Shoulder Impingement: What the Evidence Says About Diagnosis, Exercise, Injections and Surgery

Key Takeaways

• “Shoulder impingement” is a widely used label for pain around the acromion, but both the idea that bone mechanically pinches the tendon and the term itself are increasingly questioned.
• Classic clinical tests such as Neer and Hawkins-Kennedy have limited diagnostic value on their own, and rotator cuff changes on imaging are common in people without any shoulder pain.
• In placebo-controlled trials, subacromial decompression surgery did not produce clinically important benefits over sham surgery, and five-year results showed no advantage over exercise therapy.
• Exercise and corticosteroid injections produced similar results in a recent meta-analysis, and ultrasound guidance has not clearly outperformed landmark-guided injection.
• For overhead athletes, the exercise evidence is thin and dominated by expert opinion.
• In my clinical experience, most shoulder pain that starts without an injury fits the shoulder impingement picture, active physiotherapy plays the central role, and the response to cortisone injections varies from months of relief to pain that returns.

Introduction: Why Shoulder Impingement Needs a Rethink

Shoulder impingement is one of the most familiar diagnoses in musculoskeletal medicine. The explanation sounds intuitive: when you lift your arm, the tendons of the rotator cuff get squeezed under the bony roof of the shoulder (the acromion), become irritated and hurt.

That explanation is the rationale for subacromial decompression, an operation that removes bone from the underside of the acromion. Yet when that surgery was tested against a placebo operation in a large randomized trial, removing bone added no significant benefit over simply looking inside the joint [1].

This article covers what shoulder impingement actually means, why the diagnosis is being questioned, how reliable the usual tests and scans are, and what research shows for exercise, injections and surgery.

Shoulder impingement is a very common reason for visits in my practice. In my clinical experience, the large majority of shoulder pain that begins without an injury fits this picture. A smaller group of patients has something less common, such as a frozen shoulder, and some simply have a stiff neck and shoulder region. Cervical radiculopathy is another differential diagnosis, but in practice it is usually easier to tell apart from shoulder impingement, and in my view the two are rarely confused.

What Shoulder Impingement Means, and Why the Term Is Contested

“Shoulder impingement” or “subacromial impingement syndrome” is a label commonly used for pain localised around the acromion that is aggravated during or after lifting the arm. Alternative labels used in current research include subacromial pain syndrome [2] and rotator cuff related shoulder pain [3].

A 2026 narrative review in the Journal of Clinical Medicine examined whether the impingement concept should be dropped altogether [4]. In the authors’ interpretation of current evidence, narrowing of the space under the acromion is more likely a consequence of rotator cuff dysfunction than its cause, with that dysfunction thought to arise mainly from degenerative rather than compressive processes [4]. They also argue that a purely mechanical explanation may lead patients to believe that raising the arm is harmful, with potential nocebo effects, and may push clinicians towards invasive treatment even though guidelines recommend conservative care first [4].

The practical point is that the label matters less than what it implies. Telling someone their tendon is being “pinched by bone” may frame everyday movement as damaging, which is the concern the review authors raise.

Symptoms can appear in very different people. Manual workers such as installers who work with their arms at ceiling height may provoke them, but I also see the same symptoms in office workers at times. Even in what looks like a resting position at the computer, the shoulders may be held in slight abduction, which in my view can create a small static load around the supraspinatus. In these patients, I have seen the shoulder become painful without any injury, or after a minor jerk, with shoulder impingement symptoms following. In my clinical experience, a separate group develops symptoms after trauma, when an injured tendon has swollen or frayed.

Shoulder pain has other causes that need to be considered, including frozen shoulder, which we cover in detail in our article on frozen shoulder.

Diagnosis: Why Tests and Scans Can Mislead

Clinical tests

The Neer and Hawkins-Kennedy tests are the classic shoulder impingement tests. In a meta-analysis of shoulder examination tests in the British Journal of Sports Medicine, the pooled sensitivity and specificity were 72% and 60% for the Neer test and 79% and 59% for the Hawkins-Kennedy test [5]. The authors concluded that the impingement tests had limited diagnostic value and that no single shoulder test can be recommended as a definitive diagnostic tool [5].

