The rate of satisfactory results after primary rotator cuff repair
has been reported to be greater than 90% in a number of
studies1-8. However, the rate
of unsatisfactory results has been as high as 25% in some
series9-11. Previous studies have
clearly demonstrated that the results of revision rotator cuff repair
are inferior to those of primary repair12-15.
While pain relief was generally achieved in those studies, the functional
results were extremely unpredictable and, thus, the overall results
were disappointing. Factors that have been associated with poor
results following revision surgery have included detachment of the deltoid
origin, previous acromionectomy, and poor-quality rotator cuff tissue12.
We examined the results of revision rotator cuff repair in a
large series of patients in order to assess which subgroups of patients
had the greatest chance for a satisfactory functional outcome. The
functional result was examined with regard to several factors, including
the size of the cuff tear at the time of both the initial procedure
and the revision, the status of the deltoid at the time of revision,
the quality of the rotator cuff tissue, the active range of shoulder
motion prior to revision, and the number of previous operative procedures.
We reviewed the records of ninety consecutive patients in whom
a failed rotator cuff repair had been revised at our institution
by one of the three senior authors (L.U.B., E.L.F., or R.G.P.) between
1988 and 1996. Ten patients had been lost to follow-up or had died,
leaving eighty patients available for the present study. Patients with
cuff tear arthropathy and those who had undergone latissimus dorsi
or teres major transfer (who had been identified preoperatively
as having massive cuff deficiency that was not amenable to repeat
repair) were excluded from the study. The patients who were included in
the study had had an average of 1.5 (range, one to four) operative
procedures before the revision. Fifty-two patients (65%)
had had one prior surgical procedure, nineteen had had two, eight
had had three, and one had had four. Before the revision, all patients
were evaluated on the basis of a history, a physical examination,
a review of the previous operative notes, and an examination of
standard plain radiographs (anteroposterior radiographs in the scapular
plane, supraspinatus outlet radiographs, and axillary radiographs).
The study included forty-seven men and thirty-three women. The
average age of the patients at the time of the revision was fifty-nine
years (range, twenty-six to eighty-three years). The right shoulder
was involved in fifty-six patients (70%), and the left
shoulder was involved in twenty-four. The dominant shoulder was involved
in fifty-five patients (69%), and the nondominant shoulder
was involved in twenty-five. All patients complained of pain and
weakness at the time of presentation. At the time of the postoperative
interview, the patients were asked to quantify the preoperative
level of pain on a scale of 0 (no pain) to 10 (severe pain). Twenty-six
patients (33%) reported some improvement following the
first rotator cuff repair, whereas fifty-four (68%) reported
no improvement following the first procedure. The primary goal of
revision surgery was relief of pain, with restoration of function,
motion, and strength as additional objectives. The average interval between
the first procedure and the revision was twenty-six weeks (range,
three to 168 weeks).
The previous operative reports were reviewed in detail in order
to identify risk factors for failure of the first procedure. The
initial tear was found to have been massive (>5 cm) in
twenty-two patients (28%), large (>3 to 5 cm)
in twenty-eight (35%), medium (1 to 3 cm) in ten (13%),
and small (<1 cm) in thirteen (16%). No specific details
regarding tear size could be found in the operative reports of the
remaining seven patients. Forty-nine patients had had an inadequate
subacromial decompression at the time of the first procedure, as
judged both from the details in the operative reports as well as from
the intraoperative appearance of the anteroinferior aspect of the
acromion at the time of revision. Four patients had undergone a
lateral or complete acromionectomy and thirty-five had undergone
detachment of the deltoid origin from the lateral part of the acromion
during the initial procedure. Twenty-four patients were noted to
have had inadequate tendon tissue for a secure tendon-to-bone repair
at the time of the first procedure. Seventeen patients had had inadequate
tendon mobilization as indicated by tension at the site of the repair
with the arm at the side or by the need to secure the repair with
the arm in substantial abduction. Finally, nine patients were judged
to have had either inadequate rehabilitation (e.g., failure to initiate
early passive motion) or what we deemed to have been inappropriate rehabilitation
(e.g., premature strengthening and resistive exercises) following
the first procedure.
Before the revision, physical examination revealed that the active
range of motion averaged 105° (range, 0° to 180°) of elevation in
the scapular plane, 39° (range, -10° to 90°) of external rotation
with the arm at the side, and internal rotation to the eleventh
thoracic vertebra. Manual testing revealed that the average strength
in external rotation with the arm at the side was grade 2.7 of 5.
