Frozen shoulder (adhesive capsulitis): does it resolve on its own, and what helps in treatment?

What is frozen shoulder, and who gets it?

The shoulder joint capsule is normally a loose and flexible sheath. In frozen shoulder this capsule becomes inflamed, thickens, and contracts like a scar tissue that pulls; the result is a painful shoulder in which passive movement in every direction — especially in turning the arm outward — is also restricted. It is most common between the ages of 40–60 and in women; its frequency in the general population is at the level of a few percent.

Two conditions markedly raise the risk:

• Diabetes: according to meta-analyses, the risk of frozen shoulder is about 4–5 times higher in people with diabetes; roughly 10–15% of people with diabetes have frozen shoulder at a given time (point prevalence, ~13% in pooled data), and a significant portion of frozen-shoulder patients have diabetes. Diabetic cases can also run a longer and more treatment-resistant course.
• Thyroid diseases: hypothyroidism (an underactive thyroid) in particular is associated with an increased risk of frozen shoulder.

For this reason, in a frozen shoulder with no other cause, it is reasonable to review fasting glucose/HbA1c and, if needed, thyroid tests. Frozen shoulder can also develop after shoulder surgery, fracture, or long-term immobility (secondary).

Natural course: stages and the limit of the "resolves on its own" myth

The classic account defines three stages; the boundaries are not sharp, and the durations vary greatly from person to person:

The problem is: the understanding of "wait for the three stages and you recover fully" rests on small observations from the 1940s, and a systematic review has not confirmed this assumption. Although range of motion improves over time in untreated cases, it mostly does not return fully to normal; at an average follow-up of 4+ years, about 40% of patients have ongoing mild-to-moderate complaints, and a small portion (5–6%) have marked pain and functional loss. In summary: the course is good for the majority, but saying "it resolves on its own and completely in everyone" is beyond the evidence.

The key to the early period: intra-articular corticosteroid injection

The early/painful stage of frozen shoulder is the period in which inflammation is dominant — and this is also the window where corticosteroid works best:

• Meta-analyses combining randomized trials show that intra-articular corticosteroid injection significantly improves pain and function compared with placebo and physiotherapy alone especially in the first 6–12 weeks; superiority extending to six months has not been shown in every analysis.
• The benefit reaches its highest level when the injection is combined with a simple home exercise program; the return from an injection done when complaints are shorter than 1 year (in the early period) is more marked.
• Doing the injection under ultrasound guidance makes the placement reliable. In people with diabetes, corticosteroid can raise blood sugar for a few days; follow-up should be planned.

In short, in the early period, instead of "let's manage the pain with medication and wait," injection + exercise is the evidence-based first step. This limit should also be stated openly: the injection improves pain and short-term function; whether it prevents residual restriction in the long term has not been definitively shown.

Physiotherapy and exercise: at the right dose and suited to the stage

Exercise is the backbone of treatment, but its dose is adjusted to the stage. In the painful stage the aim is to reduce pain and preserve movement within pain limits; aggressive stretching in this stage can flare the picture. In the frozen and thawing stages, gradual stretching and range-of-motion work take precedence. The effect of physiotherapy alone is modest; the best result is obtained together with injection. Programs weighted toward passive devices and electrotherapy add little; the program should rest on exercises the patient can do regularly at home.

Hydrodilatation: the "expanded" form of the injection

Hydrodilatation (distension arthrography) is the procedure of stretching and expanding the contracted capsule by giving a large amount of fluid into the joint together with corticosteroid; it is done under ultrasound or X-ray guidance, without needing an operating room. Meta-analysis shows that hydrodilatation can provide a short-term temporary functional gain over standard corticosteroid injection — especially in the outward-turning movement — but that this difference is small, that its clinical importance remains uncertain, and that the two methods largely even out in the long term. For this reason, presenting it as "the natural next step after a standard injection" is beyond the evidence; it is a non-surgical option that can be considered in selected patients with resistant, marked restriction.

Manipulation and arthroscopic capsular release: what did UK FROST teach?

The randomized UK FROST trial (Lancet 2020), conducted with 503 patients at 35 centers, compared three paths: (1) manipulation under anesthesia + steroid injection, (2) arthroscopic (keyhole) capsular release, (3) early structured physiotherapy + steroid injection.

• At 12 months, Oxford Shoulder Score: capsular release 40.3; manipulation 38.3; physiotherapy 37.2 (on a 0–48 scale). The differences were statistically small, remaining below the clinically significant threshold.
• Serious adverse events were more frequent in the capsular release arm (8 versus 2); capsular release was also the most expensive option.
• Conclusion: no method was clinically superior to another. The non-surgical arm can be applied earlier and carries no surgical risk.

A limit to generalization should also be noted: UK FROST covered primary frozen-shoulder patients referred to secondary care (hospital/specialist clinic); its results should not be directly generalized to all early cases in primary care.

This does not mean surgery has no place: in patients who do not improve despite conservative treatment of appropriate duration (usually at least 3–6 months) and quality, and whose daily life is markedly restricted, manipulation or capsular release are legitimate options. But there is no evidence for "surgery first."

The treatment framework at a glance

This article is for general information; since shoulder pain can have many other causes (rotator cuff problems, arthritis, extra-articular causes), it does not replace a physician's examination and an individualized diagnosis.

For clinicians

The diagnosis is made clinically: a picture in which active AND passive movement — especially external rotation — is restricted, accompanied by a radiograph showing a normal glenohumeral joint. The radiograph is needed to rule out glenohumeral osteoarthritis and posterior dislocation, the other major causes of restricted external rotation; MRI is not routine. In a newly diagnosed primary frozen shoulder, risk-based screening for diabetes (HbA1c) is reasonable (OR ~4–5, prevalence ~13%); there is no consensus on routine TSH screening, which is recommended in the presence of clinical suspicion (a significant association has been reported in hypothyroidism). No marked difference has been shown between low-dose (20 mg triamcinolone) and 40 mg in injection; ultrasound-guided glenohumeral or rotator interval placement can be used. In the network meta-analysis by Challoumas et al., IAC + home exercise in the early period provided the most consistent short-term gain; the superiority of corticosteroid is mainly in the first 6–12 weeks and it has not been shown to prevent long-term residual restriction. An important limitation of the umbrella review by Lädermann et al., which highlights hydrodilatation as the most effective of the conservative options, is that all the meta-analyses it includes were rated low or critically low quality by AMSTAR-2; this ranking should be interpreted with caution. In UK FROST, MUA stood out in terms of cost-effectiveness while ACR ran with the highest complications and cost; it should not be forgotten that the study population was primary frozen-shoulder patients referred to secondary care, and the results should not be directly generalized to all early cases in primary care. The pragmatic interpretation is to try injection+structured physiotherapy first even in a patient referred to secondary care. In diabetic cases, a more protracted course and a tendency to recur after surgery should be kept in mind; aggressive early stretching should be avoided, and in the pain-dominant stage the load should be titrated "respecting the pain."

Frequently asked questions

Related articles

• Shoulder impingement syndrome: why the name is changing, and why exercise comes first
• Rotator cuff tear: surgery or exercise?

Scientific references

This information is for general guidance only; it does not replace an examination, a diagnosis, or your physician's individual advice. Please consult a physician for your symptoms. About the author: Doç. Dr. Oğuz Yüksel — Academic Profile.