The shoulder is a complex articulation consisting of the glenohumeral joint, scapulothoracic joint, and acromioclavicular joint. During normal shoulder elevation, there is a balance of motion between these articulations driven by the shoulder musculature; this provides tremendous mobility, allowing the arm to be placed anywhere in space. Shoulder stiffness, which has numerous etiologies, can be a debilitating and painful problem that disrupts this delicate balance. Recent research has advanced our understanding of the pathogenesis and treatment of the stiff shoulder.
Stiffness about the shoulder was described clinically more than a century ago. It remains a common and frustrating problem seen in orthopedic practice today. A number of clinical scenarios manifest with loss of motion in the shoulder, and the terminology can be confusing. Despite ongoing basic science and clinical research, the mechanisms that lead to the development of a stiff shoulder are not fully understood, and there is still no consensus on the optimal method of treatment.
Definition and Classification
Numerous schemes for classifying the stiff shoulder have been proposed over the years. In most a history of some sort of traumatic event is considered a distinguishing factor, but no system has yet been universally accepted.
We prefer to classify the stiff shoulder into two general categories: idiopathic and acquired. The hallmark of both is a significant loss of both active and passive shoulder motion. In the former there is no known intrinsic shoulder disorder, although there may be associated systemic conditions, such as diabetes mellitus. In the latter there is some sort of predisposing condition affecting the shoulder (e.g., surgery or trauma) that is felt to be the initiating factor in the process.
In this chapter we use the term frozen shoulder to describe the idiopathic condition. This group includes patients with diabetes mellitus. All other conditions fall under the category of acquired stiffness. Acquired stiffness can be subdivided to indicate the mechanism associated with the condition (e.g., posttraumatic or postsurgical) ( Fig. 19-1 ).
Primary, Idiopathic Frozen Shoulder
Shoulder stiffness is a general term that encompasses a spectrum of pathology. It was first described as a clinical entity in the late nineteenth century by both Duplay in France and Putnam in the United States. The original term for the condition was scapulohumeral periarthritis, which encompassed a number of painful afflictions of the shoulder resulting in stiffness. As knowledge has developed about the specific conditions that can cause a stiff shoulder, this term has fallen out of favor.
In 1934 Codman described a clinical pattern of muscle spasms and glenohumeral stiffness, which he termed the frozen shoulder. He stated that this condition was “difficult to define, difficult to treat, and difficult to explain from the point of view of pathology.” In many respects this is still true today.
A decade after Codman, Neviaser described a similar condition, a “chronic inflammatory process involving the capsule of the shoulder causing a thickening and contracture of this structure which secondarily becomes adherent to the humeral head.” He called it adhesive capsulitis , a term he believed better represented the underlying pathology. The term adhesive capsulitis, however, has sometimes been used indiscriminately to describe both the idiopathic frozen shoulder and shoulder stiffness related to other causes. Because these are now considered separate entities, it is probably no longer appropriate to use the same term for both.
At a symposium sponsored by the American Academy of Orthopaedic Surgeons in 1992, a workshop committee defined frozen shoulder as “a condition of uncertain etiology characterized by significant restriction of both active and passive shoulder motion that occurs in the absence of a known intrinsic shoulder disorder.”
In an attempt to obtain a consensus definition and classification of frozen shoulder/adhesive capsulitis from 190 shoulder experts from around the world, Zuckerman proposed the definition of frozen shoulder to be “a condition characterized by functional restriction of both active and passive shoulder motion for which radiographs of the glenohumeral joint are essentially unremarkable except for the possible presence of osteopenia or calcific tendonitis.” He also classified the condition as primary or secondary based on whether an etiology or associated condition could be identified. He subclassified secondary frozen shoulder as intrinsic (due to rotator cuff or biceps disorders), extrinsic (due to an abnormality remote to the shoulder), or systemic (occurring with associated conditions, such as diabetes mellitus, hyperthyroidism, hypothyroidism, or hypoadrenalism). However, only 66% of the shoulder surgeons agreed with this subclassification of secondary frozen shoulder. We find it optimal to differentiate primary, idiopathic frozen shoulder from secondary frozen shoulder due to surgery, trauma, or other causes (see Fig. 19-1 ). We do not consider shoulder stiffness associated with a systemic disease, such as diabetes, to be a secondary frozen shoulder.
Secondary Frozen Shoulder
The earliest description of shoulder stiffness following trauma was recorded by Malgaigne in 1859. He wrote with regard to minor nondisplaced, extracapsular fractures about the shoulder :
There remains long afterward a stiffness, and a difficulty in moving the shoulder; and however great care may be taken, the motion of elevation of the arm will always remain limited; at least I have in no case seen it perfectly restored, even after the lapse of 11 to 15 months.
There is less controversy and confusion surrounding the nomenclature of this clinical entity. Other commonly used terms for the same condition include posttraumatic stiff shoulder and secondary stiffness . It is unclear whether acquired stiffness can recover without intervention; this is difficult to study because of the heterogeneity of this patient population.
Some degree of shoulder stiffness is typical after bone or soft tissue injuries around the shoulder. Restriction in shoulder motion has been reported after simple contusions, subluxations, dislocations, acromioclavicular joint injuries, clavicle and scapula fractures, and particularly after proximal humerus fractures in the elderly.
Especially after repetitive, low-level trauma, localized contractures can develop and can cause motion loss in specific patterns. Isolated posterior capsular contracture is the most commonly described. Neer discussed the impingement syndrome and encouraged therapeutic stretching in forward elevation, internal rotation, and cross-body adduction. Thomas and colleagues and Ticker and colleagues made similar recommendations. In a review of 30 patients from a prospective study of refractory shoulder stiffness, 11 experienced isolated restriction in motion that was attributed to posterior capsular contracture. In each of these patients an associated pathologic process was identified, with a partial-thickness rotator cuff tear being the pathology in seven cases.
Surgical procedures are widely recognized to be a cause of shoulder stiffness. Anterior or posterior capsulorrhaphy, inferior capsular shift, and rotator cuff surgery can all result in limitation of motion, often by design. Tauro found a relatively high incidence of stiffness in patients who underwent surgery for a rotator cuff tear. Patients with more pronounced stiffness often experienced difficulty in regaining normal motion after surgery.
Idiopathic frozen shoulder implies a global glenohumeral capsular contracture that occurs in the absence of a definable traumatic event. Acquired stiffness is an extrinsic process related to some other, often traumatic, cause. These conditions might not be mutually exclusive. On rare occasions, inflammation and adhesions have been identified in the subacromial bursa in association with idiopathic frozen shoulder. Although acquired stiffness is generally the result of an extracapsular process, subsequent capsular contracture can develop.
Idiopathic Frozen Shoulder
The lack of consensus regarding nomenclature and classification for frozen shoulder has partly originated from the confusion surrounding necessary and sufficient diagnostic criteria. Many authors, however, have proposed diagnostic criteria for frozen shoulder. Considerations include patient history, relative loss of motion in various planes, motion loss compared with the contralateral shoulder, and radiographic findings.
We do not subscribe to a specific, published set of criteria, but we do consider certain characteristics to be key components in the diagnosis of a frozen shoulder. These include a lack of history of major trauma or surgery about the shoulder, significantly limited passive and active range of motion (ROM) in all planes compared with the contralateral shoulder, and normal radiographs or other advanced imaging studies.
Normal Motion and Pathomechanics
Understanding the altered mechanics associated with a stiff shoulder requires an understanding of normal shoulder mechanics. Normal shoulder motion involves several interfaces: the articulation between the humeral head and the glenoid (glenohumeral motion), the gliding surface between the proximal humerus and the coracoacromial arch (the subacromial-subdeltoid plane), the articulation between the scapula and the thorax (scapulothoracic motion), and motion through the acromioclavicular joint. Adhesions or contractures at any one of these interfaces can lead to stiff shoulder development.
Under normal circumstances, the shoulder is a remarkably mobile joint. The normal ratio of glenohumeral motion to scapulothoracic motion is 2 : 1. The normal surface area of the capsule is nearly twice that of the humeral head and the capsule is inherently loose. In cadaveric tests, Harryman and colleagues measured torsional resistance of the glenohumeral joint with an intact capsule and showed that the capsule remained essentially free of tension until a terminal degree of motion was reached ( Fig. 19-2 ). Most motions during work and activities of daily living are performed in midrange positions. As the end of a ROM is approached, increased tension is seen in the static restraints, notably in the capsule and its associated ligaments: the superior, middle, and inferior glenohumeral ligaments. The humeral head does not simply rotate within the glenoid cavity like a ball and socket joint, but translates over its surface. These ligaments act as restraints to both rotation and translation at the extremes of motion.
The anterosuperior capsule in the region of the rotator interval contains the coracohumeral and superior glenohumeral ligaments. It assumes tension with increasing external rotation with the arm in 0 degrees of abduction. The middle glenohumeral ligament and the anterior band of the inferior glenohumeral ligament become taut with maximal external rotation at 45 degrees and 90 degrees of abduction, respectively. The inferior region of the capsule becomes tighter with increasing abduction. The posterior capsule becomes tight in internal rotation with the arm at the side. With increasing angles of elevation, tension is shifted inferiorly.
Isolated contractures in specific capsular regions have been described. These have often been created intentionally during procedures for instability. A contracted capsule reduces glenohumeral motion. In a cadaveric study posterior capsular plication was applied to eight shoulders, resulting in limitation of forward elevation, internal rotation, and horizontal adduction. External rotation was unaffected. Tightening of the capsule also increased the torque required to achieve an elevated position (see Fig. 19-2 ).
Asymmetric tightness of the capsule can cause an obligate translation of the humeral head. Normally, the glenohumeral joint displays ball-and-socket mechanics, and the humeral head remains centered within the glenoid fossa except at the extremes of ROM. In a cadaveric study when the posterior capsule was surgically tightened, forward flexion caused consistent anterosuperior translation of the humeral head. Clinically, such translation can lead to impingement of the rotator cuff against the coracoacromial arch. This has been described in throwing athletes who develop isolated posterior capsular contractures. Known as glenohumeral internal rotation deficit , it can lead to significant pain and disability, especially in overhead athletes. Glenohumeral internal rotation deficit has been associated with partial articular-sided rotator cuff tears, labral tears, and even ulnar collateral ligament tears of the elbow. Usually, conservative management of this subtle shoulder stiffness can restore physiologic posterior capsular laxity and eliminate the patient’s symptoms.
A similar phenomenon can occur with overtightening of the anterior structures, such as with nonanatomic repairs for anterior instability (the Putti-Platt procedure). A chronic contracture of the anterior aspect of the shoulder is almost always associated with a deficit in external rotation compared with the contralateral shoulder. This scenario can result in greater joint reaction forces directed toward the posterior portion of the glenoid and can cause excessive posterior glenoid articular cartilage wear, posterior bony erosion, and osteoarthrosis. Matsen and colleagues have called this process capsulorrhaphy arthropathy ( Fig. 19-3 ).
