Patients and Data Collection
Between 1985 and 2000, a consecutive series of forty shoulders in thirty-seven patients underwent hemiarthroplasty and rotator cuff repair for cuff tear arthropathy or osteoarthritis of the shoulder with a massive rotator cuff tear. Patients with a diagnosis of rheumatoid or inflammatory arthritis, osteonecrosis, fracture, fracture malunion, or fracture nonunion were excluded. All of the operations were performed at our institution by the senior one of us (L.U.B.). At the time of this study, five patients had died with less than two years of clinical follow-up. One patient was lost to follow-up. These patients were excluded, leaving thirty-four shoulders in thirty-one patients with at least two years of clinical follow-up.
To be included in the study, patients had to have a massive rotator cuff tear with severe glenohumeral arthropathy and normal deltoid function. All findings were confirmed intraoperatively. Thirty of the thirty-four shoulders had three-tendon involvement (supraspinatus, infraspinatus, and subscapularis) and four had four-tendon involvement.
There were twenty women (twenty-two shoulders) and eleven men (twelve shoulders) with an average age at the time of surgery of seventy-two years (range, forty-eight to ninety years). All patients had failed nonoperative treatment including the use of nonsteroidal anti-inflammatory agents, physical therapy, and activity modification. All patients underwent a preoperative physical examination in which forward elevation and external rotation were recorded and pain was assessed as none, mild, moderate, or severe. No functional scores were recorded at that time.
Radiographs demonstrated superior migration of the humerus in all shoulders and actual contact of the humeral head with the undersurface of the acromion in thirty-one shoulders. Of the thirty-four shoulders in this study, seven had had a previous operation. Three of them had had an acromioplasty (one open and two arthroscopic) as an isolated procedure, two had had a single open rotator cuff repair with acromioplasty, and two had had multiple open rotator cuff repairs.
All thirty-four shoulders had at least two years of clinical follow-up after surgery. Range-of-motion data from these patients were collected in chart review manner, from the surgeon's clinical charts, and the average follow-up with regard to range of motion was 3.7 years (range, two to twelve years).
Of the thirty-one patients (thirty-four shoulders) in the study, twenty-two patients (twenty-five shoulders) were assessed, either in person during a clinic visit or by telephone, for long-term follow-up, which was an average length of ten years (range, four to sixteen years). The assessment included completion of the American Shoulder and Elbow Surgeons (ASES) scoring system16; answering questions about activities of daily living, satisfaction with the operation, revision surgery, or complications; and rating the pain as none, mild, moderate, or severe. The ASES scoring system is a 100-point scale composed of a visual analog pain score (50 points) and a functional score (50 points). The remaining nine patients without ASES scores and long-term data died with the prosthesis in place between the time of their last clinic visit and our data collection.
On the basis of the data collected, each patient was assessed with use of the limited goals criteria of Neer et al.4, which meant that the result was satisfactory when the patient had no or mild pain, was pleased with the outcome of the procedure, and was capable of independent self-care. Preoperative factors, including an age of more than seventy years at the time of surgery, gender, a history of surgery, and the amount of preoperative active forward elevation, were examined to determine whether they had an effect on eventual outcome. Intraoperative factors, including the ability to obtain complete coverage of the humeral head during rotator cuff repair and superior transfer of the subscapularis, were also evaluated for their effect on outcome.
