Ultrasonography of the Rotator Cuff : A Comparison of Ultrasonographic and Arthroscopic Findings in One Hundred Consecutive Cases

This is an enhanced PDF from The Journal of Bone and Joint Surgery

The PDF of the article you requested follows this cover page.
Ultrasonography of the Rotator Cuff : A Comparison of
Ultrasonographic and Arthroscopic Findings in One Hundred
Consecutive Cases

Sharlene A. Teefey, S. Ashfaq Hasan, William D. Middleton, Mihir Patel, Rick W. Wright and Ken Yamaguchi
2000;82:498.
J Bone Joint Surg Am.
This information is current as of April 29, 2011
Reprints and Permissions
order reprints or request permission
Click here to
to use material from this
article, or locate the article citation on jbjs.org and click on the [Reprints and
Permissions] link.

Publisher Information
The Journal of Bone and Joint Surgery
20 Pickering Street, Needham, MA 02492-3157

www.jbjs.org

---

Ultrasonography of the Rotator Cuff
A COMPARISON OF ULTRASONOGRAPHIC AND ARTHROSCOPIC FINDINGS
IN ONE HUNDRED CONSECUTIVE CASES*
BY SHARLENE A. TEEFEY, M.D.†, S. ASHFAQ HASAN, M.D.†, WILLIAM D. MIDDLETON, M.D.†,
MIHIR PATEL, M.D.†, RICK W. WRIGHT, M.D.†, AND KEN YAMAGUCHI, M.D.†
Investigation performed at the Mallinckrodt Institute of Radiology and the Department of Orthopaedic Surgery,
Washington University School of Medicine, St. Louis, Missouri
Abstract
for detecting full-thickness rotator cuff tears, characteriz-
Background: There has been limited acceptance of
ing their extent, and visualizing dislocations of the bi-
shoulder ultrasonography by orthopaedic surgeons in
ceps tendon. It was less sensitive for detecting partial-
the United States. The purpose of this retrospective
thickness rotator cuff tears and ruptures of the biceps
study was to determine the diagnostic performance of
tendon.
high-resolution ultrasonography compared with arthro-
scopic examination for the detection and characteriza-
The use of high-resolution ultrasonography in North
tion of rotator cuff tears.
America for the detection of rotator cuff tears has
Methods: One hundred consecutive shoulders in
achieved only limited acceptance by orthopaedic sur-
ninety-eight patients with shoulder pain who had un-
geons compared with other modalities such as magnetic
dergone preoperative ultrasonography and subsequent
resonance imaging. Uncertainty about the accuracy of
arthroscopy were identified. The arthroscopic diagnosis
this modality may have contributed to its low utiliza-
was a full-thickness rotator cuff tear in sixty-five shoul-
tion rate. Although initial studies, published in the mid-
ders, a partial-thickness tear in fifteen, rotator cuff ten-
1980s, that compared ultrasonographic and surgical
dinitis in twelve, frozen shoulder in four, arthrosis of
findings showed a high rate of accuracy (92 to 94 per-
the acromioclavicular joint in two, and a superior labral
cent in series of fifty-one and forty-seven patients5,7) for
tear and calcific bursitis in one shoulder each. All ultra-
the detection of rotator cuff tears, later studies showed
sonographic reports were reviewed for the presence or