Before the revision, twenty-nine patients were evaluated with magnetic
resonance imaging and five were evaluated with arthrography. In
all of these patients, recurrence or persistence of a defect in
the rotator cuff was demonstrated on the imaging studies. In the
rest of the patients, the diagnosis was made on the basis of the history,
the physical findings, and/or the failure of the patient’s
condition to improve following the initial repair.
The patients were evaluated at an average of forty-nine months
(range, twenty-five to 110 months) after the revision procedure.
Patients were asked to rate their pain subjectively on a scale of
0 (no pain) to 10 (severe pain). They were also asked to indicate,
with a "yes" or "no" answer,
whether they were subjectively satisfied with the results of the
revision. This response was recorded as patient satisfaction, which
was measured separately from functional outcome. Function was graded
according to the level of pain and the active range of motion. The
result was considered excellent if the patient was essentially pain-free,
had an active range of motion that was within 10° of normal in all
planes, and had resumed unrestricted activities; good if the patient
had only occasional soreness or aching, >140° of active
forward elevation and >30° of active external rotation,
and some limitation of functional activity with repetitive or strenuous
overhead activity; and fair if the patient had intermittent episodes
of pain necessitating occasional use of analgesics, 90° of active
forward elevation and 5° of active external rotation, and persistent
weakness and limitation of function with some improvement after
the revision procedure. As excellent, good, and fair ratings indicated
that the patient had substantial pain relief and elevation of the
arm above the horizontal, they were considered to indicate a "satisfactory" functional
result. The result was considered poor, or "unsatisfactory," if the
above criteria were not met.
Operative Repair and Findings
An anterosuperior deltoid-splitting approach was used, although
an effort was made to utilize the previous skin incision if it allowed
safe and appropriate access for revision. The specific operative
maneuvers varied somewhat in each patient, depending on the specific
pathological findings encountered. Fifty-three patients (66%) required
acromioplasty at the time of the revision, either because of inadequate
removal of bone or because of failure to address acromial morphology
at all during the initial procedure. An additional twenty-six patients (33%)
underwent excision of the distal part of the clavicle because of
tenderness and symptoms originating from the acromioclavicular joint.
Mobilization of the remaining rotator cuff tissue was a critical
and often challenging part of the procedure. According to the criteria
described earlier, the tear that was found at the time of the revision
was characterized as massive in twenty-four patients (30%),
large in twenty-seven (34%), medium in thirteen (16%),
and small in sixteen (20%). The cuff was mobilized and repaired
to bone at the junction of the humeral head and the greater tuberosity.
In forty-five patients (56%), an anterior interval slide2 was performed to aid in cuff mobilization.
During this procedure, the rotator interval between the supraspinatus
and the subscapularis is released to the base of the coracoid process,
thus releasing the coracohumeral ligament and allowing the torn
supraspinatus tendon to be mobilized laterally. In another thirty-three
patients (41%), a posterior release of the supraspinatus-infraspinatus
interval was performed. These releases were performed in conjunction
with a blunt release of the undersurface of the rotator cuff from
the capsule at the glenoid rim. In eight patients, the superior
one-third of the subscapularis was transferred superiorly.
Twenty-four patients had an additional procedure that involved
an attempt to mobilize and repair a damaged or detached deltoid
back to the acromion. Four of these patients had had a previous
lateral acromionectomy. In the other twenty patients, the acromion
was intact and the deltoid had apparently pulled off from the acromion postoperatively.
In addition, eight patients had problems related to suture anchors
that had been used at the time of the first procedure. Three of
these patients had loosening of one or more anchors or bone fragments
that required removal. The remaining five had prominent hardware,
which was impacted into the humeral head at the time of revision.
Postoperatively, all patients used a sling for six weeks and
followed a three-phase shoulder rehabilitation program. An abduction
brace was used only when the deltoid had been repaired. Limited
passive range of motion, including pendulum exercises, passive elevation,
and passive external rotation, was begun on the first postoperative
day. Extension and internal rotation were avoided in the early postoperative
period. Active range of motion was begun at six weeks postoperatively,
and resistive exercises were begun at three months.
Analysis of Subgroups
In addition to the group as a whole, specific subgroups were
analyzed. The results of the repair were first analyzed with respect
to the size of the cuff tear at the time of both the initial procedure
and the revision. The results were then analyzed to determine which
subgroups of patients have a better prognosis for a successful functional
outcome following revision rotator cuff repair. The subgroups were
chosen on the basis of four factors: (1) the presence of an intact
deltoid origin prior to revision, (2) the presence of good-quality
rotator cuff tissue (a subjective assessment made by the surgeon
on the basis of the absence of excessive thinning or tissue friability),
(3) preoperative active elevation of >90°, and (4) only
one prior operative procedure. The subgroups were analyzed individually
as well as in combination. The Student t test was used for statistical
analysis. The entire group was analyzed with respect to preoperative
and postoperative pain and motion, and the subgroups were compared
with each other with respect to the overall functional result.