Surgical Technique
The operation was performed through a deltopectoral approach. The superior 1 cm of pectoralis major tendon was incised to improve exposure. The coracoacromial ligament, when present, was preserved to prevent anterosuperior migration of the humerus. Seven patients who had had a prior acromioplasty and/or rotator cuff repair did not have an intact coracoacromial ligament to preserve. The subscapularis tendon was incised laterally with the capsule, tagged with sutures, and mobilized to increase excursion for superior translation at the end of the procedure. The humeral head was dislocated by a combination of external rotation and extension of the arm. In most shoulders, because of the massive rotator cuff tear and loss of the superior rotator cuff tissue, the humeral head was readily accessible. An oscillating saw was used to make the humeral head cut. After hand reaming of the humeral canal, a trial prosthesis was placed in approximately 30° of humeral retroversion. The final humeral prosthesis was cemented in all shoulders. An attempt was made in all shoulders to mobilize the residual rotator cuff and at least partially repair the posterior aspect of the rotator cuff with heavy nonabsorbable sutures, as we have described previously9. The undersurface of the acromion was inspected because the leading edge of the posterior aspect of the rotator cuff was often adherent to it. Once the leading edge of the posterior aspect of the rotator cuff was found, it was tagged with nonabsorbable sutures and was freed superomedially and posteroinferiorly, but the insertion onto the humerus was never detached. The tendons were repaired to bone with number-2 Tevdek suture (Teleflex Medical, Research Triangle Park, North Carolina) through bone tunnels. Anteriorly, the inferior part of the subscapularis was always present, with the superior aspect torn and retracted. The subscapularis was completely mobilized and reattached through bone tunnels in all shoulders. In thirteen of the thirty-four shoulders, there was enough tissue to transpose the subscapularis tendon to the superior aspect of the greater tuberosity. If superior coverage was unobtainable, focus was placed on achieving stable anterior and posterior buttresses. The deltopectoral interval was closed over a suction drain, and the skin was closed with a subcuticular suture.
Postoperatively, the patients wore a sling for three weeks, and it was taken off only for bathing and exercise. On the first postoperative day, the patients were started on passive motion exercises, consisting of pendulum exercises and external rotation exercises with a stick. Outpatient rehabilitation consisted of passive motion to 30° of external rotation and supine forward elevation to 130° for eight to twelve weeks. Active-assisted exercise and isometrics were started below the horizontal at eight to twelve weeks, depending on the patient's progress. When the patient demonstrated good control, active exercise to tolerance was initiated. All patients participated in the supervised postoperative physical therapy program for at least three months after surgery.
Statistical Methods
The paired t test was used to compare paired data from the preoperative to postoperative evaluations for all patients. A two-sample unpaired t test was used to evaluate the differences in outcomes among subsets of our larger group of patients (i.e., men compared with women and those who were less than seventy years old compared with those who were seventy years or older). A one-way analysis of variance was used to determine whether there was a significant difference in the mean postoperative ASES score among the preoperative and intraoperative factors studied. Multiple regression analysis was used to assess the association between postoperative ASES scores and the preoperative and intraoperative factors studied. A p value of <0.05 was considered significant.
Informed Consent and Institutional Review Board
Informed consent to participate in this study was obtained from all patients. Our institutional review board approved this study.
Direct clinical follow-up was obtained for all thirty-four shoulders at an average of 3.7 years (range, two to twelve years) postoperatively. The mean active forward elevation (and standard deviation) improved from 78° ± 44° (range, 20° to 165°) preoperatively to 111° ± 39° (range, 40° to 180°) postoperatively (p < 0.001). The mean active external rotation improved from 15° ± 24° (range, -30° to 75°) preoperatively to 38° ± 16° (range, 0° to 60°) postoperatively (p < 0.0001). Twenty-six of thirty-four shoulders were rated as satisfactory according to the limited goals criteria of Neer et al.4, which meant the patient had no or mild pain, was pleased with the outcome of the procedure, and was capable of independent self-care.
For twenty-five of the thirty-four shoulders, long-term follow-up data were obtained at an average follow-up of ten years (range, four to sixteen years). The mean ASES score was 67 ± 21 points (range, 35 to 100 points). The patients reported no pain in eighteen shoulders, mild pain in twelve shoulders, and moderate pain in four. No patient reported severe pain.
Complications and Reoperations
One patient with a history of four failed rotator cuff repairs had a symptomatic anterosuperior escape of the humerus after the operation. She continued to have moderate pain requiring narcotics and had limited use of the shoulder. Only one patient required a reoperation to remove an osteophyte that developed over the greater tuberosity and to débride the subacromial space. No shoulder was revised for problems related to implant failure, loosening, infection, or fracture.
Influence of Preoperative Factors
Age
For the sixteen shoulders in patients who were younger than seventy years at the time of surgery, the mean active forward elevation improved from 79° preoperatively to 106° postoperatively, and the mean active external rotation improved from 17° preoperatively to 34° postoperatively. The mean ASES score was 65 points (range, 25 to 100 points), and eleven of the sixteen patients had satisfactory outcomes according to the limited goals criteria of Neer et al.
