Study Design
We performed a prospective, randomized, double-blind clinical trial
at a single university center with three orthopaedic surgeons. Each patient
was randomly assigned to be treated with either a total shoulder arthroplasty
or a hemiarthroplasty. Both the patient and an independent evaluator remained
blinded to the group assignment for the duration of the study.
Consent
The study protocol was reviewed and approved by the institutional research
ethics review board. All patients provided written informed consent.
Justification of Sample Size
The sample size was estimated with use of an equation appropriate for
comparing two independent group
means46 (see
Appendix) and was based on a two-sided alpha level of 0.05 having 80% power to
detect a 30% change in the McGill pain score. The McGill pain score, a
validated pain-rating scale, was used to estimate sample size because, when
this study was initiated, the WOOS index, although available for use, had not
been fully validated. It was validated and published in
200145 (see
Appendix).
Patient Population
The patients included in the study had a diagnosis of primary
osteoarthritis of the shoulder, had a failure of a minimum of six months of
nonoperative treatment (including analgesics, anti-inflammatory medication,
and physiotherapy), and wished to have surgical intervention. Primary
osteoarthritis of the shoulder was defined as shoulder pain; no history of
major trauma, infection, osteonecrosis, cuff tear arthropathy, chronic
dislocation, or a secondary cause of osteoarthritis; and radiographic evidence
of joint space narrowing, osteophyte formation, and/or subchondral
sclerosis.
Exclusion criteria included a condition other than shoulder osteoarthritis
that would substantially contribute to shoulder dysfunction (e.g., cervical
spine disease), a rotator cuff tear (>1 cm), inflammatory arthritis,
post-capsulorrhaphy osteoarthritis, a major medical illness that would
substantially influence quality of life (e.g., unstable angina), an active
infection, substantial muscle paralysis, and a lack of fitness for surgery or
an unwillingness to be followed for two years. All patients underwent a
standardized preoperative assessment by a physician and a research assistant.
Preoperative radiographs (anterior and posterior glenohumeral joint,
transscapular lateral, and axillary views) and a computed tomography scan of
the affected shoulder were made prior to the surgery.
Randomization
Patients were stratified according to surgeon. Randomization was performed
intraoperatively. After it was confirmed that the patient had primary
osteoarthritis as well as good-quality glenoid bone stock that was adequate
for the performance of either a hemiarthroplasty or a total shoulder
arthroplasty, a sealed envelope containing the randomly assigned
treatment-group allocation was opened by the circulating nurse.
Surgical Intervention
A Neer Series-II modular total shoulder prosthesis or hemi-prosthesis (3M
Modular Shoulder System; 3M Canada, London, Ontario, Canada) was implanted in
all patients. Radiographic templating was performed preoperatively for all
patients to estimate the size of the prosthesis, but the final implant was
determined intraoperatively with the use of trial implants. Preoperative
prophylactic antibiotics were given to all patients, and all procedures were
performed with the patient under general anesthesia. A deltopectoral approach
was used to expose the glenohumeral joint, and the integrity of the rotator
cuff, as assessed by direct visualization and palpation, was documented. The
subscapularis was completely released to achieve maximum external rotation of
the shoulder. The release was achieved through resection of the anterior and
inferior aspects of the capsule and release of the coracohumeral ligaments so
that excursion of the subscapularis tendon was limited only by the muscle
tension.
The humerus was prepared, according to the manufacturer's instructions,
with progressive intramedullary reaming. The humeral osteotomy was performed
with the use of an intramedullary guide at approximately 30° of
retroversion. The proximal part of the humerus was broached and reamed, and
the humeral prosthesis was sized appropriately.
In each patient, the glenoid was exposed and the severity of glenoid
arthritis was documented. The version of the glenoid and the amount of
posterior erosion were confirmed intraoperatively and were compared with their
appearance on preoperative radiographs and computed tomography scans. No
attempt was made to alter the glenoid anatomy in the patients being treated
with a hemiarthroplasty. If a glenoid prosthesis was to be implanted, the
glenoid was formally exposed. We did not exclude any patient because of
glenoid morphology or posterior glenoid erosion, and eccentric glenoids were
included in both groups.
For the total shoulder arthroplasty, eccentric reaming was performed as
necessary to allow the implantation of the glenoid component in the center of
the glenoid, parallel to the neck and in the appropriate version. A slot was
made in the glenoid to accommodate the glenoid component and to allow for
cement interlock. Next, the glenoid was irrigated and was dried with
thrombin-soaked gauze (Thrombostat [100 units/mL]; Pfizer, Arnprior, Ontario,
Canada), and the component was cemented into place with thumb
pressurization.
