Frozen shoulder: why it mostly strikes women aged 40 to 60

It starts with shoulder pain that gets blamed on an awkward movement. Then the stiffness sets in, and one day you can no longer fasten your bra or reach your seatbelt. Adhesive capsulitis (the "frozen shoulder") affects women around fifty in large numbers, and hormones were the obvious suspect.

This article says what is established, what is not, and why the delay before seeking care matters. It also avoids serving you the unverifiable figures that circulate everywhere on this subject.

What is really happening in the shoulder

Adhesive capsulitis is neither wear nor simple inflammation. As set out by the reference published in 2022 in Nature Reviews Disease Primers, it is a fibroproliferative fibrosis: fibroblasts producing mainly type I and type III collagen turn into myofibroblasts, with inflammation, formation of new vessels and of new nerve endings^[1]. The result: the envelope of the joint, the capsule, thickens and retracts.

It is this mechanism that explains the decisive clinical sign. In tendinitis, it can hurt to raise your arm, but someone can raise it for you passively. In capsulitis, the shoulder is blocked in every direction, including when someone else moves it. It is not pain that limits it: it is mechanics.

The diagnosis rests largely on clinical examination. The course is long: the authors speak of a "prolonged clinical course". The classic description in three phases (painful, stiffening, recovery) is useful for understanding, but we will not attach a figure in duration to it: none of the sources verified for this article quantifies these phases, and we prefer to advance nothing we cannot support.

The link with menopause: a hypothesis, not a fact

This is the most important point in this article, and it probably goes against what you have read elsewhere.

The idea that menopause promotes frozen shoulder circulates widely. What does the only recent study to have tested it directly say? Published in 2026 in Climacteric, it is a pilot study (its title indicates that it aims to resolve design issues for future research) covering 152 postmenopausal women aged 40 to 60 on hormone therapy and 1,800 without^[6].

Capsulitis was less frequent in the treated women: 3.95% versus 7.65%. But this difference was not statistically significant: odds ratio 1.99, 95% confidence interval from 0.86 to 4.58, p = 0.10. In other words, the gap observed is compatible with chance. The authors say so unambiguously: their study does not demonstrate a difference, and larger prospective work is needed.

So we have to be clear. You cannot write that "hormone therapy protects against frozen shoulder", nor that "the drop in estrogen explains capsulitis". The overrepresentation of women aged 40 to 60 is a consistent clinical observation, but it has never been explained.

On frequency, the same requirement applies. The figure of "2 to 5% of the population" that you read everywhere could not be traced to a verifiable source, and we will therefore not publish it. The only usable reference point comes from the introduction of the large British UK FROST trial, whose authors put forward that capsulitis "affects about 10% of people in their fifties and is slightly more common in women"^[8], an order of magnitude for context, not a study result.

The real risk factors: diabetes and thyroid

Here, on the other hand, the data are solid, and far more useful in practice than the hormonal hypothesis.

Diabetes is by far the best documented factor. A meta-analysis published in BMJ Open in 2023 estimates that people with diabetes have about 3.7 times more risk of developing capsulitis in case-control studies (odds ratio 3.69; 95% confidence interval: 2.99 to 4.56)^[3]. The two available cohort studies, methodologically more robust for establishing a sequence in time, find a more modest excess risk: on the order of 30 to 70%. To be flagged honestly: seven of the eight studies included had a high risk of bias.

A Korean cohort covering nearly 3.5 million people followed for up to eight years refines the picture^[4]: the incidence rate goes from about 9.5 cases per 1,000 person-years in people with normal blood glucose to 11.9 in prediabetes, 14.9 in newly diagnosed diabetes and 24.4 in people treated for type 2 diabetes. The excess risk linked to prediabetes alone remains small, however: its significance owes much to the gigantic size of the sample.

The thyroid next. A meta-analysis of ten case-control studies totalling nearly 128,000 people finds an association between capsulitis and thyroid disease (odds ratio 1.87; 95% confidence interval: 1.37 to 2.57), driven essentially by hypothyroidism, including in its subclinical form (2.56; 1.81 to 3.63)^[5]. On the other hand, no association has been demonstrated with hyperthyroidism.

These associations do not prove causality: people followed for diabetes or a thyroid condition are also investigated medically more often. But they justify a simple question to put to your doctor: have my blood sugar and my thyroid been checked recently?

Broader epidemiological data, from the American Medicare insurance records, confirms the association with diabetes and finds an annual prevalence of about 0.35% in people aged 65 and over^[2]. That study does not cover the 40 to 60 age range that concerns us here, and its authors conclude that the trigger of this so-called idiopathic condition remains unknown.

What works, in order of evidence

The most complete synthesis available brings together 65 randomized trials and more than 4,000 participants^[7]. Only one treatment comes out of it with a superiority that is both statistical and clinically relevant: corticosteroid injection into the joint, in the short term, about one point less on a pain scale from 0 to 10 compared with no treatment or with physiotherapy alone.