For the eighteen shoulders in patients who were seventy years old or more at the time of surgery, the mean active forward elevation improved from 77° preoperatively to 116° postoperatively, and the mean active external rotation improved from 14° preoperatively to 42° postoperatively. The mean ASES score was 70 points (range, 35 to 97 points), and fifteen of the eighteen shoulders had satisfactory outcomes according to the limited goals criteria of Neer et al.
No significant difference between the shoulders in patients who were less than seventy years old and those in patients who were seventy years or more at the time of surgery was found with respect to active forward elevation (106° and 116°, respectively; p = 0.45), external rotation (34° and 42°; p = 0.13), or ASES score (65 and 70 points; p = 0.61). However, because of our sample size, power for these analyses was insufficient (power = 0.11, 0.33, and 0.08, respectively).
Previous Surgery
For the twenty-seven shoulders that had had no prior surgery, the mean active forward elevation improved from 77° preoperatively to 110° after surgery and the mean active external rotation improved from 13° preoperatively to 39° postoperatively. The mean ASES score was 69 points (range, 35 to 100 points), and twenty-one of the twenty-seven shoulders had satisfactory outcomes according to the limited goals criteria of Neer.
For the seven shoulders that had had prior surgery, the mean active forward elevation improved from 82° preoperatively to 114° after surgery and the mean active external rotation improved from 23° preoperatively to 34° postoperatively. The mean ASES score was 62 points (range, 25 to 88 points), and five of the seven shoulders had satisfactory outcomes according to the limited goals criteria of Neer et al.
Comparing shoulders in patients with no prior shoulder surgery and those with prior surgery, there was no significant difference in active forward elevation (110° and 114°, respectively; p = 0.78), external rotation (39° and 34°; p = 0.53), or ASES score (69 and 61 points; p = 0.53); however, because of our sample size, power for these analyses was insufficient (power = 0.06, 0.13, and 0.12, respectively).
Gender
For the twelve shoulders in men, the mean active forward elevation improved from 99° preoperatively to 113° after surgery, and the mean active external rotation improved from 15° preoperatively to 38° postoperatively. The mean ASES score was 79 points (range, 60 to 100 points), and ten of the twelve shoulders had satisfactory outcomes according to the limited goals criteria of Neer et al.
For the twenty-two shoulders in women, the mean active forward elevation improved from 67° preoperatively to 110° after surgery and the mean active external rotation improved from 15° preoperatively to 38° postoperatively. The mean ASES score was 60 points (range, 25 to 97 points), and sixteen of the twenty-two shoulders had satisfactory outcomes according to the limited goals criteria of Neer et al.
In a comparison of men and women, no significant difference was detected with regard to active forward elevation (113° and 110°, respectively; p = 0.80) or external rotation (38° and 38°; p = 0.99); however, because of our sample size, power for these analyses was insufficient (power = 0.06 and 0.05, respectively).
Men had significantly higher postoperative ASES scores than women (79 and 60 points, respectively; p = 0.01). Men also had significantly higher preoperative active forward elevation than women (99° and 67°, respectively; p = 0.04).
Preoperative Forward Elevation
For the eighteen shoulders with preoperative active forward elevation of <90°, the mean active forward elevation improved to 100° postoperatively and the mean active external rotation improved from 9° preoperatively to 36° postoperatively. The mean ASES score was 54 points (range, 25 to 88 points), and twelve shoulders had satisfactory outcomes according to the limited goals criteria of Neer et al.
For the sixteen shoulders with preoperative active forward elevation of =90°, the mean active forward elevation improved to 121° postoperatively and the mean active external rotation improved from 22° preoperatively to 38° postoperatively. The mean ASES score was 80 points (range, 60 to 100 points), and fourteen shoulders satisfied the limited goals criteria of Neer et al.
The mean ASES score for shoulders with active preoperative forward elevation of =90° was significantly higher than that for shoulders in patients who were unable to actively elevate the arm to 90° (80 and 54 points, respectively; p < 0.001). When the ASES score was separated into its pain and function components, the patients who could elevate the arm to =90° preoperatively had both significantly better pain relief (mean, 48 points compared with 30 points; p = 0.002) and significantly better function (mean, 31 compared with 23 points; p = 0.05) than those who were unable to elevate the arm to 90°. (An ASES pain score of 50 points is equivalent to a visual analog pain score of 0 points, or no pain16.) Intraoperatively, eight of the sixteen shoulders with preoperative forward elevation of =90° achieved complete coverage of the humeral head with rotator cuff tissue during repair. Only four of the eighteen shoulders with preoperative active forward elevation of <90° achieved complete coverage of the humeral head with rotator cuff tissue during repair.
