Abstract
Background
Glenohumeral adduction restriction (AR), which is found in rotator cuff tears, is also observed in frozen shoulder (FS). AR was examined using an adduction test and treated through adduction manipulation. We aimed to compare the incidence and severity of AR and investigate clinical characteristics and outcomes of the freezing and frozen phases.
Methods
Two hundred sixteen patients with FS were enrolled in this study; consequently, 120 were classified into the freezing phase (mean age 58 years, 37 men) and 56 into the frozen phase (mean age 55.4 years, 29 men). Using the adduction test, the patients in 2 phases were divided into 2 groups, with and without AR. The glenohumeral adduction angle (GAA) was measured radiographically. Treatments in the freezing phase were physiotherapy and/or adduction manipulation and joint manipulation and physiotherapy for the frozen phase. We recorded the visual analog scale of pain severity, EuroQol-visual analog scale, flexion, abduction, external rotation (ER), internal rotation, and American Shoulder and Elbow Surgeons and Constant scores at the baseline and at the 1-, 3-, 6-, 12-, and 24-month follow-ups.
Results
Seventy-five of 120 patients in the freezing phase were divided into group 1 without AR and 45 into group 2 with AR. Eight of 56 patients in the frozen phase were classified into group 3 without AR and 48 into group 4 with AR. AR was identified in 37.5% of patients in the freezing phase and 85.7% in the frozen phase. The mean GAA decreased from the freezing (−3.0°) to the frozen phases (−18.3°). GAA was positively correlated with ER. The treatment duration in group 1 (5.2 M) was shorter than in group 2 (7.4 M), and the percentage of transition to joint manipulation in group 1 (5.3%) was lower than in group 2 (17.8%). Complete rupture of intra-articular soft tissues was observed in group 4 but not in group 3 with magnetic resonance imaging. Clinical items, except for ER in the freezing phase (group 1 vs. 2) and frozen phase (group 3 vs. 4), were not significantly different from those at the initial visit to the 24-month follow-up appointment.
Conclusion
The incidence and severity of AR increase from the freezing phase to the frozen phase. AR correlating with ER reflects the progression of intra-articular lesions, which prolongs the treatment duration and increases the joint manipulation rate in the freezing phase. Negative AR in the frozen phase suggests mild intra-articular pathologies.
Keywords: Frozen shoulder, Freezing and frozen phases, Comparison study, Glenohumeral adduction restriction, Adduction test, Adduction manipulation
Frozen shoulder (FS) is an idiopathic and disabling disorder characterized by shoulder pain and progressive loss of active and passive range of motion (ROM). Approximately 2%-5% of the general population aged 40-60 years mainly develop FS, which is more prevalent in women than in men.4,22,38 Patients with diabetes and thyroid disease manifest a greater incidence of FS than controls without such diseases.42 However, the etiology and pathology of this disorder remain unclear. The natural history of FS has been described as a self-limiting disorder and is classified into 3 phases: a freezing phase, which results from inflammation; a frozen phase, which is characterized by joint stiffness, followed by a thawing phase over 1-3 years.11,25 A long-term follow-up study reported that 94% of patients without treatment recovered to a normal level of function and joint motion; however, other studies showed that 40%-50% of patients treated conservatively had persistent pain and stiffness after a mean follow-up of 5-11 years.3,29,36 Therefore, it is essential to accurately detect the pathological conditions of this disorder and provide appropriate treatments to every patient with FS.
The pathologies of FS in the freezing phase are inflammation and synovial proliferation, and fibrotic processes develop in the frozen phase.31 Initial inflammation is induced around the rotator interval, and fibrotic processes develop in the glenohumeral (GH) joint capsule, coracohumeral ligament (CHL), and subacromial bursa. Inflammatory cytokines, immune cells, fibrotic growth factors, and type Ⅲ collagen have been identified in the joint capsule and synovium of patients with FS. The immune cell landscape switches from macrophages to T cells, and activated fibroblasts upregulate inflammatory and fibrotic processes.26,31 Angiography of the thoracoacromial artery and magnetic resonance imaging (MRI) reveal the pathologies in FS. Abnormal vessel formation has been observed at the rotator interval in patients with FS.21 MRI findings, such as thickening of the joint capsule and CHL and fibrosis of the subcoracoid fat triangle, support the diagnosis of FS.23,32 As these pathologies progress, periscapular and rotator cuff muscles' tightness and loss of scapular and thoracic movement simultaneously develop.8,14,19,33,37
The overall objective of treating FS is to relieve pain and restore the ROM. Treatment regimens start with conservative therapy, including nonsteroidal anti-inflammatory drugs, corticosteroid injections, and physiotherapy for patients in the freezing phase.6,20 More invasive approaches, such as capsular distension, joint manipulation under anesthesia, and arthroscopic capsular release, are required for patients in the frozen phase or who fail conservative intervention.24,27,35 Nevertheless, owing to its unknown pathology, a consensus has not been reached regarding the optimal treatment or the proper timing of these procedures.
The joint capsule and CHL thickening are recognized as common pathologies in FS.32 Thickened inferior capsule reduces flexion and abduction of the shoulder joint. The CHL covered the supraspinatus and subscapularis muscles and tendons,5 the thickened CHL around the subscapularis leads to loss of external rotation (ER), and the thickened ligament around the supraspinatus develops adduction restriction (AR) of the GH joint and limits internal rotation (IR) to the back.1,10,17,18 Abduction contracture has been described as an uncommon condition associated with deltoid contracture and brachial plexus palsy.2,7 We use the term AR rather than abduction contracture to distinguish this pathology from deltoid contracture. AR has not been reported as a pathological condition of FS, being expected to frequently occur in FS because it is found in 75% of degenerative rotator cuff tears with a thickened superior capsule and decreased elasticity of the supraspinatus muscle. We devised an adduction test to assess AR of the GH joint, performing adduction manipulation to treat the condition (Fig. 1, A-C).40 Investigating the incidence and severity of AR in the freezing and frozen phases and how AR severity affects clinical courses and outcomes would provide us beneficial information for the treatment of FS. This study aimed to investigate the percentages of AR, to calculate radiographically glenohumeral adduction angle (GAA), and to compare clinical characteristics and outcomes in 4 groups with positive and negative AR in the freezing and frozen phases.
Figure 1.