In plain terms, a single positive test cannot confirm shoulder impingement on its own.

In my clinical experience, the examination findings often follow a recognisable pattern, although, as the research above shows, no single finding confirms the diagnosis. Abduction of the shoulder is painful and restricted, and sometimes flexion is too. Rotation is often tender as well, more often external than internal rotation, depending on the presentation. I tend to see two kinds of patients: some are in so much pain that they cannot abduct the arm at all, while others start to feel pain around shoulder height, and in my experience some of them find that the pain eases again once the arm is raised well above horizontal. On palpation, there may be tenderness below the front of the acromion.

Imaging

Imaging findings are just as difficult to interpret. A 2025 systematic review of rotator cuff imaging abnormalities in people without shoulder pain found that tendinopathy and/or partial-thickness tears were seen in 34% of a population-based ultrasound sample and in 65% of a small population-based MRI sample (20 shoulders) [6]. Full-thickness tears were found in 11% to 17% of two population-based ultrasound samples [6]. Across all included studies, prevalence ranged from 0% to 100%, and the certainty of the evidence was low to very low [6].

This means a scan showing tendon changes does not, by itself, prove that those changes are causing the pain.

Surgery for Shoulder Impingement: What Placebo-Controlled Trials Show

Arthroscopic subacromial decompression removes bone and soft tissue from under the acromion to “make more room” for the tendon.

In the UK CSAW trial, 313 patients were randomized to subacromial decompression, arthroscopy only (an investigation-only placebo operation) or no treatment [1]. At six months, decompression did not differ significantly from arthroscopy only on the Oxford Shoulder Score (difference −1.3 points; 95% CI −3.9 to 1.3) [1]. Both surgical groups scored better than no treatment, but the authors judged the differences not clinically important and suggested that the difference might reflect, for instance, a placebo effect or postoperative physiotherapy [1].

The Finnish FIMPACT trial randomized 210 patients aged 35–65 years with shoulder impingement symptoms lasting over three months to decompression, diagnostic arthroscopy or exercise therapy [7]. At five years, decompression gave no benefit over either diagnostic arthroscopy or exercise therapy for pain at rest or on arm activity, and no between-group difference in secondary outcomes exceeded the minimally important difference [7]. In a secondary analysis, return to work by 24 months was 88% after decompression, 88% after diagnostic arthroscopy and 90% after exercise therapy [8].

A Cochrane review pooling the placebo-controlled trials (506 participants) found high-certainty evidence that, at one year, decompression improved pain by only 0.26 points on a 0–10 scale and function by 2.8 points on a 0–100 scale compared with placebo surgery, neither of which is clinically important [9]. The authors concluded that the data do not support subacromial decompression for painful shoulder impingement [9].

These findings concern isolated decompression for shoulder impingement. They do not address surgery for other problems, such as some traumatic rotator cuff tears.

Exercise Therapy

International guidelines, as summarized in a 2026 review, recommend conservative care first [4], and in the surgical trials above decompression did not produce better outcomes than an exercise program over five years [7].

The research on specific exercise types is less precise. A 2024 meta-analysis of eight randomized trials with 387 participants found that scapular stabilization exercise interventions reduced pain more than the comparison interventions (VAS weighted mean difference −0.81; 95% CI −1.11 to −0.51 in the main analysis, −0.94 in a sensitivity analysis) and improved the Shoulder Pain and Disability Index (weighted mean difference −10.10; 95% CI −18.87 to −1.33), but did not significantly improve range of motion [2]. The certainty of the evidence was rated low for pain and very low for function [2].

We discuss the broader evidence on static loading in our article on isometric exercise for pain, and on progressive tendon loading in eccentric training for tendinopathy.

In my clinical experience, active physiotherapy plays the central role, and nearly all of my patients are referred to it, some alongside a cortisone injection or an anti-inflammatory painkiller. Many of my patients seem to benefit from isometric and eccentric exercises aimed at the tendinopathy. Some, however, are left with long-term pain. In prolonged cases I usually consult an orthopaedic surgeon. Orthopaedists do sometimes operate on shoulders, although conservative treatment is often a good option as well.

