Loss of glenohumeral motion secondary to idiopathic adhesive
capsulitis (frozen shoulder) usually responds to conservative treatment
emphasizing passive range-of-motion exercises1.
Patients with frozen shoulder refractory to nonoperative treatment
have in most cases been treated successfully with manipulation under
anesthesia1,2. Historically, those
patients in whom manipulation was unsuccessful either lived with
permanent restriction of motion or had open soft-tissue release3. More recently, arthroscopic release
of frozen shoulder refractory to nonoperative treatment has been
successful in improving range of motion and reducing pain4,5. This procedure has also been used
to treat postoperative loss of motion with minimal morbidity6.
We believed that arthroscopic capsular release of a frozen shoulder
that develops after shoulder surgery or after fracture of the shoulder
girdle might be less successful than such a release of an idiopathic
frozen shoulder. Thus, we compared the objective improvement in
range of motion and the subjective improvement in function after
arthroscopic capsular release for refractory postoperative or post-fracture
capsular contracture with those outcomes after arthroscopic release
for idiopathic frozen shoulder.
Patient Selection
From April 1994 through March 1997, 1720 new patients were evaluated
at the Shoulder and Elbow Service at the University of Pennsylvania
Medical Center because of loss of shoulder motion secondary to soft-tissue
contracture. One hundred and thirty-five of these patients (135
shoulders) underwent examination and manipulation under anesthesia because
the frozen shoulder was refractory to conservative management, which
included supervised physical therapy and a home-exercise program
of at least one year’s duration. In fifty-four of these
shoulders, a range of motion of at least 80% of that of
the contralateral shoulder was regained in all planes with manipulation
under general anesthesia, and an arthroscopic capsular release was
not performed. Of these fifty-four frozen shoulders, thirteen occurred
after fracture, seven were in patients with diabetes, and thirty-four
were idiopathic. In the remaining eighty-one patients, arthroscopic
capsular release was performed after the attempt at manipulation
under anesthesia failed.
Thirty-one patients undergoing arthroscopic capsular release were
excluded from the study. Nine of these patients had insulin-dependent
diabetes, and ten had substantial degenerative arthritis of the
glenohumeral joint. Two patients had a persistent full-thickness
rotator-cuff tear at the time of the arthroscopic capsular release,
and one patient had an anterior capsulorrhaphy at the time of an
arthroscopic posterior capsular release. One patient had a seizure
disorder with recurrent postoperative dislocations that made follow-up
data unreliable. Six patients were lost to follow-up, and two patients with
incomplete preoperative data were excluded.
The remaining fifty patients formed the cohort for this study. Thirty-three
of the fifty patients experienced loss of motion following surgery
(postoperative group). Of these thirty-three patients, nine had
undergone an arthroscopic acromioplasty; nine, an open rotator-cuff
repair for a full-thickness defect; four, an open Bankart repair;
two, an open acromioplasty; two, an arthroscopic capsular shift;
two, an open reduction and internal fixation of a fracture of the
proximal part of the humerus; two, a prosthetic shoulder replacement;
one, an open biceps tenodesis; one, an arthroscopic repair of a
lesion of the superior portion of the labrum, anterior and posterior (SLAP
lesion); and one, an open calcium-deposit excision. Six of the fifty
patients experienced persistent loss of motion at least twelve months
after a fracture about the shoulder that was treated nonoperatively
(post-fracture group). Of these six patients, two had loss of motion
after a glenoid fracture; two, after a fracture of the proximal
part of the humerus; and two, after a fracture-dislocation of the
proximal part of the humerus. None of these patients had a malunion
or joint incongruity, which were contraindications for arthroscopic capsular
release. Eleven of the fifty patients experienced loss of motion
without apparent cause (idiopathic group).
