Frozen shoulder – Knowledge

1. Why this text?

It usually starts harmlessly. A pulling sensation in the shoulder that you put down to an unfamiliar movement. Then it gets worse instead of better. After a few weeks the pain wakes you at night, and at some point you notice that you can no longer reach the top shelf of the cupboard, that you catch the seat belt only with the other hand, and that you can no longer put your jacket on in one go.

Among shoulder complaints, frozen shoulder is the one that demands the most patience. It hurts a great deal in the first phase, it then restricts you for months, and – this is the uncomfortable truth – no treatment really shortens it. What can be influenced is the pain, your sleep, the movement you retain, and how well you get through this time.

In practice we see two reactions, and both make it harder than it needs to be. One: stretching hard, "so that it opens up again", against the pain, every day. The other: not moving the arm at all, because every movement hurts. Neither route gets you there faster, and the first can even prolong the painful phase.

This text explains what actually happens in the shoulder, how a frozen shoulder is recognised and what it has to be distinguished from, how long it really lasts, what helps in the painful phase, how hard you may stretch in physiotherapy, when a procedure makes sense – and why waiting is a recognised strategy and not a capitulation. All statements are referenced; the numbers in square brackets point to the reference list at the end.

This article deals with frozen shoulder alone. If you do not yet know where your shoulder pain comes from, it is better to start with the general article on shoulder pain, which sets the different forms side by side.

One note in advance: this text does not replace individual advice. Anyone who has already had shoulder surgery, has suffered an injury or has another joint condition receives a recommendation tailored to their own situation during treatment – and that takes precedence over everything described here.

2. What a frozen shoulder is

2.1 Three names for the same thing – and what they promise

Three terms are in circulation for the same condition, and they tell different stories:

• Frozen shoulder is the name that has caught on worldwide. It describes how it feels: as though the joint had frozen solid.
• Shoulder stiffness is the plain description. It is accurate, but it names only the second half of the condition – the first half is mainly painful and not yet stiff at all.
• Adhesive capsulitis is the medical term. It is made up of "capsule", "-itis" for inflammation and "adhesive" for sticking together. It is precisely this last part that misleads: there are no adhesions in the joint that could be released [1][2].

Why this is more than quibbling over words: from the word "adhesion" follows the idea that something has to be torn open – stretch hard, move it through hard. That idea is the most common reason why people put themselves through unnecessary suffering. What is actually there is a shrunken, thickened joint capsule. It gives way again in time, but not to traction over an afternoon.

An internationally agreed definition describes frozen shoulder as a condition of unclear cause in which shoulder movement is considerably restricted both actively and passively, and in which the X-ray shows nothing apart from age-typical changes [3]. "Actively" means: you move the arm yourself. "Passively" means: someone else moves your relaxed arm. That both are restricted is the decisive feature – see section 5.1.

2.2 What actually happens inside the joint

The shoulder joint is enclosed by a capsule – a sleeve of connective tissue that seals the joint and at the same time leaves room to move. In health it is wide and folded; it has to be, so that the arm can reach over the head and behind the back.

In frozen shoulder, a two-stage process runs in this capsule [1]:

• First inflammation. Inflammatory cells gather in the capsule, small new blood vessels grow in – and with them new nerve fibres. That explains the pain of this phase: tissue that previously had hardly any pain sensors is suddenly densely supplied with them. This is why it hurts at rest and at night, not only under load.
• Then scarring. Connective tissue cells transform and produce more firm collagen – the same material scars are made of. The capsule becomes thicker, shorter, less yielding. The space inside the joint shrinks. Now the shoulder is stiff, regardless of how hard you try.

The area most affected lies at the front of the joint, the so-called rotator interval with the ligament between the coracoid process and the head of the humerus (technical term: coracohumeral ligament). If this ligament thickens, exactly the movement it normally permits is blocked: turning the arm outwards [5].

The remodelling is reminiscent of another condition, Dupuytren's contracture of the palm, in which firm cords form and gradually pull the fingers into flexion. This similarity is not a chance observation: under the microscope the tissue behaves comparably, and both conditions occur together more often than expected [4][10].

Why it all starts, nobody knows. There are hypotheses – an excessive reaction of the immune system, a metabolic influence, an inherited predisposition – but no established cause [5]. That is unsatisfying, but it has a consoling side: there is also nothing you could have done differently.

2.3 Out of the blue or as a consequence: two groups

Professionally, a distinction is made between a frozen shoulder that appears out of the blue and one with a history [1][2]:

• Primary (idiopathic). It begins without any recognisable trigger. "Idiopathic" means exactly that: without known cause. This is the more common group.
• Secondary. It follows something else – an injury to the shoulder, an operation on shoulder or chest, a longer period of immobilisation in a sling, a stroke with a paralysed arm. Accompanying conditions such as diabetes also belong here.

For everyday life this classification changes little: symptoms, examination and treatment are the same. But it makes two points visible. First, it explains why a shoulder "freezes" after a long period of rest – the capsule shrinks if it is not regularly moved into its end positions. Second, it makes it understandable why, after a shoulder injury, we begin moving as early as we are allowed to, even if only a little movement is possible.

3. Who it affects – and why

3.1 How common it is

The figure quoted everywhere is: 2 to 5 in 100 people over the course of a lifetime [1]. It rests on old and methodologically weak surveys and should be read as a rough order of magnitude, not as an established value.

More reliable are the figures on age and sex, because they come out similarly in all larger studies. Those affected are overwhelmingly people between 40 and 60; in a British follow-up of 269 shoulders the average age at onset was 53 years, and women were affected about one and a half times as often as men [17]. Before the age of 40 the condition is rare – in younger people with a stiff shoulder it is worth looking for another explanation.

3.2 Diabetes – the strongest known association

Of all accompanying circumstances, diabetes is the best studied. A summary of six comparative studies with a total of 5388 people arrived at a roughly 3.7 times higher risk of developing a frozen shoulder when diabetes is present. Two further studies that followed people over time also found an increased risk, although a smaller one [8].