Corticosteroid Injections

A 2023 meta-analysis of eight randomized trials with 946 participants compared exercise and manual therapy with corticosteroid injections for subacromial pain syndrome [10]. Shoulder disability did not differ between the two approaches at very short-, short-, mid- or long-term follow-up, and pain did not differ at the very short-, short- or long-term follow-ups assessed, with certainty of evidence ranging from very low to moderate [10]. Adding a corticosteroid injection to exercise and manual therapy did not improve shoulder disability at mid-term follow-up compared with either treatment alone [10]. Citing fewer adverse reactions, the authors suggested that exercise and manual therapy be prioritized [10].

Whether an injection should be guided by ultrasound is a common question. A systematic review of four randomized trials with 179 participants found that both ultrasound-guided and landmark-guided subacromial injections improved outcomes, with no significant differences between them; three low risk-of-bias trials did not favour ultrasound guidance [3]. The trials were small and lacked long-term follow-up [3].

Patients come in with different expectations, partly depending on how long the symptoms have lasted. Many want sick leave, which I often write if their job requires use of the shoulder; others want an anti-inflammatory painkiller, and some ask for a cortisone injection. When the shoulder is being scanned anyway, an ultrasound-guided injection given by the radiologist at the same visit is often a convenient option, but I also give cortisone injections at my own clinic, and in my view both approaches are acceptable. This is in line with the small trials summarised above, which did not show a clear difference between them.

Much also depends on what the scan shows. In my experience, radiologists often do not inject when the finding is supraspinatus tendinopathy alone, whereas an inflamed bursa is more often injected. I have seen some patients get a clear benefit within a couple of days and have less pain for several months, while in others the pain returns.

Passive treatments such as massage and manipulation are discussed more broadly in our article on manual therapy for athletes.

Shoulder Impingement in Athletes

Overhead sports such as swimming, volleyball, tennis and throwing sports place repeated demands on the shoulder, but the evidence on how best to rehabilitate these athletes is limited.

A systematic review of exercise prescription for overhead athletes with shoulder pathology found that the evidence was dominated by expert opinion and that the overall level of evidence was low [11]. The strongest available evidence supported single-plane open-chain exercises below 90° of elevation and closed-chain upper extremity exercises, whereas expert literature favoured plyometrics, kinetic chain and sport-specific training [11]. The two sources recommended considerably different exercises [11].

Shoulder impingement is seen from time to time in combat sports, and in my experience it can be a particularly difficult problem for combat sports athletes, because it interferes considerably with training. With these athletes, I tend to emphasise thorough, supportive physiotherapy, along with regular muscle care from a skilled physiotherapist, both around training sessions and outside them.

Conclusion: Shoulder Impingement in Perspective

Shoulder impingement is still a widely used label for a common type of shoulder pain, but its central mechanical explanation is increasingly questioned [4]. The usual clinical tests do not reliably confirm the diagnosis [5], and rotator cuff changes on imaging are common in people without pain [6].

The treatment evidence points in a consistent direction. Decompression surgery has not outperformed placebo surgery [1][9] and has not outperformed exercise therapy over five years [7]. Exercise and corticosteroid injections have produced similar results [10], although the evidence for specific exercise types remains of low certainty [2].

This matches what I see in practice. In my clinical experience, most patients with shoulder impingement are managed without surgery, with active physiotherapy at the centre and injections or anti-inflammatory painkillers for some, although a smaller group is left with long-term pain and is referred on to an orthopaedic surgeon.

In my view, the current evidence does not support treating shoulder impingement primarily as a structural problem that generally requires decompression surgery.

References

1. https://doi.org/10.1016/S0140-6736(17)32457-1
2. https://doi.org/10.3389/fneur.2024.1357763
3. https://doi.org/10.1002/msc.1643
4. https://doi.org/10.3390/jcm15197356
5. https://doi.org/10.1136/bjsports-2012-091066
6. https://doi.org/10.2519/jospt.2025.13611
7. https://doi.org/10.1136/bjsports-2020-102216
8. https://doi.org/10.1186/s12891-021-04768-7
9. https://doi.org/10.1002/14651858.CD005619.pub3
10. https://doi.org/10.7759/cureus.48907
11. https://doi.org/10.1136/bjsports-2016-096915