somewhat lower rates (60 to 84 percent in series of
absence of a rotator cuff tear and a biceps tendon rup-
thirty-eight, ten, and forty-nine patients1,3,9). Addition-
ture or dislocation. All arthroscopic examinations were
ally, only a few studies have compared the accuracy of
performed according to a standardized operative proce-
ultrasonography with that of arthroscopy for deter-
dure. The size and extent of the tear and the status of
mining the presence or absence of rotator cuff tears2,8,12,13
the biceps tendon were recorded for all shoulders. The
and fewer have correlated the tear size with the surgical
findings on ultrasonography and arthroscopy then were
findings2,13. Brenneke and Morgan, in a study of sixty-
compared for each parameter.
one patients, found that ultrasonography had a sen-
Results: Ultrasonography correctly identified all
sitivity of 95 percent and a specificity of 93 percent for
sixty-five full-thickness rotator cuff tears (a sensitivity
the detection of full-thickness tears2. They also found
of 100 percent). There were seventeen true-negative
that it accurately predicted the size of full-thickness
and three false-positive ultrasonograms (a specificity of
tears in 89 percent of patients who had a tear that was
85 percent). The overall accuracy was 96 percent. The
greater than four centimeters, in 43 percent of those
size of the tear on transverse measurement was cor-
who had a tear that was two to four centimeters, and in
rectly predicted in 86 percent of the shoulders with a
70 percent of those who had a tear that was less than
full-thickness tear. Ultrasonography detected a tear in
two centimeters. Wiener and Seitz, in a study of 225 pa-
ten of fifteen shoulders with a partial-thickness tear
tients, demonstrated that ultrasonography had a sensi-
that was diagnosed on arthroscopy. Five of six disloca-
tivity of 95 percent and a specificity of 94 percent for the
tions and seven of eleven ruptures of the biceps tendon
detection of full-thickness tears and a sensitivity of 91
were identified correctly.
percent and a specificity of 94 percent for predicting the
Conclusions: Ultrasonography was highly accurate
size of the tear13.
The purpose of the current study was to compare the
*No benefits in any form have been received or will be received
diagnostic performance of ultrasonography with that of
from a commercial party related directly or indirectly to the subject
arthroscopic surgery to determine its accuracy for detect-
of this article. No funds were received in support of this study.
ing rotator cuff tears and biceps tendon pathology.
†Mallinckrodt Institute of Radiology (S. A. T. and W. D. M.) and
Department of Orthopaedic Surgery (S. A. H., M. P., R. W. W., and
K. Y.), Washington University School of Medicine, 510 South Kings-
Materials and Methods
highway, Box 8131, St. Louis, Missouri 63110.
The study comprised 100 shoulders in ninety-eight
Copyright 2000 by The Journal of Bone and Joint Surgery, Incorporated
consecutive patients with shoulder pain who had under-
498
THE JOURNAL OF BONE AND JOINT SURGERY