The size of the humeral head was chosen to adequately cover the humeral
osteotomy site and to restore the height of the humeral head, soft-tissue
tension, and approximately 50% translation anteriorly and posteriorly. The
humeral component was press-fit when possible. When adequate press-fit
fixation could not be obtained, the humeral component was cemented with use of
vacuum cement mixing, a distal cement restrictor, retrograde filling of the
shaft, and cement pressurization. The subscapularis was closed with number-2
Ethibond suture (Ethicon, Somerville, New Jersey) to allow a minimum of
30° of external rotation. Patients with <30° of external rotation
had the subscapularis transferred medially to the osteotomized humeral
neck.
Postoperative Care
A sling was applied with the arm at the side. Active-assisted
range-of-motion exercises were begun on the first postoperative day in the
hospital, with emphasis on forward elevation and external rotation. External
rotation was limited according to the intraoperative assessment of the tension
of the subscapularis repair and was usually 30°. An active range of motion
was allowed at four weeks postoperatively, and strengthening exercises were
begun at eight weeks postoperatively. The patient's return to normal
activities progressed as tolerated over three to six months.
Outcomes
A research assistant who was blinded to the treatment group performed a
standardized assessment of all patients preoperatively; at six weeks
postoperatively; and at three, six, twelve, eighteen, and twenty-four months
postoperatively. At each evaluation, the patients completed a disease-specific
quality-of-life instrument (Western Ontario Osteoarthritis of the Shoulder
[WOOS] index)45,
general shoulder assessment instruments (University of California at Los
Angeles [UCLA] shoulder
scale14, Constant
score47, and
American Shoulder and Elbow Surgeons [ASES] evaluation
form48), general
pain scales (McGill pain score and visual analogue
scale)49, and a
global health measure (Short Form-36
[SF-36])50. The
range of motion was extracted from the range-of-motion domain of the Constant
shoulder scale.
The primary outcome measure was the WOOS (see Appendix), which is a
patient-generated, self-evaluated, disease-specific tool for measurement of
the quality-of-life of patients with osteoarthritis of the shoulder. The WOOS
has proven to be valid, reliable, and highly responsive in that patient
population45. Each
question is answered on a visual analogue scale, and the possible scores range
from 0 to 1900. For simplicity of presentation, the raw scores from each
domain and the total WOOS score can be converted to a percentage of a normal
score, out of 100. A score of 0 therefore signifies that the patient had an
extreme decrease in the shoulder-related quality of life, whereas a score of
100 signifies that the patient had no decrease in the shoulder-related quality
of life.
Data Analysis
Paired two-tailed t tests were used to compare differences between
preoperative and two-year postoperative scores for both groups. Unpaired
two-tailed t tests were used to assess differences between groups, and
analysis of covariance was used as well with adjustment for the baseline value
of the outcome measure. Three analyses were conducted on the two-year
post-operative data. The first was an intention-to-treat analysis in which all
patients, regardless of further treatment (i.e., cross-over), remained in
their original group. The second analysis was an efficacy analysis in which
the subjects who eventually crossed over into the other group were dropped
from the analysis. The third was a conservative efficacy analysis in which the
patient's last score before he or she crossed over to the other treatment
group was identified and carried forward through the remaining evaluations. A
p value of <0.05 was considered significant.
Demographic Data
Forty-two patients were enrolled in the study. One patient died from
a thromboembolism two days postoperatively and was excluded from the analysis
of the data. Data were therefore collected on forty-one patients; twenty-one
in the hemiarthroplasty group and twenty in the total shoulder arthroplasty
group.
Subjects
Baseline variables are summarized in
Table I. There were no
significant differences between the groups in terms of age, gender, or
preoperative range of motion or functional status.
Intraoperative Findings
All patients had osteoarthritis of the glenohumeral joint that was
confirmed intraoperatively. No patient was excluded because of an inability to
insert a glenoid component. One patient in the hemiarthroplasty group had a
full-thickness 1-cm tear of the supraspinatus tendon, which was repaired. The
rotator cuffs of the remaining patients were intact. The humeral stem was
press-fit in thirty patients (fourteen in the hemiarthroplasty group and
sixteen in the total shoulder arthroplasty group) and cemented in eleven
(seven in the hemiarthroplasty group and four in the total shoulder
arthroplasty group). The mean operative time (and standard deviation) was
118.4 ± 38.6 minutes for the hemiarthroplasty group and 157.3 ±
33.0 minutes for the total shoulder arthroplasty group (p < 0.002).
