Study Design
Following ethics approval at our institution, a retrospective study was conducted on patients who had an arthroscopic capsular release for idiopathic adhesive capsulitis five years or more previously. The primary outcome measure was defined as the effect of arthroscopic capsular release on patient-reported frequency and magnitude of pain with activities five years or more after the surgery. Secondary outcomes were the long-term effect of arthroscopic capsular release on patient-reported pain frequency and magnitude of pain at rest and at night, difficulty reaching behind the back or above the head, level of activity at work and level of sport played, overall shoulder stiffness, overall patient satisfaction, complications, and shoulder motion compared with that at the preoperative evaluation and at one, six, twelve, twenty-four, and fifty-two weeks and five years or more postoperatively. A comparison was also made between the contralateral (nonoperatively treated) shoulder and the operatively treated shoulder with respect to shoulder motion at five years or more.
Inclusion and Exclusion Criteria
The criteria for a diagnosis of idiopathic adhesive capsulitis7,10,16-20 were (1) a painful stiff shoulder for at least four weeks; (2) restriction of passive external rotation of at least 50% compared with the contralateral shoulder; (3) difficulty using the affected arm, with restriction of movement and loss of function; and (4) pain at night causing a sleep disturbance and inability to lie on the affected side.
For inclusion in this study, patients were required to have undergone an arthroscopic capsular release for idiopathic adhesive capsulitis performed by the senior author (G.A.C.M.), and a minimum five-year follow-up period. The exclusion criteria included (1) evidence of glenohumeral joint arthritis at the primary procedure, (2) a full-thickness rotator cuff tear, (3) any fracture involving the shoulder girdle, (4) diabetes, (5) a history of a motor vehicle accident as a cause of the initial injury, (6) previous surgery to the involved shoulder, and (7) an unwillingness or inability to attend long-term follow-up evaluations.
An extensive effort was made to contact the patients who met the inclusion criteria, with use of the personal contact information on record, through a mailing, and by telephone. If patients could not be reached, the patient's referring medical practitioner or physical therapist was contacted by telephone and/or an online search was made of the national telephone directory. When attendance at the clinic was not possible, a standardized pain assessment was conducted by telephone or, if appropriate, a house visit was made.
Outcome Assessment
At each clinic visit (the preoperative evaluation; follow-up at one, six, twelve, twenty-four, and fifty-two weeks; and final study evaluation), patients completed a standardized questionnaire with scales for evaluating both pain and function (based on the Shoulder Rating Questionnaire [SRQ]21, with separately scored domains for global assessment, pain, daily activities, recreational and athletic activities, work and satisfaction; the final, nongraded domain of the SRQ was not included in our study). Shoulder motion and strength were measured by examiners using previously validated techniques22,23.
Patients responded to questions pertaining to frequency of pain during activity, frequency of night pain, frequency of extreme pain, intensity of resting pain, intensity of activity pain, intensity of night pain, and overall shoulder status, using Likert scales. Patient-determined pain severity was ranked as none, mild, moderate, severe, and very severe. Pain frequency was ranked as never, sometimes, monthly, weekly, and all of the time. Patients were also asked to state their level of activity at work (none, light, moderate, or strenuous labor) and current level of sport played (none, hobby, club, or national), as well as rank the difficulty they had in performing daily tasks such as reaching above the head and reaching behind the back (none, mild, moderate, severe, or very severe). At the final visit, patients were asked, “Would you be able to point to the word that best describes the impact that arthroscopic capsular release has had on your life, in terms of improving movement and reducing pain?” and they could choose excellent, good, average, or poor.
Range of Motion
In this study, isolated passive range of shoulder motion (internal rotation, external rotation, forward flexion, and abduction) was assessed visually with use of a previously described and validated protocol22,24. At the final visit after long-term follow-up, the shoulder that had the capsular release was examined and then the contralateral shoulder was evaluated to facilitate a comparison between the two.
