Frozen Shoulder

Frozen shoulder (adhesive capsulitis) is a condition in which the capsule surrounding the glenohumeral (shoulder) joint becomes inflamed, thickened, and progressively fibrotic, contracting around the joint and restricting both pain-free and structural range of motion.^[1] It affects roughly 2-5% of the general population, most often between the ages of 40 and 60 and more often in women than men.^[2]

Two broad categories are recognized. Primary (idiopathic) frozen shoulder has no identifiable precipitating event and accounts for most cases; it is closely and consistently associated with systemic conditions — most importantly diabetes mellitus, and also thyroid disease — suggesting an underlying inflammatory or fibroproliferative process rather than a purely mechanical one.^[1] Secondary frozen shoulder follows a specific precipitating event: prior shoulder trauma, shoulder or chest wall surgery, or a period of prolonged immobilization (for example, after a fracture, stroke, or cardiac surgery).^[1] Histologically, both forms show synovial hyperplasia, angiogenesis, and progressive capsular fibrosis driven by cytokines and growth factors that promote fibroblast proliferation and collagen deposition, with particular thickening of the coracohumeral ligament and rotator interval.^[1]

The diabetes association is the single most important comorbidity-driven feature of this condition for pharmacist practice: diabetic patients are roughly five times more likely than the general population to develop frozen shoulder, and the disease in this group tends to be more severe, more resistant to treatment, and slower to resolve.^[3] This chapter returns to that association repeatedly — in risk assessment, in the choice and monitoring of corticosteroid injection therapy, and in setting realistic recovery expectations with diabetic patients specifically.

The glenohumeral joint capsule (anterior view, distended) — the structure that becomes inflamed, thickened, and fibrotic in frozen shoulder, progressively restricting both active and passive range of motion. (Henry Vandyke Carter, illustrator, in Henry Gray, Anatomy of the Human Body, 20th ed., 1918; public domain, Wikimedia Commons)

Epidemiology

Mean age of onset is approximately 55 years, with women affected at roughly a 1.4:1 ratio compared with men; the nondominant shoulder is affected somewhat more often than the dominant side.^[1]

The three-phase clinical course

^[2][4]

Taken together, the full natural history of frozen shoulder — from first onset of pain through complete or near-complete recovery — commonly spans 1 to 3 years, a point that is critical to communicate explicitly and early: patients are frequently frustrated by how slowly the condition improves, particularly during the frozen phase when pain has eased but stiffness has not, and may otherwise lose confidence in a working treatment plan and abandon it prematurely.^[2][4] Recent evidence has also challenged the older assumption that frozen shoulder always resolves completely on its own: a systematic review found only low-quality evidence of some, but not complete, improvement in range of motion after one to four years of follow-up.^[2] With appropriate treatment, approximately 80% of patients regain near-normal or normal shoulder function, but 10-20% are left with some degree of residual stiffness or discomfort.^[1]

Typical presentation

Patients describe a gradual onset of diffuse, poorly localized, deep, dull ache around the shoulder that may radiate into the upper arm, intensifying over weeks to months and aggravated by movement — especially reaching overhead, reaching behind the back, and reaching across the body.^[1] On examination, a loss of the normal arm swing during gait is often noted, and in more advanced or longstanding cases, visible disuse atrophy of the shoulder girdle muscles may be present.^[1]

• Diabetes mellitus — the single most important risk factor for this condition. Diabetic patients are roughly five times more likely than nondiabetic controls to develop frozen shoulder, with an overall mean prevalence of frozen shoulder of about 13.4% among people with diabetes, and a mean prevalence of diabetes of about 30% (95% CI, 24-37%) among people presenting with frozen shoulder; the association does not differ significantly between type 1 and type 2 diabetes, or between patients treated with insulin and those on oral hypoglycemic agents.^[3] Beyond simply increasing risk, diabetes is also associated with a more severe, more treatment-resistant, and more prolonged disease course.^[3]
• Thyroid disease (hypothyroidism or hyperthyroidism) — significantly more prevalent among patients with frozen shoulder than in controls (27.2% vs. 10.7% in one series, P=.001).^[2]
• Age 40-60 years and female sex — the classic demographic profile, though the condition can occur outside this range.^[2][1]
• Prior shoulder immobilization, trauma, or surgery (secondary frozen shoulder) — including after a fracture, stroke, or cardiac or chest wall surgery.^[1]
• The nondominant shoulder is affected somewhat more often than the dominant side.^[1]
• Other reported systemic associations include Parkinson disease and cardiac disease.^[4]

History

Characterize the pain (onset, character, nocturnal worsening, radiation into the upper arm), the pattern and rate of motion loss, any preceding trauma, surgery, or period of immobilization, and screen for diabetes and thyroid disease specifically, since both substantially change the expected course and management.^[2] Ask specifically about functional limitation with reaching overhead, reaching behind the back, and reaching across the body — the three movements most affected.