To eliminate pathologic restrictions in motion, a number of techniques, both nonoperative and operative, have been described for capsular release. Although the techniques have evolved, the basic concept of releasing localized or general contractures to improve ROM remains the same. The technical details and results of these procedures are discussed in depth later.
In 1934 Codman stated, “The subacromial bursa itself is the largest in the body and is the most complicated in structure and in its component parts. It is in fact a secondary scapulohumeral joint, although no part of its surface is cartilage.”
Matsen and Romeo defined the subacromial-subdeltoid plane interface as a sliding surface between the deep side of the deltoid, the acromion, the coracoid process and its tendons, and the superficial side of the humerus and rotator cuff ( Fig. 19-4 ). Normal biceps glide is also necessary for full glenohumeral ROM. If synovitis is present within the glenohumeral joint or biceps tendon sheath, adhesions can spot-weld the long head of the biceps tendon to the capsule or bicipital groove ( Fig. 19-5 ). Pasteur, in 1932, described peritendinous adhesions within the bicipital groove in patients with frozen shoulder. This finding has also been described since by other authors.
Adhesions in this area, which would usually be secondary to trauma or surgery, can substantially limit shoulder motion. Many orthopedic surgeons also struggle with the rehabilitation protocol after shoulder trauma, specifically with proximal humerus fractures. Immobilization can reduce the risk of fracture displacement; however, prolonged immobilization risks stiffness, which can be debilitating even if the fracture has healed in a satisfactory position.
Scapulothoracic mechanics are another important component of shoulder motion. Investigators in the past have determined that scapulothoracic motion is responsible for approximately one third of total shoulder elevation. Scapulothoracic motion has been studied in detail by Harryman and McClure. Loss of glenohumeral ROM (e.g., after arthrodesis) can result in an accommodative increase in scapulothoracic ROM. Although it is challenging to clinically measure scapulothoracic motion with accuracy, Nicholson recorded excessive scapular upward rotation (elevation) during active attempts at humeral elevation in patients with frozen shoulder. Lin and colleagues showed increased EMG activity in the upper trapezius, with relatively lower activity in the lower trapezius. These findings have potential implications in the rehabilitation of patients with frozen shoulder.
Although many plausible mechanisms for stiff shoulder development have been proposed, the etiology remains elusive. Advances in basic science research, however, have improved our understanding of the condition. Current efforts are focused on determining both an immunologic basis and the role of cell signaling and inflammatory mediators in its development.
Although recent research has advanced our knowledge of the pathophysiology of the idiopathic frozen shoulder, the exact etiology and mechanism have not yet been established. Our understanding of the disease process is that it is primarily a capsular pathology with signs of inflammation, fibroblast proliferation, and neovascularization. There may be an immunologic component, but there are no reliable laboratory tests or inflammatory markers to diagnose frozen shoulder. There is a clear association of diabetes mellitus and Dupuytren’s contracture with frozen shoulder, but the exact pathophysiologic process is mostly speculative. Current research is focused on defining the role of matrix metalloproteinases (MMPs), tissue inhibitors of metalloproteinases (TIMPs), other cytokines, and cytogenetics and their potential role in affecting the natural history of the disease in clinical practice.
Original Theories, Anatomic and Histologic Analysis
In 1896 Duplay’s original view of stiff shoulders was that they arose in the subdeltoid region. Riedel subsequently hypothesized that shoulder stiffness may be related to the joint capsule itself. In 1945 Neviaser postulated that idiopathic shoulder stiffness was a capsular pathology. He correlated the surgical, arthrographic, and histologic findings regarding the synovium and capsule of patients with painful shoulder stiffness. Histologically, he identified perivascular infiltration and capsular fibrosis; the synovial layer itself appeared normal. He proposed that shoulder stiffness was due to a tight glenohumeral joint capsule that adhered to itself and the humeral neck, coining the term “adhesive capsulitis.” Neer subsequently proposed that coracohumeral ligament contracture also led to shoulder stiffness.
Others surgeons continued to histologically analyze the synovium and capsule. Lundberg, too, did not observe any pathologic changes in the synovium. Instead, he found increased collagen density within the capsule and a pattern of glycosaminoglycan distribution that resembled a reparative process. McLaughlin found inflammation in the synovium and biceps sheath in only 10% of his cases. He posited that “acute synovitis represents one phase in the life cycle of this condition.”
Hannafin and colleagues attempted to correlate the three histopathologic phases of fibroplasia detected in biopsy samples of patients with frozen shoulder, as described by Neviaser, with findings from clinical examination and arthroscopy. The authors hypothesized that hypervascular synovitis provokes a progressive fibroblastic response in the adjacent capsule, resulting in diffuse capsular fibroplasia and contracture. Based on immunohistochemical and histologic findings, they proposed a cellular pathway that eventually results in the clinical scenario of a frozen shoulder.
Macnab analyzed biopsy samples obtained during open capsular release and identified round and lymphoid cell infiltrates within the tissue. He hypothesized that this was the result of an autoimmune response directed toward degenerative collagen particles from a hypovascular supraspinatus tendon. Bulgen and colleagues found that patients with a frozen shoulder initially had increased levels of circulating immune complex and C-reactive protein and decreased lymphocyte transformation. The same group subsequently analyzed serum immunoglobulin levels in 25 patients with frozen shoulder and found serum immunoglobulin A levels to be significantly reduced, with this reduction persisting even after clinical recovery. Lymphocyte transformation in response to phytohemagglutinin was also significantly depressed in the majority of patients. These early findings suggested an immunologic basis for the disease; however, subsequent reports have failed to support the findings or to identify immunologic tests useful in diagnosis, treatment, or predicting outcome.
Other investigators have attempted to identify factors that predispose patients to shoulder stiffness by using immunologic markers. Initial research was focused on the presence of certain histocompatibility antigens. The presence of human leukocyte antigen (HLA)-B27 was reported as being more common in patients with frozen shoulder (42%) than in controls (10%). This finding, however, was later refuted by the same authors. As of yet, no definite immunologic markers have been defined for patients with frozen shoulder.
A proliferative pathologic repair process initiated by active fibroblasts may occur in response to the inflammatory infiltration of connective tissue by mononuclear cells that produce polypeptide growth factors. Mullett and colleagues examined the response of human fibroblasts to joint aspirates from patients with frozen shoulder and a control group; significantly increased fibroblast proliferation was seen in the patients with frozen shoulder. In another study biopsy samples were taken from the rotator interval in 22 patients undergoing arthroscopic capsular release. Histologic and immunocytochemical analysis revealed the presence of fibroblasts, proliferating fibroblasts, and chronic inflammatory cells (predominantly mast cells). T-cells, B-cells, and macrophages were also present.
Association With Dupuytren Contracture
Many authors dating back to 1936 have reported an association between frozen shoulder and Dupuytren contracture of the hand, with the rate of association ranging from 18% to as high as 52%. More recent investigators have identified similarities in the fibroblastic histologic changes seen in the glenohumeral joint capsule to those seen in Dupuytren contracture.
Bunker and Anthony performed manipulation and open excisional biopsy of the coracohumeral ligament and rotator interval capsule in patients who failed to improve with nonoperative treatment of frozen shoulder. The tissue specimens revealed active fibroblast proliferation amidst thick nodular bands of collagen, accompanied by some transformation to a smooth muscle phenotype (myofibroblasts). These histologic features, with neither inflammation nor synovial involvement, are very similar to those seen in Dupuytren contracture. Fibrosis is most evident in later phases of the inflammatory response, and collagen and matrix synthesis takes place after chemotactic and cellular responses. These findings may therefore reflect a later phase of the disease, perhaps following an earlier inflammatory phase. Uhthoff and Boileau confirmed the presence of vimentin, a cytocontractile protein known to be present in Dupuytren contracture, in histologic sections of only the anterior capsular structures in patients with frozen shoulder. This protein was not seen in specimens of the posterior capsule, although type I and type III collagens were. The authors hypothesized that contracture selectively involves only the anterior capsule, whereas fibroplasia involves the entire capsule.
Association With Diabetes Mellitus
The association between idiopathic frozen shoulder and diabetes mellitus is well documented. * However, the exact pathophysiology is still not completely understood. In the 1980s Brownlee and colleagues hypothesized that hyperglycemia leads to faster glycosylation and cross-linking of collagen within the glenohumeral capsule. Even though many studies have subsequently demonstrated the association between diabetes and frozen shoulder, there has been little published research since then that has advanced our understanding of the biochemical effect of diabetes on shoulder stiffness.
* References .
Cytokinetic, Genetic, and Enzymatic Analysis
The elevation of multiple inflammatory cytokines has recently been linked to idiopathic frozen shoulder. Rodeo and colleagues compared capsular tissue samples from patients undergoing arthroscopy who had frozen shoulder, nonspecific synovitis, or a normal capsule. Their results indicated that cytokines, such as transforming growth factor (TGF-β), platelet-derived growth factor (PDGF), and hepatocyte growth factor are involved in the early inflammatory stages of frozen shoulder. PDGF is a mitogenic agent that causes fibroblastic cell proliferation, and TGF-β increases extracellular matrix production. Suzuki and colleagues further demonstrated that specific growth factors stimulate capsular fibroblasts in a canine model. They found that PDGF-AB, hepatocyte growth factor, and insulin-like growth factor type I all stimulated the migration of fibroblasts from three different parts of a canine shoulder model: the upper and lower parts of the middle glenohumeral ligament and the posterior capsule. Ryu and colleagues found increased vascular endothelial growth factor expression in the synovial tissue of a group of diabetic patients with frozen shoulder; however, its exact role remains undetermined. Elevated levels of inflammatory cytokines interleukin-6 and vascular endothelial growth factor have been found in the synovium of patients with a frozen shoulder, in addition to elevated serum levels of TGF-β.
MMPs, a family of naturally occurring proteinases that control collagen matrix remodeling, have also been implicated as a contributing factor in the pathogenesis of frozen shoulder. Hutchinson and colleagues reported the onset of frozen shoulder in 12 patients with gastric carcinoma after treatment with the synthetic MMP inhibitor, marimastat. This report suggested that the inhibition of MMPs may be associated with frozen shoulder. Bunker and colleagues examined capsular tissue from patients with frozen shoulder. The tissue was analyzed for various factors, including MMPs and their inhibitors. When compared with a normal capsule, the capsule of patients with frozen shoulder demonstrated an increase in messenger RNA for MMPs as well as MMP inhibitors, suggesting that abnormalities in MMP expression are a factor in frozen shoulder. Lubis and colleagues found that the serum MMP-1 and MMP-2 levels were significantly lower in frozen shoulder patients than in patients without a frozen shoulder, while their inhibitors’ (TIMP-1 and TIMP-2) levels were significantly higher. Other investigators have found similar MMP abnormalities in synovial tissue.
Studies employing stringent diagnostic criteria for frozen shoulder have defined the incidence in the general population as ranging from 2% to 5%. Nearly 70% of patients presenting with frozen shoulder are women, and 20% to 30% develop stiffness in the contralateral shoulder. A recent study demonstrated that 11% of all patients referred to musculoskeletal specialists for a shoulder disorder were ultimately diagnosed with idiopathic frozen shoulder. Recurrence in the same shoulder is quite unusual. An accurate incidence of acquired stiffness has yet to be determined, partly because of the variability in this patient population.