Outcomes
Two patients from the hemiarthroplasty group crossed over to the total
shoulder arthroplasty group; one did so at twelve months and the other, at
eighteen months postoperatively.
All results were obtained at the two-year follow-up evaluation.
The results presented in this paper were derived from the conservative
efficacy analysis, in which the patient's last score before he or she crossed
over was identified and carried forward through the remaining evaluations. The
overall results of the intention-to-treat and efficacy analyses were not
significantly different from those of the conservative efficacy analysis (see
Appendix).
Disease-Specific Quality of Life
At the two-year follow-up evaluation, both groups demonstrated
significant improvement in the disease-specific quality of life (p = 0.000) as
measured by all four domains of the WOOS
(Table II). However, with the
numbers available for the study, there was no significant difference between
the hemiarthroplasty and total shoulder arthroplasty groups in terms of the
overall WOOS score (p = 0.18) or the scores for any of the domains of the WOOS
(Table III).
Secondary Outcomes
Similar to the WOOS, the UCLA shoulder rating scale, ASES shoulder
evaluation form, and Constant score showed significant improvements in global
shoulder function following both types of shoulder arthroplasty
(Table IV). Again, with the
numbers available, there were no significant differences between the two
treatment groups (Table V). The
McGill pain scale, the McGill visual analogue scales (for specific domains),
the physical component of the SF-36, and the range of motion demonstrated
similar trends in improvement (Tables
IV and
V). There was, however, no
significant difference between the preoperative and postoperative scores for
the mental component of the SF-36.
Complications
Intraoperative Complications
Two patients in the hemiarthroplasty group had an intraoperative fracture,
which involved the greater tuberosity in one patient and the humeral shaft in
another. The fractured greater tuberosity was secured back to the shaft with
number-5 Mersilene tape, and the patient recovered excellent external rotation
strength. The humeral shaft fracture was treated with a long-stem prosthesis,
bone graft from the humeral head, and cerclage wiring. The fracture healed
uneventfully.
One patient in the total shoulder arthroplasty group had a nondisplaced
fracture of the greater tuberosity, for which the treatment was similar to
that for the fracture described above. The patient recovered strength without
incident. A second patient in that group had a fracture of the
anterior-inferior corner of the glenoid during reaming. This fracture was
secured with a 3.0-mm cannulated AO screw, and bone graft from the humeral
head was applied. It healed without complication.
Late Complications
Evidence of superior migration of the humeral component with rotator cuff
deficiency developed in three patients in the hemiarthroplasty group. Two
patients were doing well and continued to demonstrate improvements in quality
of life. However, one patient had increasing pain and decreased motion and
strength. Despite the disability, the patient did not wish to have more
surgery.
An infection developed in one patient two weeks after a total shoulder
arthroplasty. The patient was treated with two operative débridements
and intravenous antibiotics for six weeks, and the infection resolved. In
another patient, antero-superior instability of the prosthesis developed six
months after the total shoulder arthroplasty. Although passive motion was
excellent, the patient had poor active forward elevation and external
rotation. Ultrasonography demonstrated massive tears of the supraspinatus and
subscapularis tendons, for which the patient declined further treatment.
Analysis of Failures
Four hemiarthroplasties were considered to be failures. One patient, who is
described above, had superior humeral head migration and decreased motion. Two
years following the surgery, the patient demonstrated only a shoulder shrug of
30° of active forward elevation. This patient's quality of life was not
substantially improved postoperatively. Progressive glenoid arthrosis
developed in three other patients. They had pain that was severe enough to
require a revision to a total shoulder arthroplasty, which was performed in
two of them and was being considered by the third. One of the revisions was
performed nineteen months after the hemiarthroplasty, and the patient had a
decrease in the pain and improvements in motion and the quality of life at
twelve months following this revision. The second patient had the revision
surgery at approximately sixteen months after the hemiarthroplasty. This
patient initially had decreased pain and an improvement in the quality of
life, but the pain had increased and the quality of life had decreased by two
years following the revision.
