Condition guide · Shoulder
Frozen Shoulder
A shoulder that stiffens until it will barely turn, then slowly releases. What actually helps, what the largest surgical trial found, and why the timeline matters more than almost anything else.
In plain words. The shoulder joint sits inside a sleeve of tissue called the capsule. In frozen shoulder — properly, adhesive capsulitis — that sleeve becomes inflamed, thickened and contracted. It physically shrinks, and the shoulder loses movement whether you try to move it yourself or someone else moves it for you.
That last detail is what distinguishes it. In most shoulder problems you can still be moved through range even if it hurts. In a genuinely frozen shoulder, the movement is not there to be found.
Key points
- The restriction is mechanical. The capsule has thickened and contracted, which is why forcing through it is painful and slow to change.
- An intra-articular corticosteroid injection was the one intervention with clear short-term superiority in a review of 65 studies — better than physiotherapy alone for both pain and function. [1]
- It works best combined with exercise, not instead of it. The same review found adding a home exercise programme and physiotherapy to the injection was associated with added mid-term benefit. [1]
- Earlier is better. The authors' conclusion applies to frozen shoulder of less than one year's duration. [1]
- Neither operation beat physiotherapy. In a 503-patient trial, manipulation under anaesthesia, arthroscopic capsular release and structured physiotherapy all finished within a few points of one another — below the difference the trial was designed to detect. [2]
- The surgery carried the harm. Eight serious adverse events with capsular release, two with manipulation. [2]
What is actually happening
The capsule is normally loose and pleated, which is what lets the shoulder move further than any other joint in the body. In frozen shoulder it becomes inflamed and fibrotic: thicker, redder, and tightened down onto the joint. The space inside the joint physically reduces.
A normal shoulder capsule, and a frozen one
Left: the normal articular capsule, pale and slack. Right: the same capsule thickened and inflamed in adhesive capsulitis. Nothing is torn and nothing is worn out — the sleeve around the joint has tightened, which is why the shoulder will not turn even when someone else moves it, and why the condition resolves as the capsule settles rather than because anything is repaired.
Frozen shoulder is described in stages — a painful freezing phase, a stiff frozen phase, and a thawing phase as movement gradually returns. The stages overlap and their length varies enormously between people, which is why you should be sceptical of anyone who tells you exactly how long yours will take.
It occurs more often in people with diabetes and in thyroid disease, and it can follow a period of immobilising the arm after an injury or operation. Often there is no identifiable trigger at all.
What it feels like
- Deep, aching shoulder pain, often worst at night, and often the first thing to appear.
- Progressive loss of movement — typically reaching behind your back, reaching across your body, and turning the arm outwards.
- Difficulty with fastening a bra, reaching a back pocket, putting on a coat, or reaching for a seatbelt.
- Sharp pain when the arm is moved suddenly past the point where the capsule stops it.
- A gradual settling of the pain while the stiffness remains — the transition into the frozen phase.
How it is diagnosed
By examination. The defining finding is loss of passive external rotation — when the clinician turns your relaxed arm outwards, the movement is restricted, not merely painful. The UK FROST trial used restriction of passive external rotation of at least 50% as its entry criterion. [2]
Imaging is not needed to make the diagnosis. It is used to exclude other things when the picture is atypical — a scan cannot show you a diagnosis that is defined by how the joint moves.
How physiotherapy and rehabilitation help
A frozen shoulder is a contracted capsule, and the thing that changes a contracted capsule is repeated, tolerable movement over months. That is what physiotherapy is for here. It does not melt adhesions or accelerate a natural history; it keeps the range you have, recovers the range you can, and stops the shoulder from being guarded so completely that the rest of the arm deconditions around it.
The evidence supports that directly. A systematic review and meta-analysis of exercise therapy found range of motion, function and pain all improve, whether exercise is given alone or as part of a wider programme, and that adding exercise to a programme improves active range compared with a programme without it. Adding physical modalities on top had no beneficial effect. [3] Muscle energy techniques came out ahead of other exercise types for function.
Two honest limits from that same review: no conclusion could be drawn about long-term results, and no conclusion could be drawn about the most effective dose. [3] So a physiotherapist who tells you exactly how many repetitions over exactly how many weeks will fix this is going beyond what anyone has shown.
