Patient Satisfaction and Patient Reported Outcome Measure Data in Arthroscopic Release of Frozen Shoulder

We present a consecutive series of 99 arthroscopic capsule releases with a mean follow-up period of 28 months. This cohort gave patient reported outcome measures and satisfaction scores that reflect an excellent outcome following an anterior capsule and rotator interval release with a manipulation to achieve a full range of movement intra-operatively. Oxford Shoulder Score went from a mean of 23 pre-operatively to 44 post-operatively. At a mean of four months, satisfaction score was 9/10, 88% had returned to work, 92% reported an improvement in the quality of their sleep and 94% stated that they would have the procedure again. No difference was found between those that did receive steroid injections at the time of surgery and those that did not and there was no trend for diabetic patients to do worse than their non-diabetic contemporaries.


Introduction
Primary frozen shoulder is characterised by pain and global stiffness of the shoulder in the presence of a normal shoulder radiograph, with the histological findings of chronic inflammation and fibrosis [1][2][3]. With a prevalence of approximately 2% in the adult population [1], it is most commonly seen in middle-aged women [4] and though mainly idiopathic, it is more common in diabetics [5][6][7] who have a 40% chance of developing a frozen shoulder in their lifetime [1,5,6]. There is an association with Dupuytren's contracture [2,8,9], thyroid dysfunction [10,11] and a genetic role has been implicated [2].
Although the natural history of primary frozen shoulder is believed to be self-limiting, resolution of symptoms may take more than three years and there remains a group of patients with residual pain, stiffness and loss of function in the long-term. Patients with the most severe symptoms at onset carry a worse prognosis [3,12].
Patients who have undergone multiple interventions including bee venom acupuncture, anti-inflammatory medication, corticosteroid injection, manipulation under anaesthesia (MUA) and extensive physiotherapy prior to surgery dilute the understanding of the benefits of arthroscopic capsular release itself. Some would argue that these must be tried prior to the offer has been shown to be an effective treatment for this condition, but can require excessive force to tear the fibrotic capsular tissue resulting in reported complications such as humeral shaft fracture, brachial plexus or vascular injury, rotator cuff tear and dislocation. Much like cutting half way through a piece of paper before trying to tear it in two, controlled surgical release of the worst affected frozen shoulder tissue in the rotator interval greatly reduces the force required to achieve an effective MUA, thereby reducing the complication rate.
Ogilvie-Harris et al compared MUA with arthroscopic release in a prospective cohort of 40 patients [15]. Their results after a follow-up of between two and five years showed a similar range of movement, but the release had a better outcome in terms of pain and function. The overall outcome was excellent in 15 of 20 patients in the arthroscopic group but in only seven of 18 in the MUA group. Klinger et al. [22] performed arthroscopic release on 36 shoulders after failure of conservative treatment of 6 months duration and reported that 75% of their patients returned back to work within a mean of 8 weeks after release.
In a larger series of 73 patients Watson et al. [23] discharged all their patients with full range of movement and without pain at a mean of 8.9 weeks after release, results which have not been matched in the literature. The mean duration of conservative treatment in their series was 18 months. In a recent retrospective review of 115 patients who underwent arthroscopic release for refractory shoulder stiffness of varying aetiology, Elhassan et al. [24] reported improvement of age-and gender-adjusted Constant score from 35% to 86% at a mean follow up of 46 months. In their study only 41 patients (36%) belonged to primary idiopathic group who had failed a trial of conservative therapy for a mean of 11 months. The authors reported maximum improvement in this subgroup of patients compared to post-surgical and posttraumatic groups.
Controversy exists as to how much of the capsule to release arthroscopically since similar results have been demonstrated in studies where circumferential release has been performed compared to only a rotator interval release. In the above series [24], the authors performed both anterior and posterior release. In a retrospective comparative analysis of data from 48 consecutive patients who underwent capsular release for resistant primary or secondary frozen shoulder, Snow et al. [25] evaluated the benefit of additional posterior release over standard antero-inferior release, concluding that there was no significant difference in the overall outcome with the addition of a posterior release. Jerosch et al. [26] also reported no additional advantage with the extension to a global 360° capsular release compared with a 270° release. Cadaveric study shows that the axillary nerve runs closest to the inferior glenoid rim between the 5:30 and 6:00 o'clock position and that its closest distance from the glenoid rim varied from 10 to 25mm in the neutral arm position [27]. Pearsall et al. [19] described releasing the intraarticular portion of the subscapularis [19]; however, most studies show excellent results without subscapularis release.
Use of intra-articular administration of steroids during frozen shoulder surgery is standard practice in some centres [14,28]. In 25 patients with primary frozen shoulder undergoing arthroscopic release, Bunker et al routinely instilled 10 mL of 0.5% bupivacaine and 25 mg hydrocortisone acetate through the arthroscopic cannula before withdrawal from the joint. 88% reported dramatic improvement in pain and function within the first two weeks of release [14].
Diabetes mellitus is recognized as a poor prognostic indicator in frozen shoulder, however the reported outcome of arthroscopic capsular release in patients with this disease varies in the literature [7,23,24,29]. Some reviews reported a higher recurrence rate after arthroscopic capsular release in diabetic patients, whereas others did not find a difference in the outcome between diabetic patients and non-diabetic patients.
The importance of PROMs in elective orthopedics has been highlighted by the Department of Health in the NHS Next Stage Review and their widespread use has been recommended [30]. The use of PROMs in assessment of surgical results is enormously useful; however combining this with patient satisfaction questions may add significant value to the interpretation of outcome after surgery [2].