Adduction test of the glenohumeral joint. (A) Stating position of adduction test. The participants were tested in the lateral decubitus position on the examination table. One examiner abducted the humerus up to 90°-120°. The upward-rotated scapula was fixed with both hands by another examiner. (B) Negative adduction test. When the upper arm touched the side easily, without causing pain, the test results were negative. This showed no adduction restriction of the glenohumeral joint. (C) Positive adduction test. The adduction test was considered positive when the upper arm was pushed gently toward the side in the coronal plane and the upper arm did not touch the side with pain.
Materials and methods
Enrollment of patients
Permission was obtained from the International Review Board of the authors' institute (K-2019-04) to conduct the study. We enrolled 216 patients in our hospital with shoulder pain and active and passive ROM limitation with unknown etiology, normal radiographs of the shoulder joint, and no rotator cuff tears detected on MRI (Echelon RX, 1.5T; Hitachi, Tokyo, Japan).42 Exclusion criteria were prior intervention such as physiotherapy or intra-articular injection (n = 18), nerve compression in the cervical spine (n = 1), or bilateral shoulder joint involvement (n = 13). One hundred eighty-four patients were classified into the freezing and frozen phases, the frozen phase was defined as 3 or more of the following: flexion < 100°, abduction < 90°, ER < 10°, IR < the fifth lumbar vertebra.15 Patients who did not meet the criteria of the frozen phase were classified into the freezing phase. One hundred twenty-six patients (68.5%) were diagnosed in the freezing phase, and 58 patients (31.5%) were diagnosed in the frozen phase. In addition, patients in both phases were divided into 2 groups with and without AR of the GH joint using the adduction test. Of 126 patients, 80 (63.5%) were classified into the negative adduction test group (the upper arm touched at the side of the thorax through adduction test), and 46 (36.5%) were in the positive adduction test group (the upper arm did not touch at the side with pain through adduction test). Fifty-eight patients in the frozen phase were classified into a negative adduction test group (8; 13.8%) and a positive adduction test group (50; 86.2%; Fig. 2).
Figure 2.
Flow diagram and classification of 4 groups. ROM, range of motion; ER, external rotation; IR, internal rotation; L5, fifth lumbar vertebra.
Measurement of radiographic GAA
At the initial visit and after the treatments, the radiographic GAA of bilateral shoulders in all patients was calculated using the previously reported method.40 Adduction of the GH joint was designated “plus,” a parallel of the glenoid line and the humeral line was determined as “zero,” and abduction of the GH joint was defined as “minus” (Fig. 3, A and B). The GAAs of all patients with FS were compared with that of normal subjects, whose clinical characteristics and GAAs had published in our previous study: 30 people in their 20s (5 men and 5 women, mean age, 25.4 years), 50s (5 men and 5 women, mean age, 56.1 years), and 70s (5 men and 5 women, mean age, 73.8 years). They had no history of shoulder pain and trauma, no ROM restriction of bilateral shoulder joints, negative impingement, instability, and scapular dyskinesia test, normal X-rays, and no rotator cuff tears on MRI.40
Figure 3.
Radiological glenohumeral adduction angle. Adduction of the glenohumeral joint was designated as “plus,” a parallel of the glenoid line, and humeral line was determined as “zero,” and abduction of the glenohumeral joint was defined as “minus.” (A) Glenohumeral adduction angle (GAA) in a normal subject. The GAA in a normal subject showed 16.5° of adduction in the glenohumeral joint. (B) GAA in a patient with frozen shoulder and adduction restriction. The GAA was −24.7°, which meant 24.7° of abduction in the glenohumeral joint.
Treatment options for the frozen phase
Fifty-eight patients diagnosed in the frozen phase underwent joint manipulation under brachial plexus block within 1 month of the initial examination. A brachial plexus block was performed by an anesthesiologist using 20 mL of 1% mepivacaine hydrochloride. Before manipulation, 20 mg of triamcinolone and 10 mL of 1% lidocaine hydrochloride were injected into the GH joint. The manipulation procedure started with adduction manipulation of the GH joint to eliminate superior capsular stiffness and CHL tightness and restore supraspinatus movement (Fig. 1, A and B). This adduction procedure was repeated 3 times. Next, the patient changed their position from lateral to supine. The examiner moved the affected extremity to maximal ER at the side, flexion, and ER and IR at 90° of abduction. Subsequently, the patients changed their position from supine to lateral. IR to the back was performed as the thumb reached the highest thoracic vertebral level. Finally, the humeral head was pushed posteriorly, and the elbow was pushed anteriorly while maintaining the IR position.27 After joint manipulation, an MRI was taken to observe ruptured soft tissues and complications, such as bone bruise, fracture, and labral tear.
After manipulation, physical therapists treated the patients in the frozen phase with physiotherapy for 20 min twice a week and prescribed self-exercise. The rehabilitation programs began with patient education and assessment to provide information on FS and help patients recognize their incorrect postures and painful motions. The physical therapists measured the ROM of both shoulders; examined which muscles were tight and presented with tenderness; investigated thoracic movements, including the thoracic spine, clavicle, and ribs; and evaluated passive scapular movements.8 Based on the concept that bone movements (spine, ribs, clavicle, scapula, and humerus) are essential for shoulder motion,33 the bone movements were improved using mobilization of the costovertebral, sternocostal, sternoclavicular, and acromioclavicular joints.30 Massage was used to relax tight muscles in the shoulder, including rotator cuff muscles, deltoid, pectorals major and minor, latissimus dorsi, teres major, and biceps and triceps brachii muscles.14 With increasing shoulder motion, passive and active ROM exercises such as flexion, ER, IR, and horizontal flexion were initiated. Finally, an isometric strengthening exercise of the rotator cuff muscles was initiated when every ROM reached 80% of that in the contralateral normal shoulder.
The goals of the treatment were set as a visual analog scale (VAS) score <1.0 in activities of daily living, and the American Shoulder and Elbow Surgeons (ASES) score restored up to 90 points. Patients who reached the goals were discharged from the structured physiotherapy intervention. Subsequently, they visited the hospital every 3 months to examine whether the pain persisted and if the restored ROM was maintained over the 24-month follow-up period.
