Frozen shoulder

Introduction

Frozen shoulder is a disabling and sometimes severely painful condition that is commonly managed in the primary care setting. True frozen shoulder has a protracted natural history that usually ends in resolution. In this article we consider how to diagnose frozen shoulder and how to distinguish it from other painful shoulder conditions. We also look at the current aetiological theories and the effectiveness of conservative and operative management. We reviewed the current literature on this topic and discussed papers of historical interest with consultants in our department. We have also made reference to key papers cited in Clinical Evidence (http://www.clinicalevidence.com/).

What is frozen shoulder?

The term “frozen shoulder” was first introduced by Codman in 1934.^w1 He described a painful shoulder condition of insidious onset that was associated with stiffness and difficulty sleeping on the affected side. Codman also identified the marked reduction in forward elevation and external rotation that are the hallmarks of the disease.

Long before Codman, in 1872, the same condition had already been labelled “peri-arthritis” by Duplay.^w2 In 1945, Naviesar coined the term “adhesive capsulitis.”^w3 Although still in use, this more recent term is unfortunate since, although a frozen shoulder is associated with synovitis and capsule contracture, it is not associated with capsular adhesions.

In clinical practice, the tendency is to label any patient with a stiff, painful shoulder as a case of frozen shoulder. This should be resisted. Frozen shoulder is a specific condition that has a natural history of spontaneous resolution and requires a management pathway that is completely different from such distinct shoulder conditions as a rotator cuff tear or osteoarthritis.

Who gets it?

Frozen shoulder patients usually present in the sixth decade of life, and onset before the age of 40 is very uncommon.^w4 The peak age is 56, and the condition occurs slightly more often in women than men.1 ^w4 In 6-17% of patients, the other shoulder becomes affected, usually within five years, and after the first has resolved.1 ^w4 The non-dominant shoulder is slightly more likely to be affected.1 ^w4

Few attempts have been made to calculate the cumulative lifetime risk of frozen shoulder. In the Scandinavian population at risk, it has been estimated at a minimum of 2% per year.^{w4 w5} Recurrence is highly unusual.^w6

Clinical presentation and examination

A patient with frozen shoulder traditionally progresses through three overlapping phases (box).2

When examining any joint, it is useful to apply the well known axiom of the late Alan Apley, a popular orthopaedic speaker and teacher: “Look, Feel, Move.”

Arthroscope view of a shoulder with synovitis

Credit: CNRI/SPL

Look: On inspection, the arm is held by the side in adduction and internal rotation. Mild disuse atrophy of the deltoid and supraspinatus may be present.

Feel: On palpation, there is diffuse tenderness over the glenohumeral joint, and this extends to the trapezius and interscapular area owing to attempted splinting of the painful shoulder.

Move: In true frozen shoulder there is almost complete loss of external rotation. This is the pathognomonic sign of a frozen shoulder.1 2 ^w1-w3 Confirming that external rotation is impossible with active and passive movements is important. For example, if external rotation was easily possible with the help of the doctor, we would consider the diagnosis of a large rotator cuff tear, which would require completely different management. In frozen shoulder, all other movements of the joint are reduced, and if movement occurs this usually comes from the thoracoscapular joint.

What's the natural history of frozen shoulder?

Although the natural history of frozen shoulder is for ultimate resolution, this may not be complete. Reeves, in a prospective study of 41 patients with 5-10 years' follow-up, found that 39% had full recovery, 54% had clinical limitation without functional disability, and 7% had functional limitation.2 Shaffer et al showed that 50% of his 61 patients with frozen shoulder had some degree of pain and stiffness an average of seven years after onset of the disease.3

Secondary frozen shoulder

Frozen shoulder can be a primary or idiopathic problem or it may be associated with another systemic illness. By far the most common association of a secondary frozen shoulder is diabetes mellitus.4 ^{w7 w8} The incidence of frozen shoulder in diabetes patients is reported to be 10%-36%.4^–6 The incidence in type 1 and type 2 diabetes is similar.5 Unfortunately, frozen shoulder in diabetes is often more severe and is more resistant to treatment.7 Moren-Hybbinette et al reported on the natural history of the diabetic painful stiff shoulder and found a restriction in the range of motion in 35 (65%) of 54 shoulders at a mean follow-up of 29 months.8

Bunker et al have shown an association with Dupuytren's disease in the hand, proposing that the contracting shoulder tissue itself represents a form of fibromatosis.6 ^w9 Much more rarely, secondary frozen shoulder may be associated with conditions such as hyperthyroidism,^w10 hypothyroidism,^w11 and hypoadrenalism.^w12

Additional associations include Parkinson's disease,^w13 cardiac disease, pulmonary disease, and stroke,^w14-w16 although the pathological condition here may be different from idiopathic frozen shoulder. Clearly, in the case of stroke, shoulder stiffness may be simply the result of muscle spasticity in the shoulder region.

