Study Group
One hundred and five consecutive shoulders in ninety-seven patients
underwent surgical repair of a chronic rotator cuff tear by the
senior author (R.H.C.) between January 1, 1975, and December 31,
1983. Eight patients had staged repairs of a bilateral rotator cuff
tear. Institutional Review Board approval for this study and informed
consent of all patients to participate in the study were obtained.
The criteria for inclusion in the study were a visit by and an examination
of the patient at a minimum of two years following the operation and
additional patient contact by means of a visit, questionnaire, or
telephone interview. The time between the operation and the examination
averaged 4.2 years (range, two to seventeen years), and that between
the operation and the most recent patient contact by telephone or
questionnaire averaged 13.4 years (range, two to twenty-two years). No
patient was lost to follow-up; however, sixteen patients died during
the surveillance period, with the time to death averaging 6.5 years
(range, two to fifteen years) (Fig. 1Fig. 1).
Patient age at the time of the operation averaged fifty-eight
years (range, thirty-eight to seventy-five years). The repairs were
performed on seventy-two shoulders in men and thirty-three in women
and in sixty-five right shoulders and forty left shoulders. Only
four patients were left-hand-dominant. On the basis of the history,
the tear was found to be associated with a varying degree of trauma
in sixty-two shoulders. Twenty-five injuries were work-related. Occupational
activities varied, with thirty-four patients involved in heavy-duty
work; thirty-two, in light-duty work; and seventeen, in office work.
The remaining fourteen patients were retired. Thirty-two patients
were regularly involved in some form of sports activity.
The surgery was performed to reduce pain and to improve the ability
to carry out functional activities; the primary indication for surgical
repair typically was pain that was unresponsive to nonoperative
treatment. With use of a 5-point scale, pain was classified as none,
slight, occasionally moderate with vigorous activities, moderate,
or severe. Eighty-six shoulders were reported to be severely painful
by the patients, fifteen were moderately painful, and four were
not painful. The average duration of symptoms prior to surgery was thirty
months (range, one to 180 months). Ninety-five shoulders had had
previous treatment with analgesic or anti-inflammatory medications. Eighty-two
shoulders had been treated with monitored physiotherapy, and sixty-three
shoulders had received at least one injection (average, two; range, one
to fifteen injections) of cortisone into the subacromial bursa.
Preoperatively, active abduction averaged 128° (range, 0° to 180°),
external rotation averaged 60° (range, 0° to 90°), and median internal
rotation was to the eleventh thoracic level (range, the trochanter
to the fourth thoracic level). Strength was measured by manual muscle-testing according
to the grading system of the Medical Research Council, in which
grade 5 was considered normal; grade 4, good; grade 3, fair; and
grades 2, 1, and 0, poor. Strength in abduction was normal in forty-one
shoulders, good in thirty-three, fair in twenty-eight, and poor
in three. Strength in flexion was normal in sixty shoulders, good
in thirty-three, and fair in twelve. Strength in external rotation
was normal in forty-five shoulders, good in twenty-nine, fair in
twenty-nine, and poor in two. Strength in internal rotation was
normal in forty-six shoulders, good in thirty-five, fair in twenty-one,
and poor in three.
Radiographs made preoperatively included anteroposterior views
with the shoulder in internal and external rotation as well as an
axillary view. Preoperatively there was mild glenohumeral arthritis
in three shoulders. There was cyst formation in the greater tuberosity
in nine shoulders and sclerosis or rarefaction in the greater tuberosity
in thirty-five. Acromial changes included spurring in eighteen shoulders,
sclerosis in eighteen, and a concave appearance of the undersurface
in twenty-five. The acromiohumeral distance was less than 3 mm in
ten shoulders, between 4 and 6 mm in five shoulders, and greater
than 7 mm in fifty-three shoulders in which the measurement could
be made. Shoulder arthrograms had been made for seventy-eight shoulders,
and a full-thickness tear of the rotator cuff was confirmed in seventy-six
of them. The other two shoulders, which had a normal arthrographic
appearance, were found to have a full-thickness tear intraoperatively.