Operative Procedure and Rehabilitation
Following interscalene regional anesthesia, patients were positioned in the beach-chair position for arthroscopy. The passive shoulder motion was assessed as detailed above. After preparation and draping of the patient, an arthroscopic capsular release was initiated by inserting an arthroscope into the glenohumeral joint via a standard posterior portal. An anterior portal was established under direct vision with use of a spinal needle lateral to the coracoid process as previously described25. The portal was established just superior to the superior border of subscapularis. A spinal needle was utilized to ensure that instruments could access the inferior capsule and, if possible, the posterior capsule. Through this opening, the anterior and inferior capsule were cut lateral to the glenoid labrum with use of a 3-mm suction wand (500 Suction, CoVac 50 ArthroWand; ArthroCare, Sunnyvale, California) or with a 4-mm arthroscopic punch. The tissue in the rotator interval was released to the anterior border of the long head of the biceps muscle and medially to the base of the coracoid process under direct vision. A portion of the intra-articular subscapularis tendon was also divided15 to improve shoulder motion outcomes5. Either through the same anterior portal, or more commonly from a posterior inferior portal, the inferior and posterior aspects of the capsule were released, to achieve a complete 360° release. After the release, the arthroscope was removed, a gentle manipulation was performed, and shoulder motion was assessed. The glenohumeral joint was injected with 10 mL of Depo-Medrol with Lidocaine (40 mg/mL of methylprednisolone acetate and 10 mg/mL of lidocaine hydrochloride, with 0.9% m/v benzyl alcohol [preservative]; Pfizer, New York, NY) for pain relief. The portals were closed, and the shoulder was dressed with a soft bulky dressing. The extent of the release and the shoulder motion were recorded by the operating surgeon on a specifically designed, standardized form. Patients were discharged on the day of the surgery without a sling. They were provided with an ice pack to use on the affected shoulder for twenty minutes every two hours during waking hours for two days.
Prior to surgery, all patients met with a physical therapist who emphasized the importance of postoperative exercises, and they were instructed to start range-of-motion exercise at the shoulder on the day they arrived home from surgery. Beginning on day 1, patients met with a therapist for the first of two supervised exercise sessions in the first postoperative week. The sessions focused on maintaining passive and active-assisted shoulder motion. They were instructed to perform assisted shoulder movements every two hours at home for the remainder of the week. At the second meeting, patients were instructed to perform ten repetitions of active-assisted external rotation of the shoulder with a broom handle. During the second postoperative week, exercises comprised three sets a day of ten repetitions each for resistive retraction with a Thera-Band (Hygenic, Akron, Ohio) for external rotation, internal rotation and adduction, and free-weight resistive flexion and abduction and five repetitions of ten-second duration stretches each for flexion, horizontal adduction, and external rotation (0° and 90° of abduction). Exercises to ensure early facilitation of rotator cuff muscle strengthening were then introduced. Patients were encouraged to perform the same exercise regimen at home three times daily for the ensuing ten weeks.
Statistical Analysis
Comparisons were made between groups with continuous variables that had a normal distribution with use of two-way Student t tests, and those with categorical data were compared with use of Mann-Whitney rank-sum tests. The level of significance was set at p ≤ 0.05.
Source of Funding
The study was supported in part by the University of New South Wales (Independent Learning Project student support), the St George Hospital (research infrastructure support), and Premier Specialists (data collection).
Study Group
Between April 1997 and December 2004, the senior author performed 121 arthroscopic capsular releases for idiopathic adhesive capsulitis in 115 patients. Of the 121 shoulders, thirty-seven were excluded because of concurrent rotator cuff injury and repair; ten, for glenohumeral osteoarthritis at the time of surgery; two, for a combination of these conditions; and one, for the presence of diabetes. Of the seventy-one surgical procedures remaining, twenty-two patients were unable to participate in the long-term follow-up (eight died and fourteen could not be contacted), leaving forty-nine shoulders in forty-three patients as the study group.
Cohort Demographics
Of the forty-three patients (forty-nine shoulders), nineteen (44%) were male and twenty-four were female (56%), with a mean age of sixty-one years (range, thirty-seven to eighty-seven years) and a mean follow-up period of seven years (range, five to thirteen years) since surgery. There were eight patients (19%) who indicated that they had bilateral idiopathic adhesive capsulitis at some point, with six requiring an arthroscopic capsular release for both shoulders. Of the forty-nine shoulders, nineteen (39%) were on the right side and thirty (61%) were on the left.
Patient-Reported Pain Scores
Before surgery, most patients rated their pain with activity as severe to very severe, pain at rest and at night as moderate to severe (Figs. 1-A and 1-B; see Appendix), and overall stiffness as between moderate and quite stiff (Fig. 2; see Appendix). In addition, patients described the difficulty they experienced in reaching behind the back or over the head as severe to very severe (Fig. 2; see Appendix).
At one week after arthroscopic capsular release, patients reported relief in the severity and frequency of night pain, pain with activity, and pain at rest (Figs. 1-A and 1-B; see Appendix). Improvements were also seen immediately in relation to overall shoulder function, overall stiffness, and daily activities, such as reaching behind the back or above the head (Fig. 2; see Appendix).