---

ULTRASONOGRAPHY OF THE ROTATOR CUFF
499
dardized bilateral ultrasonography of the shoulder, per-
formed by one of two radiologists who were very expe-
rienced with the technique and who had conducted
more than 2500 examinations during a ten-year period.
The ultrasonographic examination was performed
with the patient seated on a stool and the radiologist
standing behind the patient. First, the biceps tendon was
examined in the transverse plane from the level of the
acromion inferiorly to the point where the tendon
merged with the biceps muscle. The transducer then was
rotated 90 degrees in order to examine the tendon lon-
gitudinally. Next, images of the subscapularis tendon
were made with the patient’s arm externally rotated; the
transducer was placed in a transverse anatomical orien-
tation at the level of the lesser tuberosity and was
FIG. 1-A
moved medially.
Drawing of a left shoulder, viewed from above with the arm in ex-
Images of the supraspinatus tendon were made with
tension, showing the transducer oriented in a plane parallel to the
the shoulder extended, the elbow flexed, and the hand
longitudinal axis of the rotator cuff. A = anterior, L = lateral, and P =
placed on the iliac wing. This position was necessary in
posterior.
order to expose as much of the supraspinatus tendon as
gone standardized preoperative ultrasonography and
possible from under the acromion. The transducer was
subsequent arthroscopy between January 1996 and Sep-
oriented parallel to the tendon (approximately 45 de-
tember 1997. The interval of time between the ultra-
grees between the coronal and sagittal planes) in order
sonographic and the arthroscopic examination ranged
to visualize the fibers in a longitudinal plane (Figs. 1-A
from one to 417 days (mean, sixty days). There were
and 1-B), and it was moved anteriorly to posteriorly in
fifty-four female patients and forty-four male patients,
order to visualize the supraspinatus and infraspinatus
and their ages ranged from fourteen to eighty-two years
tendons. The transducer was rotated 90 degrees in order
(mean age, fifty-six years).
to examine the tendons in the transverse plane (Figs. 2-A
The primary arthroscopic or final clinical diagnosis
and 2-B).
was a full-thickness tear of the rotator cuff in sixty-five
shoulders, a partial-thickness tear in fifteen, rotator cuff
Ultrasonographic Criteria
tendinitis in twelve, frozen shoulder in four, arthrosis
A finding of a full-thickness rotator cuff tear was re-
of the acromioclavicular joint in two, and a superior
corded when the rotator cuff could not be visualized be-
labral tear and calcific bursitis in one shoulder each.
cause of complete avulsion and retraction under the
Two patients with a full-thickness tear had a large
acromion or when there was a focal defect in the rotator
partial-thickness component.
cuff created by a variable degree of retraction of the
In general, the indications for the surgery and the ar-
torn tendon ends. In the latter case, either joint fluid or
throscopic examination included shoulder pain of more
thickened bursal tissue and the deep surface of the del-
than six months’ duration and a lack of a response to
nonoperative treatment including physical therapy, non-
steroidal anti-inflammatory medications, and at least one
cortisone injection. For the patients with a full-thickness
tear, the indications for the operation included severe
pain of more than three months’ duration despite the
nonoperative measures just mentioned. Patients with a
full-thickness tear who had a recent loss of shoulder
elevation or a recent injury (sustained less than three
months before the time of presentation) were offered the
option of an operation at earlier than three months.
Ultrasonographic Technique
All ultrasonograms were obtained in real time with
use of an ATL HDI 3000 scanner (Advanced Technolo-
gies Laboratories, Bothell, Washington) or a Siemens
Elegra scanner (Siemens Medical Systems, Issaquah,
FIG. 1-B
Washington) and a variable high-frequency linear-array
Corresponding ultrasonographic image showing the rotator cuff
transducer (7.5 to ten megahertz). All patients had stan-
(arrow) in the longitudinal plane.
VOL. 82-A, NO. 4, APRIL 2000

