial
angle, which measured
23.

Fig. 2.-33-year-old
woman with impingement
symptoms
and partial tear of rotator cuff at arthroscopy.
Arch radiograph
shows anterior
(arrowheads)
and posterior (arrows)
inferior acromial
surfaces. Mean acromial angle was 40.

AJR:165,

September

Fig. 3.-39-year-old

ACROMIAL

1995

woman

with

ANGLE

ON RADIOGRAPHS

OF SHOULDER

611

impingement
symptoms
and intact rotator cuff at arthroscopy.
(black arrows)
but curved anterior
(white arrows)
inferior surfaces
of acromion.
tip to apex at junction
with posterior
line. Acromiai
angle measured
24.
shows resection
of inferior portion of anterior
acromion
to produce
flat undersurface
(arrow).

A, Arch radiograph
shows flat posterior
B, Anterior
line drawn from anteroinferlor

C, Arch

view after

acromioplasty

Interobserver
and intraobserver
variabilities
were evaluated
with the
intencomnelation coefficient
(a test of reproducibility
of quantitative
measunements). Correlation coefficient values range up to 1 , with 0 indicating

no correlation and 1 representing a strongly positive correlation.

value of less than .05 was assumed to be statistically significant.

TABLE

Clinical

A p

Results
The average acromiab angle for the study population
was 25#{176}
(range, 8-42#{176})
(mean ofthe three independent
measurements).
We found no significant
difference
in the average
angles for
males (25#{176})
and females (26#{176})
(p = .46). The average acromial
angles were 23.6#{176}
for patients with an intact rotator cuff, 24.3#{176}
for patients with a partial team, and 28.9#{176}
for patients with a fullthickness tear. The coefficient
of correlation
between increasing
severity of cuff disease and increasing
acromiab angle was 0.27
(p < .01). Among patients with impingement,
patients with a fullthickness
team (29%)
accounted
for 43% of those with an acromiab angle of 30#{176}
or greaten.
We found a statistically
significant
correlation
between
the
average
acromiab
angle
and the preoperative
diagnosis
(Table 1). The average
angle for patients
with impingement
(27.0#{176})
was greaten than the average
angles for patients with
instability
(19.5#{176})
and patients
with trauma
(21.7#{176}),
both with
a p value of less than .05. An angle of 25#{176}
on greater
was
measured
in 63% of patients
with impingement
but in only
18% of patients with instability.
The average
acromiab angle
for the 22 shoulders
with an intact rotator cuff in patients with
impingement
was 26.5#{176}.
The difference
between
this angle
and the average
acromiab
angle in patients
with instability
also was statistically
significant
(p = .001). The coefficient
of
correlation
between
age and acromiab angle among patients
with impingement
was not statistically
significant,
being 0.23
(p = .06). The intemobservem variability
among the three independent
measurements
had a calculated
intencommelation
coefficient
of 0.90. The intraobservem
variability
for the senior
author produced
an intercommelation
coefficient
of 0.91.

1 : Clinical

and Arthroscopic

Diagnosis

and Acromial

Angle
and Arthroscopic
Diagnosis

Impingement
Full-thickness

No. of
Shoulders

tear

Partial tear
Intact
Traumatic
Instability

impingement

asignificantly

greater

than

with instability

bSignificantly

greater

70
20

27a

28

25

22
12
17

26b

99

25

30

22
20

Total
and patients

Average Acromial
Angle (#{176})

angle

in patients

with

traumatic

impingement

(p < .05 for both).

than angle

in patients

with instability

(p

.001).

Discussion
One of the main purposes
of our study was to determine
whether
an association
exists between
an increasing
acromiab

angle

and

the

prevalence

of rotator

cuff

tears,

as

has

been shown for the Bigbiani types [1 3, 8]. Our study showed
a statistically
significant
association.
A more important
finding, however,
was that the mean
acromiab
angle
in our
patients
with impingement
but with an intact rotator cuff was
significantly
greater than that in our patients
with instability.
Morrison
and Bigliani [8] were unable to show a difference
in
the distribution
of acmomial types between
control
subjects
and patients
with the impingement
syndrome
but with an
intact rotator cuff. Epstein et al. [3] showed
a slight increase
,

in the

prevalence

of type

III acromion

with isolated
impingement,
although
reach statistical
significance.
We also believed that the acromiab
orthopedic

scopic

surgeons

acmomiopbasty.