Clinical Assessment
All of the patients were evaluated preoperatively and at the time
of follow-up with use of a modified American Shoulder and Elbow
Surgeons (M-ASES) shoulder score7,
a subjective scoring tool designed to assess pain, patient satisfaction,
and function. The maximum possible overall score is 100 points,
of which 30 points are awarded for pain; 10 points, for patient
satisfaction; and 60 points, for function. Pain is scored with use
of three visual analog scales, each ranging from 0 points (severe
pain) to 10 points (no pain), in order to evaluate pain at rest,
pain with normal activities, and pain with strenuous activities.
A maximum score of 10 points is assigned for each scale, for a total
possible score of 30 points for pain. Patient satisfaction was also
measured with use of a visual analog scale that ranged from 0 points
(not satisfied) to 10 points (very satisfied). Functional outcome
was assessed with use of a twenty-item questionnaire that focused on
activities that involved shoulder function. A maximum of 60 points
was awarded for normal shoulder function, (i.e., the ability to
perform all twenty functional activities normally). For each item,
a score of 3 points indicated that the patient had no difficulty
in performing the task; 2 points indicated some difficulty; 1 point,
much difficulty; and 0 points, that the patient was unable to perform
the function. The scores for the three categories (pain, patient
satisfaction, and function) were tabulated to obtain the total shoulder
score.
Passive range of motion of the shoulder was measured with the
patient in the supine position preoperatively, intraoperatively,
and at the time of final follow-up. Elevation was measured in the
sagittal plane. External rotation was measured in the coronal plane
in adduction (with the arm by the patient’s side) as well
as in 90° of abduction. Internal rotation was measured in the coronal
plane in 90° of abduction.
Operative Technique
All patients had an interscalene block before surgery to facilitate
postoperative pain control and range of motion of the shoulder.
Interscalene catheters were used in three patients who had had at
least one unsuccessful attempt at manipulation under anesthesia
or an arthroscopic capsular release performed at another institution.
This subset of patients was admitted for one or two days of inpatient
supervised physical therapy and intermittent administration of local
anesthesia via the catheter. All other patients had outpatient surgery.
The glenohumeral joint was injected with 3 mL (18 mg) of betamethasone
at the end of the procedure.
A closed manipulation under anesthesia was performed in all patients,
first in forward flexion, by elevating the arm in the sagittal plane
while the surgeon stabilized the scapula by placing one hand along
its axillary border. The arm was then passively forward flexed in
the sagittal plane to the maximum possible extent. Next, passive
external rotation was performed in 0° of abduction, followed by
external rotation in 90° of abduction. At each position, the scapula
was manually stabilized and the rotation of the shoulder was performed
by rotation of the humerus at the elbow level, rather than at the
level of the forearm or hand, in order to protect the elbow ligaments.
Lastly, internal rotation in 90° of abduction and cross-body adduction
were performed. Arthroscopic capsular release was performed if the
manipulation under anesthesia did not restore at least 80% of
the range of motion of the normal, contralateral shoulder in all
planes.
Arthroscopic capsular release is performed with the patient in the
beach-chair position. A 4.5-mm arthroscope is placed through a posterior-superior
portal and an anterior-superior portal is made just below the long
head of the biceps tendon. A hooked electrocautery (Linvatec, Largo,
Florida) is used to release first the rotator interval, the superior
glenohumeral ligament, and the coracohumeral ligament. The superior
aspect of the capsule is released just superior to the glenoid until
the overlying supraspinatus muscle belly can be seen. At the anterior
border of the supraspinatus, a full-thickness capsulotomy is continued
through the rotator interval. At the superior border of the subscapularis,
the electrocautery is brought deep to the subscapularis, and the
anterior aspect of the capsule is incised down to the five-thirty
position, approximately 5 mm lateral to the glenoid labrum. A blunt
elevator or curved-end biting punch is then inserted in the anterior-superior
portal to dissect the inferior aspect of the capsule away from the
axillary nerve. Capsulotomy near the glenoid rim in the inferior pouch
minimizes the risk of injury to the axillary nerve because the nerve
is closest to the capsule at the midpoint between the capsule’s
glenoid and humeral insertion sites. Manipulation in forward elevation
and abduction-external rotation after anterior capsular release
often releases the inferior pouch. The arthroscope is next placed
in the anterior-superior portal, and the posterior aspect of the
capsule is released in the midportion of the capsule through the
posterior-superior portal. An accessory posterior-inferior portal
can be used to release the remaining portion of the inferior pouch. Manipulation
in internal rotation and 90° of abduction helps to complete the
posterior release.