---

500
S. A. TEEFEY ET AL.
tear or of a bursal or articular-side partial-thickness tear
and the width (perpendicular to the long axis of the cuff
fibers) of any tear that was found were recorded. The
biceps tendon was examined arthroscopically for dislo-
cation or rupture. Representative arthroscopic images
were made of all tears and other pathological findings,
such as abnormalities of the biceps tendon.
In shoulders in which a partial-thickness tear was
present or the arthroscopic findings were discrepant
from those recorded on ultrasonography, or both, a tag-
ging suture (number-1 PDS [polydioxanone]) was placed,
from the bursal side without a knot, through the sus-
pected area of the rotator cuff to guide arthroscopic
bursal imaging. In shoulders in which a full-thickness tear
was recorded on ultrasonography but was not visualized
on arthroscopy, an extensive partial-thickness tear was
FIG. 2-A
present. In these shoulders, a mini-open deltoid split (a
Drawing of a left shoulder, viewed from above with the arm in ex-
three to four-centimeter skin incision with approximately
tension, showing the transducer oriented in a plane perpendicular to
a three-centimeter deltoid split without any takedown
the longitudinal axis of the rotator cuff. A = anterior, L = lateral, and
of the deltoid origin) was performed to directly visualize
P = posterior.
the involved area of the rotator cuff and to verify the
toid muscle occupied the defect created by the tear. If
arthroscopic findings. Additionally, as all full-thickness
no tear was visualized, the deltoid muscle was com-
tears were repaired through a mini-open deltoid-splitting
pressed against the cuff with the transducer in an at-
approach, the size and extent of the tear were determined
tempt to separate the torn tendon ends at the site of a
by direct visualization. If a partial-thickness tear was re-
nonretracted tear.
corded on ultrasonography but was not seen on arthros-
A finding of a partial-thickness tear was recorded
copy, a mini-open deltoid split was not performed.
when there was minimal flattening of the bursal side of
the rotator cuff (a bursal-side partial-thickness tear) or
Data Analysis
a distinct hypoechoic or mixed hyperechoic and hypo-
The ultrasonographic and arthroscopic findings were
echoic defect visualized in both the longitudinal and
correlated with regard to the presence or absence of a
the transverse plane at the deep articular side of the
full or partial-thickness rotator cuff tear, the size and
rotator cuff (an articular-side partial-thickness tear).
extent of the tear, and the presence or absence of a dis-
The extent of the rotator cuff tear was determined
location or rupture of the biceps tendon. When there
with transverse measurements. According to empirical
was disagreement between the findings, representative
guidelines instituted prior to the inception of this
arthroscopic and ultrasonographic images were reevalu-
study, if the tear extended posteriorly 1.5 centimeters
or less from the intra-articular portion of the biceps
tendon it was recorded as involving only the supra-
spinatus tendon, whereas if it extended more than 1.5
to 3.0 centimeters it was recorded as involving both the
supraspinatus and the infraspinatus tendon. The teres
minor tendon was not evaluated when the extent of the
tear was determined.
A finding of a rupture of the biceps tendon was re-
corded when the tendon was not identified within or
medial to the intertubercular sulcus. Dislocation of the
biceps tendon was recorded when the tendon was ante-
rior or medial to the lesser tuberosity.
Surgical Technique and Criteria
All arthroscopic examinations and operative proce-
dures were performed by a single orthopaedic surgeon
who recorded all findings in a standardized manner. The
presence or absence of a rotator cuff tear and the size
FIG. 2-B
and extent of the tear, when present, were recorded.
Corresponding ultrasonographic image showing the rotator cuff
Specifically, the presence or absence of a full-thickness
(arrow) in the transverse plane.
THE JOURNAL OF BONE AND JOINT SURGERY

