Rotator cuff repair is one of the most common surgical procedures performed
in the shoulder, and the benefit of repair is well
known1-5.
Over the past decade, the treatment of rotator cuff tears has evolved from an
open procedure to an arthroscopic-assisted (mini-open) technique to an
all-arthroscopic technique. Traditional open rotator cuff repairs produce
satisfactory results when used for the treatment of nonmassive tears (<5
cm). However, this procedure has been associated with morbidity such as severe
early postoperative pain, deltoid detachment and/or weakness, and
arthrofibrosis6-8.
Mini-open repairs were developed because they had the potential advantage of
less deltoid morbidity, and they have demonstrated results that have been
similar to those of open repairs (Figs. 1-A
through
1-D)9-14.
With recent advances in arthroscopic techniques, many surgeons are now
performing complete arthroscopic repairs. The potential advantages of this
procedure include less pain, more rapid rehabilitation, the ability to treat
intra-articular lesions, smaller skin incisions, less soft-tissue dissection,
and an extremely low risk of deltoid detachment
(Figs. 2-A through 2-E). In the
short and long term, the arthroscopic approach has shown promising
results3,7,15-27.
Despite these advantages, the use of the complete arthroscopic repair is
technically demanding and requires a large-volume practice in order for a
surgeon to obtain proficiency in this
procedure28.
Because of the technical demands of this procedure, many orthopaedic surgeons
still consider the mini-open repair to be the gold standard for rotator cuff
repair29. We
hypothesized that arthroscopic rotator cuff repair produces clinical results
comparable with those of mini-open rotator cuff repair, with fewer
complications.
In order to compare the mini-open and all-arthroscopic techniques, we
performed a qualitative systematic review with use of a defined methodology to
collect the most relevant information to answer a specific clinical question.
This analysis included published literature on mini-open and all-arthroscopic
techniques in patients with full-thickness rotator cuff tears with a mean
duration of twenty-four months of follow-up. The purpose of the present study
was to compare the clinical outcomes of mini-open and all-arthroscopic
techniques of rotator cuff repair with use of a systematic review of the
published literature.
Prior to conducting a literature search, we established the study design
and specific objectives. The objectives were (1) to compare the clinical
results of arthroscopic and mini-open rotator cuff repairs with use of
shoulder outcome scales, (2) to compare the postoperative ranges of motion,
and (3) to compare the complication rates for each procedure. The inclusion
criteria were the performance of rotator cuff repair with use of arthroscopic
and mini-open techniques involving tendon-to-bone fixation (i.e., transosseous
tunnels or suture anchors) with a mean duration of follow-up of twenty-four
months. Studies of rotator cuff repair involving direct tendon repair, margin
convergence, or interposed allograft did not meet the inclusion criteria.
Studies were excluded if they involved partial repairs or revision repairs or
if >50% of the rotator cuff tears were massive (>5 cm) or involved
multiple tendons. Studies that did not provide information on the size or
tendon involvement were also excluded. Studies that compared two techniques
were only included if technique-specific data could be extracted for the
analysis. Demographic information, rotator cuff tear characteristics,
operative technical details, objective and subjective outcome measurements,
and complications were gleaned from the studies.
Literature Search
We searched Medline, CINAHL, and the Cochrane Central Register of
Controlled Trials for all literature published from January 1966 to November
2005 using the keywords shoulder, rotator cuff, rotator cuff tear, rotator
cuff repair, arthroscopic, arthroscopic-assisted, mini-open, treatment
outcome, and outcome. General search terms were chosen to
prevent the possibility of missing studies. Studies that were only presented
as abstracts were not included in the final
analysis30. To
ensure that all possible articles were considered, the references of all
relevant articles and review articles were manually cross-referenced.