Treatment options for the freezing phase
All patients in the freezing phase underwent an injection containing 10 mL of 1% lidocaine hydrochloride and 6 mg of triamcinolone: 4 mL in the rotator interval, 3 mL in the subacromial bursa, and 3 mL in the GH joint within 2 weeks after the initial visit. Patients with positive adduction test results underwent adduction manipulation of the GH joint under the local anesthesia, whereas those with negative adduction tests did not. Physiotherapy, the same as the frozen phase, was initiated after the injection and/or adduction manipulation. The treatment goal was set the same as that in the frozen phase. The patients who were resistant to the conservative treatments and whose ROM worsened to the same level as the criteria of the frozen phase underwent joint manipulation.
Outcome assessment
Age, sex, dominant arm, symptom and treatment durations, and body mass index (BMI) were recorded as baseline characteristics. Passive ROM of the affected and unaffected shoulders (flexion, abduction, ER, and IR) were measured in the sitting position using a goniometer. The VAS pain, ASES score, Constant score, and EuroQol-visual analog scale (EQ-VAS) were also assessed at the initial examination and at the 1-, 3-, 6-, 12-, and 24-month follow-up appointments.9 The treatment effectiveness and radiological GAA were calculated and compared among the 4 treatment groups: the negative and positive adduction test groups (groups 1 and 2) in the freezing phase and the negative and positive adduction test groups (groups 3 and 4) in the frozen phase.
Statistics
Data are described as mean (95% confidence interval) or median (interquartile range) values. EZR (Easy R EZR, Saitama Medical Center, Jichi Medical University, Saitama, Japan) on R Commander version 1.52 (CRAN, Vienna, Austria), a free statistical software, was used for statistical analysis.16 A two-sample test for equality of proportions without continuity correction was performed on clinical characteristics at the initial visit. A Welch Two-Sample t-test was conducted on the data in both the freezing and frozen phases at the initial visit. The IR index of the affected side was determined using the Wilcoxon rank-sum test. Multiple regression analysis was conducted with the GAA as the objective variable and VAS, EQ-VAS, ROM, and clinical items as explanatory variables. Repeated analysis of variance for the split-plot factorial design was used to compare the data from the first visit to 24 months in the negative and positive adduction tests in both phases. Multiple comparison tests were used to determine the significant differences in the measurement items. All tests were conducted as 2-sided tests, and P < .05 represented statistical significance.
Results
Grouping using freezing and frozen phases and adduction test
One hundred eighty-four patients with FS were divided into 126 (68.5%) in the freezing phase and 58 (31.5%) in the frozen phase based on ROM measurement. AR of the GH joint was present in 46 of 126 (36.5%) in the freezing phase and 50 of 58 (86.2%) in the frozen phase; the percentage of AR in the frozen phase was 2.3 times higher than that in the freezing phase. Two patients who discontinued the intervention and 6 patients who were lost during follow-up were excluded from the present study. Seventy-five patients showed a negative adduction test (group 1), and 45 patients showed a positive adduction test (group 2) in the freezing phase. Eight patients demonstrated a negative adduction test (group 3), and 48 had a positive adduction test (group 4) in the frozen phase (Fig. 2). Sex, percentage of positive adduction test, and the clinical items such as treatment duration, GAA, VAS, EQ-VAS, ROMs, and ASES and Constant scores at the baseline showed significant differences between the freezing and frozen phases (P < .001; Table I).
Table I.
Patients' characteristics of freezing and frozen phases.
| Freezing phase (n = 120) | Frozen phase (n = 56) | P value | |
|---|---|---|---|
| Age, yr | 58.0 (56.5, 59.5)∗ | 55.4 (53.1, 57.7)∗ | .057 |
| Sex, male, n (%) | 37 (30.8) | 29 (51.8) | .007† |
| BMI (kg/m2) | 22.8 (22.2, 23.4)∗ | 22.8 (21.8, 23.9)∗ | .989 |
| Side, right | 64 (53.3) | 34 (60.7) | .359 |
| Dominance, right (%) | 114 (95.0) | 54 (96.4) | .671 |
| DM or thyroid disease, n (%) | 12 (10.0) | 10 (17.9) | .142 |
| Adduction test, positive (%) | 45 (37.5) | 48 (85.7) | <.001‡ |
| Symptom duration (M) | 5.1 (4.4, 5.9)∗ | 5.4 (4.4, 6.4)∗ | .681 |
| Treatment duration (M) | 6.0 (5.2, 6.9)∗ | 10.3 (9.1, 11.6)∗ | <.001‡ |
| GAA (degree) | −3.0 (−4.8, −1.3)∗ | −18.3 (−22.0, −14.6)∗ | <.001‡ |
| VAS (mm) | 58.6 (55.1, 62.2)∗ | 70.5 (65.3, 75.8)∗ | <.001‡ |
| EQ-VAS (mm) | 49.7 (46.7, 52.6)∗ | 34.0 (30.3, 37.7)∗ | <.001‡ |
| Flexion (°) | 141.2 (137.9, 144.5)∗ | 100.6 (95.3, 106.0)∗ | <.001‡ |
| Abduction (°) | 121.1 (115.9, 126.3)∗ | 80.09 (74.4, 85.8)∗ | <.001‡ |
| ER (°) | 48.0 (44.6, 51.3)∗ | 14.8 (9.24, 20.4)∗ | <.001‡ |
| IR (vertebral level) | 7, T12 (4, 10)†,‡ | 1, S1 (0, 3)†,‡ | <.001‡ |
| ASES score | 55.5 (53.0, 58.0)∗ | 39.8 (35.7, 44.0)∗ | <.001‡ |
| Constant score | 59.7 (57.5, 61.9)∗ | 42.8 (39.1, 46.6)∗ | <.001‡ |
BMI, bone mass index; DM, diabetes mellitus; M, month; GAA, glenohumeral adduction angle; VAS, visual analog scale; EQ-VAS, EuroQol-visual analog scale; ER, external rotation; IR, internal rotation; ASES, American Shoulder Elbow Surgeons.
Values are shown as the mean (95% CI).
P < .01.
P < .001 represents significant differences.