What is actually offered, and what it is worth
- Graded mobilisation and stretching into tolerable range — the core of it. Posterior glenohumeral approaches and higher-grade mobilisations appear effective, and stretching is described as a mandatory component of any programme. [5]
- A home exercise programme — the part that does most of the work, because frequency over months matters more than what happens in a clinic once a week.
- Manual therapy added to exercise — weaker than it is often sold. A 2023 systematic review of 16 studies found non-significant effects on pain, disability and external rotation at both short- and long-term follow-up, with very low to low certainty throughout. [4] It is not useless; it is unproven as an addition, and should not be the reason a course of treatment costs more.
- Ultrasound therapy — did not prove effective. [5] If it is on your treatment plan, ask what it is being used for.
What treatment involves
The combination with the best support is an intra-articular corticosteroid injection, given early, together with a home exercise programme and physiotherapy. [1] The injection addresses the inflammatory component and buys a window of reduced pain; the exercise uses that window to maintain and regain range.
The physiotherapy arm of the UK FROST trial gives a concrete picture of what "structured" means: mobilisation techniques and a graduated home exercise programme supplemented by a steroid injection, delivered over 12 sessions in up to 12 weeks. [2] That is a real course of treatment, not a handout.
Aggressive stretching into severe pain is not the goal. The capsule is contracted tissue; it responds to consistent, tolerable loading over months, not to being forced.
Shoulder pain that is stiff in every direction is a frozen shoulder; shoulder pain that is painful in some directions but moves fully when someone else moves it is usually a rotator cuff problem, which is managed quite differently.
What the evidence supports — and what it does not
UK FROST: shoulder score at 12 months, by treatment
Supported
- Early intra-articular corticosteroid injection — the only intervention with both statistical and clinical superiority in the short term, against no treatment or placebo (mean difference −1.0 VAS points, 95% CI −1.5 to −0.5) and against physiotherapy (−1.1, 95% CI −1.7 to −0.5). [1]
- Injection plus a home exercise programme — adding simple exercises and stretches was associated with added mid-term benefit (pain −1.4 VAS points, 95% CI −1.8 to −1.1 versus no treatment or placebo). The review's own conclusion is that the injection should be accompanied by a home programme. [1]
- Structured physiotherapy with injection as first-line care — it matched both operations at twelve months, without their risks. [2]
Not supported
- Arthroscopic capsular release as a superior option — it beat physiotherapy by 3.06 Oxford points (95% CI 0.71 to 5.41) against a 5-point target, and produced eight serious adverse events. [2]
- Manipulation under anaesthesia as a superior option — 1.05 points versus physiotherapy (95% CI −1.28 to 3.39): no meaningful difference at all. [2]
- Physiotherapy alone, without addressing the inflammation — the injection outperformed physiotherapy on its own in the short term. The two work better together than either does by itself. [1]
- Waiting it out untreated because "it resolves anyway" — the evidence favours treating early, within the first year. [1]
Shoulder pain that is provoked by particular movements rather than stiff in every direction is a different problem — see shoulder impingement.
How certain is this?
Evidence grade: Moderate. UK FROST is a large, multicentre, pragmatic randomised trial with 94% follow-up at the primary endpoint, and its central finding — that none of the three treatments was clinically superior — is robust. [2] Note what that does not mean: it does not mean the treatments are useless, since all three groups improved. It means the expensive, invasive options did not buy a better result.
The injection evidence comes from a network meta-analysis, which compares treatments that were often never tested head-to-head. That is a real limitation. The review pooled 65 studies and 4,097 participants, and the authors were careful to say that despite many statistically significant results, only the corticosteroid finding reached clinical as well as statistical superiority. [1] The effect is also short-term: about one point on a ten-point pain scale.
Neither source tells us reliably how long an individual's frozen shoulder will last, or which people will recover fastest.
What to expect
Frozen shoulder is measured in months, not weeks, and that is the hardest part of it. In UK FROST every group was still being followed at twelve months, and the average Oxford Shoulder Score at that point was around 37 to 40 out of 48 — substantially improved, but not everyone fully back to normal.
Pain usually settles well before movement does. The practical consequence is that the middle of the condition can feel like a plateau — less painful, still stiff — and that is the phase in which people abandon their exercises. Continuing through it is the main thing you control.
