Adhesive Capsulitis (Frozen Shoulder)

Adhesive capsulitis, the frozen shoulder, is one of the commonest shoulder complaints in clinic: a painful condition in which the glenohumeral joint capsule becomes inflamed and then fibrotic, so the shoulder gradually stiffens and loses movement in all directions. It typically affects people aged 40 to 60, is slightly more common in women, and is strongly linked to diabetes.

The hallmark is painful, progressive loss of both active and passive movement, with external rotation usually the first and worst affected. The natural history runs through three overlapping stages: freezing (painful), frozen (stiff) and thawing (recovering).

The disease process is inflammation that turns into fibrosis. Current concept reviews describe the pathology of frozen shoulder as capsular inflammation followed by fibrosis and contracture.

Where the disease lives. The rotator interval, the triangular space at the front of the joint between the supraspinatus and subscapularis tendons, and the coracohumeral ligament within it are affected early. Contracture here limits external rotation, which is why external rotation is lost first and most severely. The anterior and inferior capsule, including the axillary recess, thickens and contracts, reducing the overall joint volume: the small, tight joint seen on an arthrogram.

The cellular story. The steps from inflammation to a stiff capsule, and what each produces:

The myofibroblast. Its presence explains why a frozen shoulder behaves like other fibrotic, contractile conditions: there is a recognised overlap with Dupuytren disease, which shares the same myofibroblast-driven fibrosis.

View larger

Diabetes is the strongest risk factor, and the reason lies in the same collagen biology that drives the capsular contracture.

Glycation. Chronic hyperglycaemia drives non-enzymatic glycation of proteins, producing advanced glycation end-products (AGEs) that accumulate in capsular collagen. AGEs form abnormal cross-links between collagen fibres, making the capsule stiffer and less compliant, and they also promote fibroblast and myofibroblast activity and a pro-inflammatory state. The result is a capsule more prone to the inflammation-then-fibrosis cascade and more resistant to remodelling.

The clinical pattern. Diabetic frozen shoulder is more common, often bilateral, more resistant to treatment, slower and less complete in its recovery, and more likely to need a procedure. Risk rises with longer diabetes duration and poorer glycaemic control, and similar glycation and fibrotic biology underlies the association with thyroid disease. A new bilateral frozen shoulder can be the presenting feature of undiagnosed diabetes.

Two classifications matter: by cause (primary versus secondary) and by clinical stage.

By cause. Primary (idiopathic) frozen shoulder has no identifiable cause, and the international (ISAKOS) terminology reserves the term "frozen shoulder" for this idiopathic stiff shoulder. Secondary frozen shoulder follows a known cause and is better called a secondary stiff shoulder; the distinction matters because the underlying cause needs treating as well, and secondary disease may behave differently from idiopathic disease. Causes include:

• Intrinsic - shoulder pathology such as a rotator cuff tear or calcific tendinitis
• Extrinsic - a remote cause such as immobilisation after a wrist fracture, stroke, or cardiac or thoracic surgery
• Systemic - diabetes and thyroid disease, which both predispose to the picture

By clinical stage. The classic staging follows the natural history. The stages overlap and the timings vary widely between patients.

• Stage 1 - Freezing (painful). Increasing pain, often worse at night, with stiffness developing. The inflammatory phase; movement is limited mainly by pain.
• Stage 2 - Frozen (stiff). Pain begins to settle but global stiffness is marked, with external rotation worst. The fibrotic phase.
• Stage 3 - Thawing (recovery). Movement gradually returns over many months as the capsule remodels.

View larger

At arthroscopy. The appearance evolves in parallel with the clinical stages: early fibrinous synovitis and inflammation in the rotator interval, progressing to capsular scar and contracture with loss of the axillary recess, and finally a mature, less inflamed contracted capsule. These findings underpin the rationale for capsular release, which divides the contracted rotator interval and capsule.

History. The patient describes shoulder pain that came on gradually with no major injury, is worse at night and on movement, and is followed by progressive stiffness. Ask specifically about the risk factors:

• Diabetes
• Thyroid disease (hypothyroidism and hyperthyroidism)
• A previous frozen shoulder on the other side
• Immobilisation after injury, surgery or a sling
• Trauma or surgery to the shoulder
• Stroke and prolonged inactivity
• Dupuytren disease

Examination. The key is passive movement: a capsular contracture restricts the passive range, which pain or a cuff tear alone usually does not.

View larger

Frozen shoulder is a clinical diagnosis. Investigations exclude other causes and support the diagnosis; they do not make it.