Glenohumeral adduction angle in 4 groups and the relationship between GAA and ROM
The GAAs of normal shoulders and patients in the 4 groups are summarized in Table II. The mean GAA of group 1 was 2.0° (unaffected side, 1.7°) at the initial visit and increased to 4.3° after the treatments. Compared with group 1, the GAA in group 2 (−11.3°) was significantly smaller (P < .001) and increased up to 0.6° (P < .05) after the treatment. The mean GAA in group 3 in the frozen phase was −0.4° (unaffected side, 3.3°) at the initial visit and 1.9° after the intervention. The mean GAA in group 4 was −21.3° (unaffected side, 2.2°) at the initial visit and −3.5° after the treatment. The mean GAA in normal subjects was 5.1° in their 20s (20 shoulders), 4.5° in their 50s (20 shoulders), and 1.6° in their 70s (20 shoulders), which were larger than the GAAs of the affected shoulders in the freezing and frozen phases.
Table II.
Glenohumeral adduction angle in normal subjects and patients with frozen shoulder.
| Normal subjects |
Freezing phase (n = 120) |
Frozen phase (n = 56) |
|||||
|---|---|---|---|---|---|---|---|
| Adduction test |
Adduction test |
||||||
| 20 yr | 50 yr | 70 yr | Negative | Positive | Negative | Positive | |
| Group | Group 1 | Group 2 | Group 3 | Group 4 | |||
| n (Shoulder) | 20 | 20 | 20 | 75 | 45 | 8 | 48 |
| GAA | |||||||
| Unaffected side (°) | 5.1 | 4.5 | 1.6 | 1.7 | 5.1 | 3.3 | 2.2 |
| Affected side (°) | - | - | - | 2.0∗ | −11.3†,‡ | −0.4 | −21.3§ |
| After treatment (°) | 4.3 | 0.6 | 1.9 | −3.5 | |||
GAA, glenohumeral adduction angle.
Normal subject data are cited from reference40.
P < .001 Comparison of GAA of the affected side at the initial visit between Groups 1 and 2.
P < .001 Comparison of GAA of the affected side at the initial visit between Groups 2 and 4.
P < .001 Comparison of GAA in Group 2 at the initial visit and after treatment.
P < .001 Comparison of GAA in Group 4 at the initial visit and after treatment.
GAA was correlated with only ER ROM. The standard regression coefficients between the GAA and ROMs were 0.16 in flexion, −0.10 in abduction, 0.56 in ER, and 0.10 in IR. The odds ratio of GAA was 0.719 and 0.83 in ER, and the discrimination accuracy rate from ER ROM to GAA was 89.2%. The relative equation between GAA and ER was expressed as follows: ER = 1.152 × GAA + 46.197.
Clinical outcomes in the freezing phase
The baseline characteristics of patients in the freezing phase, who were divided into groups 1 and 2 based on the adduction test, were summarized in Table III. Group 1 (negative adduction test) comprised 75 patients, with a mean age of 57.5 years; 24% of them were men, their BMI was 22.7, 6.7% had diabetes or thyroid disease, and the mean symptom duration was 4.9 months. Group 2 (positive adduction test) consisted of 45 patients, with a mean age of 58.8 years; 42.2% of them were men, their BMI was calculated at 23.1, 15.6% had diabetes or thyroid disease, and the mean duration of symptoms was 5.6 months. Only the proportion of men was significantly different (P = .036). The mean treatment period of the freezing phase was 5.2 months in group 1 and 7.4 months in group 2 (P = .019). Although there were no significant differences in clinical items except ER at the initial examination, ER (51°) in group 1 was greater than that (43°) in group 2 (P < .05). All clinical items, except ER, from the initial visit to the 24-month follow-up period in group 1 were not statistically different from those in group 2 (Fig. 4). Twelve (10%) of 120 patients, who were refractory to conservative treatments and their ROM loss worsened up to same level of the frozen phase, consequently, underwent joint manipulation: 4 of 75 shoulders (5.3%) in group 1 and 8 of 45 shoulders (17.8%) in group 2 (P = .028; Table III). At the final follow-up period, none of the clinical items except ER showed statistical differences between groups 1 and 2.
Table III.
Patients' characteristics of positive and negative adduction test in the freezing phase.
| Group 1; Add test negative (n = 75, 62.5%) | Group 2; Add test positive (n = 45, 37.5%) | P value | |
|---|---|---|---|
| Age, yr | 57.5 (55.5, 59.5)∗ | 58.8 (56.5, 61.2)∗ | .398 |
| Sex, male, n (%) | 18 (24.0) | 19 (42.2) | .036† |
| BMI (kg/m2) | 22.7 (21.9, 23.4)∗ | 23.1 (22.1, 24.0)∗ | .507 |
| GAA (°) | 2.0 (0.6, 3.5)∗ | −11.3 (−13.1, −9.4)∗ | <.001‡ |
| DM or thyroid disease, n (%) | 5 (6.7) | 7 (15.6) | .116 |
| Symptom duration (M) | 4.9 (4.0, 5.7)∗ | 5.6 (4.3, 6.9)∗ | .350 |
| Treatment duration (M) | 5.2 (4.4, 6.1)∗ | 7.4 (5.8, 9.0)∗ | .019† |
| Translation to MUA, n (%) | 4 (5.3) | 8 (17.8) | .028† |
Add, adduction; BMI, bone mass index; M, month; GAA, glenohumeral adduction angle; DM, diabetes mellitus; MUA, manipulation under anesthesia.
Values are shown as the mean (95% CI).
P < .05.
P < .001 represent significant differences.
Figure 4.
Comparison of clinical items between freezing and frozen phases for 24 months. Clinical items except external rotation did not show statistical differences from the initial visit to the 24-month follow-up period. VAS, visual analog scale; EQ-VAS, EuroQol-visual analog scale; ASES, American Shoulder Elbow Surgeons score; Constant, Constant score.