Intraoperative Factors
Rotator cuff repair was attempted in all thirty-four shoulders. Complete coverage of the humeral head was achieved in twelve shoulders, and partial coverage was achieved in twenty-two. In the twelve shoulders in which complete coverage of the humeral head was achieved, the mean active forward elevation improved from 106° preoperatively to 125° postoperatively and the mean active external rotation improved from 29° preoperatively to 41° postoperatively. The mean ASES score was 73 points (range, 35 to 97 points), and ten of the twelve shoulders had a satisfactory outcome according to the limited goals criteria of Neer et al.
In the twenty-two shoulders in which only partial coverage of the humeral head was achieved, the mean active forward elevation improved from 63° preoperatively to 103° postoperatively and the mean active external rotation improved from 8° preoperatively to 36° postoperatively. The mean ASES score was 64 points (range, 25 to 100 points), and sixteen of the twenty-two shoulders had a satisfactory outcome according to the limited goals criteria of Neer et al.
When patients with complete humeral head coverage at the time of surgery were compared with those with only a partial rotator cuff repair, there was a trend toward improved outcomes but none were significant. The mean active forward elevation was 125° and 103°, respectively (p = 0.14), the mean external rotation was 41° and 36° (p = 0.28), and the mean ASES score was 73 and 64 points (p = 0.30). Because of our sample size, power for these analyses was insufficient (power = 0.36, 0.15, and 0.18, respectively).
In thirteen shoulders, the subscapularis tendon was transferred superiorly and reattached to the superior aspect of the greater tuberosity. In this group, the average forward elevation improved from 73° to 105°, the average external rotation improved from 15° to 38°, the average postoperative ASES score was 73 points (range, 40 to 97 points), and ten shoulders satisfied the limited goals criteria of Neer et al. In the twenty-one shoulders in which the subscapularis was not transferred superiorly, the average forward elevation improved from 81° to 115°, the average external rotation improved from 15° to 38°, the average postoperative ASES score was 64 points (range, 25 to 100 points), and seventeen of twenty-one shoulders had satisfactory outcomes according to the limited goals criteria of Neer et al.
When shoulders with a superior transfer of the subscapularis were compared with those in which the subscapularis was not transferred superiorly, there was no significant difference in active forward elevation (105° and 115°, respectively; p = 0.48), external rotation (38° and 38°; p = 0.97), or ASES score (73 and 64 points; p = 0.27). However, because of our sample size, power for these analyses was insufficient (power = 0.11, 0.05, and 0.18, respectively).
One-way analysis of variance revealed that mean postoperative ASES scores differed significantly among the preoperative and intraoperative factors studied (preoperative forward flexion, age, gender, previous surgery, and complete humeral head coverage) (F[9, 115] = 2.05; p = 0.04). Multiple regression analysis featuring the same five studied factors as independent variables was performed and was able to account for 34% (R2 = 0.34) of the variation in the mean postoperative ASES score, with preoperative forward flexion and gender accounting for 33% (R2 = 0.33) of the variation.
This study describes the long-term clinical outcomes of hemiarthroplasty for the treatment of cuff tear arthropathy or osteoarthritis with a massive tear of the rotator cuff. We chose to include both diagnoses in our study for the simple reason that it is often very difficult to distinguish between the two. We are not aware of any study that has attempted to define any criteria, clinical or radiographic, to separate the two diagnoses. Furthermore, we are aware of no study that has made any treatment recommendations based on distinguishing between the two disorders. Neer et al. originally described rotator cuff-tear arthropathy as a separate clinical entity1 and found an incidence of rotator cuff tears in primary osteoarthritis of <5%3. Neer et al. did not use range of motion as a criterion for diagnosis, although most patients had poor elevation. We attempted in this study to make an outcome assessment and possibly develop treatment recommendations on the basis of preoperative function rather than on any pathologic or radiographic criteria. In our study, the ability to actively raise the arm to 90° preoperatively was the most useful predictor of a good outcome.