Two qualifications belong with this, because they put the figure in its place. First, the quality of almost all the included studies was weak. Second, an association says nothing about direction: it is not proven that well-controlled blood sugar prevents frozen shoulder. The suspected explanation is that permanently raised sugar alters the proteins in connective tissue and makes it stiffer.

It is nevertheless of practical importance, for two reasons: in people with diabetes the condition often lasts longer and responds less well to treatment (see section 12), and anyone over 40 who develops a frozen shoulder without knowing of any diabetes should have their blood sugar checked once.

3.3 Thyroid, Dupuytren's and other companions

Besides diabetes, further associations are described. They are less well established, but conspicuously frequent in practice:

• Underactive thyroid. In one study of people with frozen shoulder, an underactive thyroid was found more often than expected [9]. That is a reason to check thyroid values if they have never been tested.
• Dupuytren's contracture. People with the firm cords in the palm of the hand develop a frozen shoulder considerably more often than others [10].
• Obesity. In a large review of medical records, obesity was linked with the diagnosis [7].
• Heart and lung disease, Parkinson's disease, a stroke with a paralysed arm – wherever the arm is moved little over a long period [11].

None of these points is a trigger you can switch off. They are pointers to who is more likely to be affected – not explanations of why it has happened to you.

3.4 After injury, surgery and immobilisation

The secondary frozen shoulder is the only form in which prevention is conceivable at all. After a shoulder injury, an operation on shoulder or chest, and after any prolonged immobilisation, the risk is increased [1].

From this follows a simple rule for aftercare: as much movement as is allowed, as early as it is allowed. "Allowed" is meant literally – after a suture to a tendon or a fracture there are instructions that must be respected. Within those instructions, however, movement is always better than keeping still, and this is why physiotherapy after such procedures usually begins within the first few days.

3.5 What does not cause a frozen shoulder

A number of things are regularly blamed and have nothing to do with it:

• Draughts, cold, an open car window. An inflamed joint capsule does not arise from cold.
• A single wrong movement. The remodelling of the capsule takes weeks. The movement during which it first hurt did not set the process off – it made it noticeable.
• Too much overhead work or poor posture. These are explanatory patterns from other shoulder complaints. For frozen shoulder there is no evidence for them.
• A disc problem in the neck. Neck problems can radiate into the shoulder, but they do not stop someone else from turning your relaxed arm outwards.

4. Warning signs – when it is not a frozen shoulder

A stiff, painful shoulder is in the great majority of cases harmless in the sense of: unpleasant, protracted, but not dangerous. There are, however, situations that should be ruled out first, before settling in for a test of patience lasting months [15].

Seek prompt medical assessment if one of these applies:

• Fever, chills, a hot and swollen shoulder, especially after an injection or an operation on the joint. This may be an infection and needs immediate assessment.
• A fall or an accident immediately before the symptoms began. A fracture or dislocation then has to be excluded, and that requires an X-ray.
• A history of cancer, or unintended weight loss, night sweats, persistent exhaustion.
• A seizure, an electric shock or a fall followed by a blocked shoulder. In this situation the shoulder may be dislocated backwards, which feels exactly like a frozen shoulder: outward rotation is blocked. This is the classic mix-up, and it is often discovered late.
• Marked weakness, numbness or tingling in the arm that extends beyond the shoulder region.
• Pain that stays just as severe at complete rest and cannot be influenced by any position.

This list is not a reason for alarm; it is the reason why an examination comes first. If none of it is present and the picture fits a frozen shoulder, you generally need no further investigation – no MRI either.

5. How a frozen shoulder is recognised

5.1 The key sign: outward rotation

Frozen shoulder is recognised through the physical examination, not through a scan [1][12]. The decisive sign can be described in one sentence:

Movement is restricted even when someone else moves the relaxed arm – and outward rotation is affected most of all.

This is how it is tested: you sit relaxed, the elbow rests against the body and is bent at a right angle, the forearm points forwards. The examiner now turns the forearm outwards, away from the midline. In a healthy shoulder this goes a long way. In a frozen shoulder the arm meets a hard stop after a few centimetres – and it does so even when you let go completely. In the large UK FROST trial this was precisely the entry criterion: a restriction of passive outward rotation by at least half compared with the other side [33].

Why this movement in particular? Because the thickened part of the capsule lies at the front and brakes exactly this rotation. And because it cannot be worked around: if movement is missing when lifting the arm, the shoulder blade can take over part of it and hide the restriction. With outward rotation and the elbow held at the body, that is not possible.

The second point is just as important: equally restricted actively and passively. With most other shoulder complaints – a tendon problem, say – you cannot lift the arm yourself, but someone else can lift it. That is exactly what is no longer possible with a frozen shoulder. Anyone who knows this difference can almost always tell the two groups apart in practice.

5.2 What else the examination covers

The examination involves more than this one movement:

• Measuring all directions, actively and passively, and comparing them with the healthy side. The numbers later serve as a point of comparison – they show progress earlier than the way it feels.
• Testing strength. In a pure frozen shoulder, strength within the remaining range is normal or reduced only by pain. Genuine weakness points to something else.
• Examining the neck, because complaints from the cervical spine can radiate into the shoulder.
• Asking about accompanying conditions – diabetes, thyroid, previous immobilisation, previous problems in the other shoulder.

An addition for the sake of accuracy: textbooks state that the restriction always follows the same pattern – outward rotation most of all, then lifting to the side, then inward rotation. A study that measured this pattern did not find it so uniform: in some of those affected, a different direction was the most restricted [13]. For practice this means measuring all directions, not only the one you expect.

A practical remark on sequence: at the very beginning, in the first few weeks, the distinction is often not yet possible. The painful early phase looks very much like an irritated tendon insertion, and the restriction of passive movement develops only over weeks. It is therefore no oversight if the diagnosis is not yet settled at the first appointment – it is made at the follow-up.

5.3 What a scan achieves – and what it does not

No MRI is needed for the diagnosis. Frozen shoulder is a clinical diagnosis; that means it follows from what you describe and from what the examination shows [12].