Two operational details follow from this. First, this injection gives better results when it is combined with a programme of simple exercises and stretches to do at home: the additional benefit holds up in the medium term. Second, the authors conclude that it is more effective when it is offered early, within the first year of the course. Waiting for it "to pass on its own" is therefore not the best strategy.

The limits have to be stated too: the demonstrated benefit covers twelve weeks at most, and one point out of ten is real but modest. This is not a cure.

And surgery? The British UK FROST trial randomly allocated 503 adults (mean age 54, 63% women, 30% with diabetes) between three approaches: early structured physiotherapy with injection, manipulation under general anaesthesia, or arthroscopic capsular release^[8]. At one year, the gaps observed (1 to 3 points on the Oxford scale) all remained below the threshold of clinical relevance set in advance by the researchers. The authors' conclusion: none of the three interventions was clearly superior. And serious complications, although rare, occurred in the surgical arms.

Hydrodilatation, which consists of injecting fluid to distend the capsule, brings, according to a 2023 review, an at least transient improvement in function and external rotation greater than that of an injection alone, and could constitute a less expensive and more comfortable alternative to manipulation under anaesthesia^[9]. The authors remain cautious: the real clinical scope of this benefit is not established, and whether it lasts is debated.

The British Elbow and Shoulder Society updated its recommendations in 2025 using the GRADE methodology^[10]. One detail says a lot about the state of knowledge: for several questions, in the absence of trials of sufficient quality, the working group had to fall back on narrative syntheses.

What is not proven

None of the sources verified for this article supports the use, in adhesive capsulitis, of dietary supplements, oral collagen, long-term anti-inflammatories, spinal manipulation, acupuncture, shockwave therapy, platelet-rich plasma, consumer electrostimulation devices, or taking hormone therapy with the aim of treating or preventing a frozen shoulder.

A final word on forced stretching. The only thing demonstrated is the benefit of simple exercises at home alongside an injection. Nothing, in the data examined here, validates aggressive mobilization by another person during the painful phase.

When to see a doctor

Without waiting several months if your shoulder becomes painful and then progressively stiff, and above all if the stiffness affects every movement, including when someone else tries to move your arm. It is this point that tells capsulitis apart from tendinitis. The delay matters: injection gives better results within the first year of the course.

Ask for a check of your blood sugar and your thyroid if that has not been done recently. This decision rests with your doctor, but the question is worth asking.

Seek care promptly, without waiting, if the shoulder pain follows a fall or an impact, if you cannot raise your arm actively after an injury, in case of visible deformity of the shoulder, fever, redness or warmth of the joint, loss of strength, numbness or pins and needles in the arm or hand, unexplained weight loss, night sweats or a history of cancer, or intense night pain that worsens rapidly. These situations do not match the picture of a simple capsulitis.

Frequently asked questions

How do you tell capsulitis from tendinitis?

The decisive sign is passive stiffness. In tendinitis, it can hurt to raise your arm, but someone can raise it for you. In adhesive capsulitis, the shoulder is blocked in every direction, including when another person moves it: it is not pain that limits the movement, it is the retraction of the joint capsule. The diagnosis rests largely on clinical examination.

Does menopause cause frozen shoulder?

It is not demonstrated. The only recent study to have tested it directly is a pilot study published in 2026, covering 152 women on hormone therapy and 1,800 without. Capsulitis was less frequent in the treated women (3.95% versus 7.65%), but the difference was not statistically significant (odds ratio 1.99; 95% CI 0.86-4.58; p = 0.10). The authors themselves conclude that their study does not demonstrate a difference. The overrepresentation of women aged 40 to 60 is a consistent clinical observation, but it remains unexplained.

What are the real risk factors?

Diabetes is the best documented: about 3.7 times more risk in case-control studies, and an excess of 30 to 70% in the cohorts, which are methodologically more robust. A Korean cohort of nearly 3.5 million people shows risk rising with the degree of glucose abnormality, from prediabetes to treated diabetes. Hypothyroidism, including subclinical, is also associated; hyperthyroidism is not. These are associations, not demonstrated causal links.

Should a frozen shoulder be operated on?

Not as a first-line approach. In the UK FROST randomized trial, which compared early structured physiotherapy with injection, manipulation under anaesthesia and arthroscopic surgical release in 503 adults, none of the three came out clearly superior at one year: the gaps observed remained below the threshold of clinical relevance set in advance. Serious complications, although rare, occurred in the surgical arms. Surgery is discussed with a specialist if a well conducted approach has failed.

Which treatment has the most evidence?

Corticosteroid injection into the joint is the only intervention to come out with a superiority that is both statistical and clinically relevant in a synthesis of 65 randomized trials, but in the short term only (twelve weeks at most), with a gain of about one point on a pain scale from 0 to 10. It gives better results combined with simple exercises at home, and offered early, within the first year of the course.

⚠️ This article is provided for informational and educational purposes only. It does not replace medical advice, diagnosis or treatment. Always consult your doctor or a qualified health professional with any questions about your health.

How we work: our editorial method · Written and verified by Bouchra, editorial lead.

How we work: our editorial method · Written and verified by Bouchra, editorial lead.