Clinical outcomes in the frozen phase
The baseline characteristics of patients in the frozen phase, who were divided into groups 3 and 4 using the adduction test, were summarized in Table IV. Group 3 (negative adduction test) comprised 8 patients, with a mean age of 54.8 years; 75% of them were men, a median BMI was calculated at 23.9, 12.5% of them had diabetes or thyroid disease, and the mean duration of symptoms was 4.8 months. Group 2 (positive adduction test) consisted of 48 patients, with a mean age of 55.5 years, 35.7% of them were men, their BMI was 22.7, 12.5% of them had diabetes or thyroid disease, and the mean symptom duration was 5.5 months. Age, sex, BMI, percentage of diabetes or thyroid disease, symptom duration, VAS, EQ-VAS, ROM except ER, and ASES, and Constant scores at the initial visit did not show significant differences between groups 3 and 4 (Table IV and Fig. 4). ER was greater in group 3 than in group 4 at all follow-up periods. The mean treatment periods in groups 3 and 4 were almost the same (10.8 months vs. 10.3 months; P = .817). Although the complete rupture of the inferior joint capsule of all patients in group 4 was observed with MRI after joint manipulations, the soft tissues in group 3 showed no or incomplete rupture because the upper arm did not reach the maximum ROM due to periscapular muscles' tightness. No improvement in VAS, EQ-VAS, and ROM after the joint manipulation was observed in 10 of 56 patients (17.9%) in the frozen phase, 4 of 8 patients (50%) in group 3, and 6 of 48 (12.5%) in group 4 (P = .01); however, clinical items in such patients started to improve 6 months after the joint manipulation (Table IV). Clinical outcomes, except ER, showed no statistical differences between the 2 groups at every follow-up appointment (Fig. 4).
Table IV.
Patients' characteristics of positive and negative adduction test in frozen phase.
| Group 3; Add test negative (n = 8, 14.3%) | Group 4; Add test positive (n = 48, 85.7%) | P value | |
|---|---|---|---|
| Age, yr | 54.8 (48.8, 60.7)∗ | 55.5 (52.9, 58.1)∗ | .806 |
| Sex, male, n (%) | 6 (75.0) | 20 (35.7) | .156 |
| BMI (kg/m2) | 23.9 (22.0, 25.8)∗ | 22.7 (21.4, 23.9)∗ | .236 |
| GAA (degree) | −0.4 (−7.3, 6.5)∗ | −21.3 (−24.8, −17.7)∗ | <.001‡ |
| DM or thyroid disease, n (%) | 1 (12.5) | 9 (12.5) | .669 |
| Symptom duration (M) | 4.8 (2.7, 7.1)∗ | 5.5 (4.3, 6.7)∗ | .586 |
| Treatment duration (M) | 10.8 (6.1, 15.5)∗ | 10.3 (9.0, 11.5)∗ | .817 |
| Translation to MUA, n (%) | 8 (100) | 48 (100) | |
| No improvement after MUA, n (%) | 4 (50.0) | 6 (12.5) | .010† |
Add, adduction; BMI, bone mass index; M, month; GAA, glenohumeral adduction angle; DM, diabetes mellitus; MUA, manipulation under anesthesia.
Values are shown as the mean (95% CI).
P < .05.
P < .001 represents significant differences.
Discussion
The present study has 3 important findings. First, the percentage of AR in the frozen phase (85.7%) was 2.3 times higher than that in the freezing phase (37.5%), and the severity of the restriction using the GAA developed from −3.0° in the freezing phase to −18.3° in the frozen phase. The ER in groups 1 and 3 without AR was well preserved during the initial examination, and the follow-up period compared with groups 2 and 4 with AR. Second, the mean treatment period in group 1 (5.2 months) was shorter compared with that in group 2 (7.4 months), and the percentage of transitioning to joint manipulation in group 2 (17.8%) was 3 times higher than in group 1 (5.3%). Third, complete rupture of the joint capsule in group 4 was demonstrated with MRI after joint manipulation; however, the soft tissues in group 3 showed no or incomplete rupture. No improvement of ROM after the joint manipulation was confirmed in 4 of 8 patients (50%) in group 3, which was 4 times more than 6 of 48 (12.5%) in group 4. To summarize the results, preserving ER and a negative adduction test suggest low severity of intra-articular lesions (maintained capsular elasticity) in the freezing and frozen phases. In the negative AR group of the frozen phase, the cases showing no immediate efficacy of joint manipulation increased compared with those in the positive AR group.
Accurate evaluation of the pathological condition in every patient with FS is important for selecting appropriate treatments; however, AR of the GH joint has not been reported as a pathology. AR was not identified in 30 healthy subjects, their mean GAA was 4.8°, and GAA gradually decreased with age.40 In the present study, the frequency of AR increased 2.3 times from the freezing phase to the frozen phase. The average GAA in group 4 (−21.3°) was 1.9 times that in group 2 (−11.3°). Interestingly, the average GAA (−21.8°) in the frozen phase is very similar to that (−21.4°) of the degenerative rotator cuff tears.40 Aging and abnormal neovascular formation, followed by inflammatory processes, may be common causes for the development of AR in FS and degenerative rotator cuff tears.17,28 A decrease in the GAA indicates a shortening of the distance from the supraspinatus attachment on the greater tuberosity to the superior rim of the scapular cavity. Tightness of the supraspinatus muscle and tendon, thickening of the CHL, and proliferation of the superior capsule and subacromial bursa are likely to relate to AR development. Two studies measured the elasticity of the supraspinatus, infraspinatus, and CHL in FS. One study reported that the shear wave elastography (SWE) values of the supraspinatus and infraspinatus tendons increased in the freezing phase, and those of the CHL also increased in the frozen phase.37 Another study documented that the SWE of the CHL of the affected shoulder in the freezing and frozen phase was higher than that of the unaffected side.41 A thicker and stiffer CHL develops tightness of the supraspinatus during the freezing and frozen phases. Hence, the results of these studies using SWE support our speculation concerning the development of AR from the freezing to the frozen phases.
GAA measurement and the adduction test evaluate the stiffness of the superior portion of the GH joint; however, it is unclear whether the stiffness affects what kind of ROM loss. During arthroscopic surgery, the anterior, inferior, and superior joint capsules are released, and the rotator interval and subacromial bursa are eliminated simultaneously. A thickened inferior capsule reduces flexion and abduction, stiffness of the anterior capsule, including the middle GH ligament and rotator interval, limits ER, and posterior capsular thickening causes IR limitation at the side. We searched for superior capsule stiffness in FS, but no literature was found. AR appears to limit all ROM from the results of ROM measurement and multiple regression analysis. Furthermore, AR is presumed to affect IR loss to the back because adduction of the GH joint is needed when we raise our hands high behind the back. We reported 3 cases of FS with residual IR loss to the back after joint manipulation, in which arthroscopic findings revealed thickened CHL and superomedial capsule, and eliminating the soft tissues improved their IR over the seventh vertebra.18 AR progression certainly reduces the level of hand reaching the back.