Aim
The aim of this study is to report the functional outcome of arthroscopic release in primary frozen shoulder using validated patient-reported outcome measures (PROMs) and a patient satisfaction index.

Study Population
Between 2002 and 2009, 132 consecutive patients (135 shoulders) identified in a specialist shoulder clinic with a diagnosis of primary frozen shoulder unresponsive to conservative management underwent arthroscopic capsular release and manipulation. Patients with a secondary frozen shoulder, for example post-traumatic, post-surgical or from bony structural abnormalities were excluded from the study, as were all cases in which a secondary procedure was performed to include sub-acromial decompression. All patients were treated with an initial outpatient conservative regimen of intra-articular glenohumeral injection of local anaesthetic and steroid followed by supervised physical therapy. Patients who failed to improve with this treatment, with significant ongoing symptoms of pain and stiffness, were offered arthroscopic capsular release with a manipulation.

Surgical Technique
Arthroscopic capsular release was performed in the beach chair position as a day case procedure under general anaesthesia supplemented by an interscalene regional nerve block. The operations were carried out by either of two fellowship trained shoulder surgeons in the department (ASC & GCH). Capsular release was performed using a radio-frequency hook (3.0-mm 90° hooked electrode, Depuy Mitek, Raynham, MA) through an anterior arthroscopy portal with a standard posterior portal used for visualization. Release of the rotator interval, the middle glenohumeral ligament and the anterior capsule between the 2 and 5 O'clock positions was performed under direct vision with arthroscopic scissors used to complete the anterior capsule release between 5 and 6 O'clock. The subscapularis tendon was not released or violated. Arthroscopic release was followed by gentle manipulation of the shoulder using a short lever arm technique to achieve maximal range of movement.
During the study period it became routine practice to perform an intra-articular injection of corticosteroid and local anaesthetic (8ml 0.5% bupivacaine hydrochloride and 80mg methylprednisolone) to the glenohumeral joint. Post-operatively the shoulder was rested in a sling with instructions to remove it once the effect of the interscalene block had ceased in order to initiate early mobilisation.
Patients were seen by a physiotherapist before discharge and given both verbal and written instructions about shoulder exercises. Formal sessions were introduced at two weeks post operatively as per protocol. Patients were followed up in the outpatient clinic at regular intervals until suitable for discharge [31][32][33][34][35][36][37][38][39].

Data Collection
Patient demographic and comorbidity details were recorded with most Oxford Shoulder Scores (OSS) collected preoperatively. Careful record was made of operative details and any postoperative complications. All patients were followed in the outpatient clinic until discharge. As part of routine postoperative follow-up, patients completed an OSS after surgery at each follow-up. A further OSS was obtained from patients by postal questionnaire at a minimum of 1 year postoperatively. Severity was based on interpretation from Dawson et al. [12] but using the inverted range (0-48) with a greater score representing fewer symptoms.

Statistical Analysis
Paired t-tests were performed to compare pre-and postoperative OSS. Unpaired t-tests were performed to assess variation in OSS between different groups.

Results
Arthroscopic capsular release was performed on 135 shoulders in 132 patients with primary frozen shoulder but a certain number had suspicion of a concomitant impingement syndrome. After postal and telephone reminders, a questionnaire response rate of 88% (119/135) was achieved with the remaining 16 patients lost to long-term follow-up.
In 17% (20/119), subacromial decompression was also performed at the time of surgery excluding them from further statistical analysis thus leaving 99 patients suitable for inclusion in the study. 51.5% of this cohort (51/99) received a peri-operative intraarticular injection. The demographic characteristics of these subgroups were similar to that of the overall cohort.
The mean age of patient was 53 years (range 37 to 70 years) with 53 men and 46 women, of whom 20% (20/99) were diabetic and 53 were right sided operations. Mean follow-up was 28 months (range 4-88 months) with 76% (75/99) having followup greater than 12 months.
Complications were seen in 11 of our 99 patients (11%). 6 patients had ongoing pain at final follow-up, two described weakness, two described stiffness, one had recurrence and one had a possible ophthalmic artery infarct, not thought to be directly related to this surgery.
Pre-operative OSS was available for 77 of the 99 (Subgroup 1), but only 37% (37/99) had an early post-operative OSS recorded, described as scores recorded within 6 months of surgery (Subgroup 2). Thus, 2 sub-groups were identified for analysis from the overall cohort and their demographics are shown in (Table 1).