The undersurface of the anterior portion of the acromion appeared
normal in forty-five shoulders, had wear with sclerosis in eighteen,
had a concave or hooked shape in twenty-five, and had anteromedial
spurring or osteophyte formation in seventeen. Tear size was evaluated
by measuring the maximum length or diameter of the full-thickness
tendon tear after limited d衲idement of mucoid or frayed tendon
edges3-5,13,22. The tears were
categorized into five shapes (Table ITable I). The biceps tendon appeared
normal in sixty-one shoulders, had some degree of wear in fifteen, appeared
inflamed in twelve, was torn in eleven, was dislocated in three,
and appeared markedly hypertrophic in three. Biceps tearing or dislocation occurred
with a higher frequency in the shoulders with the larger tears (p
= 0.01).
Operative Procedure and Postoperative Care
The surgical approach was anterosuperior in 101 shoulders and
anteromedial in four. The procedure of tendon repair varied according
to the size and shape of the tendon tear. The repair was tendon-to-tendon
only following d衲idement in fourteen shoulders, was tendon-to-bone
only in sixteen shoulders, and was both tendon-to-tendon and tendon-to-bone
in an anatomically normal position in forty-two shoulders. A V-Y
plasty was performed in twelve shoulders. Subscapularis tendon transposition
upward was a component of the repair in fifteen shoulders, subscapularis
and infraspinatus tendon transposition upward was carried out in
five shoulders, and reinforcement of the tendon repair with a fascia
lata graft was done in one shoulder. All repairs were performed
with the arm in 15° to 20° of abduction. Anterior-inferior acromioplasty was
always performed. The distal 1 cm of the clavicle was excised in
fifty-six shoulders to remove hypertrophic changes on the undersurface
of the acromioclavicular joint that compromised the supraspinatus
outlet or to treat severe acromioclavicular arthritis. Biceps tenodesis
was performed in the three shoulders with dislocation of the biceps tendon.
The type of external support following surgery was determined
by visual analysis of the tension on the repaired tendon at the
time of surgery. A shoulder immobilizer was used for forty-four
shoulders, a low-angled (30°) humeral abduction splint was used
for fifty-seven shoulders to relieve slight tension on a tendon
repair in the posterior or posterosuperior part of the shoulder,
and a low-angled shoulder spica cast was used for four shoulders with
an extensive tendon tear in patients with a thin body habitus. External
support was continued for four to six weeks.
When the patient was treated with an immobilizer or splint, a
passive range-of-motion program for elevation and external rotation
was begun on the second postoperative day and was continued for four
to six weeks. At four to six weeks, an active-assisted motion program
and gentle isometric strengthening were started. At three months, strengthening
and stretching with use of an elastic strap were begun to eliminate
any residual motion deficits.
Rating of the Results
The results were rated according to criteria defined by Neer23 and with use of a 5-point pain-rating
scale. Patients with no pain, active abduction of 145° or greater, and
external rotation of 55° or greater were rated as having an excellent
result. Those with no pain or slight or occasionally moderate pain,
active abduction between 100° and 144°, and external rotation between
30° and 54° were rated as having a satisfactory result. When the
above criteria were not met, the result was rated as unsatisfactory.
Statistical Analysis
The comparison of ordinal or continuous variables among groups
was performed with the Wilcoxon rank-sum test when two groups were
compared and with the Kruskal-Wallis test when more than two groups
were compared. Changes in ordinal or continuous variables were assessed
with use of the Wilcoxon signed-rank test. Association of pairs
of ordinal or continuous variables was estimated with the Spearman
rank correlation coefficient, denoted by "r" in the Results section.