The majority of patients who seek care for a stiff shoulder, regardless of etiology, are between 40 and 60 years of age. It is unusual for an idiopathic frozen shoulder to develop in patients younger than 40 years, with the exception of those who have had insulin-dependent diabetes since childhood. In a large study Lundberg noted that the mean age at the time of presentation was slightly higher in men than in women. However, differences in age relative to sex or type of stiffness were not evident in Harryman and colleagues’ series of 126 patients with refractory shoulder stiffness.
Patients with diabetes mellitus are clearly at much greater risk for developing limited joint motion not only in the shoulder but also in other joints. Most studies have shown a 10% to 20% incidence of frozen shoulder in these patients, but some have indicated that the rate of incidence may be as high as 35%. The longer a patient has been taking insulin, the greater the risk of developing shoulder stiffness and the greater the resistance to all treatment modalities. Patients with childhood-onset diabetes tend to develop stiffness at an earlier age, but in those with adult-onset diabetics the age of presenting with shoulder stiffness is similar to that of the general population. It has been presumed that the severity of diabetes is associated with greater risk of developing a frozen shoulder. However, Yian and colleagues found no association between hemoglobin A1C levels and the prevalence of frozen shoulder in a population of diabetic patients, although the risk increased if patients were taking any medications for diabetes or had developed neurologic manifestations of diabetes.
Insulin-dependent diabetics with joint stiffness in the hands and other major articulations are categorized as having limited joint motion syndrome. Diabetics who have cheiroarthropathy (a waxy thickening and induration of the skin associated with flexion contractures of the fingers) as well as a frozen shoulder have higher incidences of retinopathy and bilateral shoulder involvement (77%).
Shoulder stiffness may be the initial manifestation in a patient with diabetes. Lequesne and colleagues discovered 17 patients with glucose intolerance out of 60 new patients with idiopathic frozen shoulder. A diabetes workup should be considered in any patient with newly diagnosed frozen shoulder. Because of the refractory nature of shoulder stiffness in long-term insulin-dependent diabetics, early intervention has been considered appropriate to prevent progressive disability.
Diabetes not only can put patients at risk for developing a frozen shoulder but also can affect their ultimate outcome. Mehta and colleagues prospectively compared the results of arthroscopic release for frozen shoulder in 42 patients with and without diabetes. The patients in both groups significantly improved, but the Constant score at 6 months was significantly worse in the diabetic patients, with a tendency towards persistent stiffness 2 years postoperatively. Cinar and colleagues also demonstrated that patients with diabetes did worse in terms of pain and motion compared with patients without diabetes after an arthroscopic capsular release.
Stiffness can also occur after surgery that does not involve the shoulder. Common examples include axillary node dissection and neck dissection, especially when these procedures are combined with radiation therapy. Cardiac catheterization in the axilla, coronary artery bypass grafting, and thoracotomy can also restrict shoulder range due to the severe pain after the procedure. In addition, stiffness may be triggered by interventional cardiology procedures, such as cardiac catheterization through the brachial artery or placement of an ipsilateral subpectoral cardiac defibrillator. The incidence of frozen shoulder in a population of male patients who underwent cardiothoracic surgery has been estimated at 3.3%. Frozen shoulder has been described following rehabilitation after breast cancer surgery.
In adults immobilization of the shoulder puts it at risk of becoming stiff. A significant number of referrals for shoulder stiffness occur after a period of rest, which has often been recommended by the referring physician. In a review of patients referred to Binder and colleagues, 75% had initially been told to rest the shoulder rather than to perform gentle exercises.
Cervical Disk Disease
Degenerative disk disease between C5-6 and C6-7 has been noted to be more common in patients with shoulder stiffness than in a similar age-matched control group. This association has been supported in other studies. Patients with symptomatic cervical radiculopathy and a painful shoulder, with or without loss of motion, experienced less pain and were more likely to regain pain-free motion when cervical traction was added to the prescribed exercise regimen.
Bilateral frozen shoulders are rare but have been reported in the setting of both hyperthyroidism and hypothyroidism. Wohlgethan considered hyperthyroidism, frozen shoulder, and shoulder-hand syndrome to be linked disorders. Resolution of shoulder stiffness has been shown to occur after thyroidectomy and the stabilization of thyroid hormone levels.
The association between atherosclerotic coronary vascular disease and shoulder stiffness is well documented. In a review of 133 consecutive cases of myocardial infarction, Ernstene and Kinell found 17 patients who complained of unrelenting pain in the shoulder region. Shoulder-hand syndrome, an autonomic dystrophy, may be a sequela of myocardial infarction in 10% to 30% of cases.
Bunker identified raised serum lipid levels in a group of patients with primary frozen shoulder when compared with a group of age- and sex-matched controls, a finding supported by other investigators. However, Austin and colleagues did not find an increased use of lipid-lowering medication in a frozen shoulder cohort compared with the general population.
In 1959 Johnson reported that the incidence of frozen shoulder was 3.2% in a population of sanatorium patients with tuberculosis. Saha reported that frozen shoulder occurred more frequently in patients with emphysema and chronic bronchitis, but found no correlation with either the severity or the duration of illness.
Bronchogenic carcinoma and Pancoast tumors of the lung can cause severe shoulder pain that can mimic the early phase of frozen shoulder. Other occult neoplastic tumors can be masked by symptoms attributed to a frozen shoulder; these include chest wall tumors, sarcomas, and metastatic disease. Gheita and colleagues found a 15% prevalence of frozen shoulder in 60 patients who had been diagnosed with a malignancy. Malignancy should therefore always be considered in the differential diagnosis when a patient presents with a painful, stiff shoulder that is refractory to initial conservative modalities.
Shoulder stiffness has been associated with a variety of neurologic disorders. The incidence of frozen shoulder in patients with Parkinson disease is significantly higher than in age-matched controls. In 8% of Parkinson disease patients the first symptom was shoulder stiffness, which can occur up to 2 years before the onset of generalized symptoms. Patients with cerebral hemorrhage and cerebral tumors are also at increased risk for frozen shoulder.
At least 30% of patients with hemiplegia have shoulder pain and are susceptible to the development of shoulder stiffness. Wanklyn and colleagues also found that patients who required transfer assistance were more likely to suffer from shoulder pain. Bruckner and Nye reported a 25% incidence of frozen shoulder in patients who had suffered subarachnoid hemorrhage. They found that the development of stiffness was associated with impaired consciousness, hemiparesis, intravenous (IV) infusion, older age, and depression. In patients with hemiparesis the clinician may find it difficult to distinguish between shoulder stiffness caused by capsular contracture and muscular spasticity. Shoulder-hand syndrome occurs in as many as 30% to 40% of stroke patients and can be terribly disabling. Conversely, Kang and colleagues found that the development of idiopathic frozen shoulder was associated with increased risk of stroke.
Brachial neuritis, a painful neuritis condition also known as Parsonage-Turner syndrome, has been associated with frozen shoulder. Thoracic outlet syndrome, suprascapular neuropathy, and spinal accessory nerve palsy have also been linked to the onset of frozen shoulder.
Some authors have argued that individuals with certain personality characteristics are more likely to develop frozen shoulder. This was first proposed by Codman in 1934 who described four patients with frozen shoulder who “were a little run-down without anything particular the matter.” Coventry coined the term periarthritic personality and found treatment more challenging in this group. Fleming and colleagues profiled the personality type of 56 patients with a frozen shoulder by using the Middlesex Hospital Questionnaire and found that women with frozen shoulder had significantly greater anxiety levels than did controls. Tyber found the prevalence of depression to be relatively high in a group of 55 patients with painful shoulder syndromes. He included lithium and amitriptyline in his treatment of these patients.
The notion of a characteristic personality disorder is controversial. Wright and Haq found no evidence of this when using the Maudsley Personality Inventory to test 186 patients with a frozen shoulder. Matsen and Harryman reviewed the mental health scores on the Short Form-36 (SF-36) health status questionnaire of 295 patients with either frozen shoulder or acquired stiffness. They found that patients with frozen shoulder and patients with acquired stiffness scored within 95% and 88%, respectively, of the mean score for normal age-matched controls. Debeer and colleagues prospectively evaluated 118 frozen shoulder patients and found no significant difference in personality traits between these patients and controls.
Reaction to Medication
Reports suggest that frozen shoulder can develop after treatment with certain medications. Grasland and colleagues described eight cases of frozen shoulder that developed after the treatment of human immunodeficiency virus (HIV) infection with protease inhibitors. All of these patients received the drug indinavir as part of their regimen and had no other risk factors for frozen shoulder. This association has been confirmed by other investigators. Other medications that have been linked to the onset of frozen shoulder include barbiturates, fluoroquinolones, nelfinavir, and isoniazid. The development of a frozen shoulder has also been recently reported after intramuscular vaccination administration.
A heritability of 42% for frozen shoulder has been reported, although no specific gene has been identified. Wang found over a four times greater risk of developing an idiopathic frozen shoulder in patients with a first degree relative who had developed the disease.
A recent study by Austin and colleagues found an association between idiopathic frozen shoulder and hypertension. In a cohort of patients with the diagnosis of frozen shoulder, the authors found a 50% greater prevalence of antihypertensive medications in these patients than the national average. The authors hypothesized that hypertension is a proinflammatory condition and part of the metabolic syndrome, which can increase the risk of frozen shoulder. However, further research is required to substantiate this.
The natural history of frozen shoulder continues to be debated. Many consider primary idiopathic frozen shoulder to be a self-limited disease. Codman originally stated that “recovery is always sure and can be confidently expected.” However, it is not uncommon for patients to continue to have some discomfort and restriction of motion even after the resolution of the thawing phase. Some patients show absolutely no improvement over time or show only mild improvement followed by plateauing.
Meulengracht and Schwartz monitored 65 frozen shoulder patients for 3 years and found 23% developed persistent pain and limitation of shoulder motion. Reeves monitored 41 patients for more than 4 years and identified residual stiffness in more than 60%, with 12% displaying severe restriction in motion. Interestingly, subjective and objective measures of function can differ. At a 3.5-year follow-up, Binder and colleagues found objective motion restriction in 16 of 40 patients, yet only a few complained of significant functional impairment. This result was quite different from the findings of Shaffer and colleagues, who monitored 68 patients with frozen shoulder for approximately 7 years. By objective criteria, 30% of these patients had shoulders that were restricted when compared with the opposite unaffected side, yet 50% complained of persistent pain or stiffness.
Wolf and Green analyzed the influence of several comorbidities on the self-reported functional status of patients with frozen shoulder. They found that increasing numbers of comorbidities were associated with worsening scores on the Disabilities of the Arm, Shoulder, and Hand (DASH) survey, the Simple Shoulder Test (SST), and the SF-36 health status survey.
Involvement of the dominant arm appears to be a good prognostic sign, whereas occupation, ability to work, duration of stiffness, associated injuries, and the treatment program used did not achieve significance when analyzed against other outcome measures. Patients with insulin-dependent diabetes for more than 10 years tended to have a poorer outcome.