Images do have a role, but a limited one:

• The X-ray serves to exclude. It shows osteoarthritis of the shoulder joint, a missed backward dislocation or a fracture. In frozen shoulder itself it is unremarkable – and that is precisely the information it is meant to provide.
• Ultrasound and MRI can make the thickened capsule and the thickened ligament at the front of the joint visible [14]. This finding changes nothing about treatment. These investigations make sense when an additional problem is suspected – a tendon tear, for example – or when the course is unusual.

The catch with imaging done too early is the same as at the spine and the knee: images almost always find something. Wear on the tendons, small tears, calcium deposits – all of this is common in people of this age with no symptoms at all. If such an incidental finding then appears in the report, attention wanders there, even though it has nothing to do with the stiff shoulder. How easily a finding leads you astray is something you can work out for yourself in our learning game A false sense of certainty.

5.4 What it is confused with

Four conditions look similar to frozen shoulder but call for a different approach:

On the first three you will find fuller sections in the article on shoulder pain, and on osteoarthritis in the article on osteoarthritis.

6. The three phases – useful, but not a timetable

Almost everywhere, frozen shoulder is described in three phases [1][2]:

This model is useful because it gives a language for what you experience, and because treatment follows it: in the first phase it is about pain, in the second about movement, in the third about the return to load.

But it is not a timetable. A systematic review looked specifically for evidence that the condition really does pass through these three phases and heals completely at the end. It found none. The available studies were of low quality, and the three investigations with usable longitudinal data even showed the opposite of what was expected: the greatest improvement occurred early, not late [19].

What follows from this for you? The phases work as a description, not as a promise. Nobody can tell you: "You are in phase two now, phase three begins in four months." The transitions are fluid, the duration varies widely, and the figures in the table overlap for good reason. Anyone who clings to this timetable experiences every deviation as a setback – when deviation is the rule.

7. The course – how long does it really take?

7.1 The figures from the follow-up studies

The best guide comes from a Finnish study that followed people with a spontaneously arising frozen shoulder over 2 to 27 years. Among others, it looked at a group that had received no treatment at all. In this group the condition lasted 15 months on average, with a range of 4 to 36 months. In 94 out of 100 of these shoulders, movement was in the end as good as on the other side [16].

Two things about this figure matter. The range is wider than the average: four months and three years are both "normal". And the study is retrospective, so less robust than a planned trial – it describes what was, and does not predict with certainty what will be.

7.2 What remains in the end

The second large follow-up comes from the United Kingdom and covers 269 shoulders in 223 people, on average 4.4 years after symptoms began [17]:

• 59 out of 100 had a normal or near-normal shoulder.
• 41 out of 100 reported residual symptoms. In the vast majority of those (94 per cent) these were mild – most often a residual pain that did not get in the way of daily life.
• 6 out of 100 had substantial symptoms with pain and functional loss.

Another finding of the same work matters for the conversation: those with the most severe symptoms at the outset also had the poorer long-term outlook. An older follow-up over an average of seven years arrived at a comparable picture – about half of those affected reported mild residual symptoms without any substantial restriction of everyday life [18].

7.3 How robust these figures are

Honesty belongs here. The sentence "frozen shoulder heals by itself" appears in many patient guides, and in that absolute form it is not supported. The systematic review already mentioned found no study that had cleanly demonstrated complete resolution without treatment; the available work showed a clear but incomplete improvement over one to four years [19].

The honest formulation is therefore: almost everyone improves considerably, many recover fully, and in some a residual stiffness or residual pain remains that usually does not get in the way of everyday life. That is less catchy than "heals by itself" – but it is what the data support.

7.4 What this means for you

Three practical conclusions follow from these figures:

• Count in months, not in weeks. That sounds discouraging, but it is the single most important piece of information. If you reckon on a year, a quiet month is part of the course. If you reckon on six weeks, the same month feels like failure.
• Judge progress at intervals of weeks. From day to day it fluctuates too much. A comparison with six weeks ago is meaningful; a comparison with yesterday is not.
• Use measurements rather than impressions. How far you can turn the arm can be measured and written down. Everyday life provides its own markers: the shelf you can reach again, the seat belt you can catch yourself again.

8. The painful phase: what helps now

8.1 The corticosteroid injection into the joint

In the first, very painful phase, a corticosteroid injection into the shoulder joint is the most effective single measure. This is one of the few clear statements in this condition, and it comes from a comprehensive review of 65 studies with over 4000 participants. Only the injection into the joint was superior to the other treatments in the short term, both statistically and noticeably – for pain as well as for function, and both compared with no treatment and compared with physiotherapy alone [20].

How much does it achieve? A meta-analysis of eight controlled trials puts the pain relief compared with a sham treatment at about 1.3 points on a 10-point scale after four to six weeks, around 1 point after three to four months and still about 0.6 points after half a year – with the certainty of the statement decreasing over time. At the same time, movement and questionnaire scores improved [21].

Important for your expectations: the injection does not shorten the condition. It provides a window with less pain – and that window is valuable, because within it sleep, movement and exercise become possible again. This is exactly why the combination performs best in the reviews: injection plus a simple home programme, not the injection alone [20].

8.2 Dose, ultrasound, repetition

Three practical questions come up regularly:

• How much corticosteroid? One trial compared the two usual doses (20 and 40 milligrams of triamcinolone) with a sham injection. Both doses worked better than the sham injection, but did not differ from each other. The authors' conclusion: choose the lower dose [22]. That is good news, because corticosteroids have dose-dependent side effects – among them a temporary rise in blood sugar, which has to be considered in diabetes.
• Given with or without ultrasound? Ultrasound guidance hits the joint more reliably. For the outcome experienced by the patient the difference is small.
• How often? A repeat after some weeks is possible if the first injection clearly helped and the effect is wearing off. Series of many injections make no sense: repeated corticosteroids can damage tendon tissue, and if two injections have achieved nothing, the third will not either.

Side effects are rare and mostly harmless. In the meta-analysis mentioned, around 2 in 100 of those treated had temporary facial flushing, and less often dizziness, nausea or a brief increase in pain [21].