Examination — the key distinguishing finding

The single most important examination finding in frozen shoulder is restriction of both active and passive range of motion, particularly external rotation with the arm at the side.^[2] This is the key feature that distinguishes frozen shoulder from rotator cuff tendinopathy or a rotator cuff tear, in which passive range of motion is typically preserved even when active motion is limited by pain or weakness — a clinician moving the arm passively can usually still achieve a fuller arc of motion in rotator cuff disease, but not in frozen shoulder.^[2] Motion is typically lost in a specific order: external rotation first, followed by abduction, internal rotation, and forward flexion.^[1] Additional findings include diffuse (not focal) tenderness to palpation, loss of the normal arm swing during gait, and possible disuse muscle atrophy in longstanding cases; distal neurological and vascular findings should be normal.^[1]

Diagnostic injection test

When the distinction from rotator cuff pathology remains uncertain, a subacromial injection of local anesthetic (for example, lidocaine 1%) can help: persistent restriction of passive motion despite adequate anesthesia of the subacromial space supports frozen shoulder (a true capsular restriction, not a pain-limited one), while improved passive motion after anesthesia points toward a pain-limited rotator cuff process instead.^[1]

Differential diagnosis

^[2]

Investigations

Frozen shoulder is a clinical diagnosis; laboratory testing is not needed to confirm it, but a fasting glucose or A1c and a TSH are reasonable in a patient without a known diagnosis of diabetes or thyroid disease, given how strongly both are associated with frozen shoulder.^[2]

Imaging is generally not required for a typical presentation.^[2] Plain radiographs are useful to exclude glenohumeral osteoarthritis, a pathologic fracture, avascular necrosis, or calcific rotator cuff tendinopathy, but are less helpful for the capsule itself, since soft tissue is not well visualized on x-ray.^[2] MRI can show supportive findings — thickening of the coracohumeral ligament, rotator interval synovitis or fat infiltration, and a reduced axillary recess volume — but these findings are supportive rather than definitive, and MRI should be reserved for atypical presentations or to evaluate for another cause of shoulder pathology rather than used routinely.^[2][1] This mirrors the broader Choosing Wisely Canada recommendation against ordering advanced imaging as an initial investigation for shoulder pain generally — start with a clinical diagnosis and conservative management, and reserve imaging for a presentation that fails to fit the expected pattern or fails to improve as expected.^[5]

• Control pain, particularly during the freezing phase when it is typically most severe.
• Preserve and progressively restore range of motion and shoulder function.
• Prevent prolonged disability and minimize the functional impact of the frozen phase.
• Set realistic, explicit expectations about the natural history (commonly 1-3 years) early, to preserve the patient's engagement with an appropriate but genuinely slow-acting treatment plan.
• In diabetic patients specifically, anticipate a more severe and prolonged course, and manage corticosteroid injection therapy with appropriate blood glucose counselling and monitoring.
• Avoid unnecessary imaging, and avoid escalating to invasive procedures before an adequate conservative trial.
• Identify genuine treatment failure (refractory disease) and refer appropriately for hydrodilatation, manipulation under anesthesia, or arthroscopic capsular release, rather than persisting indefinitely with an ineffective conservative plan.

Structured physiotherapy and range-of-motion exercise

A supervised or guided program of stretching and range-of-motion exercise is the mainstay of frozen shoulder treatment across all three phases, though its emphasis shifts with the phase: gentle, pain-limited range-of-motion and pendulum exercises during the painful freezing phase, to avoid provoking a flare, giving way to more assertive, progressive stretching to regain motion once the frozen phase's pain has plateaued.^[2] Manual mobilization techniques combined with a home exercise program have fair evidence for improving range of motion and function.^[2] Common home exercises include pendulum swings, finger-walk/wall-climbing stretches, cross-body (towel) stretches, and external-rotation doorway stretches; a physical or occupational therapist should individualize the specific program and its progression to the patient's phase and tolerance.