Hand and colleagues have reported the long-term outcomes of 269 frozen shoulders in 223 patients after a mean follow-up of 4.4 years (range, 2 to 20 years). Of these patients, 59% had normal or near normal shoulders; the remaining 41% reported some ongoing symptoms, but in 94% of the patients, these persistent symptoms were mild. The patients with the most severe symptoms at the onset had the worst long-term prognosis. At a mean follow-up of 9 years (range, 2 to 27 years), Vastamäki and colleagues found that in 94% of the patients the shoulder had improved to near normal with benign neglect. In these patients the duration of symptoms lasted a mean of 15 months.
In general, a high index of suspicion is necessary to make a definitive diagnosis for primary idiopathic frozen shoulder because other pathologic entities can manifest with shoulder stiffness. A careful history and physical examination are essential. Radiographic and laboratory studies may be helpful in certain instances.
As there is no reliable laboratory test or imaging modality for the condition, primary idiopathic frozen shoulder is ultimately a clinical diagnosis. Usually the patient’s history and examination provide the only information required to diagnose frozen shoulder. The diagnosis should be considered in patients that present with some degree of pain associated with the corresponding loss of both passive and active ROM. Patients with glenohumeral arthritis often present with similar symptoms, but usually the onset of pain and stiffness occurs over years for arthritis instead of weeks to months as with frozen shoulder. A single true anteroposterior radiograph of the shoulder can easily differentiate idiopathic frozen shoulder from glenohumeral arthritis.
When considering the diagnosis of idiopathic frozen shoulder, it is essential to consider the patient’s age, gender, and additional risk factors as outlined in the previous section. Of these, the most relevant are age between 40 and 60 years, female sex, and having diabetes. This pattern can be so reliable that the Chinese refer to idiopathic frozen shoulder as wishi jian , which means “fifties shoulder.” If a patient lacks these three risk factors, then other diagnoses should strongly be considered.
Frozen shoulder is classically characterized by three stages: freezing, frozen, and thawing ( Fig. 19-6 ).
Stage 1: Freezing
This initial stage is characterized by pain. Symptoms have usually been present for several months or less. Typically, patients have what is often described as achy discomfort at rest and severe pain with attempted movements, especially sudden movements. Difficulty with sleeping is an almost universal complaint. Generally, patients limit their use of the affected extremity as discomfort worsens, resulting in loss of function. At this point many patients seek medical attention and the diagnosis can easily be missed. Indeed, many patients are encouraged to immobilize their shoulders, which exacerbates the problem. Reeves wrote that this phase generally lasts between 2 and 9 months.
Stage 2: Frozen
During the frozen phase, the pain tends to abate, but motion becomes severely limited in all planes. Even simple tasks, such as turning off the light and washing hair, become chores. Sleeping is usually problematic. A frozen shoulder is diagnosed in most patients when they enter this stage. Although it can last between 3 and 12 months, the frozen stage can become refractory and last longer.
Stage 3: Thawing
In the final stage ROM slowly returns. As motion improves, residual discomfort generally resolves. Return of flexibility can take months to years. Motion restrictions often persist. Generally, these restrictions are mild and do not cause significant impairment, but patients should be counseled about this at an early stage.
Differentiating Primary From Secondary Frozen Shoulder
Secondary frozen shoulder does not follow a predictable course like idiopathic frozen shoulder. There are, however, predictable patterns of motion loss after certain injuries. A rotator cuff strain typically causes restricted forward elevation, internal rotation, and cross-body adduction. This is thought to be related to contracture of the posterior capsule. Articular-sided partial rotator cuff tears are also associated with posterior capsular contractures. Nonanatomic instability repairs result in diminished external rotation. Immobilization after proximal humerus fractures tends to cause a global loss of motion as adhesions form in the subacromial-subdeltoid plane.
The hallmark of a frozen shoulder is the corresponding loss of both passive and active ROM. A major reason why practitioners that are not experienced with shoulder pathology miss the diagnosis is that they do not test ROM. Instead, the examiner often moves directly to more provocative maneuvers to assess the rotator cuff, which are often positive in frozen shoulder patients. This simple step of testing ROM in the initial evaluation is paramount to avoid subsequent testing and treating being misdirected instead of focusing on the stiffness.
Along with a detailed history, a careful and complete shoulder examination is the key to diagnosing a frozen shoulder. A complete cervical examination should be performed, including neurologic testing of the extremities. The shoulder should be inspected for signs of trauma or previous surgery, which may provide clues for the diagnosis of acquired stiffness. Important landmarks, such as the acromioclavicular joint and bicipital groove, are palpated for tenderness.
ROM, both active and passive, should be tested in all planes and recorded as objectively as possible. This should be repeated in the contralateral shoulder. These measurements are important not only for diagnostic purposes but also to monitor the response to treatment. It is also essential to differentiate glenohumeral motion from humeroscapular motion as many patients with glenohumeral stiffness can compensate with scapulothoracic motion. This is best done by examining the patient in the supine position with the arm free so that the scapula is being compressed against the chest wall through gravity. Alternatively, the examiner can stabilize the scapula with one hand while passively moving the arm with the other. The same measurements should be made at every visit (ideally by the same examiner) and concisely documented until the patient has demonstrated satisfactory improvement. Strength testing of the rotator cuff is performed using standard manual motor testing as well as special tests (belly press, lift-off, lag signs, hornblower’s sign). Provocative tests for impingement, acromioclavicular joint pathology, labral tears, instability, and biceps pathology may also be indicated depending on the history and clinical suspicion.
A patient with routine shoulder stiffness does not need laboratory studies for diagnosis or management. A complete blood count, C-reactive protein, and erythrocyte sedimentation rate (ESR) may be indicated if there is concern about infection. Decreasing ESR has been observed in patients with successful treatment. In more recent studies, however, ESR has not proved to be reliable or useful when evaluating the patient or monitoring the response to medication. A fasting glucose level may be ordered if undiagnosed diabetes mellitus is a concern. There are other biochemical mediators associated with frozen shoulder, such as MMPs, TIMPs, TGF-β, and interleukin-6, but their current role in clinical practice is only experimental.
Radiographs are routinely performed for any patient with a stiff shoulder. Specific conditions to be ruled out include fracture, malunion, arthrosis, osteonecrosis, and chronic dislocation. The specific views that are obtained vary depending on the clinical scenario, but a minimum of two orthogonal views is mandatory, including anteroposterior and axillary lateral views. Additional films, such as cervical spine radiographs, may be warranted depending on the clinical suspicion. Radiographs are essential as it can often be difficult to differentiate primary, idiopathic frozen shoulder from glenohumeral arthritis based on history and examination alone.
Routine shoulder radiographs in patients with shoulder stiffness and no history of trauma or surgery are typically normal. Osteopenia of the humeral head may be seen and is probably related to disuse of the extremity. In approximately 50% of 74 cases of frozen shoulder, Lundberg and Nilsson reported bone loss within a short period. They considered this to be the result of an inflammatory process and not due to disuse alone.
In 1957 Kernwein and colleagues performed arthrographic studies in 12 patients with frozen shoulder. At open biopsy, they found that the capsule and coracohumeral ligament were very contracted, thickened, and inelastic. They also noted the presence of subacute inflammation. Later, Neviaser described the arthrographic findings of adhesive capsulitis, which included decreased joint capacity, obliteration of the reflected axillary fold, and variable filling of the bicipital tendon sheath. No correlation between arthrographic findings and treatment outcome has been found. In routine practice today, arthrography is rarely used.
A bone scan is rarely indicated in the evaluation of a frozen shoulder because these have not proved to be useful in diagnosis, management, or prognosis. Positive bone scans have been reported in as many as 96% of frozen shoulder patients. However, no association has been demonstrated between bone scan activity and the severity of disease, duration of symptoms, arthrographic findings, or ultimate outcome. A bone scan should be ordered only if there is suspicion of a neoplastic process.
Magnetic Resonance Imaging
As techniques and access have improved, magnetic resonance imaging (MRI) has become widely used for evaluating shoulder disorders, including stiffness. Emig and colleagues reported on the MRI characteristics of 10 patients with a frozen shoulder compared with patients without a frozen shoulder. They found the joint capsule and synovium in the frozen shoulder patients to have a combined thickness greater than 4 mm, but they did not observe any significant differences in the volume of intra-articular fluid seen on the MRI scan or in the thickness of the rotator cuff and rotator interval capsule. In a similar study Sofka and colleagues evaluated 46 shoulders with the clinical diagnosis of idiopathic frozen shoulder and attempted to correlate the clinical stage with the MRI characteristics. They found a mean thickness of the axillary pouch of 7.5 mm; this was significantly greater than the axillary pouch measurements in other stages, with means of less than 5.5 mm. Normal capsular and synovial thickness has been found to be less than 3 mm. Abnormalities within the rotator interval are also common. Mengiardi, Gerber, and colleagues described the “subcoracoid triangle sign” as obliteration of the normal-appearing fat between the coracoid process and the coracohumeral ligament on the sagittal oblique images.
The use of IV gadolinium can improve the diagnostic capability of MRI in patients with frozen shoulders. Connell and colleagues noted soft tissue density showing variable enhancement in the rotator interval and partially encasing the biceps anchor on MRI after gadolinium administration. They also noted thickening and gadolinium enhancement of the axillary pouch. Carrillon and colleagues performed IV gadolinium administration followed by MRI and demonstrated enhancement of the synovial lining in patients with frozen shoulder. Tamai and Yamato found enhancement of the joint capsule, which was not seen in patients with subacromial impingement. However, a study that compared findings on MR arthrography in patients with and without frozen shoulder did not identify any specific diagnostic findings for frozen shoulder.
Despite interest shown in it, MRI has not proved to be essential in the diagnosis or management of patients with stiff shoulders, and its routine use in this setting is not supported. Specific indications include concern for underlying rotator cuff integrity and the possibility of a soft tissue or bone neoplasm.
Several investigators have reported on the use of ultrasound in the evaluation of frozen shoulder. Ryu and colleagues noted that the main sonographic feature of frozen shoulder was a constant limitation of the sliding movement of the supraspinatus tendon against the scapula. They reported sensitivity of 91%, specificity of 100%, and accuracy of 92% when comparing ultrasound with arthrography as the gold standard for diagnosis. As with MRI, ultrasound is not a required diagnostic study and is usually indicated only for evaluating the rotator cuff in patients whose physical examination suggests the possibility of a rotator cuff tear coexisting with shoulder stiffness. With the use of ultrasound, Homsi and colleagues found that coracohumeral ligament thickness was greater in idiopathic frozen shoulder patients than in controls.
Before performing arthroscopy, the diagnosis of stiffness should already have been made and perhaps confirmed by examination under anesthesia. Arthroscopy allows evaluation and possible treatment of additional pathology, such as rotator cuff tears, impingement, biceps disease, and articular cartilage lesions.