8.3 Corticosteroids as tablets

If an injection is not possible or not wanted, a short course of corticosteroid tablets can help. A Cochrane review found an improvement in pain, movement and function – but only in the short term; beyond about six weeks the advantage could no longer be demonstrated [23]. The usual approach is a course over a few weeks with subsequent tapering, prescribed and supervised by a doctor.

8.4 Painkillers

Anti-inflammatory painkillers (technical term: non-steroidal anti-inflammatory drugs, NSAIDs – these include ibuprofen and diclofenac) and paracetamol are frequently used. For frozen shoulder in particular their effect is poorly studied; they are not a remedy against the condition, but against the pain [6].

That is nevertheless a sensible goal. The measure is not "pain-free" but: enough relief to sleep and to move the arm within the permitted range. The choice, the dose and the duration are decided by the doctor – NSAIDs in particular are not unproblematic in stomach, kidney and heart conditions and alongside certain other medicines.

8.5 Sleeping

Disturbed sleep is for many the greatest burden of the first phase – greater than the restriction itself. What helps most often in practice:

• Lying on your back or on the healthy side, with a pillow between the chest and the affected arm, so that the arm rests slightly away from the body and supported towards the front.
• When lying on the healthy side, place a second pillow in front of your abdomen and rest the affected arm on it rather than letting it fall forwards.
• Warmth in the evening – a heat pack, a warm shower – loosens things and makes falling asleep easier. Warmth heals nothing, but it helps you fall asleep, and that counts.
• Timing a painkiller so that its effect is strongest in the first half of the night. This too belongs in the conversation with your doctor.

9. Physiotherapy – and the question of how hard to stretch

9.1 What it achieves

Let us begin with what the studies do not show: physiotherapy does not shorten a frozen shoulder. A Cochrane review analysed 32 studies with 1836 participants on manual therapy and exercise. The most important result: compared with the corticosteroid injection, the combination of manual therapy and exercise performed worse in the short term – after seven weeks, 46 out of 100 of those treated reported treatment success, against 77 out of 100 in the injection group. At six and at twelve months the differences between groups were no longer meaningful. The certainty of the evidence was largely low [24].

A more recent summary of 33 studies reaches a similarly restrained picture: movement, function and pain improve with exercise programmes, but the differences between the various programmes are small. Supervised exercise tended to be better than a purely home-based programme; additional devices and applications brought no advantage. Which dose is the right one, the authors could not answer [25].

What physiotherapy is then good for:

• For making sense of it. Knowing what you have, what is still to come and how long it lasts changes how you handle it more than any technique does. Expectation and confidence are demonstrably linked with the outcome [30].
• For dosing. The most common mistakes are too much and too little. Both can be steered by the reaction in the 24 hours after exercising.
• For preservation. Whatever movement is there should be kept; whatever strength would be lost does not have to be lost.
• For everyday life. Dressing, driving, working, sleeping – there are solutions for these, and they matter more for quality of life than one more degree of movement would.
• For monitoring. If the course differs from what was expected, regular measurement makes that apparent.

9.2 How hard may you stretch?

This is the most important question in this text, and the answer contradicts what most people expect.

A Dutch study compared two routes in 77 people with frozen shoulder. One group received intensive physiotherapy with passive stretching and vigorous mobilisation beyond the pain threshold. The other group received supervised restraint – explanation, exercises expressly within the pain threshold, nothing else. After two years, 89 out of 100 in the restrained group had achieved a normal or near-normal, pain-free shoulder function, against only 63 out of 100 in the intensively treated group. In the restrained group, moreover, nearly two thirds had reached this result within twelve months [26].

This study has a weakness you should know about: participants were not randomised but allocated, and the groups may therefore have differed. Its result is, however, so clear and fits so well with everything else that it has changed practice.

Does that mean gentle treatment is always better? No, and a distinction is needed here. Another study with 100 participants compared vigorous with cautious mobilisation, each over twelve weeks. Both groups improved considerably; the more vigorous group had small advantages in outward rotation and in the questionnaires. Overall the differences were small [27]. The decisive difference from the first study: here the symptoms had been present for at least three months, so the painful early phase was over.

This is how the two fit together – and it is the practical rule:

• In the painful phase (pain at rest and at night) you move within the pain threshold. The aim is to preserve the movement you have – not to force new movement. Vigorous stretching in this phase increases the pain and achieves nothing.
• In the stiff phase (pain clearly reduced, restriction in the foreground) you may work noticeably at the limit. A distinct pull at the end of range is fine, as long as it settles quickly after exercising.

9.3 The rule you can go by

For home use, one simple rule has proved itself, and it makes the difference between "demanding" and "too much" tangible:

What counts is not how hard it pulls while you exercise, but how you feel 24 hours later.

• Next morning the pain is as before or better, and movement is unchanged or better → the dose is right.
• Next morning the pain is worse, or you slept worse, or you get less far than the day before → it was too much. Do not stop; make the next session shorter and gentler.

This rule applies here in particular, because the inflamed capsule reacts very sensitively to over-stretching. More often and shorter beats seldom and long: five minutes five times a day works more reliably than half an hour once a day.

9.4 Which exercises make sense

There is no exercise programme that has been shown to be superior in frozen shoulder. What the programmes in the studies have in common is their simplicity:

• Pendulum swings. Lean forward with support, let the arm hang loosely and describe small circles by shifting your body weight – the arm stays passive throughout. Comfortable in the painful phase.
• Assisted lifting. Lying down or seated, guide the affected arm with the healthy hand or a stick. That way the healthy arm does the moving and the painful muscles do not have to work.
• Outward rotation at the door frame. Elbow at the body, forearm forwards, place the back of the hand against the door frame and slowly turn away from it. The movement that is missing most – which is why it belongs in the programme.
• Hand behind the back. With a towel held from above by the healthy hand, gently guide the affected arm upwards. You need this movement constantly in daily life, from dressing to personal care.
• Wall walking. Walk the fingers up the wall to regain lifting in small steps. The feedback is clearly visible – the height on the wall is a measurement.