While it was not surprising that our study confirmed that many
patients with osteoarthritis of the shoulder have pain, loss of motion, and
loss of
function4-6,
it also demonstrated that osteoarthritis of the shoulder leads to more global
dysfunction. Substantial impairments were noted in all four domains of the
WOOS—i.e., physical symptoms, sports/recreation/work, lifestyle, and
emotions. Thus, shoulder osteoarthritis may substantially affect all aspects
of health51, and
these impairments are
quantifiable45.
Importantly, these results suggest that the validity of previous assessments
in which such evaluations were not performed may be limited, and they support
the continued use of comprehensive disease-specific quality-of-life
measurement tools in the evaluation of orthopaedic
disorders45.
Both total shoulder arthroplasty and hemiarthroplasty significantly
decreased shoulder pain and improved function and quality of life. However, we
found no significant difference between hemiarthroplasty and total shoulder
arthroplasty with regard to any of the outcomes evaluated at two years,
although there was a trend for superior results in the total shoulder
arthroplasty group.
Our study differs from previous
studies29,31,34-36,43
in several important ways. Our purpose was to measure disease-specific quality
of life as it pertains to patients with osteoarthritis of the shoulder. In
many of the previous studies, physician-generated questionnaires were utilized
as the primary outcome measures; therefore, the items in the measurement tools
are those that physicians deem to be important and not necessarily those that
are important to patients. Our study was also prospective, randomized, and
double-blinded (patient and evaluator), thus decreasing the bias in both the
clinical assessments and the patient self-evaluations.
Whenever a study demonstrates an insignificant difference, it is important
to establish the power of the investigation in order to determine if the final
conclusion is valid (i.e., that there is truly no significant difference
between treatment
groups)46. Our
preliminary sample-size estimates were based on a group of ten patients who
were assessed with the McGill pain score, a validated outcome measure,
following shoulder arthroplasty. However, our post-investigation sample-size
estimates, based on the same 30% difference in the McGill pain score with
a = 0.05 and ß = 0.20, suggest that a sample size of 855
hemiarthroplasties and 382 total shoulder arthroplasties would be necessary to
determine this difference. (It is necessary for the numbers in the two groups
to be unequal because the variances of the groups are unequal; i.e.,
hemiarthroplasties have a larger variance than total shoulder arthroplasties.)
Obviously, these numbers are much different from our pre-investigation
estimates, and they were due to the wide variance in the McGill pain scores.
However, the use of a highly responsive outcome measurement tool such as the
WOOS index can decrease this number. For example, when the WOOS outcome data
were used, the power of the study was 0.29. Although this power to detect a
30% difference in the WOOS scores is low (ideally, the power of a clinical
trial should be =0.80), only eighty-six hemiarthroplasties and forty-seven
total shoulder arthroplasties were expected to be needed.
A recent
meta-analysis52 of
randomized clinical trials comparing total shoulder arthroplasty with
hemiarthroplasty confirmed the trends demonstrated in the current study. After
two years of follow-up of 117 patients (fifty-three treated with
hemiarthroplasty and sixty-four treated with total shoulder arthroplasty), the
mean UCLA score following total shoulder arthroplasty (26.5 ± 4.1
points) was found to be superior to that following hemiarthroplasty (22.4
± 5.9 points; p < 0.0001). In addition, the outcomes in the domains
of pain and function were superior after the total shoulder
arthroplasties.
Three hemiarthroplasties were considered to be failures secondary to the
development of glenoid arthrosis, and two were eventually converted to a total
shoulder arthroplasty. Despite these two cross-overs, there was no significant
difference between the results of the intention-to-treat analysis and those of
the efficacy analysis. This is because one of these patients had substantial
improvement in the WOOS score (13.4 points before the revision and 67.3 points
after it), whereas the other patient had a less successful revision (a WOOS
score of 59.5 points before the revision and 52.4 points after it). Mixed
results following revision total shoulder arthroplasty in patients with
glenoid arthrosis were previously reported by Sperling and
Cofield38.
In conclusion, patients with osteoarthritis of the shoulder demonstrated
significant impairment in the quality of life preoperatively. Total shoulder
arthroplasty and hemiarthroplasty resulted in significant improvement in the
quality of life in all domains of health. Although there was a trend toward
superior outcomes following total shoulder arthroplasty, there was no
significant difference between the two groups within the limits of the study.
Longer-term follow-up is necessary to continue to evaluate disease-specific
quality of life following these procedures.
The details of the sample-size calculation, the WOOS shoulder index, and
the intention-to-treat analysis and the efficacy analysis 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). ?