Overall Results
The results were graded on the basis of pain, range of motion,
and function. Overall, fifty-five patients (69%) had a
satisfactory result: twenty-six (33%) had an excellent
result, twenty (25%) had a good result, and nine (11%)
had a fair result. Twenty-five patients (31%) had an unsatisfactory
(poor) result. The average pain rating improved from 7.4 points
preoperatively to 3.0 points postoperatively (p < 0.001).
Sixty-nine patients (86%) reported pain relief after the
revision, while eleven (14%) reported no substantial pain
relief. Overall, fifty-six patients (70%) stated that they
were satisfied with the results of the revision rotator cuff repair.
The average active elevation in the scapular plane improved from
105° preoperatively to 130° postoperatively (p < 0.005),
for an average increase of 25°. The average active external rotation
with the arm at the side improved from 39° preoperatively to 53°
postoperatively, for an average increase of 14° (p < 0.005).
The average active internal rotation improved only slightly, from
the eleventh thoracic level preoperatively to the tenth thoracic
level postoperatively. Manual muscle-testing revealed that the average
strength in external rotation with the arm at the side improved
from a grade of 2.7 of 5 preoperatively to a grade of 3.5 of 5 postoperatively.
Although the majority of the original procedures had been performed
at other institutions, data on the initial size of the tear were
available for seventy-three patients. These data were analyzed with
the understanding that various surgeons may not use the same definitions
of "massive," "large," "medium," and "small" tears. These
results are summarized in Tables I and II. Patients with a medium or small
tear at the initial procedure had a significantly better functional
outcome following the revision than those with a large or massive tear
at the initial procedure (p < 0.005). The tear size at the
time of the revision was not associated with the functional outcome.
Results Based on Deltoid Status
Forty-five patients (56%) had an intact deltoid origin before
the revision, and thirty-five (44%) did not (Tables III and IV). Seventy-eight
percent of the patients with an intact deltoid origin had a satisfactory
functional result after the revision, compared with only 57% of
those with a compromised deltoid origin. The overall results were significantly
better in patients with an intact deltoid origin (p < 0.05).
Results Based on Cuff Status
Fifty-six patients (70%) had good-quality rotator cuff tissue,
and twenty-four (30%) did not (Tables III and IV). Seventy-seven
percent of the patients with good-quality rotator cuff tissue had
a satisfactory functional result after the revision, compared with
only 50% of those who did not have good-quality rotator
cuff tissue.
Results Based on Preoperative Active Motion
Forty-three patients (54%) were able to actively elevate the
arm above the horizontal in the scapular plane before the revision,
and thirty-seven (46%) were not (Tables III and IV). Seventy-nine percent
of the patients who could actively elevate the arm above the horizontal
preoperatively had a satisfactory result after the revision, compared
with 57% of those who could not. The results in patients
who could actively elevate the arm above the horizontal were significantly
better than those in patients who could not (p < 0.005).
Results Based on Number of Prior Procedures
Fifty-two patients (65%) had undergone only one prior procedure
before the revision, and twenty-eight (35%) had undergone
more than one prior procedure (Tables III and IV). Seventy-three percent of the
patients who had undergone only one prior procedure had a satisfactory
result after the revision, compared with 61% of those who
had undergone more than one prior procedure.
Results Among Patients Who Met All Four Criteria
Seventeen (21%) of the eighty patients in the present study
met all four of the previously mentioned criteria: an intact deltoid
origin, good-quality rotator cuff tissue, active elevation above
the horizontal before the revision, and only one prior procedure.
All seventeen patients had a satisfactory result. Patients who met
all four criteria had significantly better results than the overall
group (p < 0.01) and the other clinical subgroups of patients
who met only one of the criteria (p < 0.05).
Reoperation after a failed operative repair of a torn rotator
cuff is a technically challenging procedure, and few series have
been reported in the literature. DeOrio and Cofield reported a satisfactory
functional result for only four (17%) of twenty-four patients
and recommended that reoperation after a failed cuff repair should be
performed only in selected cases13.