Changes in binary outcomes were assessed with use of the McNemar test
for matched pairs. The method of Kaplan and Meier was used to estimate
survival free of a reoperation. The log-rank test was used to compare
survival between groups. A significance level of 0.05 was used for
all tests.
Pain
Surgical repair was successful in reducing or eliminating pain
(p < 0.0001). At the time of the latest follow-up, fifty-five
shoulders were not painful, thirty-one were slightly painful, ten
were occasionally moderately painful with unusually vigorous activities,
eight were moderately painful, and one was severely painful. Eight
of the nine moderately or severely painful shoulders underwent revision surgery,
as described below. Night pain was also markedly decreased: it was
present in ninety shoulders preoperatively and in twenty-one shoulders postoperatively
(p < 0.0001). Men had somewhat less postoperative pain than
women (p < 0.04), with the median pain rating being none for
men and slight for women (Table IITable II). Although there was a
trend for a higher degree of postoperative pain with a larger tear
size, this association was not significant (p = 0.08).
Range of Motion
There was improvement in active abduction (p < 0.001) and
external rotation (p < 0.007). Active abduction improved an
average of 26°, to 154° (range, 20° to 180°). Postoperative active
abduction was 0° to 59° in four shoulders, 60° to 89° in four shoulders,
90° to 119° in eight shoulders, 120° to 149° in twelve shoulders,
and 150° or greater in seventy-seven shoulders. External rotation
with the arm at the side improved an average of 9°, to 69° (range,
20° to 90°). Median internal rotation improved one spinal level,
from the eleventh to the tenth thoracic vertebra.
Older age was associated with less postoperative active abduction
(p = 0.005, r = -0.33), external rotation (p = 0.004, r = -0.28),
and internal rotation (p = 0.04, r = -0.41) (Table IITable II).
Postoperative active abduction was slightly better in men (p = 0.03),
who had median abduction of 173° compared with 165° in women. A
greater postoperative range of motion correlated with a better preoperative
range of motion for abduction (p = 0.004, r = 0.28), external rotation
(p = 0.04, r = 0.20), and internal rotation (p = 0.04, r = 0.28).
The size of the rotator cuff tear correlated inversely with postoperative
motion (Fig. 2Fig.
2), as shoulders with a large or massive tear had lesser ranges
of active abduction (p = 0.0001), external rotation (p = 0.002),
and internal rotation (p = 0.0009). The repair method also influenced
the final range of motion, as the patients who had had subscapularis
and infraspinatus transposition tended to have less active abduction
(p = 0.05) and external rotation (p = 0.05). Distal clavicular excision
had little effect on postoperative motion in abduction (p = 0.07),
external rotation (p = 0.47), or internal rotation (p = 0.04). Similarly,
a biceps tear or tenodesis did not significantly influence the final active
abduction (p = 0.33), external rotation (p = 0.21), or internal
rotation (p < 0.54).
Strength
Postoperatively, muscle strength in abduction was normal in fifty-three
shoulders, good in thirty-five, fair in fifteen, and poor in two.
Strength in flexion was normal in sixty-five shoulders, good in
twenty-six, fair in twelve, and poor in two. Strength in external
rotation was normal in sixty-two shoulders, good in thirty, fair
in eleven, and poor in two. The improvements in abduction strength
(p < 0.03) and external rotation strength (p < 0.002) were
significant. Older age was associated with less postoperative strength
in abduction (p = 0.08, r = -0.17), flexion (p = 0.02, r = -0.23),
and external rotation (p = 0.03, r = -0.22). Preoperative strength
was associated with postoperative strength, as patients with stronger
muscle groups preoperatively also had stronger abduction (p = 0.002,
r = 0.31), flexion (p = 0.006, r = 0.27), and external rotation
(p = 0.003, r = 0.29) postoperatively. Tear size affected final strength,
with less strength in abduction (p = 0.003, r = -0.33), flexion
(p = 0.001, r = -0.31), and external rotation (p = 0.01, r = -0.27)
in patients with a large or massive tear. Distal clavicular excision
was associated with less strength in abduction (p < 0.03), flexion
(p = 0.0003), and external rotation (p = 0.0008). However, more
patients with a larger tear required distal clavicular resection
(p < 0.02). Distal clavicular resection was carried out in seven of
the sixteen shoulders with a small tear, eighteen of the forty with
a medium tear, twenty-one of the thirty-eight with a large tear,
and ten of the eleven with a massive tear. A biceps tear or tenodesis
was present in one of the sixteen shoulders with a small tear, three
of the forty with a medium tear, five of the thirty-eight with a
large tear, and five of the eleven with a massive tear. A biceps
tear or tenodesis had no significant effect on postoperative strength
in abduction (p < 0.052), flexion (p < 0.23), or external
rotation (p = 0.12).
