Frozen Shoulder: The Condition, and Why “It Gets Better on Its Own” Is Not Quite True

You have probably been told frozen shoulder resolves on its own within one to three years. That is the standard line, it is on most patient information sheets, and it is more optimistic than the long-term evidence supports.

A systematic review looking specifically at this question found no evidence supporting the theory that patients progress through recovery phases to full resolution without treatment. What the long-term studies show instead is that most people end up functionally acceptable, and a substantial minority never get fully back to normal.

That is worth knowing at the start, because it changes how you think about treatment. If it always resolved completely, waiting would be free. It does not, so waiting has a cost.

What is actually happening in the shoulder

The capsule around your shoulder joint, normally a loose bag that lets the joint move freely, becomes inflamed and then thickens and contracts. The joint physically shrinks.

The process runs in two overlapping phases. First inflammation, which is where the severe pain comes from: synovial overgrowth, new blood vessel formation, and new nerve growth into tissue that should not have many nerves in it. Then fibrosis, where the capsule lays down new collagen and tightens.

The area most affected is the rotator interval, at the front of the shoulder, and the coracohumeral ligament, which thickens noticeably. Arthrograms show the joint volume physically reduced.

The landmark tissue study, examining capsule taken from 12 patients, found active fibroblast proliferation with those cells transforming into myofibroblasts, which are contractile cells that literally pull tissue tighter. The collagen was predominantly type III, meaning new. The researchers noted the tissue was histologically indistinguishable from Dupuytren’s disease in the hand.

The name is wrong, and it matters

You will see this condition called adhesive capsulitis. Both halves of that name are questionable.

There are no adhesions. The same tissue study that found the myofibroblasts found no adhesions at all. The restricted movement is explained by the capsule contracting, not by things sticking together. Even Neviaser, who coined the term in 1945, was describing capsule peeling away from bone rather than adhesions, and reportedly neither found nor intended to convey that adhesions existed.
It is not straightforwardly capsulitis either. The same study found inflammatory cells scanty and confined to around blood vessels.

The ISAKOS Upper Extremity Council has explicitly rejected the term adhesive capsulitis for that reason, and recommends simply frozen shoulder.

One genuine disagreement to flag: the AAOS patient page still states that thick bands of tissue called adhesions develop. That conflicts directly with the histology literature. If your surgeon describes adhesions and your physiotherapist says there are none, that is the argument you have walked into.

The three stages, and whether they are real

Everyone is taught the same three stages: freezing, frozen, thawing. Here is what different authoritative sources give for the durations.

Source	Freezing	Frozen	Thawing
AAOS	6 weeks to 9 months	4 to 6 months	6 months to 2 years
StatPearls	2 to 9 months	4 to 12 months	variable
One review	around 3 months	3 to 9 months	9 to 18 months

Notice how much they vary. That variation is itself informative.

The systematic review that looked for evidence of the staged model examined seven studies covering 239 shoulders. It found moderate quality evidence that improvement happens early and then slows, which is a continuous curve rather than discrete phases.

Imaging does not confirm staging either. Only two MRI findings correlate with stage at all, and both also appear later in the condition, so MRI alone cannot define which stage you are in. Studies that claim to stage patients are almost always doing it by how long symptoms have lasted, not by anything objective.

A 2020 review recommended abandoning the three-phase, self-limiting characterization entirely.

Practical version: the stages are a useful way of describing a shoulder that hurts more at the start and stiffens more later. They are not a timetable, and you should not expect to feel yourself crossing a line from one to the next.

Who gets it

Around 2 to 5 percent of the population. Peak onset in the fifties and sixties, with a mean around 55. Rare outside that band. Women more than men, at roughly 1.4 to 1.6 to one. The non-dominant shoulder is affected more often.

There is a genetic component: having a twin who has had it raises your risk two to three fold.