Neviaser described four arthroscopic stages of adhesive capsulitis and proposed that these stages could be used to guide treatment planning :
A mild erythematous synovitis
Acute synovitis with adhesions in the dependent folds of the synovial lining
Maturation of adhesions with less reactive synovitis
Chronic adhesions without synovitis
Arthroscopic treatment of shoulder stiffness is discussed in detail later in this chapter.
Frozen shoulder can be a disabling problem for many patients due to the functional limitations and associated pain. Although the literature indicates that most will improve without treatment, many patients are not willing to accept that it may take 2 years for significant resolution of their symptoms (see Fig. 19-6 ). Most patients present to their physician expecting not only a diagnosis but also a treatment plan to expedite recovery. As with most orthopedic pathology, initial treatment is usually conservative, consisting of activity modification, antiinflammatory medications, and physical therapy. If these treatments fail, there are several more aggressive options, such as manipulation under anesthesia, open surgical release, and arthroscopic release. Many recent publications have focused on the efficacy of arthroscopic procedures to treat frozen shoulder, with promising results.
Understandably, patients become frustrated by continued pain and impaired function. Many are unable to work for some period of time. It is often difficult for patients to understand the condition and accept the uncertainty of resolution. Although there is no consensus on optimal treatment, there appears to be a consensus that some form of treatment, nonoperative or operative, is indicated in any patient with a stiff shoulder.
Nonoperative treatment is the initial approach in almost all patients. This can include oral medications, physical therapy, injections, or other modalities. When patients fail to respond to nonoperative treatment, operative intervention may be indicated. This includes manipulation under anesthesia, surgical release (open or arthroscopic), or some combination of these two treatments.
Various types of analgesics and nonsteroidal antiinflammatory drugs (NSAIDs) have been used as supportive measures to manage the pain and inflammatory component of the stiff shoulder. Their effectiveness in alleviating the pain component has been well established. Some authors have reported greater improvement in pain in patients who were managed with nonsalicylate analgesics than in those administered NSAIDs. Others have reported that oral steroids produce significantly better improvement in ROM, function, and pain relief in patients with frozen shoulder, but these benefits were not maintained beyond 6 weeks. Greater improvement can be expected in patients who combine the intake of such medications with a regular exercise program. Some physicians prescribe short-term, low-dose oral steroids for frozen shoulder, as these can act as a powerful antiinflammatory. Although risks associated with prescribing these medications are small, physicians should still discuss them with patients. We do not routinely prescribe oral steroids for patients as intra-articular steroid injections have been shown to be superior.
Physical therapy is a mainstay of initial treatment for shoulder stiffness that has been present for less than 3 to 6 months or in patients who have had no previous treatment. Patients are usually started with active-assisted ROM and gentle passive stretching exercises. Most exercises are easier in the sitting or supine position. Repetition is probably more important than doing one long session. We recommend short sessions of 5 to 10 minutes repeated several times per day, with the goal of trying to push the affected shoulder just slightly past the point of pain during each session. Sometimes it is helpful to have a daily progression chart that the patient can follow. Generally, improvement is seen in small increments. As a result, patients sometimes do not acknowledge their improvement and can become discouraged and frustrated. Patients are encouraged to take NSAIDs and apply heat to the affected shoulder before exercise and to apply ice after exercise. This regimen can help reduce discomfort and improve compliance with the exercises.
Other modalities, such as microwaves, short waves, and heat lamps are sometimes considered as additional measures in the rehabilitation process, although they have not proved to be particularly beneficial in any specific phase of shoulder stiffness. Similarly, adding ultrasound to the physical therapy exercises has not shown any additional benefit to therapy alone. Other modalities, such as electrophysiotherapy, massage, or more atypical modalities, such as hyperbaric oxygen and magnetotherapy have been used, but no additional benefit from these could be proved, and none of these treatments fits into a standard shoulder rehabilitation algorithm.
Adding stimuli to distract the mind of the patient from painful sensations has been shown to be helpful during physical therapy exercises. In one study, using audio analgesia as an adjunct to mobilization exercises in patients with chronic frozen shoulder resulted in significant improvement in the recovery of motion and a reduction in the number of treatments necessary for recovery. Using the same rationale, transcutaneous nerve stimulation has been used in patients with frozen shoulder undergoing physical therapy. Rizk and colleagues used transcutaneous nerve stimulation on a group of 28 patients to control their pain during the application of progressive abduction traction. Jurgel and colleagues reported on 10 patients with frozen shoulder who were treated with 4 weeks of rehabilitation that combined exercises with massage and electrical therapy and were able to demonstrate that this protocol improved shoulder ROM in all directions except rotation. They compared this group to a group of 28 patients treated with heat and therapeutic exercises and observed that the increase in ROM was significantly greater in the former group. This may be explained by the strong attachments between collagen fibers, which show high resistance to suddenly applied tension but which also tends to creep when prolonged tension is applied.
Exercises should be performed gently. Forceful stretching maneuvers are contraindicated, especially in the early phases of frozen shoulder. All forms of light resistance strengthening should be postponed until the patient has recovered ROM.
For patients to see improvement from therapy, they must assume primary responsibility for their condition. They should understand and participate actively in the prescribed exercises and be prepared to tolerate some amount of discomfort. Although it can take several months, a closely supervised physical therapy program will lead to improvement in pain and ROM in up to 90% of patients with chronic frozen shoulder. In addition, physical therapy has been shown to improve health-related quality of life.
Nicholson reported on the therapeutic advantage of performing passive stretching in abduction in addition to active ROM exercises. Watson-Jones recommended 3 minutes of active stretching each hour and demonstrated that of 226 frozen shoulders treated with this stretching exercise program alone, only 5% failed to regain satisfactory ROM within 6 months. In another study, 90% of 75 patients who completed a nonoperative program for frozen shoulder achieved satisfactory results, with only 7 patients requiring more aggressive intervention. Russell and colleagues found that a hospital-based group physical therapy program was more effective for restoring ROM and functional scores than individual therapy or a home-based program.
Dudkiewicz and colleagues reported long-term follow-up (mean, 9.2 years; range, 5.5 to 16 years) on 54 frozen shoulder patients who were managed conservatively with physical therapy and NSAIDs. Good long-term outcomes were reported with significantly improved motion in all the measured directions.
Occasionally, physical therapy does not help to improve symptoms in patients with frozen shoulder and at times can even exacerbate them. In one report only 60% of patients who received physical therapy, along with other modalities, achieved the ability to sleep pain-free after 5 months of treatment. Hazelman reported on a group of patients who received physical therapy alone; one third of them experienced an increase in pain, and only half of this group experienced significant improvement with exercises. Diercks and Steven conducted a prospective study on 77 patients with frozen shoulder to compare the effect of intensive physical therapy and manual stretching versus home exercises within the pain limits (“supervised neglect”). At 24 months’ follow-up, 89% of patients treated with supervised neglect had normal or near-normal painless shoulder function with a Constant score greater than 80. However, only 63% of patients in the group that received intensive physical therapy reached a Constant score of 80 or higher.
Lin and colleagues demonstrated the existence of trapezius muscle imbalance with overactivity of the upper compared with the lower trapezius muscle in patients with frozen shoulder. They concluded that this imbalance in the trapezius muscle might contribute to scapular substitution in compensation for impaired glenohumeral motion and thus rehabilitation of the trapezius should be included in the therapy protocol for patients with frozen shoulder.
Several different types of injections have been described for treating shoulder stiffness. The simplest types include intra-articular and subacromial injections. Most of the reports on intra-articular injections involve patients with frozen shoulder and not those with acquired stiffness. Intra-articular injections can be administered along with distention arthrography (brisement), combining chemical and mechanical modalities of treatment. Other types of injections include periarticular (trigger point) injections and nerve blockade injections (of the suprascapular and upper and lower subscapular nerves).
It has been proposed that shoulder stiffness begins with an inflammatory phase followed by scar formation. The theoretical rationale for an intra-articular steroid injection is therefore to inhibit this inflammatory phase, reduce pain, and prevent further development of stiffness. The reported effectiveness of intra-articular steroid injections varies widely.
In most instances intra-articular injections are administered in combination with a physical therapy protocol. For this reason, it is difficult to determine the efficacy of this treatment as an isolated modality. Some investigators have suggested that intra-articular corticosteroid injections provide little or no benefit in the management of shoulder stiffness. Conversely, other investigators have shown beneficial effects from intra-articular steroid injections, mainly of improvement in pain but not in ROM. When an injection and exercise regimen was compared with analgesics alone, greater improvement was observed with the former.
Widiastuti and Sianturi compared oral triamcinolone intake versus intra-articular triamcinolone injection and concluded that intra-articular steroid administration provided significantly faster improvement when compared with the oral route. Lorbach and colleagues demonstrated similar findings. Williams and colleagues compared exercises and repetitive intra-articular injections of hydrocortisone acetate to serial stellate ganglion blocks, but they were unable to demonstrate significant differences between the treatment groups.
Others have demonstrated intra-articular steroid injections to be more advantageous than trigger point injections. In one study 25% of patients with frozen shoulder benefited from an intra-articular injection, whereas none of the patients who received trigger point injections experienced significant relief. In a study by Carette and colleagues corticosteroid injection plus exercise led to significantly greater improvement than exercise alone or placebo treatment. However, according to Thomas’s group this improvement may be related to a decrease in pain rather than a real improvement in ROM. Ryans and colleagues demonstrated in a randomized, blinded, placebo-controlled study that an intra-articular steroid injection was effective in improving shoulder-related disability, which, in combination with home exercise therapy, led to improvement in shoulder external rotation 6 weeks after treatment.
Subacromial injections might also have a role. A prospective, randomized, observer-blind study comparing subacromial steroid injection and physiotherapy demonstrated that injections were as effective as physical therapy alone or in combination. Some investigators have suggested the administration of a combination of injections into both the joint and the subacromial space. However, in comparative studies, no significant differences in pain relief and shoulder ROM at follow-up examination were demonstrated after combined injections. This is not unexpected because in frozen shoulder and many cases of acquired stiffness, much of the pathology is intra-articular. Oh and colleagues demonstrated no significant difference in pain, Constant score, or ROM between intra-articular and subacromial injections at more than 6 weeks after injection. These authors concluded that subacromial injections are a reasonable alternative to glenohumeral injections. We routinely perform fluoroscopic glenohumeral injections for all idiopathic frozen shoulder patients because we believe that much of the acute pain associated with the condition may be from glenohumeral synovitis, which is clearly evident in our arthroscopic experience.
In our practice we believe that during the early phase of frozen shoulder, patients might not be able to tolerate physical therapy exercises because of pain. In such cases an intra-articular steroid injection might provide enough relief for the patient to begin an exercise program. We agree with the approach of Weiss and Ting and recommend that all intra-articular injections be performed under radiologic guidance to confirm intra-articular placement. In one study physicians were monitored during the intra-articular injection of radiopaque dye, and the majority of them were unsuccessful in delivering the fluid into the shoulder joint.
When recommending an intra-articular steroid injection, it should be kept in mind that this treatment is not completely benign and complications have been reported. The deleterious effects of intra-articular steroids on tendon metabolism and articular cartilage have been shown in numerous studies. A case of fatal clostridial myonecrosis after an intra-articular injection of steroids in the shoulder has also been reported. Harryman and Lazarus reported on six cases of chronic sepsis after shoulder arthrography and steroid injections. Shoulder fusion was performed in some of these cases.