---

ULTRASONOGRAPHY OF THE ROTATOR CUFF
501
TABLE I
were discrepant from those recorded on ultrasonogra-
FULL-THICKNESS ROTATOR CUFF TEARS:
phy, a tagging suture was placed through the suspicious
ULTRASONOGRAPHIC VERSUS ARTHROSCOPIC FINDINGS*
area of the rotator cuff to guide arthroscopic bursal im-
Arthroscopy
aging. In three of these shoulders, ultrasonography re-
Positive
Negative
Total
vealed a full-thickness tear but a partial-thickness tear
Ultrasonography
was detected on arthroscopy. In the other three shoul-
Positive
65
3
68
ders, a partial-thickness tear was recorded on ultra-
Negative
0
17
17
sonography but the cuff was normal on arthroscopy.
Total
65
20
85
*The values are given as the numbers of shoulders. When true-
Size and Extent of the Tears
positive indicated a full-thickness tear and true-negative, no tear, ul-
Of the sixty-three full-thickness rotator cuff tears
trasonography had a sensitivity of 100 percent (sixty-five of sixty-five),
a specificity of 85 percent (seventeen of twenty), a positive predictive
that were analyzed for these parameters, twenty-six were
value of 96 percent (sixty-five of sixty-eight), a negative predictive
value of 100 percent (seventeen of seventeen), and an accuracy of 96
percent (eighty-two of eighty-five).
ated jointly to explain the discrepancy.
Only the full-thickness tears were analyzed with
regard to their size and extent. The subscapularis was
classified only as intact or torn. Two of the sixty-five
shoulders with a full-thickness tear were excluded from
the analysis; one shoulder had a very limited range of
motion and indeterminate findings regarding the extent
of the tear on ultrasonography, and the other shoulder
had had the arthroscopic examination one year after the
ultrasonographic study.
Results
Detection of Rotator Cuff Tears
Ultrasonography correctly identified all sixty-five
full-thickness rotator cuff tears that were diagnosed on
arthroscopy (Figs. 3-A and 3-B, and Table I). There were
no false-negative studies. Ultrasonography incorrectly
identified a full-thickness rotator cuff tear in three shoul-
FIG. 3-A
ders that were found to have a partial-thickness tear on
Arthroscopic image showing a small full-thickness tear of the left
arthroscopy; one of the three tears was extensive (more
supraspinatus tendon (arrow).
than 50 percent of the cuff thickness) and involved the
entire supraspinatus tendon.
Ultrasonography correctly identified seven of fif-
teen partial-thickness rotator cuff tears that were di-
agnosed on arthroscopy (Figs. 4-A and 4-B, and Table
II). In three additional shoulders, it identified a full-
thickness rather than a partial-thickness tear. Because
a tear was identified, these studies were considered to
be true-positive. There were five false-negative studies.
Ultrasonographic visualization of the rotator cuff was
limited by a decreased range of motion in two of these
shoulders, and arthroscopy showed only mild fraying of
the supraspinatus tendon in a third. There were three
false-positive ultrasonograms, one of which showed an
ill defined hypoechoic region, suggestive of a partial
tear, on the deep capsular side of the cuff near its inser-
tion. Another of the false-positive studies showed subtle
flattening of the bursal side of the supraspinatus tendon.
Ultrasonography correctly predicted the absence of a
FIG. 3-B
tear in seventeen of twenty shoulders that had no evi-
Corresponding ultrasonographic image showing the small tendon
tear. Fluid separates the torn tendon ends (arrow). The image is ori-
dence of a tear on arthroscopy.
ented in a plane perpendicular to the longitudinal axis of the tendon.
In six shoulders for which the arthroscopic findings
The biceps tendon is to the left of the tear (arrowhead).
VOL. 82-A, NO. 4, APRIL 2000