All patients underwent arthroscopic evaluation of the subacromial
space. The subacromial space was debrided whenever scar tissue was
visualized within it. If adequate range of motion, especially in
external rotation, cannot be obtained after capsular release, subacromial
scarring is often the cause. The subacromial space should be explored
and debrided to free the rotator cuff. Often heavy scarring connects
the acromion and the deep deltoid fascia to the underlying rotator cuff.
Complete débridement of this scar is important to regain full
range of motion. The coracoacromial ligament is routinely released
at the base of the coracoid. Subacromial débridement was
performed in six of the thirty-three shoulders in the postoperative
group. All six shoulders in the post-fracture group required formal
débridement of the subacromial space to regain maximum
range of motion. Two of the eleven shoulders in the idiopathic group
had débridement of the subacromial space, not to improve
motion but to treat bursal surface rotator-cuff tears.
Postoperative Treatment
Beginning on the day of surgery, all patients performed a home-exercise
program. They were instructed to perform each of the five exercises
for two minutes six times a day (for a total exercise time of one
hour per day). This program included assisted passive forward elevation
with the patient supine, passive external rotation with the arm
by the side, passive external rotation with the arm in abduction,
assisted passive internal rotation up the back, and assisted passive
cross-body adduction. An overhead pulley was used to initiate active-assisted
forward elevation. The patients were also encouraged to use the
involved arm in activities of daily living. Strengthening with use
of Therabands (Hygenics, Akron, Ohio) was not begun until the patient
had minimal pain with the range of motion of the shoulder (generally
four to six weeks after surgery).
Statistical Analysis
Statistical analysis was performed with use of one-way analysis
of variance to determine whether the scores differed significantly
among the three groups. If a difference was noted, t tests were
used to compare groups and to compare baseline and follow-up measurements
within each group. An analysis of covariance was performed to determine
whether the follow-up scores differed among the groups after adjustment
for baseline scores. All analyses were performed with use of Statistical
Analysis System software (version 6.12; SAS, Cary, North Carolina).
Postoperative Group
After arthroscopic capsular release, the mean duration of follow-up
for the thirty-three patients with postoperative contracture was
twenty months (range, twelve to forty-six months). There were no
complications related to the arthroscopic procedure. Four patients
continued to have stiffness, which required open capsular release
with a z-plasty lengthening of the subscapularis. All four of these
patients were involved in a Workers’ Compensation claim.
Two had had a previous open rotator-cuff repair, and two had had
a previous Bankart repair. The mean shoulder score and range of
motion improved significantly for the thirty-three patients in this group
(Table I).
Post-Fracture Group
The mean duration of follow-up for the six patients with contracture
after a fracture was twenty months (range, twelve to forty-six months).
There were no complications and no failures in this group. The mean
shoulder score and the mean range of motion in all planes improved
significantly (Table I).
Idiopathic Group
The mean duration of follow-up for the eleven patients with idiopathic
frozen shoulder was twenty months (range, twelve to thirty-two months).
There was one failure secondary to biceps tendinitis, and the patient
underwent a biceps tenodesis with full recovery. No complications
were observed in any of these patients. Again, significant improvement
was seen in the mean shoulder score and in the mean range of motion
in all measured planes (Table I).
Concomitant Surgery
An arthroscopic acromioplasty was performed concomitantly with
the capsular release in three patients in the idiopathic group and
in three patients in the postoperative group, all of whom had a
documented impingement lesion on the coracoacromial ligament.
Comparison Among Groups
Analysis of covariance was performed to compare, among the different
groups, each of the follow-up scores, adjusted for the baseline
preoperative scores. There was no difference in baseline preoperative
scores among the groups (p < 0.39). At the time of follow-up,
the mean pain score was 25 points (out of a possible 30 points)
for the post-fracture group and 25 points for the idiopathic group.