When shoulder pain is not a frozen shoulder — seek assessment
Seek prompt medical attention if you have:
- Shoulder pain with chest pain, breathlessness, sweating, or pain spreading into the jaw or arm — this can be cardiac and needs emergency care.
- A shoulder that looks deformed, or sudden loss of movement after a fall or injury.
- Fever, a hot swollen joint, or feeling generally unwell with shoulder pain.
- Marked weakness rather than stiffness — particularly inability to lift the arm at all after an injury.
- Numbness or pins and needles down the arm, or symptoms in both shoulders at once.
- Unexplained weight loss, a history of cancer, or night pain that is severe and unrelenting.
This list is not exhaustive. If you are worried about a symptom, seek professional advice.
Common questions
Do I need surgery?
On the current evidence, usually not. The UK FROST trial randomised 503 people to manipulation under anaesthesia, arthroscopic capsular release or structured physiotherapy with an injection. At twelve months, all the differences between them were smaller than the 5-point difference the trial set out to detect, so none was clinically superior — and the surgical arms accounted for all ten serious adverse events. [2]
Should I have the steroid injection?
It has the best short-term evidence of any single treatment here, and it outperformed physiotherapy alone for both pain and function. [1] The important qualifier is that the review's conclusion pairs it with a home exercise programme, and applies to frozen shoulder of under a year's duration. Whether it is right for you — particularly if you have diabetes, since steroids raise blood glucose — is a conversation with your clinician.
How long will this last?
Longer than you want, and no one can give you a reliable personal figure. What we can say from the evidence is that people in a large trial were still improving at twelve months, and had reached an average shoulder score of roughly 37 to 40 out of 48 by then. [2] Anyone quoting you an exact duration is going beyond what is known.
Should I push into the pain when stretching?
No. The restriction is contracted capsular tissue, and forcing it tends to increase pain and guarding without speeding recovery. The physiotherapy that performed as well as surgery in UK FROST was mobilisation plus a graduated home programme over twelve weeks — graduated being the operative word. [2]
Why did this happen to me?
Often there is no identifiable reason. It is more common in people with diabetes and thyroid disease, and can follow a period when the arm was kept still after an injury or operation. It is not caused by anything you did wrong.
References
- Challoumas D, Biddle M, McLean M, Millar NL. Comparison of treatments for frozen shoulder: a systematic review and meta-analysis. JAMA Network Open. 2020 Dec 1;3(12):e2029581. doi:10.1001/jamanetworkopen.2020.29581 PMID 33326025 Systematic review and network meta-analysis
- Rangan A, Brealey SD, Keding A, et al. Management of adults with primary frozen shoulder in secondary care (UK FROST): a multicentre, pragmatic, three-arm, superiority randomised clinical trial. Lancet. 2020 Oct 3;396(10256):977–989. doi:10.1016/S0140-6736(20)31965-6 PMID 33010843 Randomised controlled trial
- Mertens MG, Meert L, Struyf F, et al. Exercise Therapy Is Effective for Improvement in Range of Motion, Function, and Pain in Patients With Frozen Shoulder: A Systematic Review and Meta-analysis. Archives of Physical Medicine and Rehabilitation. 2022 May;103(5):998–1012.e14. doi:10.1016/j.apmr.2021.07.806 PMID 34425089 Systematic review and meta-analysis
- Kirker K, O'Connell M, Bradley L, et al. Manual therapy and exercise for adhesive capsulitis: a systematic review with meta-analysis. Journal of Manual & Manipulative Therapy. 2023 Oct;31(5):311–327. doi:10.1080/10669817.2023.2180702 PMID 36861780 Systematic review and meta-analysis
- Costantino C, Nuresi C, Ammendolia A, et al. Rehabilitative treatments in adhesive capsulitis: a systematic review. Journal of Sports Medicine and Physical Fitness. 2022 Nov;62(11):1505–1511. doi:10.23736/S0022-4707.22.13054-9 PMID 35179326 Systematic review
About this guide
If you need assessment
This page explains. It does not diagnose.
A genuinely frozen shoulder is defined by how the joint moves when someone else moves it — which cannot be established from a description or a scan. Several other shoulder problems look similar early on and are treated differently. Physiotherapist India publishes information and takes no bookings. Ask a qualified physiotherapist to examine your own case before acting on anything written here.