On dosing: several short sessions spread over the day, a few repetitions of each movement, each end position held for a few seconds, always following the rule from section 9.3. What you learn in the practice is less the exercise itself than how to steer it.

9.5 Do not forget strength

Amid all the attention on movement it is regularly forgotten that a shoulder used little for months loses strength. This loss is not a consequence of the condition itself but of the sparing – and it often delays the return to work and sport more than the stiffness does.

Strengthening is almost always possible, even with severely restricted movement:

• Without movement. Press against a firm resistance – the hand against a wall, the forearm against the door frame – and hold the tension for a few seconds. The joint does not move, the muscle works nonetheless.
• In the range that is free. Most people with a frozen shoulder can do a fair amount with the elbow held at the body, even when lifting overhead is impossible.
• All around. Elbow, hand and shoulder blade remain unaffected and should be used.

The fundamentals of building strength are in the article on strength training.

9.6 What achieves little or nothing

• Electrotherapy, ultrasound, laser, magnetic field. A Cochrane review of these applications in frozen shoulder found no established benefit; the few studies were small and of low quality [28]. In the more recent summary, adding such devices to an exercise programme likewise brought no advantage [25].
• Vigorous stretching against the pain in the painful phase – see section 9.2.
• "Releasing adhesions". There are no adhesions to release. Treatments justified in these terms rest on a false picture of what is happening in the joint [2].
• Complete rest. The other extreme. The capsule shrinks further, strength is lost, and the return takes longer.

9.7 What is unresolved

Honesty requires this: the evidence on physiotherapy for frozen shoulder is weak. The studies are small, the programmes differ, follow-up is usually short, and the best dose is unknown [24][25]. We know reasonably well what does not help (vigorous stretching in the early phase, device-based applications) and what reduces pain in the short term (the injection). What exactly physiotherapy contributes to the long-term outcome is not cleanly established.

That is not an argument against treatment – but it is one against big promises and against long series of identical sessions without a checkable goal.

10. Waiting as a strategy in its own right

From the foregoing follows an option that is rarely spelled out: it is defensible, after the explanation and any injection, to opt for a home programme and to do without regular therapy sessions.

An American trial examined exactly this. 61 people with frozen shoulder were randomly assigned either to physiotherapy or to supervised waiting; both groups were offered a corticosteroid injection, which almost all took up. Over twelve months both groups improved considerably, and did so with no discernible difference in pain and function at any time point. Physiotherapy meanwhile cost about ten times as much [29].

The trial is small and comes from a single centre, so its result should not be overstretched. But it supports an attitude that suits this condition: waiting here is not neglect.

What we opt for in practice depends on the situation. Supervised waiting with a home programme makes sense when the symptoms can be clearly assigned, sleep works reasonably well and you feel able to carry the programme through yourself. Regular appointments make sense when the pain is severe and the dose has to be adjusted repeatedly, when a second condition is involved, when work or sport place high demands – or when the uncertainty itself becomes a burden.

11. When nothing moves: the procedures

11.1 When they come up at all

Procedures only come into question when the restriction is substantial, clearly hinders everyday life and persists over many months despite non-surgical treatment – the figure usually quoted is about 18 months [6]. Before that point the rule is: the course needs time, and a procedure does not reliably shorten it.

11.2 The three procedures

• Distending the joint with fluid (hydrodilatation). Under imaging guidance, fluid is injected into the joint together with corticosteroid until the capsule stretches. No incision, no general anaesthetic. A Cochrane review found short-term improvements in pain, movement and function compared with a sham treatment, but could not clarify whether the procedure is superior to other treatments [31]. A later meta-analysis was more sober: the effect on pain was small, the effect on everyday function was not demonstrable, and the amount of fluid injected made no difference [32].
• Manipulation under anaesthesia. Under a short general anaesthetic the arm is moved in a guided way until the shrunken capsule gives way. This takes a few minutes and needs no incision. The procedure is old, widely used, and its effectiveness has never been tested against a sham treatment [34]. Known risks are fractures, tendon tears and dislocations – rare, but real.
• Arthroscopic capsular release. Through small incisions the thickened capsule is divided in a targeted way with instruments, usually at the front in the rotator interval. The procedure allows the extent to be controlled and is preferred in people with diabetes or after unsuccessful manipulation. A review of 22 studies found, however, no established advantage over manipulation under anaesthesia – with consistently low study quality [35].

11.3 The trial that compared all three

The British UK FROST trial is the largest investigation of this question. At 35 hospitals, 503 people with a primary frozen shoulder were randomly assigned to one of three routes: manipulation under anaesthesia, arthroscopic capsular release, or an early structured physiotherapy programme – all three with a corticosteroid injection and follow-up treatment [33].

After twelve months it looked like this (measured with the Oxford Shoulder Score, a questionnaire from 0 to 48 points in which higher values are better):

The largest gap – between arthroscopic release and physiotherapy – was about 3 points. The researchers had set 5 points in advance as meaningful. None of the three procedures was clinically superior to the others. The most cost-effective was manipulation under anaesthesia; the highest risks lay with arthroscopy.

That is a remarkable result: two operations, a year of follow-up, and at the end physiotherapy stands practically level. For the decision this means there is no obviously best choice – rather a weighing of risk, effort, waiting time and the question of how long you can still live with your present state.

11.4 After the procedure

What a procedure creates is a window of time: the capsule gives way, the joint can be moved further. Whether this gain is kept is decided in the weeks that follow.

This is why movement therapy begins immediately – as a rule on the same or the next day – and is carried out frequently at first, several times a day in small sessions. A corticosteroid injection during the procedure dampens the newly triggered inflammation. The first weeks are the most demanding, because a lot of exercising is needed while the shoulder is freshly irritated.

12. When diabetes is involved

People with diabetes not only develop a frozen shoulder more often [8], their course is also more stubborn. This shows even in the results after a procedure: in a comparison of 21 people with and 21 without diabetes after arthroscopic capsular release, the scores in the diabetes group were considerably worse at six months, and at two years 15 of 21 there had regained full movement, against 19 of 21 in the comparison group [36].