More recent studies by one of us (L.U.B.) and colleagues12 and by Neviaser and Neviaser14 yielded more encouraging results
after reoperation, although the results were inferior to those that
have been reported following primary repair. A number of factors
were found to be associated with unsatisfactory results: detachment
of the deltoid origin, breaching of the acromion as a result of
a lateral or complete acromionectomy, poor-quality rotator cuff
tissue, or a massive cuff deficiency necessitating tissue transfer
in order to obtain adequate coverage of the humeral head.
In the present study, we found that patients with an intact deltoid,
good-quality rotator cuff tissue, preoperative active elevation
of >90°, and a history of only one prior repair had uniformly
satisfactory results. In addition, patients who had a small or medium
tear at the time of the initial procedure had a tendency to fare
better after revision surgery than patients who initially had a
large or massive tear.
Detachment of the deltoid secondary to either acromionectomy
or excessive release during a superolateral or anterosuperior approach
to the shoulder has been repeatedly associated with inferior results
after both primary and revision rotator cuff repairs12-14,16,17. We currently prefer to
use a deltoid-splitting approach. The deltoid split is begun approximately
5 mm anterior to the acromioclavicular joint and is extended directly laterally
past the anterolateral corner of the acromion, splitting the muscle
in line with its fibers for a distance of 3 cm to 4 cm distally.
This approach leaves a strong, healthy cuff of tissue that allows
for a secure repair of the deltoid split, thereby decreasing the
chances of postoperative detachment2,18.
On physical examination, detachment of the deltoid is readily appreciated
as a palpable defect of variable size, beginning at the anterior
or anterolateral edge of the residual portion of the acromion (Fig. 1). Intraoperatively,
the muscle is commonly found to be scarred down to the underlying
humerus and rotator cuff. The resultant change in length and the
extensive adhesions greatly limit its ability to serve as an important
muscle in active elevation of the glenohumeral joint. The present
study confirms the difficulty of restoring shoulder strength and
function when there has been previous deltoid detachment.
In the present series, the results of revision rotator cuff repair
were better when the patient had an adequate amount of healthy rotator
cuff tissue as well as an intact deltoid. The rotator cuff muscles
are critical to smooth, strong, coordinated glenohumeral motion,
and they function not only as a motor but also as dynamic stabilizers
that maintain a stable glenohumeral joint as the deltoid raises
the arm into abduction. Previous reports have detailed several procedures
designed to compensate for large cuff deficiencies, including advancement of
the supraspinatus19,20; transfer
of the superior third of the subscapularis21,22,
the teres major23, or the latissimus
dorsi24; transposition of the
long head of the biceps tendon25;
and use of fascial autograft26,
freeze-dried allograft17, or synthetic graft27.
In general, the results of these procedures have all been inferior
to those observed when the tendons can be mobilized and restored
anatomically to achieve a secure tendon-to-bone repair18. Careful mobilization of existing
rotator cuff tissue by means of anterior and posterior interval
releases, with blunt release of adhesions from both the superficial
and deep surfaces, has decreased our need to resort to tissue transfer
and grafting procedures.
Our finding that active elevation above the horizontal prior
to revision further improved results among patients with an intact
deltoid and good-quality rotator cuff tissue may have been due to
a number of factors. In addition to having adequate muscle and tendon
substance to power glenohumeral joint motion, these patients may
have been less prone to substantial adhesion formation. They also
may have had healthier rotator cuff muscles that were less likely
to be damaged by the repeated trauma of surgery. These patients
also may have been more highly motivated and more diligent in terms
of their rehabilitation regimen. Since these patients had a greater
prerevision range of active motion than did the rest of the patients
with an intact deltoid and good-quality rotator cuff tissue, they
certainly also had less progress to make in order to achieve a good
or excellent result. The gains in motion made by the patients who
could not actively elevate the arm above the horizontal preoperatively
were much greater than those made by the patients who could, in
whom the active range of motion essentially remained unchanged after
the procedure.
Although the current study focused on functional results, adequate
pain relief remains the primary objective of both revision and primary
rotator cuff surgery. As noted in this series as well as others,
substantial pain relief can still be achieved in patients who obtain
less-than-optimal active motion and strength12-14,28.
This seems to be especially true when physical examination and radiographic
studies demonstrate strong evidence of persistent impingement.
In summary, operative treatment of a failed rotator cuff repair
can yield good functional results, particularly for carefully selected
patients. A patient with an intact deltoid origin, good-quality
rotator cuff tissue, good active prerevision motion, and only one
prior procedure is a particularly strong candidate for revision
surgery. Substantial pain relief can also be achieved even in patients who
have only modest functional results. Careful patient selection and
appropriate preoperative counseling can help to maximize patient
satisfaction with this often challenging surgical undertaking.