Plain radiographs. Usually normal in primary disease, which supports the diagnosis. Their main value is to exclude glenohumeral osteoarthritis, a locked posterior dislocation and calcific deposits, so always get an axillary or modified axillary view.

Blood tests. Check glucose or HbA1c in every new case, since undiagnosed diabetes is common, and consider thyroid function. There is no blood test for frozen shoulder itself.

Ultrasound. Mainly to look for other pathology, a cuff tear or calcific disease. It can show thickening at the rotator interval and restricted movement, but it is operator-dependent and not required for the diagnosis.

View larger

View larger

View larger

MRI and arthrography. The supportive findings are a thickened coracohumeral ligament and rotator interval, a thickened inferior capsule and axillary recess, and on arthrography a reduced joint volume. Reserve MRI for when the diagnosis is unclear or another problem is suspected.

View larger

View larger

View larger

View larger

View larger

Several conditions cause a painful, stiff shoulder, and the passive external rotation sign and a radiograph separate them. The most important mimic is glenohumeral osteoarthritis, which also restricts passive external rotation; the radiograph distinguishes the two. The most dangerous to miss is a locked posterior dislocation, which also abolishes external rotation but is easily missed on an AP film, so always obtain an axillary view, especially after a seizure or electric shock.

View larger

The ladder. Treatment escalates. Frozen shoulder is usually self-limiting, so most patients are managed without surgery, with procedures reserved for resistant or severely disabling disease. The stage chooses the rung: an injection helps most in the painful freezing stage, while stretching, hydrodilatation and release are aimed at the stiff frozen stage.

Education and analgesia. Explain the natural history and the long but generally favourable course; this alone changes expectations and reduces distress. Analgesia and NSAIDs control pain, particularly in the painful freezing stage.

Intra-articular corticosteroid injection. Most effective early, when pain dominates: it gives faster pain relief and earlier improvement than physiotherapy alone, although the long-term difference narrows. Pooling 32 randomised trials (1836 participants), six weeks of combined manual therapy and exercise probably gives less short-term improvement than a glucocorticoid injection, a difference of about 26 points on a 100-point pain scale at seven weeks, which is large, but the differences were no longer clinically important by six and twelve months.

What those two findings mean together. The injection does not change where the patient ends up; it changes how quickly they get there and how much pain they endure on the way. That is a strong reason to inject in the painful freezing stage, when pain is the dominant complaint and the patient cannot engage with physiotherapy, and a weak reason to inject a shoulder that has already moved into the stiff frozen stage, where pain is no longer the limiting factor. Physiotherapy remains the sustained mainstay either way (Cochrane review, moderate-quality evidence).

Physiotherapy. Gentle, progressive stretching within a comfortable range. An early injection plus structured physiotherapy is the backbone of treatment and is what most patients need.

Hydrodilatation (capsular distension). A larger volume of fluid (saline, local anaesthetic and steroid) is injected into the joint under image guidance to stretch and distend the contracted capsule, sometimes to the point of capsular rupture, combining the anti-inflammatory effect of steroid with a mechanical stretch. It is a reasonable next step for the stiff, resistant shoulder before considering surgery. Adding hydrodilatation and bursal steroid to mobilisation and physiotherapy improves pain, function and range compared with physiotherapy alone in the freezing stage; the trial behind that claim is discussed under Controversies.

Manipulation under anaesthesia. The contracted capsule is stretched and torn while the patient is anaesthetised, usually combined with a steroid injection and followed by intensive physiotherapy. It is used for the resistant frozen stage. The risks are those of an uncontrolled tear, fracture (especially in osteoporotic or diabetic bone), dislocation, and cuff or labral injury, so it must be done carefully. In the large UK FROST trial it was the most cost-effective of the secondary-care options.

Arthroscopic capsular release. The contracted rotator interval and capsule (the coracohumeral ligament and anteroinferior capsule) are divided surgically, a controlled release rather than the uncontrolled tear of a manipulation. It is favoured for the most resistant cases, especially diabetic patients and those who have failed other measures, and is followed by stepwise rehabilitation. It is more invasive, and in UK FROST it carried more serious adverse events than manipulation (8 against 2); its Oxford Shoulder Score at one year was statistically but not clinically better (by 2.01 points over manipulation and 3.06 over physiotherapy, both below the trial's target differences of 4 and 5 points), so it is not a routine first surgical step.

Principle. A controlled, complete release of the contracted capsule, replacing the uncontrolled tearing of a manipulation. Release the diseased tissue, restore range on the table, then protect the result.