Four points follow from this for treatment:

• Adjust expectations. On average it takes longer. Knowing that beforehand is better than experiencing it along the way as a setback.
• Corticosteroids with care. An injection into the joint raises blood sugar for a few days, corticosteroid tablets more strongly and for longer. Both are usually manageable, but they must be announced and blood sugar monitored more closely during that time.
• Keep an eye on the other shoulder, because a course affecting both sides is more common in diabetes.
• Control blood sugar well. That better control shortens a frozen shoulder is not proven. It is worthwhile for many other reasons.

13. The other shoulder

One question comes up almost every time: can this happen on the other side too?

It can. In the British follow-up, 20 out of 100 of those affected reported symptoms in both shoulders – one after the other, not at the same time [17]. This is more common in diabetes.

Equally important is the other half of the answer: in the same shoulder the condition practically never returns. In that same study there was not a single recurrence among 269 shoulders [17]. Anyone who has once been through this time does not have to go through it a second time in that shoulder.

14. Everyday life: sleeping, dressing, working, sport

Because the condition lasts so long, everyday life decides quality of life – more than any single treatment. What has proved useful:

• Dressing. The affected arm into the sleeve first, and out of it last when undressing. Shirts and blouses buttoned at the front rather than jumpers over the head. For women: a bra fastening at the front, or fasten it at the front and then turn it round.
• Personal care. Long-handled brushes and combs; keep shower gel and shampoo at hand height, not on the top shelf.
• Driving. Guide the seat belt over the shoulder with the healthy hand. When reversing, use the camera or the mirrors rather than twisting your upper body.
• Working. Arrange everything frequently used within reach below shoulder height. For overhead work: rebuild the task rather than gritting your teeth. A time-limited medical certificate with a clear restriction ("no activities above shoulder height") is often more useful than complete incapacity for work.
• Sport. Anything that works without the shoulder carries on – cycling with upright handlebars, walking, hiking, leg training, trunk exercises. What has to pause are sports with a throwing, striking or falling risk and sports with heavy overhead load. This break is temporary.
• Carrying. Carry loads close to the body with the elbow at your side; split shopping between both sides or use a rucksack.

And one point that often gets lost: this condition wears you down. Months of lost sleep and a restriction nobody can see take their toll on your mood. If you experience that, it is not a weakness of character but an expected consequence – and a good reason to raise it.

15. What you can do yourself

• Move daily, several times, briefly. Five minutes five times a day beats half an hour in one go.
• Stick to the 24-hour rule from section 9.3. It replaces every discussion about how far you may go.
• Look after your sleep. In the painful phase this is the single most important measure – if need be with an injection and with a painkiller taken in the evening as agreed.
• Keep your strength. Even with a stiff shoulder the arm can be tensed, and elbow, hand and the rest of the body remain fully usable anyway.
• Measure rather than estimate. Note down two or three everyday markers once a week: can I reach the shelf? Can I reach my back? How many nights did I sleep through?
• Have blood sugar and thyroid checked if that has never been done.
• Count in months. This expectation protects you from daily disappointment – and disappointment is the biggest disruptive factor in this condition.

16. Where, how often and how long therapy takes place

In Switzerland a medical prescription covers nine sessions [37]. For a frozen shoulder that is often enough for the first stretch: making sense of it, getting the pain under control, setting up the home programme, practising the dosing. If the restriction persists, or after a procedure, a second prescription follows.

Because the condition lasts so long, the distribution matters more than the number. With this diagnosis it is often more sensible to stretch the sessions over months – closer together at first, then with larger intervals for checking and adjusting – than to use them up in six weeks and then be left without support.

The decisive point is the same as everywhere: what happens between appointments decides the outcome. The session is the steering, not the treatment.

And a criterion for the interim assessment: after six to eight weeks something should have moved – less night pain, more degrees of outward rotation, more everyday markers within reach. If nothing moves, we change the approach or recommend a fresh medical assessment. What makes no sense is carrying on unchanged because sessions are left.

17. Eight common misunderstandings

• "Something is stuck together in there and has to be released." There are no adhesions. The capsule is inflamed, thickened and shrunken – a remodelling of the tissue, not a sticking together [1][2].
• "Stretch hard and you will be through it sooner." In the painful phase the opposite holds: in a comparative study, restraint led to better results after two years than intensive stretching [26].
• "I need to have an MRI." The examination makes the diagnosis. A scan serves to exclude other causes, not to prove a frozen shoulder [12].
• "Corticosteroids damage the joint." Individual injections into the joint are the most effective short-term measure in this condition; side effects are rare and mostly harmless [20][21]. It is only frequent repetition that is unfavourable.
• "If I wait, it will stay stiff." In the great majority of people movement largely returns; in one follow-up without treatment it was 94 out of 100 [16].
• "An operation solves the problem quickly." In the largest comparative trial, no procedure was clinically superior to the others at twelve months – not even the two operations [33].
• "I brought this on myself with a wrong movement." The remodelling of the capsule runs over weeks. The movement during which it first hurt did not set it off.
• "Afterwards the shoulder is ruined." Frozen shoulder leaves behind no osteoarthritis and no damage to the cartilage. What can remain is a residual stiffness – usually of no consequence for everyday life [17].

18. When to get in touch with us

An assessment makes sense when:

• a shoulder hurts increasingly over weeks and becomes stiffer at the same time, especially at night;
• you do not know whether it is a frozen shoulder or something else – the distinction determines the whole approach;
• you are unsure how hard you may exercise, and swing between too much and too little;
• sleep has been disturbed for weeks;
• the restriction affects work or sport and a plan for adapting is needed;
• a procedure is on the table and you would like to sort out the arguments;
• you have had surgery and need to set up the aftercare.

What to expect in our practice: first, making sense of it – does the picture fit a frozen shoulder, is passive outward rotation restricted, are there warning signs, are accompanying conditions known. Then honest information about the expected time frame, because experience shows this relieves the most. Then the work: a home programme with clear dosing, the ranges of movement recorded in writing, strength work within the possible range, solutions for sleep, dressing and work. We measure over time and take stock after six to eight weeks. If an injection or a procedure is up for discussion, we prepare that conversation with you – the decision is made together with your doctor.