Data Abstraction
The data were abstracted from each of the studies that met the inclusion
criteria by two independent reviewers (S.J.N. and M.K.S.). The demographic
data that were collected included the type of study; the level of evidence;
the number of patients enrolled; the number of patients at the time of the
latest follow-up; the age, gender, and dominant extremity of the patients; the
duration of follow-up; and the duration of symptoms. The characteristics of
the rotator cuff tear, including size, width, length, and area, were also
collected. Studies that did not describe rotator cuff tears according to size
were only included if they provided the number of tendons involved.
Intraoperative data were recorded, including the surgical technique,
tendon-to-bone fixation, the number of points of fixation, and concomitant
procedures. The percentage of satisfied or very satisfied patients for each
group was collected. Preoperative and postoperative data included range of
motion, strength, clinical outcome scales
(Constant-Murley16,
Neer Shoulder Assessment
Scale31, University
of California at Los Angeles
[UCLA]32, American
Shoulder and Elbow Surgeons
[ASES]33, Simple
Shoulder Test
[SST]23, Japanese
Orthopaedic Association
[JOA]34, Short
Form-36 [SF-36]35,
and Visual Analog Scale
[VAS]36), and
complications were extracted. The complications were subcategorized according
to orthopaedic complications (revision, arthrofibrosis, ruptured biceps
tendon, infection, hematoma, deltoid avulsion, postoperative impingement
syndrome, heterotopic ossification, nerve injury, painful suture, and
hypertrophic scar) and medical complications (pneumonia, myocardial
infarction, and deep venous thrombosis). Additional relevant information, such
as information obtained from postoperative radiographic images, was also
included from each study when appropriate. The data are presented in tabular
format (see Appendix), and no statistical comparisons were performed as part
of the systematic review.
Literature Search
Of the 3445 articles that were identified, 2576 were written in the English
language and involved human subjects. The abstracts of these 2576 studies were
reviewed to determine the appropriateness to the present study as determined
by the inclusion and exclusion parameters. Forty-five articles were
appropriate for the analysis. Ten articles were rejected because they had a
majority of rotator cuff tears that were massive (>5 cm in
size)6,8,15,19,37-42.
Eleven studies were excluded because the mean duration of follow-up was less
than twenty-four
months10,43-52.
Four studies were excluded because they combined more than one method of
treatment but the treatment groups were not
extractable44,53-55.
Two studies were excluded because the size of the tears were not
reported26,56.
The study by Gartsman et
al.21 included the
same cohort as another published
study23, and the
more recent study23
was included in the final analysis. Wilson et
al.27 compared
arthroscopic rotator cuff repair with staple fixation and suture anchor
fixation, but the staple group required arthroscopic removal of the staples;
thus, only the suture anchor group was included in the arthroscopic group.
Another study compared arthroscopic rotator cuff repair of full-thickness and
partial-thickness tears, and only the full-thickness rotator cuff repairs were
included in the arthroscopic
group57. In five
studies that compared arthroscopic and mini-open rotator cuff repairs, each
treatment group was extractable and therefore each was included in the
arthroscopic group or the mini-open group,
respectively34,58-61.
Among the seventeen studies, there was a total of twenty-two cohorts in the
final analysis: eleven in the arthroscopic group and eleven in the mini-open
group.
Demographic Data
The study design, level of evidence, total number of patients, number of
patients at the time of follow-up, and percentage of effective follow-up were
included in the analysis (see Appendix). Demographic data, including the
percentage of involvement of the dominant extremity, the mean age, the mean
duration of follow-up, the percentage of male patients, and the duration of
symptoms (in months) was recorded. According to the DeOrio and
Cofield24
classification system for rotator cuff tear size, the treatment groups were
defined in terms of the percentage of small tears (<1 cm), medium tears (1
to 3 cm), large tears (>3 to 5 cm), and massive tears (>5 cm). Studies
that provided information in terms of tendon involvement were separated into
the percentages of single-tendon and multiple-tendon tears.