Diabetes: the strongest association by far

From a meta-analysis screening 5,411 articles and including 18:

13.4 percent of people with diabetes develop frozen shoulder
People with diabetes are 5 times more likely to get it than people without
30 percent of frozen shoulder patients have diabetes

You will often read that type 1 diabetes carries a higher risk than type 2. The meta-analysis did not support that: 5.8 percent versus 12.4 percent, with confidence intervals so wide the difference was not statistically significant. Insulin treatment versus tablets made no significant difference either.

Genetic analysis has identified diabetes as a causal risk factor rather than merely an associated one, with advanced glycation end products proposed as the mechanism.

Other associations

Thyroid disease, both under and overactive
Dupuytren’s disease, with one source reporting up to 60 percent of frozen shoulder patients having a history of it, though that figure is high and repeated between secondary sources rather than well anchored
Raised cholesterol, odds ratio 1.37
Cervical spondylosis, odds ratio 2.48
Parkinson’s disease, at 12.7 percent versus 1.7 percent in controls
Immobilization after surgery or fracture

Interestingly, higher body mass index appeared slightly protective in one analysis, and atherosclerosis was not an independent risk factor.

Primary versus secondary

Primary means it appeared without an obvious trigger. It is strongly linked to systemic conditions such as diabetes and thyroid disease.

Secondary means it followed something identifiable: an injury, an operation, or a period in a sling. That version behaves differently enough to deserve its own guide, which is shoulder stiffness after surgery and injury.

A survey of shoulder specialists found 85 percent agreed with dividing the condition this way, and 84 percent agreed primary frozen shoulder is a distinct clinical entity.

How it is diagnosed

Clinically. History and examination. Routine X-rays offer little over that, according to the British Elbow and Shoulder Society pathway.

The defining feature is restriction of both active and passive movement with normal-looking X-rays. That is the single most useful thing to understand about this diagnosis.

If your shoulder hurts but someone else can move your arm through full range, you do not have frozen shoulder. You probably have a rotator cuff problem. If someone else cannot move it either, and they can feel a hard mechanical block at the end rather than you guarding against pain, that is frozen shoulder.

External rotation goes first. Turning your forearm outward with your elbow at your side is typically the most restricted movement and the earliest to go, followed by abduction, then internal rotation, then flexion.

The largest trial in this field used a specific threshold for entry: passive external rotation restricted to less than 50 percent of the other shoulder. That is a reasonable working definition.

Pain is typically at the front of the shoulder, over the biceps groove, and worse at night.

Imaging is exclusionary rather than diagnostic. X-rays rule out arthritis, which would show joint space narrowing that frozen shoulder does not. MRI, if done, may show thickening of the coracohumeral ligament and the capsule at the bottom of the joint, and loss of the fat triangle under the coracoid.

The prognosis question

This is the part most sites get wrong, and it is the reason this article exists.

What you are usually told

The AAOS says it generally gets better over time without surgery, with full recovery taking up to three years. The NHS says pain and stiffness will usually go away eventually, but that this takes months and sometimes years. StatPearls says around 80 percent regain near-normal function, with 10 to 20 percent left with residual stiffness and discomfort.

What the long-term follow-up studies actually found

Study	Follow-up	Finding
Reeves, 49 patients	5 to 10 years	Around 40 percent full recovery. More than half had some limitation of movement. 7 percent functionally restricted.
Binder, 40 patients	mean 44 months	40 percent still objectively restricted, though reporting little functional impairment.
Shaffer, 61 patients	mean 7 years	50 percent reported ongoing pain or stiffness. 60 percent had reduced range of motion. 11 percent functional deficit.
Hand, 269 shoulders	mean 4.4 years	59 percent normal or near-normal. 41 percent with ongoing symptoms, of which 94 percent mild and 6 percent severe.
Korean cohort, 234 shoulders	mean 42 months	72 percent subjectively satisfied, but 40 percent had persistent motion deficits.