What is the mechanism for the positive correlation between GAA and ER? The correlation indicates that the stiffness of the CHL around the supraspinatus and subscapularis is presumed to develop simultaneously. In the freezing phase, ER in group 1 without AR was 51°, while 43° in group 2 with AR. A similar tendency was observed in the frozen phase: ER was 37° in group 3 without AR and 11° in group 4 with AR. In other words, the groups with AR tended to have a smaller ER. The anterosuperior capsule and CHL in the rotator interval affect the tension with ER motion of the GH joint.10 The CHL envelops both supraspinatus and subscapularis tendons,1,5 and the stiffness of the former portion can be evaluated with the adduction test. The thickened latter portion of the CHL limits ER. Therefore, the thickening of the CHL in the supraspinatus and subscapularis is presumed to progress simultaneously and to develop intra-articular pathologies. The entire CHL release with arthroscopic capsular release can promote smooth movement of the supraspinatus and infraspinatus tendons around the base of the coracoid process and glenoid neck. The release of the CHL under the coracoid process can promote smooth movement of the subscapularis tendon.12 The entire CHL release contributes to greater restoration of ER and IR than in patients without the entire CHL release.13 These findings support our understanding concerning the positive correlation between GAA and ER.
The clinical significance of the positive correlation between GAA and ER ROM is that it can predict the severity of intra-articular lesions. In the freezing phase, the treatment duration in group 1 (5.2 months) was shorter than in group 2 (7.4 months), and the percentage of patients requiring joint manipulation was 4 of 75 patients in group 1 (5.3%) and increased 3 times more to 8 of 45 in group 2 (17.8%). These results can be explained by the incidence of diabetes or thyroid disease and the presence of AR. Diabetes or hypothyroidism is known to reduce clinical outcomes and requires a longer treatment period in FS.42 Severe AR and ER loss mean the development of intra-articular lesions in group 2 than in group 1; therefore, treatment periods became longer, and the transition of joint manipulations was more frequent in group 2 compared with group 1. In the frozen phase, the MRI findings revealed no or incomplete rupture of the inferior joint capsule in group 3 without AR, despite complete rupture in group 4 with AR. We understand that the elasticity of the inferior capsule in group 3 may be maintained, and the capsule in group 4 was thickened and stiff. The treatment periods in groups 3 (10.8 months) and 4 (10.4 months) were the same. The reason is that extra-articular lesions in group 3 may be more predominant than intra-articular lesions, intra-articular lesions in group 4 were eliminated by joint manipulation, and only extra-articular pathologies remained. As described above, limited shoulder elevation depends on the thickened inferior joint capsule, ER loss is due to stiff CHL of the subscapularis portion and rotator interval, and IR deficit is caused by thickening of the posterior capsule. Causes of ROM restriction in FS are difficult to determine because of the presence of both intra- and extra-articular lesions. MRI is useful for observing the thickened joint capsule and CHL; however, the stiffness of these soft tissues cannot be assessed.32 Despite severe ROM restriction in the frozen phase, preservation of ER ROM and a negative adduction test is presumed to indicate mild intra-articular lesions (maintaining capsular elasticity) and more predominant extra-articular pathologies. Steroid injections and physiotherapy should be given priority in such patients; conversely, joint manipulation or arthroscopic capsular release is recommended for patients demonstrating ER < 10° and a positive adduction test.
Ten of 56 patients (17.9%) in the frozen phase did not restore their ROM early after the joint manipulations: 4 of 8 shoulders (50%) in group 3 and 6 of 48 shoulders (12.5%) in group 4. The incidence of no early restoration with joint manipulation in this study is consistent with those in previous studies: in 47 of 246 shoulders (19.1%) by Thomas et al and 141 of 792 shoulders (17.8%) by Woods et al.34,39 We treated these patients with improvement of the thoracic spine, ribs, and scapula movement and relaxation of the pectoralis major, latissimus dorsi, teres major, biceps brachii, triceps, and rotator cuff muscles; however, physiotherapy could not restore the ROM.14,19,37 Interestingly, their ROM began to recover 6 months after joint manipulation. To our knowledge, there have been no studies involving patients who fail to improve after joint manipulation, nor have their causes been clarified. Since intra-articular lesions were eliminated through joint manipulation, extra-articular pathologies were improved with physiotherapy, and restoration of ROM started 6 months after the joint manipulation, we presume that the rotator cuff tendon tightness would delay ROM improvement. Some patients without AR in the frozen phase tend to take an unexpected treatment course even though they undergo joint manipulation or arthroscopic capsular release. Another clinical significance of the adduction test might be possible to select patients for whom such treatments do not immediately respond.
The present study has some limitations. First, the pathology underlying AR was not fully delineated. Therefore, we need to macroscopically and microscopically observe the upper part of the GH joint and subacromial bursa in an anatomic cadaver study. Second, we enrolled 200 patients, but the number of patients in group 3 (n = 8) was too small for data analysis. Four hundred patients should be required to collect accurate data to compare the freezing and frozen phases. Third, not intentionally, the percentage of men was 30.8% in the freezing phase and 51.8% in the frozen phase and 24% in group 1 and 42.2% in group 2. The higher proportion of women in the freezing phase and group 1 may have influenced the clinical outcomes because the incidence of diabetes is lower in women than in men. Finally, the classification of the freezing and frozen phases depended on the degree of ROM loss. We used the classification that the frozen phase was defined as 3 or more of the following ROM limitations: flexion <100°, abduction <90°, ER < 10°, IR < the fifth vertebra, and patients who demonstrated better ROM were classified into the freezing phase. If the classification methods had been different, the findings of the present study may have changed. The degree of ROM limitations would be a better classification criterion because MRI alone may not detect the phases of FS.32
Conclusions
The incidence and severity of AR increase from the freezing phase to the frozen phase. The clinical significance of the positive correlation between GAA and ER is that it can predict the progression of intra-articular lesions. The decrease of GAA and loss of ER in the freezing phase prolong the treatment duration and increase the joint manipulation rate. Physiotherapy is the first choice for patients in the frozen phase with ER preservation and a negative adduction test. Conversely, joint manipulation or arthroscopic capsular release is recommended for those demonstrating severe ER restriction and a positive adduction test.
Disclaimers
Funding: No funding was disclosed by the authors.