In the group as a whole, twenty-six of the thirty-four shoulders in our patients achieved satisfactory results according to the limited goals criteria of Neer et al., and there were no prosthesis revisions. Although our sample size was small, age and a history of surgery did not appear to influence the results, although our power to detect differences was insufficient. A power of 0.8, meaning an 80% chance of avoiding a type-II error, is considered standard. When the effect of gender was analyzed, women had more severely affected shoulders from the beginning. When preoperative active forward elevation was evaluated, there were significant differences between the groups with respect to pain and function.
We believe that for patients with an intact coracoacromial arch who were able to actively raise the arm to =90° before surgery, hemiarthroplasty and rotator cuff repair can be a durable and reliable procedure. This group had very reliable pain relief and function, and fourteen of the sixteen shoulders were rated as satisfactory according to the limited goals criteria of Neer et al. In eight of the sixteen shoulders, complete coverage of the humeral head with rotator cuff tissue was achieved intraoperatively.
For the shoulders with preoperative elevation of <90°, pain relief and function were less predictable but twelve of the eighteen patients still achieved satisfactory results according to the limited goals criteria of Neer et al. The low rate of complications and the absence of any implant failure make this procedure a viable option. Werner et al.17 treated a similar group of patients, who all had preoperative active elevation of <90°, with a reverse shoulder prosthesis. They described the patients with elevation of <90° as having pseudoparesis. At an average of 3.2 years, the complication rate was 50%; a reoperation was required in one-third of all patients, including 18% of those who had had a primary operation. Other studies with use of a reverse shoulder prosthesis, and all with mean lengths of follow-up under six years, have described complication rates of between 5% and 17% for rotator cuff-deficient patients18-21.
Field et al.14 and Sanchez-Sotelo et al.13 noted failure of hemiarthroplasty in patients with a previous acromioplasty because of anterosuperior escape of the humeral head with a disrupted coracoacromial arch. We had only one patient with this complication; however, only seven of thirty-four shoulders had had prior surgery. We agree with those authors that every attempt to preserve the coracoacromial arch should be made and that patients with preexisting anterosuperior escape instability are not good candidates for hemiarthroplasty.
The mean increase in forward elevation for all patients was 33°, and the mean increase in external rotation was 23°, both significantly improved over preoperative levels. In two studies of the reverse shoulder prosthesis17,18, the mean elevation increased 55° and 58°, while the mean external rotation improved only 8° in one study and decreased by 5° in the other.
Although range-of-motion data were available for comparison of the preoperative and postoperative levels, no functional scoring or visual analog pain data had been recorded preoperatively. The ASES scoring system16, which we used for long-term follow-up, was not developed until almost a decade after the first patients were treated. The inability to quantify the patients' self-reported functional assessments from preoperative to postoperative evaluation is a potential drawback of this study and of many longer-term studies.
As in our previous study, we attempted at least partial rotator cuff repair in all shoulders9. There was a trend, although it was not significant, for an improved outcome when complete repair was achieved. We never detach the rotator cuff posteriorly, but rather try to mobilize medial and superior rotator cuff tissue as much as possible. We do, however, detach the subscapularis and transfer it superiorly in an attempt to increase elevation and rotation, although this did not seem to affect the outcome in our study. The rotator cuff repair portion of the procedure took a considerable amount of time, and, in the opinion of the senior surgeon (L.U.B.), remains important to the success of the operation. In other studies14,22-24, surgeons deliberately made no attempt to repair any remaining tissue in order to minimize soft-tissue trauma. The senior surgeon believes that the failure to mobilize and approximate some torn rotator cuff only leads to further instability.
On the basis of the long-term results in our study, we believe that hemiarthroplasty can provide good long-term pain relief, functional improvement, and overall satisfaction for patients with cuff tear arthropathy or osteoarthritis with a massive rotator cuff tear. We think that it is the procedure of choice for patients who can actively elevate the affected arm to 90° preoperatively as it provides very reliable results in this group. For patients who cannot elevate the arm to 90°, the low prevalence of complications and the high rate of satisfaction with the outcome of the hemiarthroplasty should be weighed against other options such as reverse shoulder arthroplasty that may provide greater functional benefits at the risk of a higher complication rate. 