These gains in the relief of shoulder pain severity and frequency continued for one to two years. Patients rated their pain severity (at night, at rest, and with activity) between moderate and mild, and the pain frequency (at night, at rest, and with activity) between monthly and sometimes (Figs. 1-A and 1-B; see Appendix). Daily activities, such as reaching behind the back, remained steady at the moderate level, and reaching above the head continued to improve to the mild level between weeks 6 and 52 after surgery (Fig. 2; see Appendix).
Patient-reported pain scores at the time of the seven-year follow-up continued to improve to between mild and none (p < 0.001; Fig. 1-A; see Appendix). There was a significant reduction in pain frequency: during activity (p < 0.001) and during sleep (p < 0.01), and the frequency of extreme pain levels (p < 0.05) decreased to between sometimes and never (Fig. 1-B; see Appendix). Overall shoulder function was rated between fair and good (p < 0.001), and the severity of shoulder stiffness was between a little and none (p < 0.001; Fig. 2; see Appendix). Patient-reported difficulty with reaching behind the back (p < 0.001) and reaching above the head (p = 0.01) decreased to between mild and none (Fig. 2; see Appendix).
At the time of the long-term follow-up, patient-reported levels of activity at work were significantly better (p ≤ 0.05) compared with preoperative rankings, and the current level of sports played was not significantly different compared with preoperative levels or those at short-term follow-up (Fig. 2; see Appendix).
When asked the question “Would you be able to point to the word that best describes the impact that arthroscopic capsular release has had on your life, in terms of improving movement and reducing pain?” 85% responded that it had been excellent or good; 12%, average; and only 3%, poor.
Range of Motion
Of the forty-three patients (forty-nine shoulders), twenty-five patients (thirty-one shoulders) were able to attend the seven-year follow-up evaluation to have both the shoulder that had capsular release and the contralateral shoulder examined. Old age (two patients), family commitments (three patients), work commitments (eight patients), and distance (five patients) were reasons cited by the patients as to why they could not return for shoulder examination.
There was a significant improvement in shoulder motion at seven years compared with the initial presentation (p < 0.001) and at the completion of the arthroscopic capsular release (p < 0.001; Figs. 3-A, 3-B, and 3-C; see Appendix).
In comparison with the results at the time of short-term follow-up, there were significant improvements in shoulder motion at the seven-year follow-up evaluation with regard to abduction (p < 0.001), forward flexion (p < 0.001), and internal rotation (p = 0.05). However, external rotation (p = 0.6) showed no significant difference with respect to the short-term follow-up, although there was no overall regression in shoulder movement (Figs. 3-A, 3-B, and 3-C; see Appendix).
At a mean of seven years after surgery, the contralateral shoulders (in the patients who had not had surgery bilaterally) had passive range of motion (a mean and standard deviation of 173° ± 8° of abduction, 77° ± 6° of external rotation, 174° ± 9° of forward flexion, and internal rotation to T9 ± 1) that was almost identical to that in the shoulders after capsular release (a mean of 172° ± 10° of abduction, 74° ± 7° of external rotation, 174° ± 7° of forward flexion, and internal rotation to T9 ± 2); the difference was not significant (Table I).
Complications
There were no intraoperative, postoperative, or long-term complications in the forty-three patients (forty-nine shoulders). Specifically, there was no osteoarthritis, recurrence requiring a rerelease, axillary nerve dysfunction, infection, or shoulder instability. Three patients were diagnosed as having so-called impingement in the shoulder that had the capsular release, which was treated with a corticosteroid and local anesthetic injection in the subacromial space at seventy-one, eighty-two, and ninety-five months, respectively, after the operation. One patient had osteoarthritis (which had not been noted at the index procedure) in the involved shoulder at eighty-nine months postoperatively, but this condition was also present in the contralateral shoulder, the neck, and both knees. Two patients had undergone a rotator cuff repair at forty-three and fifty-six months postoperatively for injuries that had occurred after the capsular release.
This study confirmed that arthroscopic capsular release resulted in significant reduction in pain severity and frequency at one, six, twelve, and fifty-two weeks after surgery and that these improvements persisted, and were even more pronounced at a mean of seven years (range, five to thirteen years) after capsular release, with no short or long-term complications. At a mean of seven years (range, five to thirteen years) after the operation, there were also no deficits in shoulder motion compared with the contralateral, noninvolved shoulder, and the improvements in shoulder movement gained as a result of the surgery and maintained at fifty-two weeks persisted and even improved over the long term.