during

pmesurgicab

Acromioplasty,

processes

this
angle

did

not

may be useful

planning

or

in patients

increase

to

for an arthro-

resection

of

the

TUllE

612

undersurface
of the anterior
acromion,
is a common
procedune that usually alleviates
impingement
pain and is important
in the treatment
of rotator cuff teams [6, 7, 1 0]. It can be difficult
during
arthroscopy,
however,
to evaluate
how much bone
needs to be mesected. This factor is important
because excessive

acmomioplasty

can bead to fracture

or can

weaken

the del-

toid muscle
origin and insufficient
acromioplasty
can fail to
relieve symptoms
[6, 10]. Several articles have stressed
the
importance
of evaluating
anterior acromiab shape on the arch
view during pmesumgicab planning
[6-8]. The amount
of bone
that needs to be nesected to produce
a flat acromion
can be
demonstrated
by extending
the posterior
line drawn during
measurement
of the acromiab angle (Fig. 3).
The average
acmomiab angle in our series was smaller than
the 37#{176}
average
reported
by Malbon et al. [5]. There may be
several reasons
for this result. First, Mabbon et al. measured
their angle at the junction
of the anterior
and posterior
acromiab axes, whereas
we used the inferior
acromial
cortex.
They also nadiogmaphed
their cadavenic
specimens
with the
X-ray beam directed
perpendicular
to the glenoid
cavity, a
view similar
but not identical
to the arch view used in our
study. Finally, the average
age of their cadavers
was probably greater than that of our patients.
The significance
of this
factor is controversial,
because
some authors
have found a
correlation
between
increasing
age and subacromiab
spurring [ii
12], whereas
others
have found
no correlation
between
patient age and Bigliani type in asymptomatic
contnob subjects
[3]. Among
our patients
with impingement,
the
correlation
between
increasing
age and acromiab angle did
not reach statistical
significance.
As in our series, Mabbon et al. [5] reported
no scapula with
an acromiab angle of 0#{176}
to correlate
with a true Bigliani type I.
They also did not mention
any curved,
or type II, acromion
processes
and were abbe to identify
anterior
and posterior
axes in all of their specimens.
Our series included
several
curved anterior
acromion
processes,
and we had to define a
modified
method
for determining
the angle in these shoubdens. The prevalence
of truly curved anterior
acromion
processes
is controversial.
Morrison
and Bigliani
[8] reported
that 41% of their patients
had a curved acromion.
Epstein et
al. [3] classified
38% of their patients
as having
a type II
acromion
but included
some patients
who actually
had an
,

ET

AL.

AJR:165.

September

1995

apex in the middle third of the acromion.

Zuckemman
et ab.
(American
Shoulder
and Elbow Surgeons
meeting,
Novemben 1 993) found that only 22% of acromion
processes
did not
have an apex and that these included
both curved and flat
acmomion processes.
We chose to include in this study only patients
with surgicab correlation
and therefore
did not measure
the acromial
angle in asymptomatic
control
subjects.
We used patients
with instability
as our comparison
group in several of the statisticab analyses
because
their distribution
for anterior
acromiab morphology
has been reported
to be similar to that of
asymptomatic
control subjects
[3, 8]. The arch view is not
part of the usual radiographic
series obtained
at our institution for patients
with instability.
This view was available
in
these 16 patients
(17 shoulders)
either because
they had
secondary
impingement
that at first dominated
the clinical
picture on because
the initial ordering
physician
was unable
to distinguish
between
instability
and impingement.
We excluded
patients with an os acmomiabe from our study
because
ossa acromiabe
are independently
associated
with
rotator cuff tears [1 3]. However,
we did not control for acromioclavicuban
joint osteophytes,
another
major factor associated with teams [1 3, 14]. No method for objectively
measuring
the size of such osteophytes
is widely accepted,
and their
contribution
to the Bigliani
types is controversial
[ii
12].
Although
some authors
classify
shoulders
with an osteophyte as type Ill, Bigliani and several
other authors
do not
consider
a spun a criterion for a hooked acromion
[1 2, 4, 8].
In some patients in our study who had acromioclaviculam
joint
osteophytes
and a flat anterior
acromiab
undersurface,
the
contribution
of the osteophytes
to rotator cuff disease
would
not be accounted
for by the acromiab angle (Fig. 4).
Patients
with impingement
symptoms
after a significant
traumatic
episode
were placed in a separate
group for some
of our analyses.
Significant
trauma can cause a rotator cuff
team even in shoulders
with a flat acmomion [13]. The pmevabence of rotator
cuff tears in our patients
with traumatic
impingement
(58%) was similar to that in our patients
with
impingement
but no trauma,
despite
a significantly
smaller
average
acromial
angle.
We did not measure the acromiab angle in any patients who
only had a scapular
V-view radiograph.
Haygood
et ab. [2]
,

Fig. 4.-44-year-old
man with Impingement
symptoms
and full-thickness
tear of rotator cuff
at arthroscopy.
A, Arch radiograph
shows
posterior
(wavy
arrows)
and dominant
anterior
(straight
arrow)
inferior acromial
surfaces.
Acromial
angle measured 23.
Note additional
secondary
anterior
acromial undersurfaces
(arrowheads)
and anterior
spur (curved arrow).
B, Anteroposterior
radiograph
shows spur at
anteroinferior
and medial
corner
of acromion
(arrows).