Patient Satisfaction and Activities
Patients were asked to rate the status of the shoulder after
the surgery as much better, better, the same, or worse. Seventy-seven
shoulders were considered to be much better by the patient; twenty, better;
five, the same; and three, worse postoperatively than they had been
before the surgery. There was a significant association between
the patient's response and the tear size (p = 0.003). Fourteen of the
sixteen shoulders with a small tear, thirty-four of the forty with
a medium tear, twenty-three of the thirty-eight with a large tear,
and six of the eleven with a massive tear were considered to be
much better by the patient. Following the surgery, 89% of the patients
were able to return to work without modifications and 8% returned
to a modified work position. Similarly, 78% of the patients were
able to resume their sports activities, which included golf (twelve
patients), tennis (eight), volleyball (two), basketball (one), racquetball
(one), and darts (one).
Rating of the Results
With use of the Neer system23,
sixty-eight shoulders were rated as excellent; sixteen, as satisfactory;
and twenty-one, as unsatisfactory. There was often a combination
of factors leading to an unsatisfactory result. The result was considered
unsatisfactory because of inadequate pain relief in nine shoulders,
because of limited active abduction in eleven, and because of limited external
rotation in one. There was no association between age and the result
rating (p = 0.14). Men had better result ratings (p = 0.01). The
tear size significantly affected the result rating (p = 0.0002). The
result was excellent for thirteen of the sixteen shoulders with
a small tear, thirty-two of the forty with a medium tear, twenty-one
of the thirty-eight with a large tear, and two of the eleven with
a massive tear. All of the fifteen patients with subscapularis transfer
had a reduction of pain, to no pain or slight pain. There were nine
excellent results, one satisfactory result, and five unsatisfactory
results in this group. Four of the five shoulders with an unsatisfactory
result lost active abduction, ranging from 10° to 90°.
Complications and Reoperations
Four patients had four medical complications: pneumonia, deep
venous thrombosis, myocardial infarction, and postoperative depression.
Two patients had a superficial wound infection, which was successfully
treated with oral antibiotics. In the patient with the fascia lata
graft, a seroma developed at the graft donor site, which drained
for ten days.
There were eight reoperations. Two shoulders, one with a small
tear and one with a medium tear, had excision of a hypertrophic
bursal scar and revision acromioplasty. In another shoulder, with
a massive tear, severe glenohumeral arthritis developed two years
following the rotator cuff repair and prosthetic shoulder arthroplasty
was required. Five patients with clinically apparent failure of
rotator cuff tendon-healing and persistent postoperative pain underwent
rerepair of the rotator cuff tear; the tear at the initial operation
was small in one of these patients, medium in two, and massive in
two. Injury was thought to be the cause of the tearing in two of these
patients; aggressive physiotherapy, in one; and impaired healing
due to renal failure and ongoing dialysis, in one. The cause was
unknown in the fifth patient. Seven of the eight patients who had
a reoperation had some pain relief after it, and one patient continued
to have severe pain. At the final evaluation, three of the shoulders
had an excellent result; one, a satisfactory result; and four, an
unsatisfactory result. A reoperation was needed more commonly by
the shoulders with a massive rotator cuff tear (p = 0.003).