Conflicts of interest: The authors, their immediate families, and any research foundations with which they are affiliated did not receive any financial payments or other benefits from any commercial entity related to the subject of this article.
Footnotes
The study was approved by the Internal Review Board of Kuwano Kyoritsu Hospital (K-2019-04).
References
- 1.Arai R., Nimura A., Yamaguchi K., Yoshimura H., Sugaya H., Saji T., et al. The Anatomy of the coracohumeral ligament and its relation to the subscapularis muscle. J Shoulder Elbow Surg. 2014;23:1575–1581. doi: 10.1016/j.jse.2014.02.009. [DOI] [PubMed] [Google Scholar]
- 2.Banerji D., De C., Pal A.K., Das S.K., Ghosh S., Dharmadevan S. Deltoid contracture: a study of nineteen cases. Indian J Orthop. 2008;42:188–191. doi: 10.4103/0019-5413.40256. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Binder A.I., Bulgen D.Y., Hazleman B.L., Roberts S. Frozen shoulder. A long-term prospective study. Ann Rheum Dis. 1984;43:361–364. doi: 10.1136/ard.43.3.361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bunker T.D. Frozen shoulder: unravelling the enigma. Ann R Coll Surg Engl. 1997;79:210–213. [PMC free article] [PubMed] [Google Scholar]
- 5.Clark J.M., Harryman D.T., 2nd Tendons, ligaments, and capsule of the rotator cuff. Gross and microscopic anatomy. J Bone Joint Surg Am. 1992;74:713–725. [PubMed] [Google Scholar]
- 6.De Sire A., Agostini F., Bernetti A., Mangone M., Ruggiero M., Dinatale S., et al. Non-surgical and rehabilitative interventions in patients with frozen shoulder: umbrella review of systematic reviews. J Pain Res. 2022;15:2449–2464. doi: 10.2147/JPR.S371513. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Eismann E.A., Little K.J., Laor T., Cornwall R. Glenohumeral abduction contracture in children with unresolved neonatal brachial plexus palsy. J Bone Joint Surg Am. 2015;9:112–118. doi: 10.2106/JBJS.N.00203. [DOI] [PubMed] [Google Scholar]
- 8.Endo K., Hamada J., Suzuki K., Hagiwara Y., Muraki T., Karasuno H. Does scapular motion regress with aging and is it restricted in patients with idiopathic frozen shoulder? Open Orthop J. 2016;10:80–88. doi: 10.2174/1874325001610010067. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.EQ-5D-5L/About. https://euroqol.org/eq-5d-instruments/eq-5d-5l-about Available at:
- 10.Ferrari D.A. Capsular ligaments of the shoulder. Anatomical and functional study of the anterosuperior capsule. Am J Sports Med. 1990;18:20–24. doi: 10.1177/036354659001800103. [DOI] [PubMed] [Google Scholar]
- 11.Grey R.G. The natural history of “idiopathic” frozen shoulder. J Bone Joint Surg Am. 1978;60:564. [PubMed] [Google Scholar]
- 12.Hagiwara Y., Ando A., Kanazawa K., Koide M., Sekiguchi T., Hamada J., et al. Arthroscopic coracohumeral ligament release for patients with frozen shoulder. Arthrosc Tech. 2017;7:e1–e5. doi: 10.1016/j.eats.2017.07.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hagiwara Y., Kanazawa K., Ando A., Sekiguchi T., Yabe Y., Takahashi M., et al. Clinical outcomes of arthroscopic pan-capsular release with or without entire coracohumeral ligament release for patients with frozen shoulder. JSES Int. 2020;4:826–832. doi: 10.1016/j.jseint.2020.08.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hollmann L., Halaki M., Kamper S.J., Haber M., Ginn K.A. Does muscle guarding play a role in range of motion loss in patients with frozen shoulder? Musculoskelet Sci Pract. 2018;37:64–68. doi: 10.1016/j.msksp.2018.07.001. [DOI] [PubMed] [Google Scholar]
- 15.Itoi E., Arce G., Bain G., Diercks R., Guttman D., Imhoff A.B., et al. Shoulder stiffness: current concepts and concerns. Arthroscopy. 2016;32:1402–1414. doi: 10.1016/j.arthro.2016.03.024. [DOI] [PubMed] [Google Scholar]
- 16.Kanda Y. Investigation of the freely available easy-to-use software ‘EZR’ for medical statistics. Bone Marrow Transpl. 2013;48:452–458. doi: 10.1038/bmt.2012.244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Karasuno H., Hamada J., Yano Y., Tsutsui H., Hagiwara Y., Endo K., et al. Adduction manipulation of the glenohumeral joint versus physiotherapy for atraumatic rotator cuff tears: a randomized controlled trial. J Clin Med. 2023;12:4167. doi: 10.3390/jcm12124167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Koide M., Hamada J., Hagiwara Y., Kanazawa K., Suzuki K. A thickened coracohumeral ligament and superomedial capsule limit internal rotation of the shoulder joint: report of three cases. Case Rep Orthop. 2016;2016 doi: 10.1155/2016/9384974. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Kurashima W., Iijima Y., Sasanuma H., Saito T., Takeshita K. Evaluation of muscle stiffness in adhesive capsulitis with Myoton PRO. JSES Int. 2022;7:25–29. doi: 10.1016/j.jseint.2022.08.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lowe C.M., Barrett E., McCreesh k, Burca N.D., Lewis J. Clinical effectiveness of non-surgical interventions for primary frozen shoulder: a systematic review. J Rehabil Med. 2019;51:539–556. doi: 10.2340/16501977-2578. [DOI] [PubMed] [Google Scholar]
- 21.Okuno Y., Oguro S., Iwamoto W., Miyamoto T., Ikegami H., Matsumura N. Short-term results of transcatheter arterial embolization for abnormal neovessels in patients with adhesive capsulitis: a pilot study. J Shoulder Elbow Surg. 2014;23:e199–e206. doi: 10.1016/j.jse.2013.12.014. [DOI] [PubMed] [Google Scholar]
- 22.Pal B., Anderson J., Dick W.C., Griffiths I.D. Limitation of joint mobility and shoulder capsulitis in insulin- and non-insulin-dependent diabetes mellitus. Br J Rheumatol. 1986;24:147–151. doi: 10.1093/rheumatology/25.2.147. [DOI] [PubMed] [Google Scholar]