AJR:165,

September

reported

evaluating
scapular

that

4i%

of scapular

anterior
V-view

ACROMIAL

1995

does

acromial
not have

V-views

shape.

were

Unlike

the caudal

ANGLE

ON RADIOGRAPHS

not diagnostic

the arch

tilt that

for

view,

is necessary

to optimally
profile the undersurface
of the acromion
[1 6]. A
tilt of 10#{176}
but occasionally
of as much as 15#{176}
was necessary
in
our study to obtain a satisfactory
arch view, in which the infenor acromiab
cortex
was projected
separately
from
the
humeral head, glenoid cavity, and base of the scapular spine.
Other authors
also have searched
for a more accurate
method
of assessing
anterior
acromiab shape than the Bigbiani classification
as imaged
on the arch view, and some
have found that oblique
sagittab MR images
may be more
accurate
[3, 4, 1 5]. Other studies have not been abbe to confirm this result, however
[2, 16].
A weakness
of our acromiab angle measurements
is that,
because ofthe heterogeneity
of acromiab shapes, some patients
had curved or several separate anterior acromial undersurlaces
(Figs. 3 and 4). The variability in placing the apex and/or anterior
acromial line in these patients appeared to be the main cause of
our interobserver
and intraobserver
variabilities.
Although
our
intercorrebation
coefficients
were acceptable
at 0.90 and 0.91,
,

respectively,

a value

of at least

0.95

is required

for a test

to be

OF SHOULDER

613

REFERENCES
1. Bigliani LU, Morrison
DS, April EW. Morphology
of the acromion
and its
relationship
to rotator cuff tears (abstr). Orthop Trans 1986:10:459-460
2. Haygood TM, Langlotz CP, Kneeland JB, lannotti JP, Williams GA, Dalinka
MK. Categorization
of acromial shape: interobserver variability with MR
imaging and conventional
radiography.
AJR 1994:162:1377-1382

3. Epstein RE, Schweitzer ME, Fnieman BG, Fenlin JM, Mitchell DG. Hooked
acromion: prevalence on MA images of painful shoulders.
Radiology
1993:187:479-481
4. Gagey N, Aavaud E, Lassau JP. Anatomy
of the acromial arch: correlation
of anatomy
and magnetic
resonance
imaging.
Surg Radio! Anat 1993:

15:63-70
5. Mallon WJ, Brown HR. Vogler JB, Martinez
S. Radiographic
and geometnc anatomy
of the scapula.
Clin Orthop 1992:277:142-1
54
6. Sampson
TG, Nisbet JK, Glick JM. Precision
acromioplasty
in arthroscopic subacromial
decompression
of the shoulder.
Arthroscopy
1991:

7:301-307
7. Snyder SJ. Evaluation
and treatment
of the rotator cuff. Orthop Clin North
Am 1993:24:173-1
92
8. Morrison DS, Bigliani LU. The clinical significance
of variations
in acromial
morphology
(abstr). Orthop Trans 1987:11:234
9. Neer CS. Anterior acromioplasty.
J Bone Joint SurgAm
1972:54-A:41-50
10. Ryu RKN. Arthroscopic
subacromial
decompression:
a clinical
review.
Arthroscopy
1992:8:141-147
11. Hardy DC, Vogler JB, White RH. The shoulder
impingement
syndrome:
prevalence
of radiographic
findings and correlation
with response
to then-

considered
to have excellent reproducibility.
In summary,
the acromial
angle is an objective
and fairly
reproducible
measure
of anterior
acromiab shape. The angle
is useful is identifying
patients
with a greater
likelihood
of
having a rotator cuff tear and in distinguishing
patients
with
primary impingement
from those with instability.

apy. AJR 1986:147:557-561

12. Ogata 5, Uhthoff HK. Acromial
enthesopathy
and rotator cuff tear. Clin
Orthop 1990:254:39-48
13. Bigliani LU. Impingement
syndrome:
aetiology
and overview.
In: Watson
MS. ed. Surgical
disorders
of the shoulder.
New York: Churchill Livingstone, 1991:237-246
14. Petersson
CJ, Gentz CF. Ruptures
of the supraspinatus
tendon: the significance
of distally pointing
acromioclavicular
osteophytes.
Clin Orthop

ACKNOWLEDGMENTS

15. Farley TE, Neumann

CH, Steinbach
LS, Petersen
SA. The coracoacromial arch: MR evaluation
and correlation
with rotator cuff pathology.
Skeleta! Radio! 1994:23:641-645
16. Farmer TH, Peh WC, Tony WG. Assessment
of acromial shape with MA
imaging (abstr). Radiology
1994:1 93(P):153

1983:174:143-148

We thank Beth Washa for secretarial assistance and Lawrence

Zaborski for performing statistical analyses during the preparation of
the manuscript.