Other types of intra-articular injections have garnered interest. Tamai and colleagues demonstrated that intra-articular injection of hyaluronate led to the suppression of synovitis, suggesting it has an antiinflammatory property. In a randomized study intra-articular injection of sodium hyaluronate plus steroid was compared with steroid alone for the treatment of frozen shoulder, and both groups improved substantially. Rovetta and colleagues showed that an intra-articular injection of sodium hyaluronate was not as effective in improving ROM in the first 3 months after administration as an intra-articular injection of steroid or physical therapy. In another study intra-articular hyaluronic acid was injected into painful shoulders due either to frozen shoulder or to osteoarthritis. In a short-term follow-up both groups improved substantially in joint comfort and mobility.
Intra-articular injection of enzymatic proteases, such as α-chymotrypsin and hyalase have been used to break up glenohumeral capsular fibrosis. A few studies have reported the results of intra-articular injection of α-chymotrypsin and hyalase combined with physiotherapy, but beneficial effects of this treatment modality were not conclusive. These enzymes are rarely used in current practice.
Payr first reported capsular distention , also known as distention arthrography or brisement , in 1931. This is a technique that relies on rupturing the glenohumeral capsule by fluid injection. By injecting progressively higher volumes of fluid into the glenohumeral joint, the pressure eventually builds up until it is high enough for the capsule to be disrupted. This is evident when the pressure of the injection decreases significantly. When capsular disruption occurs, it usually involves the weakest point in the capsule, the biceps tendon sheath, or the subcoracoid bursa.
During the initial injection phase, an arthrogram is often obtained to confirm that the fluid is being injected inside the joint. Some surgeons inject 60 to 100 mL before manipulation in order to produce hydrostatic distention of the joint capsule. More fluid is then injected until the intra-articular pressure increases progressively from 800 mmHg up to a maximum of 1500 mmHg. Variations of this technique include the injection of saline with local anesthesia and arthroscopic distention.
Capsular distention has received wide attention because it can be performed in a radiology suite or in an office setting and carries low risk. Several authors have claimed that it is a simple procedure to perform and that it is an effective mechanism for achieving lasting symptomatic relief in patients with frozen shoulder. However, a review of the literature reveals that the results of this modality are variable, as with all other methods of treatment of frozen shoulder. In addition, steroid injections or manipulations are often performed in conjunction with distention arthrography, which makes comparison between reports and the evaluation of its effectiveness very difficult.
Most cases of distention arthrography reported in the literature have been performed for frozen shoulder. In one study the outcomes of distention arthrography for frozen shoulder and acquired stiffness were compared, with better rates of recovery after 6 months observed in the frozen shoulder group.
In a double-blind prospective study involving 45 patients with frozen shoulder, a comparison was made between distention and nondistention arthrography combined with intra-articular steroid injection. A significant improvement in nocturnal pain was demonstrated in both groups during the first 3 months following the procedure, but no significant difference in ROM and degree of pain relief could be demonstrated between the two groups after 3 months. Rizk and Pinals performed brisement on 16 patients by injecting 30 mL of fluid that contained 20 mL of contrast material, 2 mL of steroid, and 8 mL of lidocaine. This led to capsular rupture in all patients, as evidenced by dye extravasation into the subscapular recess or subacromial space. Thirteen patients experienced immediate relief of pain, with this improvement maintained for 6 months. In two patients who did not experience pain relief, the rupture of the capsule occurred at the level of the distal bicipital sheath. A more recent study by Amoretti and colleagues showed similar results. In addition, Khan and colleagues demonstrated that distention arthrography with intra-articular steroid administration and physical therapy was superior to physical therapy alone.
The patients who experience good results after brisement are those with less severe restriction of ROM and a moderate joint distention before rupture. Improvement in ROM is expected after repeated injections. In a study by Piotte and colleagues the efficacy of repeated distention arthrographies performed at 3-week intervals was evaluated. They showed that two distention arthrographies with steroid, combined with a home exercise program, significantly improved shoulder impairments and disability. However, they were not able to demonstrate a benefit from a third brisement.
When brisement is compared with manipulation under anesthesia (MUA), the reports are variable. One study by Sharma and colleagues demonstrated significantly better results after distention arthrography. However, other studies have shown that distention arthrography compares unfavorably with MUA.
Periarticular Trigger Point Injection and Nerve Blockade
Trigger points are well-localized painful areas around the subscapularis tendon and the periscapular muscles. It is believed that the tenderness of these points is secondary to fibrositis or myofascial inflammation. In a study by Steinbrocker and Argyros that included 45 patients with no control group, 95% of patients restored 85% of their function after multiple injections into the joint capsule, bicipital tendon, supraspinatus tendon, and subdeltoid bursa. Other studies have shown that injections into sites other than those mentioned by Steinbrocker and Argyros, including the periarticular region, did not provide long-term benefit or pain relief.
The idea of performing nerve blockade to improve the pain component of frozen shoulder started with Wertheim and Rovenstine and later Kopell and Thompson who realized that the suprascapular nerve innervates the joint capsule and could be the main pain generator in patients with frozen shoulder. They demonstrated that suprascapular nerve blockade performed on 20 patients led to substantial pain relief within 24 hours in most of the patients.
For the same reasons, other authors have recommended a suprascapular block with a local anesthetic and steroid to relieve pain. In a randomized case-control study patients with a diagnosis of frozen shoulder were randomized into either a suprascapular nerve blockade group or a placebo cohort. Significant improvement in pain was demonstrated in the treated patients at 1 month after injection. Another study compared suprascapular nerve blockade to intra-articular corticosteroid injection and found improved pain relief and function in the nerve block group. Repeated suprascapular nerve blockade has also been shown to improve patients’ tolerance to deep pressure on the shoulder and to improve passive ROM in frozen shoulder associated with reflex sympathetic dystrophy.
Jankovic and van Zundert studied five patients with frozen shoulder who did not respond to conventional treatment. All these patients underwent a combination of subscapularis nerve blockade along with infiltration of trigger points and all experienced significant pain relief. The authors concluded that the combination of nerve block and trigger point infiltration of the subscapularis muscle might have both therapeutic and diagnostic value for the treatment of frozen shoulder.
In a study by Sun and colleagues patients were randomized into two groups to receive physical therapy alone or therapy plus acupuncture, with significant improvement seen in the Constant scores of the patients in the acupuncture group. In another study by Lin and colleagues 150 patients were randomized into three groups and treated with regional nerve blocks, electroacupuncture, or a combination of both treatments. The combination of both treatments resulted in significantly better ROM and longer-lasting pain relief than either treatment alone. Another study showed that undergoing six acupuncture sessions led to complete recovery in patients with frozen shoulder who had experienced limited success with conventional physical therapy.
The injection of subcutaneous calcitonin and mobilization therapy has been studied prospectively in 50 patients with shoulder stiffness secondary to a number of different etiologies. A significant effect on pain reduction was observed in patients with posttraumatic shoulder stiffness when compared with stiffness from other causes. Other investigators using this modality have documented similar improvement in patients with frozen shoulder and other inflammatory conditions.
Coventry surmised in 1953 that radiation therapy did not bring much benefit in the treatment of chronic frozen shoulder. In a study by Quin radiation therapy was compared with heat and ultrasound but no treatment advantage was found. In another prospective study of 233 patients improvement was found in only 26% of patients. However, in other large series radiation therapy was shown to improve pain in up to 70% of patients with shoulder stiffness, although the long-term risks were not evaluated. Radiation therapy might have a role as a treatment option in patients with stiff shoulder-related heterotopic ossification to prevent recurrence after excision. Otherwise, the use of radiation in the United States for the treatment of idiopathic frozen shoulder has been abandoned.
Patients who do not regain satisfactory ROM or fail to demonstrate progress after 3 to 6 months of nonoperative care may be candidates for operative intervention. Around 5% to 20% of patients who continue to have significant loss of shoulder motion will remain functionally disabled. Management options include MUA (with or without arthroscopy) or surgical release (open or arthroscopic). Patients should understand that any form of operative intervention will usually be followed by intensive physical therapy. The decision to proceed with surgical treatment is controversial as some surgeons consider frozen shoulder to be a self-limited process. However, many patients are not willing to accept the prolonged pain and stiffness associated with the natural history of the disease, and so one of the main premises of operative intervention is to expedite the recovery process even if long-term outcomes may be equivalent.
Manipulation Under Anesthesia
Historically, closed MUA has often been recommended as the next step after the failure of nonoperative treatment in patients with stiff shoulder. However, cases of acquired stiffness in which extra-articular adhesions have developed, as well as changes to the articular anatomy, might not respond well to manipulation. Manipulation might be attempted in the acute phase of motion loss when therapy has failed to result in regained motion.
Patients with worsening symptoms after at least 3 months of an appropriate nonoperative exercise program are candidates for MUA. Most reports of MUA have been for patients with frozen shoulder, and the results of this intervention are varied. In one study a greater degree of improvement was achieved in patients who were symptomatic for more than 6 months before manipulation compared with those with a shorter duration of symptoms. This is probably related to the timing of the intervention in relation to the inflammatory phase of frozen shoulder. This phase is usually painful, and as pointed out by Neviaser and Neviaser, any operative intervention in this stage is likely to exacerbate the patient’s symptoms and motion loss because of increasing capsular injury. Manipulation should be postponed until pain is experienced only at the extremes of motion, which indicates resolution of the inflammatory phase.
In reports by Neviaser and DePalma the immediate effects of manipulation on structures about the shoulder were observed with arthroscopic and open surgery. Tears were seen in the subscapularis muscle and tendon, the anterior and inferior capsule, the supraspinatus tendon, and the long head of the biceps tendon. In addition, because many patients with frozen shoulders have osteopenia of the proximal humerus, fractures may occur with manipulation. Harryman reported on a referred patient with a complete brachial plexus palsy after MUA that resulted in anteroinferior dislocation. The reported cumulative risk of an adverse event after MUA is less than 1%.
Although complications can arise, other authors have contended that manipulation is a safe and reliable treatment if performed appropriately. Atoun and colleagues manipulated 32 shoulders and showed that manipulation was not associated with rotator cuff tears, fractures, dislocations, or nerve palsies in their cohort.
As with patients with severe osteopenia, MUA is not recommended for patients with long-term diabetes mellitus (lasting longer than 20 years) because these patients are extremely resistant to this method of treatment. In two studies with short follow-up of approximately 6 months, recurrent stiffness was reported to be between 5% and 20%. However, in a study with a longer follow-up of patients with long-term, insulin-dependent diabetes, Janda and Hawkins reported an unacceptably high rate of recurrent stiffness after MUA.