- 23.Picasso R., Pistoia F., Zaottini F., Marcenaro G., Miguel-Pérez M., Tagliafico A.S., et al. Adhesive capsulitis of the shoulder: current concepts on the diagnostic work-up and evidence-based protocol for radiological evaluation. Diagnostics (Basel) 2023;13:3410. doi: 10.3390/diagnostics13223410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Rangan A., Brealey S.D., Keding A., Corbacho B., Northgraves M., Kottam L., et al. Management of adults with primary frozen shoulder in secondary care (UK FROST): a multicentre, pragmatic, three-arm, superiority randomised clinical trial. Lancet. 2020;396:977–989. doi: 10.1016/S0140-6736(20)31965-6. [DOI] [PubMed] [Google Scholar]
- 25.Reeves B. The natural history of the frozen shoulder syndrome. Scand J Rheumatol. 1975;4:193–196. doi: 10.3109/03009747509165255. [DOI] [PubMed] [Google Scholar]
- 26.Ryan V., Brown H., Minns Lowe C.J., Lewis J.S. The pathophysiology associated with primary (idiopathic) frozen shoulder: a systematic review. BMC Musculoskelet Disord. 2016;17:340. doi: 10.1186/s12891-016-1190-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Saito T., Hamada J., Sasanuma H., Iijima Y., Saitsu A., Minagawa H., et al. Clinical outcomes and cost-effectiveness of manipulation under brachial plexus block versus physiotherapy for refractory frozen shoulder: a prospective observational study. JSES Int. 2023;7:2410–2419. doi: 10.1016/j.jseint.2023.07.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Sasanuma H., Sugimoto H., Iijima Y., Kanaya Y., Saito T., Takeshita K. Blood flow evaluation by dynamic magnetic resonance imaging of symptomatic rotator cuff tears and frozen shoulders. J Shoulder Elbow Surg. 2018;27:e372–e379. doi: 10.1016/j.jse.2018.05.042. [DOI] [PubMed] [Google Scholar]
- 29.Shaffer B., Tibone J.E., Kerlan R.K. Frozen shoulder. A long-term follow-up. J Bone Joint Surg Am. 1992;74:738–746. [PubMed] [Google Scholar]
- 30.Tachihara H., Hamada J. Characteristic movement of the ribs, thoracic vertebrae and scapula while elevating the upper limbs - influences of age and gender on movements. Open Orthop J. 2019;13:170–176. doi: 10.2174/1874325001913010170. [DOI] [Google Scholar]
- 31.Tamai K., Hamada J., Nagase Y., Morishige M., Naito M., Asai H., et al. Frozen shoulder. An overview of pathology and biology with hopes to novel drug therapies. Mod Rheumatol. 2024;34:439–443. doi: 10.1093/mr/road087. [DOI] [PubMed] [Google Scholar]
- 32.Tamai K., Hamada J., Nagase Y., Morishige M., Naito M., Asai H., et al. Can magnetic resonance imaging distinguish clinical stages of frozen shoulder? A state-of-the-art review. JSES Rev Rep Tech. 2024;4:365–370. doi: 10.1016/j.xrrt.2024.05.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Theodoridis D., Ruston S. The effect of shoulder movements on thoracic spine 3D motion. Clin Biomech (Bristol) 2002;17:418–421. doi: 10.1016/s0268-0033(02)00026-8. [DOI] [PubMed] [Google Scholar]
- 34.Thomas W.J., Jenkins E.F., Owen J.M., Sangster M.J., Kirubanandan R., Beynon C., et al. Treatment of frozen shoulder by manipulation under anaesthetic and injection: does the timing of treatment affect the outcome? J Bone Joint Surg Br. 2011;93:1377–1381. doi: 10.1302/0301-620X.93B10.27224. [DOI] [PubMed] [Google Scholar]
- 35.Van Der Stok J., Fitzsimons M., Queally J.M., O'Donnell T. Does capsular distension and a short period of countertraction improve outcome following manipulation under anesthesia for the treatment of primary adhesive capsulitis of the glenohumeral joint? J Shoulder Elbow Surg. 2022;31:772–781. doi: 10.1016/j.jse.2021.08.032. [DOI] [PubMed] [Google Scholar]
- 36.Vastamäki H., Kettunen J., Vastamäki M. The natural history of idiopathic frozen shoulder: a 2- to 27-year followup study. Clin Orthop Relat Res. 2012;470:1133–1143. doi: 10.1007/s11999-011-2176-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Wada T., Itoigawa Y., Yoshida K., Kawasaki T., Maruyama Y., Kaneko K. Increased stiffness of rotator cuff tendons in frozen shoulder on shear wave elastography. J Ultrasound Med. 2020;39:89–97. doi: 10.1002/jum.15078. [DOI] [PubMed] [Google Scholar]
- 38.Walker-Bone K., Palmer K.T., Reading I., Coggon D., Cooper C. Prevalence and impact of musculoskeletal disorders of the upper limb in the general population. Arthritis Rheum. 2004;51:642–651. doi: 10.1002/art.20535. [DOI] [PubMed] [Google Scholar]
- 39.Woods D.A., Loganathan K. Recurrence of frozen shoulder after manipulation under anaesthetic (MUA): the results of repeating the MUA. Bone Joint J. 2017;99-B:812–817. doi: 10.1302/0301-620X.99B6.BJJ-2016-1133.R1. [DOI] [PubMed] [Google Scholar]
- 40.Yano Y., Hamada J., Hagiwara Y., Karasuno H., Tamai K., Suzuki K. A new pathophysiology of atraumatic rotator cuff tears: adduction restriction of the glenohumeral joint. JSES Int. 2020;4:333–340. doi: 10.1016/j.jseint.2020.02.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Zhang J., Zhang L., Guo F., Zhang T. Shear wave elastography of the coracohumeral ligament with frozen shoulder in different stages. J Ultrasound Med. 2022;41:2527–2534. doi: 10.1002/jum.15942. [DOI] [PubMed] [Google Scholar]
- 42.Zuckerman J.D., Cuomo F. In: The shoulder: a balance of mobility and stability. Masten F.A., Fu F.H., Hawkins R.J., editors. American Academy of Orthopedic Surgery; Chicago: 1993. Froze shoulder; pp. 253–267. ISBN No. 978-0892030910. [Google Scholar]