To our knowledge, no other studies have evaluated the outcomes of arthroscopic capsular release for idiopathic adhesive capsulitis at more than two years after the index procedure. We found that patients continued to benefit from the arthroscopic capsular release at a mean of seven years postoperatively, with continued significant improvement in overall shoulder function, relief of shoulder stiffness, and relief of difficulty in reaching behind the back or above the head. Another important finding was the continued significant relief in terms of the severity and frequency of pain with activity, at rest, and when trying to sleep. These data are consistent with short-term outcome studies on arthroscopic capsular release for adhesive capsulitis26,27.
A number of studies28-33 have evaluated the short-term results of nonoperative management of adhesive capsulitis. At two years, most patients who had nonoperative treatment had significant improvement, but the shoulders were not normal10,16. There have been few studies evaluating the long-term natural history of adhesive capsulitis10,34.
Other studies have offered insights into the functional and shoulder motion outcomes after arthroscopic capsular release for shoulder stiffness. Ide and Takagi19 evaluated thirty patients with primary adhesive capsulitis and twelve with secondary adhesive capsulitis who had undergone an arthroscopic capsular release accompanied by a subacromial decompression and used a continuous passive motion machine postoperatively. The patients continued to have improved shoulder movement and functional outcomes at a mean follow-up of 7.5 years, a finding consistent with our results.
Hand et al.7 studied a cohort of 223 patients with a diagnosis of idiopathic adhesive capsulitis who were treated with a variety of options, including benign neglect, steroid injection, physiotherapy, manipulation under anesthesia, and arthroscopic capsular release, and were followed for an average of 4.4 years (range, two to twenty years). With a focus on functional outcomes associated with the Oxford Shoulder Score, their results revealed that patients who had reported unbearable symptoms in the first six months had a significantly worse outcome compared with those who reported severe, moderate, or mild symptoms at presentation. However, the plurality of treatments made it difficult to compare the intervention strategies. One of the interesting findings in our study was the absence of limited shoulder motion in the operatively treated shoulder compared with the contralateral shoulder at seven years. This is in contrast to the study by Shaffer et al.10, who reported that 50% of the patients who had been treated nonoperatively for adhesive capsulitis continued to have mild pain or stiffness seven years after the initial symptoms and to have, on the average, limited shoulder motion compared with the contralateral shoulder and study-generated controls.
Potential complications of this arthroscopic capsular release include axillary nerve injury, infection, and iatrogenic chondral injury from the insertion of the arthroscope35. In the present study, these complications were not observed. No patient had iatrogenic arthritis, and only one patient had arthritis in the involved shoulder, but that patient also had arthritis in other joints.
It is difficult to determine if the long-term outcomes after capsular release are better than the results after manipulation with the patient under anesthesia. Farrell et al.18,who examined eighteen patients after they had had manipulation under anesthesia for idiopathic adhesive capsulitis, reported significant improvements in the range of motion in >90% of patients, with 85% of patients without pain at an average of fifteen years (range, 8.1 to 20.6 years) after treatment. Weber et al.36 also showed that 73% of patients showed a perfect recovery 4.7 years after they had manipulation under anesthesia for adhesive capsulitis, which is consistent with our results.
In a study with an average follow-up period of 13.5 months, Segmüller et al.37 conducted a patient satisfaction survey in relation to the overall satisfaction with arthroscopic capsular release among twenty-four patients. A total of 88% of the patients were very satisfied with the procedure. We used a similar visual scale when we asked the patients to evaluate the impact that an arthroscopic capsular release had had on their life at the time of the long-term follow-up evaluation, and most patients (85%) reported their outcome as “good or excellent.”
The strengths of our study are the relatively long follow-up period, the precise inclusion and exclusion criteria for idiopathic adhesive capsulitis, the fact that both patient-reported and examiner-reported data had been regularly collected during the treatment process, and that a single surgeon with extensive experience in arthroscopic shoulder surgery performed all of the procedures. However, this may also imply that the results of this study cannot necessarily be extended to other forms of painful, stiff shoulders or to other surgeons. Weaknesses of the study include an absence of a control group, and the fact that the examiner, while independent, was not blinded with respect to the nature of the index procedure. In addition, a substantial number of the originally treated patients were not available for the final study evaluation. It is possible that the improvements noted by the examiners and the patients between the evaluation at one year and that at a mean of seven years may represent a perception on the part of the patients and the examiners.
In conclusion, this study shows that, after five to thirteen years, patients with an arthroscopic capsular release for idiopathic adhesive capsulitis had a continued, complication-free, significant improvement, with restoration of overall shoulder function, relief of shoulder stiffness, and relief of difficulty in reaching behind the back and above the head. There was continued significant relief in the severity and frequency of shoulder pain with activity, at rest, and when trying to sleep. Shoulder motion was comparable with that of the contralateral shoulder.