This study confirms that standard repair of rotator cuff tendons
combined with anterior-inferior acromioplasty yields consistently
good and enduring outcomes in a high percentage of patients. Eighty-four
(80%) of the shoulders in our series had an excellent or satisfactory
outcome at an average of thirteen years after the rotator cuff surgery.
This study supports previously published findings indicating that
the preoperative range of motion, preoperative strength, and size
of the tendon lesion affect the eventual outcome1-3,5,8,10,15,22,24-29.
Because of the prospective nature of this study and the consistent
methods of collection of the data on patient characteristics, pathological
findings at the operation, method of operative treatment, and follow-up,
a vigorous statistical analysis could be accomplished to assess
the relationship of various parameters with the outcome of the rotator
cuff repair (Table IITable
II). At the latest follow-up evaluation, the result was excellent
or satisfactory for 94% and 85% of the small and medium tears, respectively,
compared with 74% and 27% of the large and massive tears, respectively.
It is clear that these standard techniques have deficiencies when
used for the repair of massive tears, and introduction of newer,
experimental repair methods for improving the functional outcome
in this group may be justified13,22.
There have been reports suggesting that older patients tend to
have larger tears and that the quality of the cuff tissue, its attachment
to bone, and the potential for a durable repair deteriorate with
age or disuse6,29. In this study,
age was related to tear size, with older patients having larger
tears. We found no clear explanation for the adverse effect of female gender
on the outcome parameters. Lesser preoperative active motion, preoperative
weakness, distal clavicular excision, and transposition repair techniques
were also associated with larger tears, confirming the critical
importance of tear size in determining the outcome of this operation.
Other factors, such as the etiology of the tear, side of the repair,
hand dominance, time from the beginning of symptoms to surgery,
shape of the acromion, location of the tear, biceps tenodesis, and
type of postoperative immobilization, did not have a significant effect
on outcome.
Anterior-inferior acromioplasty was performed in all shoulders
in this series and continues to be an important part of rotator
cuff repair in our practice. We agree with other authors that acromioplasty,
by creating space for the rotator cuff tendons, provides better
surgical exposure and thus improves the quality of the surgical
repair, and that, by relieving impingement, it offers protection
to the tendons during healing, facilitates rehabilitation, and lessens
postoperative pain4,23,30,31.
We now rarely perform distal clavicular excision as part of rotator
cuff repair, as we believe that it contributes little, if any, benefit.
We reserve this procedure for patients with symptoms, signs, and the
radiographic appearance of degenerative changes at the acromioclavicular
joint. If osteophytes on the inferior surface of the acromioclavicular
joint compromise the supraspinatus outlet, they are trimmed. Currently
we perform biceps tenodesis only if greater than 50% of the diameter
of the tendon is torn or the tendon is dislocated medially. We believe
that tendon-transposition repair techniques are occasionally useful
as a part of the surgeon's armamentarium; however, they are infrequently
used in current practice. We continue to individualize the type
of postoperative limb support. A shoulder immobilizer is usually
used both to relieve pain and to remind the patient to avoid early
active use of the shoulder. For selected patients with somewhat
weakened tendon material adjacent to the tear and a more extensive
tear, a splint with a low angle of abduction is used to support
the weight of the arm and hopefully to enhance healing of the repair
with less stress on the tissues. This is particularly true for patients
with a substantial portion of the repair in the posterior aspect
of the rotator cuff. Similarly we continue to use early passive
motion with a delay in active motion for four to six weeks.
It seems clear that surgical repair of symptomatic rotator cuff
tears is a highly satisfactory operation with long-term benefit
to the patient. To improve the results, and particularly to enhance
postoperative active motion and strength, better methods must be
developed for the repair of larger tendon tears.