---

502
S. A. TEEFEY ET AL.
found on arthroscopy to involve only the supraspinatus
TABLE II
tendon and to be less than 1.5 centimeters wide, and
PARTIAL-THICKNESS ROTATOR CUFF TEARS:
thirty-seven involved both the supraspinatus and the in-
ULTRASONOGRAPHIC VERSUS ARTHROSCOPIC FINDINGS*
fraspinatus and were more than 1.5 centimeters wide. In
Arthroscopy
addition, seven shoulders had a tear of the subscapularis
Positive
Negative
Total
tendon. Transverse measurement with ultrasonography
Ultrasonography
correctly predicted the extent of the tear in twenty-one
Positive
10
3
13
(81 percent) of the twenty-six shoulders with an isolated
Negative
5
17
22
Total
15
20
35
tear of the supraspinatus tendon. In three shoulders,
ultrasonography overestimated the width of the tear by
*The values are given as the number of shoulders. When true-
positive indicated a partial-thickness tear and true-negative, no tear,
0.5 centimeter or less and in two, by 1.1 and 1.3 centime-
ultrasonography had a sensitivity of 67 percent (ten of fifteen), a
ters. In the latter two shoulders, arthroscopy confirmed
specificity of 85 percent (seventeen of twenty), a positive predictive
the presence of a full-thickness tear of the supraspinatus
value of 77 percent (ten of thirteen), a negative predictive value of 77
but also showed an extensive partial-thickness tear
percent (seventeen of twenty-two), and an accuracy of 77 percent
(more than 50 percent of the cuff thickness) extending
(twenty-seven of thirty-five).
into the infraspinatus tendon, which had been inter-
preted as a full-thickness tear on ultrasonography.
Transverse measurement with ultrasonography cor-
rectly predicted the extent of the tear in thirty-three (89
percent) of the thirty-seven shoulders that had a com-
bined tear of the supraspinatus and infraspinatus ten-
dons. In the four remaining shoulders, ultrasonography
underestimated the extension of the tear into the in-
fraspinatus tendon by one to 1.5 centimeters. In three of
these shoulders, arthroscopy showed that the infraspina-
tus component was a midsubstance extension of the tear
medial to the supraspinatus portion of the tear. Ultra-
sonography correctly identified six of the seven tears of
the subscapularis tendon. Overall, ultrasonography cor-
rectly predicted the size and extent of the tear in 86 per-
cent of the shoulders with a full-thickness tear.
Five of the thirty-seven shoulders had a massive tear
with retraction of the torn tendon underneath the acro-
mion. The edge of the torn tendon could not be visual-
ized at the time of the ultrasonographic study.
FIG. 4-A
Dislocation of the Biceps Tendon
Arthroscopic image showing a small partial-thickness tear of the
Ultrasonography correctly identified five of six dislo-
right supraspinatus tendon (arrow).
cations of the biceps tendon that were diagnosed on ar-
throscopy. The one false-negative study, which showed
an absence of the biceps tendon, was interpreted as
demonstrating a rupture rather than a dislocation. There
were ninety-four true-negative ultrasonograms and no
false-positive ultrasonograms.
Rupture of the Biceps Tendon
Ultrasonography correctly identified seven of eleven
ruptures of the biceps tendon that were diagnosed on
arthroscopy. There were four false-negative studies. Two
of the false-negative ultrasonograms showed the nor-
mal echogenic fibrillar pattern of the tendon within the
groove. There was one false-positive ultrasonogram, and
there were eighty-eight true-negative ultrasonograms.
FIG. 4-B
Discussion
Corresponding ultrasonographic image showing the small hypo-
High-resolution shoulder ultrasonography has not
echoic tendon tear, located on the deep capsular side of the cuff (ar-
been widely utilized by orthopaedic surgeons to diagnose
row). The image is oriented in a plane parallel to the longitudinal axis
of the tendon.
and characterize rotator cuff and biceps tendon pathol-
THE JOURNAL OF BONE AND JOINT SURGERY

