Shoulder pain referred from cardiac or abdominal sources accounts for a significant proportion of missed diagnoses in primary care. The phrenic nerve (C3–5) shares dermatomal referral with the shoulder, meaning any sub-diaphragmatic irritant (haemorrhage, abscess, ectopic pregnancy) can present as shoulder tip pain. Pancoast tumours are particularly important — they arise at the lung apex, are not visible on standard CXR views without apical cuts, and cause a characteristic triad of arm pain, Horner's syndrome, and hand weakness. NICE NG12 mandates 2WW referral for any unexplained shoulder or arm pain with features suggestive of malignancy. Septic arthritis of the glenohumeral joint has a mortality of up to 15% and must be aspirated urgently.

The movement restriction pattern is the single most useful diagnostic discriminator. True frozen shoulder (adhesive capsulitis) causes global restriction of both active AND passive movements — passive external rotation is the first and most severely restricted. In rotator cuff tears, passive movement is preserved but active is restricted. Painful arc at 60–120° abduction is pathognomonic of subacromial impingement — the greater tuberosity impinges on the coracoacromial arch in this range. Diabetes is the strongest risk factor for frozen shoulder; prevalence is 10–20% in people with diabetes (vs 2–5% general population). Night pain is important — it should prompt active consideration of malignancy and RC tear, both requiring imaging.

Accurate classification prevents inappropriate treatment and imaging. Subacromial impingement is the most common diagnosis but is frequently over-diagnosed — cervical referred pain is commonly mistaken for shoulder pathology. The key differentiator is passive shoulder range: if passive movement is full and painless, the shoulder joint itself is not the source. Frozen shoulder is frequently misdiagnosed and overtreated with cortisone without physiotherapy — the condition is self-limiting but runs a predictable 18–24 month course, and treatment decisions should account for this. Full-thickness rotator cuff tears require MRI before any decision about surgical repair — ultrasound is a reasonable first investigation but has operator dependency. ACJ pathology is identified by the scarf test and point tenderness at the joint line, not generalized shoulder pain.

The passive vs active range distinction is the single most important examination finding for shoulder diagnosis. Neck examination is non-negotiable — up to 20% of apparent shoulder pain presentations originate from the cervical spine (C5/C6 radiculopathy particularly). The drop arm sign (inability to slowly lower the arm from 90° abduction) has high specificity for full-thickness supraspinatus tear. The Hawkins-Kennedy test has the best sensitivity for subacromial impingement (~79%) but poor specificity — use in combination with Neer. Infraspinatus wasting in the posterior shoulder fossa is highly significant and indicates either RC tear or suprascapular nerve entrapment — do not miss this. Axillary nerve function must be assessed after every shoulder dislocation before and after reduction.

Over-investigation of shoulder pain is a major NHS resource issue — most subacromial impingement and frozen shoulder presentations require no imaging and should proceed directly to physiotherapy. However, ultrasound is appropriate and cost-effective when a rotator cuff tear is clinically suspected (weak empty can, drop arm sign, age >60 with acute-on-chronic pain). Ultrasound has excellent specificity for full-thickness tears. MRI is more sensitive for partial tears but should generally be requested by orthopaedics when surgical planning is needed. CXR with apical views is critical when Pancoast tumour is in the differential — standard PA CXR misses ~30% of apical lesions. ESR in PMR is typically markedly elevated (>40, often >80) and combined with dramatic prednisolone response is diagnostic.

The vast majority of shoulder pain is safely managed in primary care — subacromial impingement, mild RC tendinopathy, and frozen shoulder should all receive a trial of physiotherapy before referral. However, confirmed full-thickness RC tears require orthopaedic review — surgical repair is more successful when performed within 6 months of tear, as fatty infiltration of the muscle belly becomes irreversible. PMR should be initiated with prednisolone 15mg OD in primary care (NICE NG167) without waiting for rheumatology — the response is both therapeutic and diagnostic. Recurrent dislocation requires Bankart repair assessment — each dislocation risks further labral damage and Hill-Sachs lesion enlargement, increasing instability.

Subacromial corticosteroid injection is the most effective short-term intervention for impingement syndrome (NNT ~3 for short-term pain relief) but effect is not durable beyond 6 weeks without physiotherapy. NICE evidence synthesis shows exercise therapy has equivalent outcomes to injection at 12 months — injections are adjuncts that facilitate engagement with rehabilitation. Topical NSAIDs are recommended first-line for localised musculoskeletal pain before systemic NSAIDs (lower GI/renal/CV risk). Hydrodilatation for frozen shoulder (arthrographic distension) has NICE guidance support and is more effective than injection alone in the frozen phase. Prednisolone for PMR should be started without delay — the ESR/CRP are supportive but not diagnostic, and treatment should not be withheld pending rheumatology in a patient with typical features.

The landmark CSAW trial (2017, BMJ) demonstrated that arthroscopic subacromial decompression surgery was no better than structured exercise physiotherapy at 12 months for subacromial impingement — exercise is the reference standard treatment. Postural retraining is particularly important in desk workers who develop impingement from prolonged forward-head posture and rounded shoulders, which mechanically reduce the subacromial space. Sleep disruption from shoulder pain is a significant contributor to quality of life impairment and should be addressed — simple positional advice (pillow support) makes a measurable difference. Glycaemic control in diabetes is one of the few modifiable risk factors for frozen shoulder and is an important co-management target.

The 6-week review is critical — failure to improve at 6 weeks after a corticosteroid injection and physiotherapy should prompt diagnostic re-evaluation and imaging, not just another injection. Injection without physiotherapy has poor long-term outcomes — the exercise programme must be reinforced at every review. PMR relapse occurs in 40–50% of patients during steroid taper and presents as return of bilateral girdle stiffness and rising ESR/CRP — patients should know to return if this happens. Safety-netting for GCA is particularly important in PMR patients — the estimated risk of GCA developing in a PMR patient is 10–15%. Visual symptoms in a PMR patient are a same-day emergency (same-day prednisolone 60mg and ophthalmology referral). Failure to improve after adequate conservative management should always prompt re-evaluation for missed diagnosis, particularly malignancy.