Two to 5% of patients referred to orthopedic clinics suffer from adhesive capsulitis of the shoulder (ACS), also known as frozen shoulder [1,2]. In 1945, Neviaser initially proposed this term for chronic inflammation of the joint capsule with fibrosis and adhesion leading to pain and limited range of motion (ROM) of the shoulder joint [3,4]. The dominant demographic characteristics of this disease are 40–60-year-old women with or without any comorbidity (diabetes, thyroid disorders, or cardiovascular diseases) [5,6]. The disease manifests itself in three consecutive phases: a freezing stage when there is a gradual increase in pain with nocturnal peaks and stiffness. Continuation of the disease process involves shoulder joint stiffness. The final thawing phase can last from a few weeks to several years and comprises gradual improvement in shoulder function and decrease in pain intensity [7,8]. The time of the patient's visit among these phases, the wide range of clinical symptoms, and various underlying factors have led to the suggestion of various treatments [1,9,10].

Most patients respond well to conservative treatments [11,12], and most orthopedists recommend such a procedure for at least 6 months [5,6]. Manipulation under anesthesia is another option for orthopedists, but it can lead to iatrogenic fracture [13,14]. Open (more invasive) and arthroscopic surgeries have had satisfactory results [5,6], although open surgical procedures involve more intraoperative bleeding and morbidity and possibly longer hospitalization than do arthroscopic procedures [15,16]. However, arthroscopic techniques are not free of complications, with the possibility of iatrogenic damage to the axillary nerve and chondrolysis [17,18].

ACS is a multifactorial disease whose demographic characteristics, etiology of disease onset (idiopathic, after surgery, and following trauma), and comorbidities can each affect the process and severity of the disease. It is unclear which of the above factors is most impactful in the onset and severity of the disease. The aim of this study was to evaluate the objective and subjective criteria and the response rate to arthroscopic release treatment in ACS patients refractory to conservative treatment. The secondary purpose was to investigate the effects of underlying factors of age, sex, comorbidities, and disease etiology on patient response rates. Our study hypothesized that, in patients with ACS, the arthroscopic release method leads to effective improvement in shoulder function and significant reduction in pain.

The protocol for this study was approved by the Institutional Re­view Board of Guilan University of Medical Sciences (IRB No. 725). Owing to the retrospective design, the requirement for in­formed consent was waived. All collection data forms were blind­ed and without the patients names.

In this retrospective study, all patients with unilateral stiffness of the shoulder refractory to conservative treatment (the second stage of the disease or the frozen stage) who underwent arthroscopic releasing surgery of the capsule from 2011 to 2016, were evaluated for eligibility. The inclusion criteria of the study were (1) normal findings on radiographic images of the shoulder joint; (2) pain during shoulder movement; (3) no improvement or progress after at least 6 months of nonsurgical, conservative treatment; (4) restriction in forward flexion to less than 100° and reduction of external and internal rotation to less than 50% of the normal limit [19,20]; and (5) patient follow-up for at least 2 years. The patients with evidence of degenerative changes in the shoulder joint (glenohumeral arthritis) or rotator-cuff tear and patients with uncontrolled diabetes (based on laboratory tests) were excluded. All patients underwent magnetic resonance imaging (MRI) to rule out other shoulder joint pathologies, and all underwent nonsurgical treatment for at least 6 months. Nonsurgical treatment included administration of analgesics and oral corticosteroids, intra-articular corticosteroid injections, and shoulder physiotherapy.

A total of 73 patients was included in the study, of which 15 were lost to follow-up, and seven were excluded because of glenohumeral arthritis found on MRI. The mean age and follow-up period of the 51 eligible patients who completed all follow-up sessions (before surgery, 6 months after surgery, and final visit) were 49.1±5.6 years and 49.3±13.6 months, respectively. Table 1 shows other demographic information of the patients. Table 2 shows the etiology of ACS by age, sex, and diabetes.

In general, all outcome scores (VAS, CS, SST, and ROM) improved over time. The passage of time was an effective factor in improving outcomes (Table 3); also, there was no statistically significant difference between the outcome scores over these time periods (before surgery, 6 months after surgery, and the final visit) based on sex, age, or diabetes (p>0.05).

