Most shoulder pain doesn't need early scans — what the evidence says

What just happened

A systematic review published in JAMA Internal Medicine in August 2026 has pulled together evidence from guidelines, randomised trials, and long-term cohort studies to reach a clear conclusion about shoulder pain: most of it gets better without imaging, and early scans are more likely to muddy the clinical picture than to clarify it.

Lead researcher Dr Romi Haas from Monash University and colleagues reviewed evidence across thousands of patients. Their central finding: around 95% of shoulder pain presentations in general practice do not require X-ray, ultrasound, or MRI. In one 10-year cohort study of 526 adults with new shoulder concerns, around 80% achieved satisfactory improvement with watchful waiting or NSAIDs alone.

Medical Republic and newsGP both covered the review this week, with RACGP’s Dr Joshua Hatch noting that current guidelines already support this approach: six weeks of rehabilitation first, then imaging only if symptoms haven’t responded.

Both-and

The scan problem is structural

Shoulder pain is the third most common musculoskeletal presentation in Australian general practice. It is also one of the most over-investigated. Imaging gets ordered early — often at the patient’s request, or out of habit, or because a structural label feels more reassuring than “let’s wait and see.”

The problem is anatomy. Nearly every adult over 40 has rotator cuff changes on MRI, whether or not they have shoulder pain. Subacromial thickening, partial tears, tendinopathic signal changes — these are the marks of a life lived in a body. When a scan picks them up in someone who also has pain, the temptation is to link the two. But a finding present in most pain-free adults is a poor explanation for pain in this particular person.

The structural label problem compounds this. Terms like “bursitis,” “tendinopathy,” and “rotator cuff tear” carry an implicit message — there is something broken that might need fixing. The review’s authors advocate moving toward the term “subacromial pain” instead, which acknowledges the symptom without implying that a structural lesion is the central driver, or that surgery is the next logical step.

Unnecessary imaging also carries direct costs. Around 95% of general practice shoulder pain presentations don’t require scans, yet unnecessary imaging consumes significant resources — and, more importantly, delays the appropriate first step: education, reassurance, and graduated loading.

When scans are the right call

This review does not argue against imaging in all circumstances. The clinical picture matters, and red flags exist for good reason.

Imaging is appropriate where there is significant trauma with possible fracture, signs of infection (septic arthritis presents with warmth, systemic upset, and disproportionate pain), concern about malignancy, or persistent symptoms after six weeks of appropriate management without the expected recovery trajectory.

The RACGP’s Dr Hatch was direct: “Patients with rest pain or disproportionate symptoms warrant referral.” Rest pain — pain that wakes you at night or is present at complete rest — is not a feature of typical subacromial pain. It warrants a closer look and possibly specialist assessment.

On treatment: the review found that corticosteroid injections may provide 4–8 weeks of meaningful relief for moderate-to-severe subacromial pain. Anatomy-guided injection was found to work as well as image-guided injection for this purpose — which further reduces the clinical argument for early imaging. And surgery for subacromial pain without full-thickness rotator cuff tears showed no meaningful benefit over non-operative management.

The scan-then-refer reflex

One practical pattern worth naming: early scans often lead to specialist referrals that might not otherwise occur. A GP orders an MRI, the report describes a rotator cuff tear (present in a large proportion of the over-40 population), the patient expects a surgical solution, a referral follows. None of this may be in the patient’s best interest. The review is a useful tool for the conversation that breaks that chain before it starts.

My two cents

There is real pressure in a general practice consultation to order a scan. Patients often arrive expecting one. “I just want to know what’s wrong” is a completely reasonable thing to want — and also something imaging may not be able to answer in any clinically useful way for most shoulder presentations.

What this review gives GPs is clearer grounding for the reassurance conversation: most shoulder pain gets better, the timeline is weeks to months, and a scan finding is unlikely to change the management plan or speed the recovery. Six weeks of appropriate rehabilitation is both the well-supported first step and a genuinely useful frame to offer a patient who is frustrated and wants action.

For someone with shoulder pain right now: the absence of a scan is not the absence of being taken seriously. General practice has well-developed clinical tools for shoulder assessment — range of motion, provocation tests, neurovascular examination — that can identify the small proportion of cases where imaging is genuinely needed, and appropriately manage the large proportion where it is not.

The broader principle — that a structural finding on a scan is not the same as finding the cause of pain — applies well beyond the shoulder. It is increasingly the consensus position in musculoskeletal medicine for low back pain, knee pain, and hip pain as well.

Verdict: yes — if your GP recommends rehabilitation rather than an early scan for shoulder pain, that is what the research supports.

---

Sources cited

1. ‘Little benefit’ in early scans for shoulder pain. newsGP, RACGP, 18 August 2026. https://www1.racgp.org.au/newsgp/clinical/little-benefit-in-early-scans-for-shoulder-pain
2. GPs, give imaging the cold shoulder. The Medical Republic, 18 August 2026. https://www.medicalrepublic.com.au/gps-give-imaging-the-cold-shoulder/128273