Patient education — the natural history is the single most important teaching point

Explain the three-phase course and the realistic overall timeline (commonly 1-3 years from onset to full or near-full recovery) explicitly and early.^[2][4] Patients who are not told this in advance often become discouraged partway through the frozen phase, when pain has improved but stiffness has not, and may stop an effective treatment plan believing it has failed — framing the frozen phase's persistent stiffness as an expected part of a genuinely slow but generally favourable course is itself a therapeutic intervention.

Activity modification and heat

Modify, rather than completely avoid, painful overhead or behind-the-back activities during the freezing phase; complete immobilization is not recommended and can worsen stiffness. Heat before stretching, and ice after activity, are reasonable, low-risk adjuncts that many patients find helpful for comfort, though dedicated evidence in frozen shoulder specifically is limited.

Diabetes and thyroid optimization

Because diabetes is so strongly linked to both the development and the severity/duration of frozen shoulder, reinforcing engagement with the patient's existing diabetes management plan is a genuinely relevant adjunct, not a tangential point — better glycemic control does not reverse capsular fibrosis that has already developed, but the association between diabetes and a more severe, prolonged, treatment-resistant course is well established.^[3] Correcting a previously undiagnosed or undertreated hypothyroidism is similarly reasonable where relevant.

Acupuncture

Limited evidence suggests acupuncture combined with physiotherapy may produce faster short-term pain relief than physiotherapy alone; it is a reasonable adjunct where available and acceptable to the patient, not a substitute for the exercise program.^[2]

Diabetes — the central comorbidity for this condition

Diabetic patients are roughly five times more likely to develop frozen shoulder, and their disease course tends to be more severe, more treatment-resistant, and slower to resolve.^[3] Consider fasting glucose or A1c screening in any patient presenting with frozen shoulder who does not already have a known diabetes diagnosis.^[2] Corticosteroid injection remains a reasonable and effective option in diabetic patients, but counsel closer blood glucose self-monitoring for the first 2-3 days after injection, and advise contacting the prescriber or diabetes care team if readings are significantly elevated.^[6]

Pregnancy

Favor non-pharmacologic measures — physiotherapy, activity modification, and heat — throughout pregnancy. Preferred analgesic acetaminophen. Avoid NSAIDs from approximately 20 weeks of pregnancy onward Health Canada has confirmed a risk of fetal renal impairment and low amniotic fluid (oligohydramnios) with NSAID use from about 20 weeks, and NSAIDs are contraindicated in the third trimester (beyond 28 weeks) for this reason.^[9] Pregnancy-specific safety data for intra-articular corticosteroid injection are limited; reserve for genuinely disabling symptoms that have not responded to conservative measures, using the lowest effective dose.

Breastfeeding

Acetaminophen and ibuprofen are considered compatible with breastfeeding. A single local corticosteroid injection at standard shoulder-injection doses is not expected to meaningfully affect a breastfed infant.

Pediatrics

Frozen shoulder is rare in children and adolescents; new shoulder stiffness in a child warrants a broader differential — including juvenile idiopathic arthritis, infection, and tumour — rather than an assumption of frozen shoulder, and should prompt referral.

Older adults

Frozen shoulder can occur in older adults, often together with degenerative rotator cuff disease or glenohumeral osteoarthritis — confirm the specific pattern of restricted passive range of motion before attributing shoulder stiffness to frozen shoulder rather than osteoarthritis. Favor acetaminophen or the lowest effective NSAID dose for the shortest duration where systemic NSAID risk (gastrointestinal, renal, cardiovascular) is a concern.

Renal and hepatic impairment

Avoid or dose-reduce NSAIDs in significant renal impairment. Reduce the acetaminophen total daily dose in significant hepatic impairment. No specific renal or hepatic dose adjustment is needed for a single local corticosteroid injection at standard doses.

Arrange follow-up at intervals matched to the phase-based natural history rather than expecting rapid change — reassess pain and range of motion roughly every 4-6 weeks initially, extending the interval once the trajectory (whether improving, plateaued, or refractory) becomes clear.