19. In summary

Frozen shoulder is a condition of the joint capsule: first an inflammation that hurts severely, then a scarring that makes it stiff. It usually comes without any recognisable trigger, mainly affects people between 40 and 60, and occurs considerably more often in diabetes. Nothing is stuck together – despite the technical term "adhesive capsulitis".

It is recognised in the examination, not on a scan: movement is restricted even when someone else moves the relaxed arm, and outward rotation is affected most. An X-ray serves to exclude other causes; an MRI is generally not needed.

The course is protracted and ends well for almost everyone. In the untreated group of one follow-up study it lasted 15 months on average, with a range of 4 to 36; a good four years after onset, close to 6 in 10 had a normal or near-normal shoulder, and most residual symptoms were mild. The notion of a fixed timetable of three phases is a description, not a law.

What is treated is not the capsule but the pain and the time. In the painful phase a corticosteroid injection into the joint is the most effective single measure, best combined with a simple home programme. Stretching hard against the pain does not help in this phase and tends to do harm; in the stiff phase you may work noticeably at the limit. What matters is not how much it hurts while exercising, but how it is 24 hours later.

Procedures only come into consideration after many months, and the largest comparative trial found no clinically meaningful difference after a year between two operations and a structured physiotherapy programme. So by waiting you miss nothing.

The most useful question here, too, is not "when will it be gone?" but: what exactly can I not do right now, and what is the next step back? In this condition the answer is almost always: dampen the pain, protect your sleep, move every day – and give the capsule the time it needs.

References

All Digital Object Identifiers (DOIs) were individually checked against the Crossref register. The links in the reference list lead via the DOI service to the publishers' pages. Some of these are located outside Switzerland and the EU. When you click, your IP address is transmitted to the respective provider – this does not happen on our own site.

[1] Millar NL, Meakins A, Struyf F, et al. Frozen shoulder. Nature Reviews Disease Primers. 2022;8(1):59. https://doi.org/10.1038/s41572-022-00386-2

[2] Lewis J. Frozen shoulder contracture syndrome – Aetiology, diagnosis and management. Manual Therapy. 2015;20(1):2–9. https://doi.org/10.1016/j.math.2014.07.006

[3] Zuckerman JD, Rokito A. Frozen shoulder: a consensus definition. Journal of Shoulder and Elbow Surgery. 2011;20(2):322–325. https://doi.org/10.1016/j.jse.2010.07.008

[4] Bunker TD, Anthony PP. The pathology of frozen shoulder. A Dupuytren-like disease. The Journal of Bone and Joint Surgery. British volume. 1995;77-B(5):677–683. https://doi.org/10.1302/0301-620X.77B5.7559688

[5] Kraal T, Lübbers J, van den Bekerom MPJ, et al. The puzzling pathophysiology of frozen shoulders – a scoping review. Journal of Experimental Orthopaedics. 2020;7(1):91. https://doi.org/10.1186/s40634-020-00307-w

[6] Haas R, Ibounig T, Buchbinder R. Management of Shoulder Pain in Primary Care: A Review. JAMA Internal Medicine. 2026. Published online August 17, 2026. https://doi.org/10.1001/jamainternmed.2026.3135

[7] Kingston K, Curry EJ, Galvin JW, Li X. Shoulder adhesive capsulitis: epidemiology and predictors of surgery. Journal of Shoulder and Elbow Surgery. 2018;27(8):1437–1443. https://doi.org/10.1016/j.jse.2018.04.004

[8] Dyer BP, Rathod-Mistry T, Burton C, van der Windt D, Bucknall M. Diabetes as a risk factor for the onset of frozen shoulder: a systematic review and meta-analysis. BMJ Open. 2023;13(1):e062377. https://doi.org/10.1136/bmjopen-2022-062377

[9] Schiefer M, Teixeira PFS, Fontenelle C, et al. Prevalence of hypothyroidism in patients with frozen shoulder. Journal of Shoulder and Elbow Surgery. 2017;26(1):49–55. https://doi.org/10.1016/j.jse.2016.04.026

[10] Smith SP, Devaraj VS, Bunker TD. The association between frozen shoulder and Dupuytren's disease. Journal of Shoulder and Elbow Surgery. 2001;10(2):149–151. https://doi.org/10.1067/mse.2001.112883

[11] Wang K, Ho V, Hunter-Smith DJ, Beh PS, Smith KM, Weber AB. Risk factors in idiopathic adhesive capsulitis: a case control study. Journal of Shoulder and Elbow Surgery. 2013;22(7):e24–e29. https://doi.org/10.1016/j.jse.2012.10.049

[12] Kelley MJ, Shaffer MA, Kuhn JE, et al. Shoulder Pain and Mobility Deficits: Adhesive Capsulitis. Journal of Orthopaedic & Sports Physical Therapy. 2013;43(5):A1–A31. https://doi.org/10.2519/jospt.2013.0302

[13] Rundquist PJ, Ludewig PM. Patterns of motion loss in subjects with idiopathic loss of shoulder range of motion. Clinical Biomechanics. 2004;19(8):810–818. https://doi.org/10.1016/j.clinbiomech.2004.05.006

[14] Zappia M, Di Pietto F, Aliprandi A, et al. Multi-modal imaging of adhesive capsulitis of the shoulder. Insights into Imaging. 2016;7(3):365–371. https://doi.org/10.1007/s13244-016-0491-8

[15] Rangan A, Hanchard N, McDaid C. What is the most effective treatment for frozen shoulder? BMJ. 2016;354:i4162. https://doi.org/10.1136/bmj.i4162

[16] Vastamäki H, Kettunen J, Vastamäki M. The Natural History of Idiopathic Frozen Shoulder: A 2- to 27-year Followup Study. Clinical Orthopaedics & Related Research. 2012;470(4):1133–1143. https://doi.org/10.1007/s11999-011-2176-4