---

ULTRASONOGRAPHY OF THE ROTATOR CUFF
503
ogy. This limited acceptance may be due in part to the
ponent on arthroscopy. The partial-thickness compo-
paucity of shoulder ultrasonographic studies in the ortho-
nent was misinterpreted as a full-thickness tear. In both
paedic literature compared with magnetic resonance im-
of these shoulders, a focal defect was produced by com-
aging studies, the frequent lack of local radiological
pression of the deltoid muscle against the rotator cuff
expertise, and difficulty in recognizing the relevant anat-
with the transducer, an integral part of our examination.
omy and pathology on hard-copy ultrasonographic im-
Like full-thickness tears, partial-thickness tears involv-
ages. In addition, wide ranges of sensitivity (57 to 100
ing more than 50 percent of the cuff substance appear to
percent) and specificity (50 to 100 percent) have been
demonstrate a focal defect (a criterion used to define a
reported, in series ranging from ten to 225 patients, for
full-thickness tear) when the deltoid muscle is com-
the ultrasonographic detection of rotator cuff tears, caus-
pressed into the tear. While this maneuver increased the
ing further uncertainty about the true accuracy of this
sensitivity of ultrasonography for detecting small, non-
modality1-3,5,7-9,12,13. Investigators who reported poor results
retracted, full-thickness tears, it lowered the specificity;
for the diagnosis of rotator cuff tears used ultrasono-
ultrasonography may not be able to differentiate exten-
graphic criteria that either are no longer accepted or have
sive partial-thickness tears from full-thickness tears.
been refined, employed a scanning technique that has
In three shoulders in which ultrasonography under-
since been modified to improve visualization of the cuff,
estimated the extent of the tear, arthroscopy showed a
and used older equipment and transducers with a lower
medial midsubstance extension of the supraspinatus
frequency than is currently available1,3,8,9.
tear into the infraspinatus tendon. The midsubstance
In the present study, the accuracy of shoulder ultra-
component of the tear was not detected when we
sonography was reinvestigated in the context of modern
viewed only the more lateral aspect of the rotator cuff
refinements in the scanning technique, improvements in
near its insertion, which demonstrates the importance
the resolution capabilities of the equipment, and clarifi-
of proper positioning of the arm to visualize the rotator
cation of the criteria for diagnosing a rotator cuff tear.
cuff not only at its insertion but more medially.
In contrast to many of the earlier studies, in which the
Our ability to detect partial-thickness rotator cuff
findings on ultrasonography were correlated with those
tears with ultrasonography was limited; however, two of
on arthrography or open surgery, we compared the find-
the five shoulders that had a false-negative study had a
ings on ultrasonography with those on arthroscopy,
decreased range of motion (the patient was unable to
which is a procedure with several potential advantages.
externally rotate and extend the shoulder past the level
Magnified arthroscopic images can provide accurate
of the buttock) that prevented a thorough evaluation of
intra-articular and bursal visualization of the rotator
the cuff, and in a third the partial-thickness tear that was
cuff and, in contrast to arthrography, can delineate
identified on arthroscopy consisted only of mild fraying
partial-thickness and midsubstance tears as well as
of the supraspinatus tendon, which may not be detect-
intra-articular pathology of the biceps tendon.
able with ultrasonography. While Brenneke and Mor-
We found that ultrasonography was highly accurate
gan also reported a low sensitivity for the detection of
for detecting full-thickness rotator cuff tears and for
partial-thickness tears2, two other recent studies demon-
characterizing their extent in the transverse plane. It led
strated a sensitivity of more than 90 percent12,13.
to a misdiagnosis of a full-thickness tear in only three
Biceps tendon abnormalities frequently are associ-
shoulders, all of which had a partial-thickness tear on ar-
ated with rotator cuff tears. In the current study, the prev-
throscopy, with one of the tears involving more than 50
alence of rupture of the biceps tendon was 11 percent
percent of the cuff substance. Our sensitivity rate of 100
and that of dislocation was 6 percent. The dislocations,
percent and our specificity rate of 85 percent compare
whether anterior or medial to the lesser tuberosity, were
favorably not only with the rates reported in recent pre-
recognized easily on ultrasonography; we correctly diag-
vious studies on ultrasonography (in which sensitivity or
nosed five of the six dislocations. On the other hand, we
specificity, or both, has been as high as 95 percent12,13)
identified only seven of the eleven biceps tendon rup-
but also with those reported in numerous magnetic res-
tures. Adhesion of a ruptured biceps tendon at the articu-
onance imaging studies6,8,10,11.
lar entrance to the groove was the most likely cause of a
Only a few studies have evaluated the use of ultra-
false-negative ultrasonogram. Two of the false-negative
sonography for determining the size and extent of the
ultrasonographic studies showed the normal echogenic
tear2,13. Brenneke and Morgan reported that ultrasonog-
fibrillar pattern of the tendon within the groove, creating
raphy was accurate for predicting the size of large tears
the false impression of an intact tendon.
but less so for moderate and small tears2. We found that
Our study was limited by its retrospective design;
ultrasonography was very accurate in predicting the
however, when the operative and ultrasonographic
extent of any tear in the transverse plane. Our findings
findings were in disagreement, representative ultra-
substantiate those reported by Wiener and Seitz13. Two
sonographic hard-copy and arthroscopic images were
of the shoulders in which we overestimated the extent
reviewed jointly to explain the discrepancy. Addition-
of the tear by more than one centimeter had a full-
ally, prior to the inception of this study, standardized
thickness tear with an extensive partial-thickness com-
criteria for determining the presence, location, and ex-
VOL. 82-A, NO. 4, APRIL 2000