Manipulation can be performed under either regional or general anesthesia, although anesthesia with local injection combined with hydrostatic distention has also been reported. It is essential during the manipulation that complete muscle paralysis be achieved. A regional block is advantageous because the patient can witness the improved ROM immediately after surgery. The immediate postoperative pain is also eliminated, which allows immediate participation in ROM exercises. The block can be administered as a single injection or with placement of an indwelling interscalene catheter. A single block of 0.5% bupivacaine provides anesthesia for approximately 12 hours and can be repeated on postoperative days 1 and 2 for continued analgesic effect. If an interscalene block is used, the patient can be maintained on a continuous drip of bupivacaine that provides continuous pain relief during the patient’s hospital stay and substantially improves the patient’s ability to comply with the intensive physical therapy exercises postoperatively. The addition of systemic steroids has not been shown to have any lasting benefit.
Codman and Neviaser both described attaching the arm to the bed in a position of abduction and external rotation in order to maintain the motion obtained by the manipulation. This maneuver can require significant narcotic analgesia and primarily stretches the anteroinferior portion of the capsule, with the remaining sections of capsule left in a position that can lead to recurrence of the contracture. Other authors have recommended traction to stretch residual contractures and improve motion.
To perform the manipulation, the patient is placed supine on the operating room table. Complete muscular paralysis facilitates the procedure and minimizes the risk of fracture. The axillary border of the scapula is stabilized, allowing isolated glenohumeral movement. A constant controlled force is applied to the proximal end of the humerus while the scapula is kept stable; any sudden force increases the risk of fracture or injury to shoulder soft tissues. Gradual traction and flexion are carried out first until an audible and palpable release is heard and felt, indicating rupture of the inferior capsule. The arm is then manipulated into adduction across the patient’s chest to stretch or rupture the posterior capsule and restore internal rotation. We perform these manipulations first in order to minimize the theoretical risk of fracture due to the increased torque produced with the rotational movements that follow. The arm is then moved to the side of the body and is held at the supracondylar level while the forearm is rotated very gently into external rotation, taking care not to exert an excessive force that would risk elbow ligament injury. Abduction in the scapular plane is performed with additional internal and external rotation to further release the anterior and posterior capsule. Some steps may be gently repeated until ROM similar to that of the contralateral shoulder is achieved.
Charnley warned against beginning the manipulation with an abduction force because he believed that external rotation had to be achieved first to prevent dislocation. Nevertheless, many authors recommend starting with an initial abduction force. A good prognostic sign is the production of crepitus followed by immediate full ROM. If crepitus does not occur, increased force should not be applied. Repeat manipulation may be indicated if the recovered motion is not symmetrical to the opposite side or if motion loss recurs within a short period postoperatively. Some authors have reported good results with translational rather than rotational manipulative forces and recommend their application to lessen the chance of fracture.
Although its use has not been proved to enhance the outcome in patients with stiff shoulders, many authors recommended corticosteroid injections after MUA to reduce pain, diminish local inflammation, and decrease the chances of early healing of the disrupted capsule. In a study of more than 100 patients who underwent MUA, codeine was the only medication required to control pain in the immediate postmanipulation period. In a study by Weiser in which MUA was performed under a local anesthetic, there was full recovery in 60% of patients with three to five treatment sessions. In another study by Thomas and colleagues, 30 patients with frozen shoulders were randomly allocated to one of two groups, one with manipulation and intra-articular steroid injection and the other with intra-articular steroids alone. Of the patients managed with MUA and steroid injection, 80% had decreased pain at follow-up; 40% achieved good ROM, whereas only 47% of patients who received steroid alone had decreased pain, and only 13% achieved good ROM.
The reported outcomes of MUA with or without steroid injection have been highly variable. Good results have been reported in most reviews, except with the long-term diabetic population that is resistant to this treatment and is at a high risk of developing recurrences. Haines and Hargadon reported on 78 patients with a mean preoperative duration of symptoms of 2 to 6 months who underwent MUA and steroid injection. At a mean follow-up of 12 weeks, 83% of the patients regained 80% of their glenohumeral motion. Hill and Bogumill reported on 14 frozen shoulder patients with a mean preoperative duration of symptoms of 5.4 months, who underwent MUA and steroid injection. At a mean follow-up of 22 weeks, 75% of the patients were pain-free. Kivimaki and Pohjolainen reported on 24 patients who underwent either MUA alone or MUA with steroid injection and found same results in both groups at 4 months’ follow-up.
Weber and colleagues reported on 43 patients with frozen shoulder and a mean preoperative duration of symptoms of 6 months who underwent MUA along with inpatient physiotherapy. At a mean follow-up of 4.7 years, 73% of the patients had gained full recovery of function. Harmon reported on MUA in three separate follow-up studies 2 to 3 years after manipulation, in which 64% to 94% of patients achieved painless motion. Dodenhoff and colleagues reported on 37 patients with a mean preoperative duration of symptoms of 8 months who underwent MUA. They found that 94% patients were satisfied with their result, but 12.8% retained significant persistent disability.
Othman and Taylor reported on 74 patients who underwent MUA and found that significant improvement in range and comfort occurred as early as 3 weeks postoperatively. Sharma and colleagues reported on 32 patients who underwent either MUA or distention with local anesthesia and found significantly better results in the distention group. In a study by Reichmister and Friedman 97% good results were reported in patients who underwent MUA, but 8% of their patients required a second manipulation. Melzer and colleagues reported on 110 patients, comparing those who underwent physiotherapy with medication, as needed, with those who underwent MUA; they concluded that physiotherapy was better than MUA. In a study by Farrell and colleagues MUA was performed on 25 patients with frozen shoulder. The authors reported significant improvement in pain and function, which was maintained for 15 years postoperatively.
Ekelund and Rydell reported on 22 patients who underwent distention arthrography with local anesthetic and steroid followed by MUA and found that 91% of patients had no or slight pain and 83% had almost normal ROM. Similar studies have been performed with local intra-articular anesthesia and manipulation, with the authors claiming successful pain relief and recovery of motion in almost two thirds of patients, with a greater percentage of patients relieved of discomfort alone. Lundberg noted that MUA did not affect the natural history or the time course of the disease.
Manipulation and Arthroscopy
MUA usually produces capsular rupture, leading to improved motion, but does not affect intra-articular synovitis. Synovitis may be a major precipitating factor in the development of shoulder stiffness and a contributing factor leading to the recurrence of contracture. In a study by Pollock and colleagues 83% of patients with frozen shoulder who were treated with MUA and arthroscopic debridement achieved a successful result, although patients who were recalcitrant to treatment underwent additional release of the coracohumeral ligament. In a study by Andersen and colleagues 79% of patients who had MUA followed by arthroscopic debridement were pain-free, and 75% had normal ROM at 12 months of follow-up. In a recent study Rill and colleagues demonstrated that frozen shoulder patients with pain and stiffness refractory to conservative means did well following manipulation and arthroscopic release.
Open Surgical Release
Patients who have acquired stiffness resulting from contracture of the extra-articular soft tissues, such as after a Putti-Platt procedure, are unlikely to benefit from MUA. They might not be good candidates for arthroscopic release because the contracture involves both the subscapularis tendon and the capsule. Open surgical excision of the scar and release of extra-articular adhesions, with or without Z-plasty of the capsule and subscapularis tendon, may be the treatment of choice in this population.
Historically, Codman described open lysis of adhesions in the subacromial and subdeltoid bursae. In patients who experience recurrent stiffness after MUA, Neviaser found it necessary to perform an arthrotomy through an anterior axillary approach, releasing periarticular adhesions and the contracted articular capsule from the humeral head. Leffert recommended targeting the structures responsible for restricted motion with a surgical release when patients failed to improve after 6 months of a nonoperative treatment regimen. Lippman suggested the open release of adhesions around the long head of the biceps tendon to liberate the shoulder from stiffness. Simmonds proposed complete excision of the long head of the biceps tendon in cases of severe frozen shoulder. He noted no intra-articular adhesions between the surface of the joint and the articular capsule when he performed these procedures; however, the operative management that he described did not lead to improvements in ROM. Matsen and Kirby recommended open surgical release of the capsule in patients with shoulder stiffness if they failed to improve after 6 months of home exercise therapy. They found this approach to be safer than MUA for patients with osteoporosis and recalcitrant stiffness.
Harmon performed soft tissue release of contracted tissues about the joint in eight patients who failed to improve after gentle manipulation and found the outcome to be similar to that in patients who underwent closed manipulation alone, except two patients who had intractable stiffness. He also treated 30 cases of stiffness with acromioclavicular joint and acromial excision and was able to restore active abduction to around 160 degrees by the third postoperative month.
Kernwein and colleagues performed open release of the capsule and coracohumeral ligament in 4 of 12 patients with frozen shoulder. Tissues were found to be markedly thickened and inelastic. In a study by Ozaki and colleagues the authors reported their findings during the surgical release of 17 patients who had recalcitrant shoulder stiffness. They found that the contracted coracohumeral ligament in the rotator interval was the major tether restricting glenohumeral motion, and they noted that the long head of the biceps tendon was inflamed and stenosed beneath this contracted structure. Nobuhara and colleagues, who performed rotator interval capsular release on 21 shoulders and obtained remarkable improvement in ROM, also appreciated the role of the rotator interval in stiff shoulders.
In one study 25 of 75 patients who failed therapy and MUA underwent open surgical release. At surgery, the coracohumeral ligament was found to be fibrotic and contracted and was excised. Of these 25 patients, 20 achieved good or excellent results; outcomes were worse in patients with long-term diabetes.
As an alternative to the open release of contracted tissues, Baumann described denervation of the ventral aspect of the shoulder capsule, reporting that this method caused a progressive decrease in or total elimination of shoulder pain in 85% of 20 shoulders treated in this fashion. However, this is the only report about this particular technique.
Another population who develops internal rotation contracture is that of patients who had a stroke; as a result, their internal rotator muscles, including the pectoralis major and the subscapularis muscles, spasm and eventually shorten and contract. Braun and colleagues incised the pectoralis major muscle, excised the subscapularis tendon, and preserved the anterior capsule in 13 patients who had this problem. Ten of these 13 patients regained complete pain relief, 20 degrees of external rotation, and 90 degrees of abduction.
When we perform an open surgical release, our preference is to place the patient in the beach chair position. An interscalene nerve block is performed usually along with general anesthesia. Patients are admitted postoperatively for 48 hours and undergo supervised physical therapy exercises.
We favor the deltopectoral approach unless a prior incision dictates another approach. Layered dissection is performed to excise scar in all tissue planes. The shoulder is placed in abduction and slight flexion to relax the deltoid. Adhesions in the subdeltoid plane are released, taking special care not to injure the axillary nerve, which can be palpated in the substance of the deltoid 3 to 5 centimeters distal to the lateral aspect of the acromion. Once all adhesions have been removed from the subdeltoid region, the humerus may rotate freely beneath the deltoid and the dissection is extended medially to release the adhesions from the subacromial space.
Next, the interval between the conjoined tendon and subscapularis muscle is addressed. The axillary nerve is identified and protected throughout the procedure to allow thorough releases to be performed safely. Adhesions are released by a combination of blunt and sharp dissection. The subscapularis tendon is identified and the rotator interval is released from the humerus to the coracoid.