Steps. The release proceeds in order:

1. Examination under anaesthesia to document the restricted range before release.
2. Arthroscopy of the glenohumeral joint; inflamed synovium and a contracted, scarred capsule with a lost axillary recess are typical.
3. Release the rotator interval, including the coracohumeral ligament; this restores external rotation.
4. Release the anterior and inferior capsule in a controlled fashion, protecting the axillary nerve inferiorly.
5. A gentle manipulation at the end if needed, then a steroid injection.

View larger

View larger

View larger

View larger

Aftercare. The main reason a release fails is recurrent stiffness, so immediate, supervised physiotherapy to maintain the gains is essential. In diabetic patients, counsel about a slower, less complete recovery and a higher chance of needing further treatment.

Complications arise both from the disease and from its treatment.

Rehabilitation is central, both as primary treatment and after any procedure. Home exercise between sessions is essential; a frozen shoulder is treated mostly by the patient at home.

Matched to the stage. Early on, the goal is to maintain range without provoking the inflamed capsule; later, the goal is to regain range as the capsule remodels.

After manipulation or release. Start immediate, intensive, supervised physiotherapy to hold the gains; the result is lost if the capsule re-scars.

Good but slow. Most patients improve substantially, whether treated conservatively or with a procedure. Primary frozen shoulder is usually self-limiting, but recovery commonly takes 1 to 3 years, and a minority are left with some lasting restriction that is often well tolerated.

Treatments converge. In secondary care, manipulation, capsular release, and structured physiotherapy with steroid injection all produced similar shoulder scores at 12 months, so the choice rests on risk and cost as much as on outcome.

• Diagnosis is consistently clinical. Systematic reviews of clinical practice guidelines for shoulder disorders agree that the initial assessment rests on history and examination, that MRI is usually not needed early, and that radiographs are used mainly to exclude other causes. The international (ISAKOS) terminology distinguishes primary frozen shoulder (idiopathic) from a secondary stiff shoulder (known cause).
• Conservative care first, side by side. Guidance from bodies such as the AAOS (US), BOA/BESS (UK), and equivalent European groups converges on the same ladder: education, analgesia, an early intra-articular steroid injection, and stretching physiotherapy, with hydrodilatation, manipulation, and arthroscopic capsular release reserved for resistant disease.
• Where guidance differs is mainly the threshold and timing for escalating to a procedure and the relative emphasis on hydrodilatation versus manipulation versus release - reflecting local resources and expertise rather than disagreement about the underlying disease.
• Best randomised evidence (UK FROST) shows the secondary-care options converge by one year, so many systems favour the least invasive, most cost-effective route, with surgery reserved for selected patients - especially the more resistant diabetic shoulder.
• Global practice variation largely reflects access (availability of image-guided injection, theatre time, and arthroscopy) rather than differences in the principles of stage-matched, escalating care.

Manipulation or arthroscopic release for the resistant shoulder. The two best sources disagree, and a good answer says so. UK FROST found that release carried more serious adverse events than manipulation, and its small Oxford Shoulder Score advantage at one year (2.01 points) was statistically significant but below the 4-point clinically important difference, making manipulation the most cost-effective secondary-care option. The ISAKOS consensus review reaches the opposite recommendation: for refractory stiffness it prefers arthroscopic release over manipulation, precisely because manipulation is an uncontrolled tear carrying fracture, dislocation and cuff-injury risk.

Reconciling them. Both positions are defensible because they weigh different harms. FROST counts adverse events attributable to an operation; ISAKOS weighs the catastrophic-but-uncommon manipulation complication in fragile bone. Manipulation is reasonable as a first surgical step in a shoulder with normal bone quality, while a controlled release is preferred where the bone or capsule makes an uncontrolled tear dangerous: the diabetic, the osteoporotic, and the shoulder that has already failed a manipulation. State the disagreement rather than picking a side blindly.

Where the evidence is genuinely thin. The cause of primary frozen shoulder remains unknown; the ISAKOS review describes capsular fibrosis and inflammation with chondrogenesis without establishing why it starts. The optimal timing of every intervention is inferred from stage rather than tested directly, and the trials that support hydrodilatation are small. The strongest is a single-blind randomised trial of just 70 patients in the freezing stage, where hydrodilatation with bursal steroid added to physiotherapy beat physiotherapy alone on pain, SPADI, range and self-assessment, with benefit sustained at one, two, four and six months. That is encouraging and durable but not definitive, and it has not been compared head-to-head against injection alone in a large trial.

The honest limit on prognosis. "Self-limiting" over one to three years is a long time to be in pain. No good trial demonstrates that any treatment changes the endpoint, only the speed of getting there, and that is worth saying plainly to a patient rather than promising resolution.

The following studies underpin the modern management of frozen shoulder.