After a corticosteroid injection in a diabetic patient specifically, advise checking blood glucose more often than usual for the first 2-3 days post-injection, and to contact their prescriber or diabetes care team for significantly elevated readings.^[6]

Track functional milestones — dressing independently, reaching overhead, reaching behind the back — alongside a simple pain score, since motion and function, not pain alone, define a meaningful response; explicitly revisit the expected phase and timeline at each visit to reinforce realistic expectations.

Distinguish the expected, slow-but-still-progressing course typical of the frozen phase from genuine treatment failure (see Treatment Failure/Next Step) before escalating therapy.

1. Re-confirm the diagnosis: re-examine specifically for the active-versus-passive range-of-motion pattern and screen for the red flags and differential-diagnosis clues above before assuming treatment failure.
2. Confirm an adequate conservative trial actually occurred — structured physiotherapy genuinely attempted, an NSAID trial completed, and at least one corticosteroid injection given if not already tried, especially during the freezing phase.
3. If pain remains the dominant limiting symptom despite physiotherapy and an NSAID, and no corticosteroid injection has yet been given, offer one.
4. If pain and stiffness remain inadequately controlled after one corticosteroid injection and a genuine conservative trial (commonly 3-6 months), consider a second injection or move to hydrodilatation.
5. For disease that remains refractory despite conservative measures and injection therapy for 6-12 months, refer for manipulation under anesthesia or arthroscopic capsular release, presenting both as reasonable options with different risk/benefit tradeoffs rather than a single clearly superior choice.
6. In a diabetic patient with a slower or more resistant course than typical, do not assume treatment failure prematurely — the expectation should already have been set that diabetic frozen shoulder often takes longer and may respond less completely to a single injection.
7. Persistent shoulder pain or stiffness that does not fit the expected phase-based pattern at any point should prompt reconsideration of the differential (rotator cuff tear, glenohumeral osteoarthritis, cervical radiculopathy) rather than continued frozen-shoulder-specific escalation.

• Ordering an MRI or other advanced imaging as an initial investigation rather than making the diagnosis clinically and reserving imaging for atypical presentations or failure to improve as expected.^[5]
• Assessing only active range of motion — missing the key finding (restricted passive range of motion, especially external rotation) that distinguishes frozen shoulder from rotator cuff pathology.^[2]
• Attributing new shoulder stiffness that follows acute trauma to frozen shoulder without first considering an acute rotator cuff tear.
• Not screening an undiagnosed patient for diabetes and thyroid disease, given how strongly both are associated with frozen shoulder.
• Injecting corticosteroid into a diabetic patient's shoulder without counselling on the expected transient rise in blood glucose and the need for closer monitoring in the days after injection.^[6]
• Understating the realistic 1-3 year natural history — this undermines adherence once the frozen phase's persistent stiffness sets in despite improving pain.
• Escalating directly to manipulation under anesthesia or surgical capsular release before a genuine trial of physiotherapy and at least one corticosteroid injection.
• Injecting corticosteroid in the presence of active local or systemic infection.
• Treating hydrodilatation as a clearly superior alternative to corticosteroid injection rather than an option with a modest, mostly short-term, and still-uncertain long-term benefit.^[7]
• Assuming complete immobilization of the shoulder is protective — it typically worsens stiffness; activity modification, not complete rest, is the right approach.

How to use this table: doses are adult doses and require patient-specific confirmation, including renal/hepatic adjustment. Select any drug name to open its complete Pharmasist Drug Database page.

NSAIDs and Analgesics

Doses below are adult doses and require patient-specific confirmation, including renal/hepatic adjustment. Evidence specific to frozen shoulder is limited; use the lowest effective dose for the shortest duration needed for comfort.

Intra-Articular Glenohumeral Corticosteroid Injection

Practitioner-administered. The pharmacologic intervention with the strongest evidence in frozen shoulder, especially given early in the freezing phase and combined with physiotherapy. In diabetic patients, counsel closer blood glucose self-monitoring for the first 2-3 days after injection given the risk of a transient, usually short-lived, rise in blood glucose.

Refractory Disease — Hydrodilatation Regimen

Reserve for disease that remains refractory after an adequate conservative trial (physiotherapy, NSAID, and at least one corticosteroid injection), typically 3-6 months. Image-guided, practitioner-administered.