There is one study on the other side worth taking seriously. Vastamäki followed 51 untreated patients for between 2 and 27 years and found 94 percent recovered range of motion equal to their other shoulder, with mean symptom duration of 15 months. That is the strongest evidence for the self-limiting claim.

Two caveats on it. It was retrospective. And even in that group, only 51 percent were completely pain-free at rest, at night and on exertion.

How to hold all this

A consistent pattern runs through the data: measured stiffness persists more often than patients complain about it. Reeves and Binder both found restriction with little functional complaint. Hand found 41 percent symptomatic but the vast majority of those mild.

So the honest version is three sentences. Most people end up functionally fine and get on with their lives. A substantial minority, somewhere between 40 and 60 percent depending on the study, keep some measurable loss of movement long term. And a small group, around 6 percent, are left with genuinely troublesome pain and disability.

Self-limiting is an overstatement. Usually manageable is closer.

Will it happen to the other shoulder?

Around 14 percent have both shoulders affected at once, and up to 20 percent develop it in the other shoulder at some point. Risk of it occurring in the other shoulder within five years is reported at 6 to 17 percent.

Recurrence in the same shoulder is rare. Hand’s study of 269 shoulders found no recurrent cases in a shoulder that had already had it.

One figure to be careful with: one review states sequential bilateral occurrence up to 50 percent, far above everything else. Treat that with skepticism.

What predicts a worse course

Severity at onset. Those with the worst symptoms at the start had the worst long-term outcome, at high statistical significance.
Longer symptom duration before treatment. An independent predictor of poor outcome.
Diabetes. A systematic review of 28 longitudinal studies found people with diabetes generally had worse outcomes. Every one of the 10 pain studies favored non-diabetics, and 15 of 18 multidimensional score studies did. Range of motion results were less consistent. Certainty of evidence was low to moderate, with 21 of the 28 studies at high risk of bias.
Longer duration of diabetes specifically.
Thyroid dysfunction.

What else can look like this

Neviaser’s own warning is worth repeating: every patient with a painful shoulder and apparent limitation of motion does not have adhesive capsulitis.

Things that must not be missed:

Glenohumeral arthritis, which shows on X-ray
Locked posterior dislocation, which can be missed for months and looks exactly like a stiff shoulder
Pancoast tumor, a lung cancer at the apex that presents as shoulder pain
Septic arthritis, particularly if immunosuppressed
Polymyalgia rheumatica
Early Parkinson’s disease
Complex regional pain syndrome, suspected if there are skin, temperature or sweating changes in the hand, particularly after a stroke or heart attack
Tumors of the shoulder girdle, which have been documented presenting as frozen shoulder in published case series

One useful distinguishing feature: stiffness from disuse or after an injury tends to improve over time, whereas frozen shoulder stiffness gradually increases during the early months.

When to get checked

See a doctor for shoulder pain and stiffness that persists, or pain severe enough to limit movement. The British Elbow and Shoulder Society suggests review if there is no progress at 6 to 12 weeks, or if pain worsens.

Seek attention sooner for: fever with a hot swollen shoulder, unremitting night pain in someone with a history of cancer, night sweats, weight loss, an unexplained lump, or any neurological change in the arm or hand.

What to do about it

Treatment is its own subject, and there is a large randomized trial that answers most of the important questions about it. That is covered in frozen shoulder treatment: what the evidence shows.

The short version: a steroid injection is one of the few shoulder treatments with genuinely good evidence behind it, physiotherapy intensity should be matched to how irritable your shoulder is rather than pushed as hard as possible, and the three main secondary care options produce results that are statistically different but not clinically different from each other.

Related conditions

Shoulder stiffness after surgery or injury · Rotator cuff tendinopathy · Rotator cuff tears · Shoulder impingement

This article is for general education and is not medical advice. If your shoulder is stiff and not improving, or if you have any of the warning signs above, see a doctor or physical therapist.