We divided patients into three etiological groups of idiopathic, after surgery, and following trauma. The results showed that the outcome scores improved over time based on etiology (P_time<0.05). In addition, the group effect was significant (P_group<0.05), as was the group-by-time interaction effect. The variations between groups differed by visit (P_{int.time×group}<0.05). The results of the post hoc test performed with the Tukey method showed that, before surgery, only the SST level in the after-surgery group was lower than that of the idiopathic group (p<0.05). At 6 months after surgery, the levels of VAS, CS, and SST were poorer in the after-surgery group than in the other groups (p<0.05). Also, in the ROM examination at this time, only forward flexion and internal and external rotation in the after-surgery group were significantly smaller than those of the idiopathic group (p<0.05). At the final follow-up, the outcome scores in the after-surgery group were poorer than in the other two groups (p<0.05). The outcome scores of the idiopathic group and following trauma group were not significantly different at any time point (p>0.05) (Table 4).

This study showed that patients, regardless of underlying factors, had significant improvements in pain, functional scores (CS and SST scores), and ROM in the short-term (6 months) and final (at least 2 years) follow-up after treatment. Also, Jerosch et al. [21] and Ranalletta et al. [22] achieved good results after arthroscopic release in patients with ACS. The CS in our study improved from 48.2 to 74.4 and 77.02 at 6 months and the final follow-up, respectively; these scores improved from 41 to 83 in the Jerosch study [21] and from 42 to 86 in the Ranalletta study [22].

The response to treatment and patient satisfaction rate (SST score) indicated successful surgery. The SST score increased from 3.1 to 7.5 at the 6-month follow-up and to 9.4 at the final follow-up. Segmüller et al. [23] in a 13.5-month follow-up and Le Lievre and Murrell [24] in a 7-year follow-up assessed patient satisfaction rate, noting good to excellent satisfaction in 88% and 85% of patients, respectively.

From an anatomopathological point of view, the most important feature of the disease is the presence of scars in the rotator interval involving the supra glenohumeral and coracohumeral ligaments [25]. In the present study, rotator interval release and subsequent global capsular release improved ROM in the shoulder joint. Cohen et al. [26] and Berghs et al. [27] achieved similar results.

There is a debate in the literature about the rate of arthroscopic release. Some consider release of the coracohumeral ligament and the rotator interval to be sufficient [28], whereas some surgeons support the release of other joint structures such as the inferior and posterior capsules or advocate global capsular release [29]. In this study, we first released the interval rotator, followed by that of existing contracted remnants in the anterior and/or the posterior area if needed. However, Ranalletta et al. [22] released only the anteroinferior portion of the capsule and obtained suitable therapeutic results. In future prospective studies that measure patient limitations in various shoulder movements, it might be possible to estimate the desired extent of capsular release before surgery. The potential complications of this surgical procedure include infection, iatrogenic injuries causing chondral lesions, and axillary nerve damage, none of which were found in our study.

In this study, we also examined the impact of underlying factors of sex, diabetes, and etiology of the disease on the recovery rate, functional outcome, and ROM improvement. Sex did not affect patient response to treatment, and women showed equal improvement in performance and ROM to men. To our knowledge, no study has found these factors to affect the effectiveness of arthroscopic treatment in patients with ACS. In a study that also examined the underlying factors in manipulative treatment under anesthesia, Theodorides et al. [30] found that the functional scores of women improved more than those of men. However, their study did not note the etiological cause of ACS. We divided patients into three etiological groups of idiopathic, after surgery, and following trauma. In patients with idiopathic underlying cause and in patients with trauma etiology, pain and function scores (CS and SST) improved significantly and more effectively than for the group of post-surgery patients. Also, the ROM in patients who had ACS because of a previous shoulder surgery showed weaker recovery than that in the other two groups. These differences might be attributed to the manipulations following the previous surgery that triggered inflammatory cascades and arthritic processes in that area. Because the shoulder joint does not have high blood supply like other ball and socket joints, patient healing might have been delayed even after the joint was released. Another reason for the differences in group outcomes could be the formation of adhesion bands in several areas of the shoulder joint complex.

One of the strengths of our study was the careful selection of patients with no structural problems. Also, because different people have different ROMs and joint strength, we compared each involved shoulder with the contralateral one of the same individual. Another strength of this study is the use of subjective scores for the shoulder using the CS and SST questionnaires along with objective scores. Some studies consider only patient-reported criteria and subject the study to subjective/objective bias. Another advantage of our study was the performance of all surgeries by the same expert surgeon with 8 years of experience in arthroscopic shoulder surgery to avoid the bias of multiple surgeons and the differences in degree of expertise. However, performance by a single surgeon limits the generalizability of the study. In addition, the small sample size, lack of a control group, and retrospective nature of the study are weaknesses.

In patients with ACS refractory to conservative treatment, arthroscopic releasing surgery of the shoulder involves rare complications and effective response in terms of pain relief and increased function and in the appropriate increase of shoulder ROM. The surgeon should be aware that patients with adhesive capsulitis following a previous surgery might have a weaker response to treatment compared with that of those with idiopathic or injury etiology.