There were no randomized controlled trials (Level I) or prospective cohort
studies (Level II) in either group. Five of the eleven reports in the
arthroscopic group were retrospective cohort studies (Level III), and five of
the eleven reports in the mini-open group were retrospective cohort studies
(Level III). The effective follow-up ranged from 64.9% to 100% in the
arthroscopic group and from 60% to 100% in the mini-open group.
The age, percentage of male patients, and percentage of dominant extremity
involvement were similar between the two groups. All studies had a mean
duration of follow-up of at least twenty-four months. In only one of the
eleven arthroscopic studies and five of the eleven mini-open studies was the
mean duration of follow-up greater than or equal to forty-eight months.
Although studies were excluded if the majority of tears were massive (>5
cm) or if multiple tendons were torn, the percentage of massive tears or
multiple-tendon tears differed only slightly between the two groups.
Surgical Technique
All patients in all studies in the arthroscopic group underwent an
all-arthroscopic rotator cuff repair with suture anchor fixation (see
Appendix). Four of the eleven studies in the mini-open group evaluated repair
involving the use of suture anchor fixation, and the others evaluated repair
involving the use of sutures placed in transosseous tunnels. Paulos and
Kody12 used suture
anchors to augment the transosseous tunnels or used transosseous tunnels
alone. In the retrospective cohort study by Kim et
al.58, the surgeon
attempted arthroscopic rotator cuff repair for all patients, but, when
arthroscopic repair could not be performed, the operation was converted to the
mini-open technique. The patients in each group had similar clinical
scores.
Subacromial decompression was performed in all patients in nine of the
eleven arthroscopic studies and in all eleven mini-open studies. In the
remaining two arthroscopic studies, subacromial decompression was performed in
79%7 and
94%62 of the cases.
Distal clavicle excision was performed in
81%63 and 89%
27 of the cases in
two of the arthroscopic studies, and biceps tenodesis was an adjunctive
procedure in 82% of the cases in one
study62.
Rehabilitation Protocol
For the five retrospective cohort studies in each group, the postoperative
rehabilitation was the same for the arthroscopic and mini-open groups (see
Appendix), and therefore, performance bias is limited for these studies. In
the remaining six arthroscopic studies, patients began active shoulder
range-of-motion exercises as early as three weeks and as late as nine weeks
after surgery, and strengthening was initiated by six weeks after surgery. In
the other six mini-open studies, active shoulder range-of-motion exercises
were begun at four to six weeks and strengthening was initiated between six
and eight weeks or after the sling was
removed12.
Range of Motion (Forward Elevation and External Rotation)
Only four of the eleven arthroscopic studies and five of eleven mini-open
studies recorded range of motion as a separate outcome (see Appendix). The
mean postoperative forward elevation ranged from 149.0° to 169.6° for
the arthroscopic group and from 155.0° to 173.0° for the mini-open
group. In the study by Warner et
al.60, the
arthroscopic and mini-open cohorts were compared retrospectively and there was
no difference between the two groups. The mean postoperative external rotation
ranged from 50.0° to 85.7° for the arthroscopic group and from
50.0° to 66.0° for the mini-open group.
Postoperative Shoulder Scores (UCLA, ASES, and Satisfaction)
The UCLA score was the one most commonly used for both the arthroscopic
group (eight studies) and the mini-open group (ten studies) (see Appendix).
The UCLA shoulder score was also expressed as excellent (34 or 35 points),
good (29 to 33 points), fair (25 to 28 points), and poor (=24 points) in
seven arthroscopic studies and eight mini-open studies. Only one of the seven
arthroscopic studies had <90% good or excellent results, compared with five
of the eight mini-open studies. All studies had a mean postoperative UCLA
score of >30.