---

504
S. A. TEEFEY ET AL.
tent of a rotator cuff tear in the transverse plane had
sensitivity would have decreased markedly had we in-
been established, and the statistical analysis was based
cluded all patients with normal ultrasonograms.
on the original interpretation of the ultrasonographic
We found that ultrasonography was a highly accu-
study rather than on a retrospective review of the
rate and reliable technique for detecting full-thickness
images.
rotator cuff tears and biceps tendon dislocations in pain-
Although diagnostic arthroscopy was performed in
ful shoulders. The high accuracy is in part attributable to
an unblinded fashion, the surgeon’s knowledge of the
improved image resolution, optimization of the scanning
ultrasonographic results prior to the operation was ad-
technique, and reliance on well defined criteria. How-
vantageous to the patient as it led to a more focused
ever, more than with almost any other imaging modality
evaluation of the rotator cuff, particularly when the ar-
that is employed to evaluate the shoulder, the success of
throscopic findings did not correlate with the ultrasono-
an ultrasonographic examination depends heavily on the
graphic report. In all shoulders for which a discrepant
experience of the operator.
ultrasonographic finding was reported, the area in ques-
In summary, our findings indicate that shoulder ultra-
tion was tagged with a suture intraoperatively to allow
sonography can be a valuable noninvasive procedure for
focused intra-articular and bursal-side viewing of the
imaging of the rotator cuff. Not only is it comparable with
cuff.
magnetic resonance imaging in terms of accuracy for de-
Patients with a normal ultrasonogram who had reso-
tecting full-thickness tears; it provides bilateral informa-
lution of the symptoms did not have arthroscopy and
tion, is better tolerated, allows patient viewing of real-
were not included in the study. Hence, it is possible that
time information, and is less expensive. Improvements in
the actual number of false-negative studies may have
image resolution have allowed for more intuitive ana-
been greater than what our study showed. On the other
tomical and correlative pathological interpretation of the
hand, patients with normal ultrasonograms but persis-
hard-copy images by orthopaedic surgeons. Increased
tent symptoms frequently had arthroscopy and thus were
awareness of the important role that ultrasonography
included in the study. Since patients with persistent
can play in the diagnosis of rotator cuff pathology may
symptoms are more likely to have a tear than patients in
foster acceptance and increase the availability of this im-
whom the symptoms have resolved, it is unlikely that the
aging modality to the orthopaedic community.
References
1. Brandt, T. D.; Cardone, B. W.; Grant, T. H.; Post, M.; and Weiss, C. A.: Rotator cuff sonography: a reassessment. Radiology, 173: 323-
327, 1989.
2. Brenneke, S. L., and Morgan, C. J.: Evaluation of ultrasonography as a diagnostic technique in the assessment of rotator cuff tendon tears.
Am. J. Sports Med., 20: 287-289, 1992.
3. Burk, D. L., Jr.; Karasick, D.; Kurtz, A. B.; Mitchell, D. G.; Rifkin, M. D.; Miller, C. L.; Levy, D. W.; Fenlin, J. M.; and Bartolozzi, A. R.:
Rotator cuff tears: prospective comparison of MR imaging with arthrography, sonography, and surgery. AJR: Am. J. Roentgenol., 153: 87-
92, 1989.
4. Clark, J. M., and Harryman, D. T., II: Tendons, ligaments, and capsule of the rotator cuff. Gross and microscopic anatomy. J. Bone and Joint
Surg., 74-A: 713-725, June 1992.
5. Hodler, J.; Fretz, C. J.; Terrier, F.; and Gerber, C.: Rotator cuff tears: correlation of sonographic and surgical findings. Radiology, 169: 791-
794, 1988.
6. Iannotti, J. P.; Zlatkin, M. B.; Esterhai, J. L.; Kressel, H. Y.; Dalinka, M. K.; and Spindler, K. P.: Magnetic resonance imaging of the shoul-
der. Sensitivity, specificity, and predictive value. J. Bone and Joint Surg., 73-A: 17-29, Jan. 1991.
7. Mack, L. A.; Matsen, F. A., III; Kilcoyne, R. F.; Davies, P. K.; and Sickler, M. E.: US evaluation of the rotator cuff. Radiology, 157: 205-
209, 1985.
8. Nelson, M. C.; Leather, G. P.; Nirschl, R. P.; Pettrone, F. A.; and Freedman, M. T.: Evaluation of the painful shoulder. A prospective com-
parison of magnetic resonance imaging, computerized tomographic arthrography, ultrasonography, and operative findings. J. Bone and
Joint Surg., 73-A: 707-716, June 1991.
9. Paavolainen, P., and Ahovuo, J.: Ultrasonography and arthrography in the diagnosis of tears of the rotator cuff. J. Bone and Joint Surg., 76-A:
335-340, March 1994.
10. Quinn, S. F.; Sheley, R. C.; Demlow, T. A.; and Szumowski, J.: Rotator cuff tendon tears: evaluation with fat-suppressed MR imaging with
arthroscopic correlation in 100 patients. Radiology, 195: 497-500, 1995.
11. Rafii, M.; Firooznia, H.; Sherman, O.; Minkoff, J.; Weinreb, J.; Golimbu, C.; Gidumal, R.; Schinella, R.; and Zaslav, K.: Rotator cuff lesions:
signal patterns at MR imaging. Radiology, 177: 817-823, 1990.
12. Van Holsbeeck, M. T.; Kolowich, P. A.; Eyler, W. R.; Craig, J. G.; Shirazi, K. K.; Habra, G. K.; Vanderschueren, G. M.; and Bouffard, J. A.:
US depiction of partial-thickness tear of the rotator cuff. Radiology, 197: 443-446, 1995.
13. Wiener, S. N., and Seitz, W. H., Jr.: Sonography of the shoulder in patients with tears of the rotator cuff: accuracy and value for selecting
surgical options. AJR: Am. J. Roentgenol., 160: 103-107, 1993.
THE JOURNAL OF BONE AND JOINT SURGERY

---