[17] Hand C, Clipsham K, Rees JL, Carr AJ. Long-term outcome of frozen shoulder. Journal of Shoulder and Elbow Surgery. 2008;17(2):231–236. https://doi.org/10.1016/j.jse.2007.05.009

[18] Shaffer B, Tibone JE, Kerlan RK. Frozen shoulder. A long-term follow-up. The Journal of Bone & Joint Surgery. 1992;74(5):738–746. https://doi.org/10.2106/00004623-199274050-00013

[19] Wong CK, Levine WN, Deo K, et al. Natural history of frozen shoulder: fact or fiction? A systematic review. Physiotherapy. 2017;103(1):40–47. https://doi.org/10.1016/j.physio.2016.05.009

[20] Challoumas D, Biddle M, McLean M, Millar NL. Comparison of Treatments for Frozen Shoulder: A Systematic Review and Meta-analysis. JAMA Network Open. 2020;3(12):e2029581. https://doi.org/10.1001/jamanetworkopen.2020.29581

[21] Sun Y, Zhang P, Liu S, et al. Intra-articular Steroid Injection for Frozen Shoulder: A Systematic Review and Meta-analysis of Randomized Controlled Trials With Trial Sequential Analysis. The American Journal of Sports Medicine. 2017;45(9):2171–2179. https://doi.org/10.1177/0363546516669944

[22] Yoon SH, Lee HY, Lee HJ, Kwack KS. Optimal Dose of Intra-articular Corticosteroids for Adhesive Capsulitis: A Randomized, Triple-Blind, Placebo-Controlled Trial. The American Journal of Sports Medicine. 2013;41(5):1133–1139. https://doi.org/10.1177/0363546513480475

[23] Buchbinder R, Green S, Youd JM, Johnston RV. Oral steroids for adhesive capsulitis. Cochrane Database of Systematic Reviews. 2006;(4):CD006189. https://doi.org/10.1002/14651858.CD006189

[24] Page MJ, Green S, Kramer S, Johnston RV, McBain B, Chau M, Buchbinder R. Manual therapy and exercise for adhesive capsulitis (frozen shoulder). Cochrane Database of Systematic Reviews. 2014;(8):CD011275. https://doi.org/10.1002/14651858.CD011275

[25] Mertens MG, Meert L, Struyf F, Schwank A, Meeus M. 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;103(5):998–1012.e14. https://doi.org/10.1016/j.apmr.2021.07.806

[26] Diercks RL, Stevens M. Gentle thawing of the frozen shoulder: A prospective study of supervised neglect versus intensive physical therapy in seventy-seven patients with frozen shoulder syndrome followed up for two years. Journal of Shoulder and Elbow Surgery. 2004;13(5):499–502. https://doi.org/10.1016/j.jse.2004.03.002

[27] Vermeulen HM, Rozing PM, Obermann WR, le Cessie S, Vliet Vlieland TPM. Comparison of High-Grade and Low-Grade Mobilization Techniques in the Management of Adhesive Capsulitis of the Shoulder: Randomized Controlled Trial. Physical Therapy. 2006;86(3):355–368. https://doi.org/10.1093/ptj/86.3.355

[28] Page MJ, Green S, Kramer S, Johnston RV, McBain B, Buchbinder R. Electrotherapy modalities for adhesive capsulitis (frozen shoulder). Cochrane Database of Systematic Reviews. 2014;(10):CD011324. https://doi.org/10.1002/14651858.CD011324

[29] Martin SD, Dean MC, Eberlin CT, et al. AAOS 2024 best paper in the shoulder and elbow classification: watchful waiting provides higher value with similar functional outcomes to physical therapy for frozen shoulder: a prospective randomized controlled trial. Journal of Shoulder and Elbow Surgery. 2025;34(6):e455–e467. https://doi.org/10.1016/j.jse.2024.09.017

[30] Chester R, Jerosch-Herold C, Lewis J, Shepstone L. Psychological factors are associated with the outcome of physiotherapy for people with shoulder pain: a multicentre longitudinal cohort study. British Journal of Sports Medicine. 2018;52(4):269–275. https://doi.org/10.1136/bjsports-2016-096084

[31] Buchbinder R, Green S, Youd JM, Johnston RV, Cumpston M. Arthrographic distension for adhesive capsulitis (frozen shoulder). Cochrane Database of Systematic Reviews. 2008;(1):CD007005. https://doi.org/10.1002/14651858.CD007005

[32] Saltychev M, Laimi K, Virolainen P, Fredericson M. Effectiveness of Hydrodilatation in Adhesive Capsulitis of Shoulder: A Systematic Review and Meta-Analysis. Scandinavian Journal of Surgery. 2018;107(4):285–293. https://doi.org/10.1177/1457496918772367

[33] 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. The Lancet. 2020;396(10256):977–989. https://doi.org/10.1016/S0140-6736(20)31965-6

[34] Kraal T, Beimers L, The B, Sierevelt I, van den Bekerom M, Eygendaal D. Manipulation under anaesthesia for frozen shoulders: outdated technique or well-established quick fix? EFORT Open Reviews. 2019;4(3):98–109. https://doi.org/10.1302/2058-5241.4.180044

[35] Grant JA, Schroeder N, Miller BS, Carpenter JE. Comparison of manipulation and arthroscopic capsular release for adhesive capsulitis: a systematic review. Journal of Shoulder and Elbow Surgery. 2013;22(8):1135–1145. https://doi.org/10.1016/j.jse.2013.01.010

[36] Mehta SS, Singh HP, Pandey R. Comparative outcome of arthroscopic release for frozen shoulder in patients with and without diabetes. The Bone & Joint Journal. 2014;96-B(10):1355–1358. https://doi.org/10.1302/0301-620X.96B10.34476

[37] Verordnung des EDI über Leistungen in der obligatorischen Krankenpflegeversicherung (KLV), SR 832.112.31, Art. 5 (Physiotherapie). Zusammengefasst in: FMH. Verordnung von Physiotherapie im KVG-Bereich – Frequently Asked Questions. Stand 31.10.2022. fmh.ch (PDF)