Seven of the eleven arthroscopic studies and four of the eleven mini-open
studies included ASES scores; the mean postoperative scores ranged from 83.0
to 95.0 and 81.0 to 95.0, respectively. Four retrospective cohort
studies34,58,59,61
compared UCLA scores, two retrospective cohort
studies59,61
compared ASES scores, and none of them were able to demonstrate a significant
difference between the two groups. The percentage of patients who were either
satisfied or very satisfied after rotator cuff repair appeared to be similar,
with a range of 90% to 100% in the arthroscopic group and 86% to 100% in the
mini-open group.
Complications
There were fourteen complications after 473 procedures (prevalence, 3.0%)
in the arthroscopic group and twenty-seven complications after 411 procedures
(prevalence, 6.6%) in the mini-open group (see Appendix). Revision rotator
cuff repair was reported in three cases in three studies in the arthroscopic
group and in six cases in four studies in the mini-open group. Arthrofibrosis
was reported in five cases in the arthroscopic group, compared with nine cases
in the mini-open group. Postoperative symptoms consistent with impingement
occurred in one case in the arthroscopic group and in six cases in the
mini-open group. No medical complications were reported in any of the
studies.
The treatment of rotator cuff pathology has evolved with an improved
understanding of rotator cuff anatomy, more sophisticated instrumentation, and
advances in surgical technique. The most effective method of surgical repair
is controversial given that both arthroscopic and mini-open rotator cuff
repairs have been shown to produce satisfactory clinical results. There has
been growing interest in arthroscopic rotator cuff repair, and it is believed
to be at least as effective as mini-open rotator cuff repair with the added
advantages of reduced surgical morbidity, reduced postoperative stiffness,
and, potentially, a more rapid return to baseline shoulder function once
rotator cuff healing has
occurred29. The
present study is a qualitative description of the clinical results of
published articles on arthroscopic and mini-open rotator cuff repairs. On the
basis of the observations in the present study, there are no apparent
differences between arthroscopic repair and mini-open repair in terms of range
of motion or clinical scores after a mean of twenty-four months of follow-up,
but there may be a trend toward increased complications associated with
mini-open repair.
There appeared to be a higher percentage of complications in the mini-open
group, including revision, arthrofibrosis, and postoperative impingement;
however, the mini-open studies also tended to have longer follow-up, which
might allow for a greater number of complications. In the retrospective cohort
studies, there were approximately two times the number of revisions and cases
of arthrofibrosis in the mini-open group. Specifically, there were four
revisions and six cases of arthrofibrosis in the mini-open group, compared
with two revisions and three cases of arthrofibrosis in the arthroscopic
group.
Arthroscopic repairs are thought to be better able to reproduce rotator
cuff anatomy because the three-dimensional evaluation allows for the
recognition of tear configuration, thereby allowing the surgeon to formulate a
strategy that is most appropriate for that particular
pattern61,64,65.
In contrast, the visualization during a mini-open procedure is limited by the
size of the lateral split, which may not allow adequate access to the rotator
cuff and can compromise one's ability to perform necessary surgical releases,
perhaps resulting in less-optimal
repairs61,66.
Severud et al.59
described four patients who underwent a mini-open repair who had development
of fibrous ankylosis (defined as <120° of forward flexion by twelve
weeks postoperatively). Splitting of the deltoid and surgical retraction can
result in postoperative pain and may account for the increased prevalence of
postoperative
arthrofibrosis67.
There were six cases of postoperative impingement in the mini-open group,
compared with only one in the arthroscopic group. Four patients required
repeat subacromial decompression. The other two patients were found to have
acromioclavicular joint degeneration, one at seven to twelve
months68 and the
other at two to five
years68 after the
initial operation, requiring débridement and acromioclavicular joint
resection, respectively. It is impossible to determine if these patients had
unrecognized acromioclavicular joint pathology at the time of the rotator cuff
repair or if the degeneration developed after surgery.