The shoulder is examined at this point to evaluate the effectiveness of the release. In patients with severe internal contractures the subscapularis tendon as well as the anterior capsule is usually shortened and scarred. In such a case, the subscapularis tendon and capsule can be released in the coronal plane using Z-plasty. With this method, the superficial half of the tendon remains attached to the muscle, and the deep half remains attached to the lesser tuberosity. Stay sutures may be placed in the superficial tendon and are used to pull on the muscle, aiding the dissection and the excision of scars tethering the subscapularis muscle to the surrounding tissues. This often requires identification of the axillary nerve and placement of a vessel loop around it to protect it. However, we rarely now perform this procedure to lengthen the tendon. Instead, a complete release of the subscapularis from the anterior glenoid neck and subcoracoid space as well as anteriorly, after identifying the axillary nerve, can achieve sufficient excursion of the tendon to restore physiologic external rotation.
The posterior and inferior capsules are released further if abduction and internal rotation are still limited. A humeral head retractor is placed to retract the head posteriorly to expose the posterior capsule, which is then released from inferior to superior. The shoulder is then placed in maximal external rotation with the arm at the side of the body, and the subscapularis is repaired in a lengthened position using large nonabsorbable braided sutures.
Once the repair is completed, the shoulder is reexamined and the safe zone of early passive ROM is determined. This information is relayed to the physical therapist who will be performing the passive ROM exercises on the patient. In postoperative rehabilitation early, passive rather than active ROM is started until soft tissue healing is substantial at around 4 weeks.
Disadvantages of open surgical release include postoperative pain and the need to protect a repaired or lengthened subscapularis tendon, which explains the less predictable results reported using this technique in patients with acquired stiffness. Many of these patients have a concomitant injury to the joint or more extensive soft tissue damage that can limit motion recovery. Successful outcomes are determined by the quality of the surrounding tissue envelope and the status of the articular surfaces.
Arthroscopic Capsular Release
Conti, from France, performed the first arthroscopic partial release of a stiff shoulder capsule in 1979. He used a trocar and forceps to release the rotator interval, and he injected intra-articular steroid and performed gentle manipulation. He reported on 18 patients on whom he performed this technique; 16 of them recovered fully within 3 weeks and the other two recovered within 3 to 6 months.
In 1980 Wiley and Older reported on a new arthroscopic technique that they had performed in 10 patients with stiff shoulders. In this technique they performed an arthroscopic examination of the glenohumeral joint, combined with joint distention to stretch the tight capsular constraints and then performed a manipulation. All 10 patients were relieved of their symptoms after the procedure.
In 1986 Ogilvie-Harris and Wiley used blunt instruments and lacerated the anterior capsule to release a tight glenohumeral joint in patients with recalcitrant stiffness secondary to frozen shoulder. They reported satisfactory results in most of the 81 patients on whom they performed this technique. Their results were less satisfactory in patients who had diabetes.
In 1994 Pollock and colleagues reported on 30 shoulders treated with manipulation under interscalene block, followed by diagnostic arthroscopy, debridement of the glenohumeral joint and subacromial space, and sectioning of the coracohumeral ligament. Twenty-five (83%) of the patients had satisfactory results, but only 64% of the patients with diabetes had satisfactory results.
Warner introduced the modern technique of arthroscopic capsular release for primary, idiopathic frozen shoulder in 23 patients in 1997. He performed a complete rotator interval release in addition to complete sectioning of the anterior capsule, middle glenohumeral ligament, and anteroinferior glenohumeral ligament. Significant gains in motion were immediately achieved at the time of surgery, and these were sustained at follow-up of a minimum of 2 years. The technique has evolved to include release of the posterior capsule. Chen and colleagues found that extending the release posteriorly beyond the anteroinferior glenohumeral ligament to the posteroinferior glenohumeral ligament improved short-term ROM, but there was no significant difference at 6 months. We routinely extend the capsular release posteriorly to include posteroinferior glenohumeral ligament in order to relieve any internal rotation deficit, which, in our experience, is present in nearly all patients with significant idiopathic adhesive capsulitis.
The morbidity associated with complete arthroscopic capsular release is minimal when compared with open release, with similar rates of recurrent stiffness and earlier pain relief. Complications reported include complete axillary nerve palsy, anterior dislocation occurring immediately postoperatively, and contracture recurrence, which can occur in up to as many as 11% of patients.
Harryman and colleagues demonstrated a remarkable improvement in six of nine health status scores on the SF-36 general health survey and excellent recovery of function as shown by results of SST after arthroscopic capsular release. They did not observe any significant differences between diabetic and nondiabetic patients, although refractory stiffness developed in three patients with insulin-dependent diabetes. Warner and colleagues reported similar results, with a mean increase of 48 points in Constant score. Similar improvement in Constant scores have been reported by Gerber and colleagues, Jerosch, Bennett, and Massoud and colleagues.
Nicholson reported the results of arthroscopic capsular release in 68 patients with shoulder stiffness secondary to five different etiologies :
Impingement syndrome with secondary stiffness
He was able to demonstrate that arthroscopic capsular release was equally effective in relieving pain and restoring function and motion across the five groups of patients.
Gerber and colleagues demonstrated improvement in ROM, pain scores, and Constant scores in 45 shoulders that underwent arthroscopic capsular release for frozen shoulder. They observed that patients with idiopathic frozen shoulder did better after arthroscopic capsular release than the posttraumatic and postsurgical patients. Similarly, Holloway and colleagues reported on 50 patients who underwent arthroscopic capsular release for frozen shoulder with excellent postoperative outcomes. However, patients with shoulder stiffness secondary to a postsurgical etiology achieved less improvement in pain, subjective function, ROM, and overall satisfaction.
Most investigators have reported successful and durable results of arthroscopic capsular release for frozen shoulder. However, more than 50% of patients reported by Segmuller and colleagues were found to have persistent stiffness in internal rotation at a mean follow-up of 13.5 months after arthroscopic inferior capsulotomy performed with cutting diathermy. Despite this residual limitation of function, 88% of the patients were satisfied with their outcome, and 87% had good to excellent results according to their Constant scores.
Better results, with up to 95% of patients achieving complete, painless ROM, were reported in a study by Yamaguchi and colleagues when capsulotomy was combined with the placement of an intra-articular pain pump containing 0.5% bupivacaine. However, recent evidence of the chondrotoxic effect of intra-articular bupivacaine has led many surgeons to refrain from the use of these pumps.
Many recent studies have also demonstrated good results. Elhassan, Higgins, Warner, and colleagues demonstrated that arthroscopic release was an effective procedure for shoulder stiffness regardless of the etiology, but patients with postsurgical stiffness did not fare as well as patients with idiopathic or posttraumatic frozen shoulder. Jerosch and colleagues concluded that arthroscopic capsular release is effective for treating both primary and secondary frozen shoulder, according to their findings at a mean follow-up of 3 years. Le Lievre and colleagues found that arthroscopic capsular release is a durable procedure, with pain relief and motion maintained at a mean follow-up of 7 years, with motion essentially equivalent to that of the contralateral shoulder. Waszczykowski and colleagues found that there was no significant decrease in strength in patients followed up a minimum of 2 years after an arthroscopic capsular release.
After its initial description by McLaughlin, subscapularis tendon release has been reported by several authors as part of a complete capsular release. Ide and colleagues reported on 42 patients with shoulder stiffness secondary to various etiologies. All the patients underwent complete capsular release with release of the intra-articular subscapularis tendon if the patient had loss of external rotation. The results at 7.5 years follow-up were 84% excellent, 7% good, and 9% poor. No complications were reported. Pearsall and colleagues reported on 35 patients who underwent capsular and subscapularis release; only one of these patients had possible subscapularis insufficiency postoperatively.
Similarly, Diwan and colleagues presented a case-controlled cohort study of 40 patients divided into two cohorts. The first cohort of 18 patients underwent a standard arthroscopic anteroinferior release of the capsule. The second cohort of 22 patients underwent the same release along with an extended posterior release and release of a portion of the intra-articular part of the subscapularis tendon as well as a modified physical therapy program. Both groups showed significant reduction of pain by 1 week postoperatively, but improved ROM was seen in the second group. No instability or subscapularis weakness, as determined by the lift-off test, was seen in either group.
Liem and colleagues demonstrated that concomitant release of the intra-articular portion of the subscapularis in addition to an arthroscopic capsular release demonstrated good clinical results without a resultant internal rotation deficit. However, despite the results of this study, we do not recommend routine subscapularis release for frozen shoulder because a thorough and safe arthroscopic release usually restores adequate motion without the need to transect the subscapularis.
Cost-Effectiveness and Value
In the current political and socioeconomic climate of the United States, demonstrating clinical utility alone is no longer sufficient to support the effectiveness of a procedure. A greater emphasis is being placed on value, which takes into account not only patient outcomes but also cost. Maund and colleagues have contended that the poor quality of the data precludes a definitive cost-effectiveness analysis among the various treatments of frozen shoulder. Steroid injection and physical therapy was the only treatment that demonstrated a statistically significant clinical benefit in the short term. In response Dattani and colleagues evaluated the cost and outcomes of treating 100 patients with an arthroscopic capsular release, concluding that arthroscopic treatment was a cost-effective procedure that can restore health-related quality of life in most patients within 6 months of surgery. Currently, there is insufficient evidence available to make a compelling determination of the most cost-effective way to manage frozen shoulder.
In a prospective study by Hsu and Chan of 75 patients in 1991, shoulder arthroscopy with distention or manipulation led to better outcomes than physical therapy alone. They favored arthroscopic distention over manipulation because it provided valuable insight into the intra-articular pathology and was more controllable.
In another study comparing manipulation with arthroscopic release in patients with resistant frozen shoulder, Ogilvie-Harris and colleagues also showed significantly better results with arthroscopic release at 2 to 5 years of follow-up. Their arthroscopic release technique included synovectomy in the rotator interval, division of the anterior capsule and the superior glenohumeral ligament, and division of the subscapularis tendon and the inferior capsule. Patients with diabetes initially did worse, but the final outcome for these patients was similar to that for patients without diabetes. The authors concluded that despite the worse outcomes initially reported in diabetic patients with frozen shoulder, these patients still do well after arthroscopic release if they undergo early intervention. Ogilvie-Harris and Myerthall demonstrated no pain and symmetrical motion in 13 of 17 diabetic patients with frozen shoulder who had undergone arthroscopic release.
There have been several recent publications evaluating the efficacy of arthroscopic capsular release for frozen shoulder, and the trend appears to be shifting towards shoulder surgeons performing arthroscopic releases rather than manipulations for recalcitrant frozen shoulder. However, Grant and colleagues were unable to demonstrate any significant difference when comparing arthroscopic releases to closed manipulation. They noted that the quality of evidence was low, and that high-quality studies are needed to definitively compare the procedures. Walther and colleagues retrospectively reviewed the results of 54 patients treated with arthroscopic subacromial decompression and manipulation, arthroscopic subacromial and glenohumeral release, or selective glenohumeral releases only, and found no significant difference between the surgical strategies. Currently, there is insufficient definitive evidence to recommend a single treatment strategy over another.