Oxford Shoulder Scores
In the 99 patients with long-term follow-up (mean 28 months) the mean postoperative OSS was 42 (median 46, range 4-48). 35% (35/99) had a postoperative OSS of 48/48, indicating normal shoulder function. 64% (63/99) had an OSS of 44 or  In the 77 patients with pre-operative and post-operative OSS (after greater than a year) a highly clinically and statistically significant improvement in OSS was seen. Mean pre-operative OSS was 23 (+/-7.65SD), which improved to 44 (+/-6.62SD) post-operatively (p<0.0001, 95% CI -23.72 to -19. 16) In the 37 patients with pre-operative and both early and late post-operative OSS, an early improvement in OSS was seen within a mean post-operative period of 4 months which was sustained and further improved at long term follow-up (mean 37 months). This was statistically significant P<0.0001 (Table 2).
For further analysis the 77 patients were split into two groups by the severity of their pre-operative OSS. A severe group (OSS ≤ 23) (n=38) and a mild to moderate group (OSS 24-43) (n=39). When pre and postoperative OSS were compared, there was a significant difference, suggesting a poorer, but still clinically acceptable, outcome in the presence of more severe symptoms at the time of surgery (Table 3).
Similar satisfaction levels were found in Subgroup 1 who had pre-operative OSS recorded. Table 4 summarizes the results of the satisfaction questionnaire.

Intra-articular injection
There was no statistical difference in post-operative OSS or satisfaction between those patients who received an intraarticular injection and those who did not (Table 4).

Diabetes
In these 99 patients there was no significant difference in post-operative OSS between diabetic (mean OSS 43) and nondiabetic patients (mean OSS 42) (p =0.6232). There also appeared to be no significant difference in the severity of symptoms preoperatively between diabetics (mean OSS 23) and non-diabetics (mean OSS 23) (p = 0.893). As a group, their responses to the satisfaction questionnaire questions appeared no different to the overall response (Table 4) and there was no trend for these patients to do worse than their non-diabetic contemporaries.

Discussion
Although some frozen shoulders may spontaneously resolve, patients presenting to our specialist shoulder clinics are generally frustrated with the longevity of their symptoms.
Our study demonstrates good relief of symptoms with anterior capsule and an interval release with an MUA, reducing the risk of axillary nerve or subscapularis tendon damage that may ensue with a more extensive release.
We are able to establish that arthroscopic capsular release for primary frozen shoulder gives significant improvement in shoulder symptoms and function when assessed using a jointspecific PROM such as the OSS and produces a very high level of patient satisfaction. 94% of patients would have the procedure again if required.
In our series, the patients in whom early post operative OSS were available showed a significant early improvement in their OSS with a further improvement at a mean of 30 months post release to a near maximal OSS. These results show that most of the improvement in symptoms occurs within the first few weeks but further improvement occurs several years after the release. This potential for early postoperative recovery is reinforced by      the response to our patient questionnaire where the reported median time for symptom improvement was 3 months. At a recent discussion at the national meeting for shoulder surgeons, early recovery was seen to be the main thrust for surgical intervention and suggestions were made that we should be scoring our patients as early as two weeks post-operatively.
Our complication rate was comparable to that noted in the literature, with only a suspected ophthalmic artery infarct appearing out of the ordinary, but clearly unrelated.
Same day discharge is preferred by most of our patients, and the excellent postoperative pain relief achieved with interscalene block together with concomitant oral analgesia encourages early discharge from hospital.
Throughout the literature, varying comments are made regarding outcomes in diabetics but we did not find a significant difference for those suffering from diabetes mellitus. Their preoperative & post-operative OSS, satisfaction and improvement was similar to non-diabetics.
Long-term outcome studies of primary frozen shoulder have reported poor outcome to be related to severity of symptoms on initial presentation [2]. In our study a significantly worse postoperative OSS was found in those patients with severe pre-operative OSS (OSS ≤ 23) compared to those with mild to moderate preoperative OSS (OSS 24 to 43). However there was still a clinically acceptable improvement in patients with severe pre-operative OSS with a very high satisfaction amongst all groups.
Though some centers choose routinely to administer steroid intra-operatively, others would suggest the risk of infection is too great. We saw no infections in our cohort and believe that no harm has come as a result of their use in this study, however the data would suggest there was no gain either.

Conclusion
Very few complications were recorded in our series, which concurs with most of the previously published reports.
Arthroscopic capsular release has been established as a highly successful procedure. This study shows excellent results in a large group of patients with the specific diagnosis of primary frozen shoulder, treated with an arthroscopic release, as reported by the patient rather than the clinician.
We have actively moved to offering our patients questionnaires as early as two weeks post-operatively and look forward to publishing further work on early recovery as reported by the patient. Both short and long-term outcomes are clearly of interest here.
Our operative technique is not only safe and reproducible but also allows faster overall recovery from pain and restricted range of movement, permitting this age group to return to their desired activities and their work sooner.