Frozen shoulder has also been reported subsequent to non-shoulder surgical procedures, such as cardiac surgery,^w17 cardiac catheterisation through the brachial artery,^w18 neurosurgery,^w19 and radical neck dissection.^w20

Laboratory investigations and radiology in frozen shoulder

There are few specific laboratory tests or radiological markers for frozen shoulder, and the diagnosis is essentially clinical. Immunological studies (such as human leucocyte antigen B27), C reactive protein, and erythrocyte sedimentation rate are all normal^w21-w23 and would be measured only to exclude other conditions. Most orthopaedic surgeons would not investigate a frozen shoulder beyond a plain x ray. When plain radiographs of the frozen shoulder are taken they may well be reported as normal, although they may show periarticular osteopenia as a result of disuse.9 ^w24

Contrast technetium-99m diphosphonate bone scan shows an increased uptake on the affected side in 92% of patients compared with the opposite side or with controls.9 Arthrography shows characteristic findings of limitation of capacity of the shoulder joint (5-10 ml compared with 25-30 ml in the normal joint) and a small or non-existent dependent axillary fold.9 10 ^w24 However, in most units, arthrography is a historical investigation in frozen shoulder. Magnetic resonance imaging may show a slight thickening in the joint capsule and the coracohumeral ligament.^w25

Pathogenesis

The aetiology of frozen shoulder remains unclear. The disease process particularly affects the anterosuperior joint capsule and the coracohumeral ligament.11 Arthroscopy shows a small joint with loss of the axillary fold and tight anterior capsule, mild or moderate synovitis, and no adhesions.12 ^{w4 w26}

Neviaser and Neviaser have described an arthroscopic four stage classification for the frozen shoulder,10 and Hannafin et al^w5 have described a correlation between the arthroscopic stage, the clinical examination, and the histological appearance of the tissues.

Disagreement prevails about whether the underlying pathological process is an inflammatory condition,^{w5 w27} a fibrosing condition,6 or even an algo-neurodystrophic process.^w28

Evidence shows a synovial inflammation with subsequent reactive capsular fibrosis. A dense matrix of type I and type III collagen is laid down by fibroblasts and myofibroblasts in the joint capsule. Subsequently, this tissue contracts.

Increased growth factors, cytokines, and expression of matrix metalloproteinases in capsular biopsy specimens obtained from patients with primary and secondary frozen shoulder indicate that these are involved in the inflammatory and fibrotic cascades seen in frozen shoulder.^{w27 w29}

Cytokines and growth factors are involved in the initiation and termination of repair processes in musculoskeletal tissues through regulating fibroblasts, and the remodelling process is controlled by matrix metalloproteinases and their inhibitors.^{w29 w30} An association between frozen shoulder and Dupuytren's disease has been identified,6 13 and this may be related to matrix metalloproteinase inhibitors.^w31

How should I treat it?

Educating patients helps to reduce frustration and encourages compliance. An explanation that the condition will spontaneously resolve and stiffness will greatly reduce helps. However, it is important to emphasise that the full range of motion may never recover. Ideally, the treatment of frozen shoulder should be tailored to the stage of the disease.

Treatment during the adhesive phase

Intra-articular steroid injections are not indicated in the adhesive phase as the inflammatory stage of the disease has passed. More aggressive stretching exercises will be tolerated and should be the focus of treatment, with the aim of regaining the range of motion. Low load, prolonged stretches produce plastic elongation of tissues as opposed to the high tensile resistance seen with high load, brief stretches.22 ^w33

Conclusions

Frozen shoulder is a common, sometimes painful, but ultimately self limiting, condition that is usually managed in the primary care setting with a combination of analgesics, injections, and physiotherapy. Formal investigations are usually normal, and the diagnosis is essentially clinical. Most cases can be managed in the primary care setting. Educating patients plays an important part in the management of the condition. A minority of patients require referral to an orthopaedic specialist, where manipulation under anaesthesia is the most common method of treatment. Arthroscopic surgical release has proved itself to be useful in refractory cases. Irrespective of the treatment given, a high proportion of patients with frozen shoulder do not regain a full range of motion.