One of the difficulties in comparing arthroscopic and mini-open repairs is
identifying which primary
outcome5 (clinical
score, range of motion and strength, pain, patient satisfaction, rate of
complications, or postoperative evidence of rotator cuff healing) defines
success. Eight of the eleven arthroscopic studies and ten of the eleven
mini-open studies involved the use of the UCLA shoulder score to assess
clinical outcome. While many studies included range of motion (assessed on a
5-point scale) as part of the UCLA score, only four arthroscopic studies and
five mini-open studies included separate range-of-motion data. Three studies
included visual analog scores outside of the UCLA score. All but one study
from each group assessed complications. Five arthroscopic and six mini-open
studies assessed patient satisfaction. Only one study from each group included
a postoperative imaging study to evaluate the healing of repaired rotator cuff
tendons as an outcome. Aside from the UCLA score, the outcome measures varied
considerably, making comparisons difficult within and between groups. At the
time of the literature search, radiographic analysis after rotator cuff repair
was not routinely performed, but more recently published studies have
incorporated postoperative imaging as an objective outcome
measurement19,62,69,70.
Selection Bias
Because the majority of the studies were case series of either arthroscopic
or mini-open repairs, we set strict inclusion and exclusion criteria to
provide homogeneity between the two groups to limit the potential for
selection bias. As previously mentioned, no randomization was performed in any
of the studies, but the studies that were included in the final analysis had
similar patient ages, percentages of male patients, and percentages of
involvement of the dominant extremity. Numerous studies in the literature on
open, mini-open, and arthroscopic procedures have shown that tear size is an
important determinant of outcome and
healing3,58,62,70-73.
We excluded any clinical study on massive rotator cuff tears and any study in
which >50% of the patients had large and massive rotator cuff tears.
Studies that did not provide information on tear size characteristics were
also excluded. Studies that provided information on the number of tendons torn
were included if >50% of the patients had an isolated supraspinatus tendon
tear6,60,63,71.
Although the clinical results of the repair of massive rotator cuff tears
may be satisfactory, postoperative imaging studies have demonstrated rates of
recurrent defects to be as high as 68% in mini-open
studies72 and as
high as 94% in arthroscopic
studies19. There is
a growing body of evidence suggesting that although patients with failed
repairs demonstrate good pain relief and the ability to perform activities of
daily living in the short term, their outcomes may deteriorate over
time19. In
comparison with patients with healed tendons, patients with failed rotator
cuff repairs have decreased range of motion and strength, which has been a
consistent finding following both open and arthroscopic
procedures19,62,69,70,72.
Performance Bias
Performance bias may occur in studies in which a disproportionate number of
concomitant procedures are performed. Subacromial decompression was performed
in all patients in eight of the eleven arthroscopic studies and in all eleven
mini-open studies. We do not believe that this represents a substantive
difference leading to performance bias. In the arthroscopic group, there were
two studies with an unusually high percentage of patients who had concomitant
procedures for the treatment of acromioclavicular joint
pathology27,63.
One other study had a large proportion of combined arthroscopic rotator cuff
tears and treatment of biceps tendon
pathology62. On the
basis of the clinical outcomes, these three studies performed similarly to the
rest of the arthroscopic studies and therefore remained in the final analysis.
Variation in the rehabilitation protocol is another potential variable that
may influence performance bias. The retrospective cohort studies eliminated
performance bias by implementing the same rehabilitation for each group. There
were only minimal differences in the rehabilitation for the case series. Thus,
performance bias was minimized.
Exclusion Bias
Of the studies in the final analysis, seven of the eleven arthroscopic
studies and seven of the eleven mini-open studies had >80% follow-up and
all studies had a mean duration of follow-up of twenty-four months. There is a
potential for exclusion bias for any study in which patients were lost to
follow-up, but especially for those four studies in each group that had
<80% follow-up.
Detection Bias
Eight of the eleven arthroscopic studies and ten of the eleven mini-open
studies involved the use of the UCLA Shoulder Score as the primary outcome
measure. Among the retrospective cohort studies, no significant differences
were noted between the arthroscopic and mini-open groups in terms of the UCLA
score34,58,59,61
and the ASES
score58,59,61.