For the postoperative group, the mean pain score was 17 points,
which was significantly lower than the scores for the other two
groups (p < 0.03) (Fig. 1). The mean patient satisfaction
score at the time of follow-up was also lower for the postoperative
group (5 points) compared with that for the idiopathic and post-fracture
groups (9 points for each) (p < 0.004) (Fig. 2), and the mean
functional score was lower as well (35 points for the postoperative
group compared with 49 points and 53 points for the post-fracture
and idiopathic groups, respectively; p < 0.002) (Fig. 3). Finally, the
total subjective shoulder score was significantly lower for the
postoperative group (57 points) compared with that for either the
post-fracture (82 points) or the idiopathic group (86 points) (p < 0.0007).
The idiopathic group had the greatest mean range of motion at the
time of follow-up, whereas the postoperative group had the smallest
mean range of motion in all planes, but the range of motion did
not differ significantly among the three groups. Mean forward elevation
at the time of follow-up was 163° for the idiopathic group, 157°
for the post-fracture group, and 154° for the postoperative group.
Mean external rotation with the arm at the side was 58° for the
idiopathic group, 53° for the post-fracture group, and 44° for the
postoperative group. Mean external rotation in 90° of abduction
was 98° for the idiopathic group, 91° for the post-fracture group,
and 87° for the postoperative group. Mean internal rotation in 90°
of abduction was 37° for the idiopathic group, 35° for the post-fracture
group, and 26° for the postoperative group.
Arthroscopic capsular release has become a reliable method for
restoring range of motion in patients with idiopathic frozen shoulder
for which physical therapy and manipulation have failed5. A nearly normal range of motion
and good outcome scores can be obtained with this procedure. In
the current study, range of motion significantly improved compared
with preoperative values in each group and did not differ significantly among
the three groups. Arthroscopic capsular release is therefore a reliable
treatment for improving range of motion in patients with postoperative,
post-fracture, or idiopathic frozen shoulder. In 1997, Warner et
al. reported on eighteen patients with postoperative capsular contracture
that was treated with arthroscopic capsular release6. They used the same operative technique
and postoperative management as were used in our study, and they
demonstrated the same results with regard to improvement in range
of motion.
Although patients with postoperative frozen shoulder gained significant
improvement in both the range of motion and the shoulder scores
after arthroscopic capsular release, pain, patient satisfaction,
function, and overall scores were all significantly lower than those
for the other two groups. We believe that the residual pain and
functional limitations experienced in the postoperative group were
caused by the initial injury or the initial surgery, resulting in
problems beyond loss of motion. Four of thirty-three patients with
postoperative contracture required formal open capsular release
and z-plasty lengthening of the subscapularis. All four had had
prior open surgery that resulted in both intra-articular capsular
scarring as well as extra-articular scarring.
The presence of a subacromial spur was evaluated in all patients.
In only one case requiring subacromial débridement for
scarring was a subacromial decompression performed. On the basis
of these findings, we do not believe that persistent impingement
was the cause of the persistent pain in the patients in the postoperative
group. Because of the small sample size, when the data were stratified
we were unable to define which specific component of shoulder pathology resulted
in a less favorable outcome after arthroscopic capsular release
for postoperative shoulder stiffness.
The results of arthroscopic capsular release were similar in the
post-fracture and idiopathic groups. It should be emphasized that
none of the patients in the post-fracture group had avascular necrosis
or posttraumatic arthritis. The results in the post-fracture and
idiopathic groups should not be considered attainable in patients
with a malunion or articular incongruity or a history of shoulder
surgery.
Caution is necessary in predicting the outcome of arthroscopic
capsular release in patients with postoperative capsular contracture.
Arthroscopic capsular release can result in significant improvement
in range of motion of the shoulder, but because of underlying concomitant
pathology the results with regard to pain and function can be less
favorable than those in patients with idiopathic frozen shoulder.