Among the case series, there was no appreciable difference between
arthroscopic and mini-open studies in terms of range of motion, the UCLA
score, the ASES score, and satisfaction.
The present study had many strengths related to a design that resulted in
homogeneity between the two study groups. With use of strict inclusion and
exclusion criteria, there were eleven arthroscopic and eleven mini-open groups
that had similar patient ages, percentages of male patients, percentages of
involvement of the dominant extremity, percentages of effective follow-up,
durations of follow-up (mean, twenty-four months), and distributions of
rotator cuff tears. We attempted to exclude any study with a potential
confounding factor such as less than twenty-four months of follow-up, a
majority of massive tears, a failure to define tear sizes, partial tears,
revision cases, mixed cohorts, and fixation other than tendon-to-bone. The
final analysis included 473 patients in the arthroscopic group and 411
patients in the mini-open group, and most studies had the same primary
outcome.
Whether qualitative or quantitative, systematic reviews are limited by the
quality of the published studies. After reviewing the literature, there were
no published randomized controlled trials (Level I) or prospective cohort
studies (Level II) that met the study criteria at the time of the literature
search. Because of a lack of randomized clinical trials, a quantitative
systematic review, or meta-analysis, could not be performed, indicating the
need for an improvement in the quality of published studies on the treatment
of rotator cuff repairs, with a focus on prospective, randomized clinical
trials with validated outcome scores and postoperative imaging studies. There
were studies in the final analysis with <80% effective follow-up, which may
subject the present report to exclusion bias.
In terms of the surgical technique, all of the arthroscopic studies
involved the use of suture anchor fixation whereas the majority of the
mini-open studies involved the use of transosseous tunnels for tendon-to-bone
fixation. The majority of studies evaluating rotator cuff repair investigated
either tendon-to-bone healing with use of transosseous
tunnels18,74-78
or ex vivo biomechanical analyses of suture
anchors17,72,79-81.
To date, we are aware of no studies that have compared the effect of these two
techniques on tendon-to-bone healing.
Finally, there was also heterogeneity in the proportion of concomitant
procedures performed, but with no apparent effect on clinical outcomes. By
including these studies, we were able to maximize the overall number of
patients.
The rehabilitation protocol was the same for the retrospective cohort
studies, but although there were no clinical differences, there was a slight
difference in the rate of complications. We believe that the difference in
arthrofibrosis rates, for instance, is likely related to the technical aspects
of surgery. For the case series, there was minor variation in the
rehabilitation, which potentially could have affected the outcomes, but it did
not seem to affect the analysis.
Aside from the UCLA Shoulder Score, the outcomes varied greatly from study
to study, thus limiting the number of variables in the analysis.
Range-of-motion data were incomplete and therefore could not be used reliably
to compare the two groups. Many studies involved the use of other
shoulder-scoring systems that could not be readily compared. Overall, there
was a lack of objective outcome measures at the time of the analysis, but the
rate of healing may be an interesting outcome to compare between the two
techniques.
In conclusion, this systematic review demonstrates that both arthroscopic
and mini-open rotator cuff repair can result in significant improvement from
baseline in terms of shoulder function and clinical outcome, with relatively
low complication rates. Although we could not identify a difference between
the two techniques in terms of range of motion or function, there may be a
slightly increased rate of complications associated with the mini-open repair.
We do not recommend one technique over the other; instead, we believe that
both techniques are effective and that the surgeon should use the technique
that produces the most reliable result with the least complications in his or
her hands.
Extensive tables presenting data from all of the included studies are
available with the electronic versions of this article, on our web site at
(go to
the article citation and click on "Supplementary Material") and on
our quarterly CD-ROM (call our subscription department, at 781-449-9780, to
order the CD-ROM). ?