Condition guide · Shoulder and upper limb
Shoulder Impingement
The operation designed to fix impingement has now been compared against fake surgery and followed for ten years. It made no difference. This page is about what that means for the diagnosis itself.
In plain words. "Impingement" describes an idea rather than a finding: that the tendons of the rotator cuff are being pinched under the bony arch above them, and that shaving away part of that arch would relieve it. The pain is real and common. The explanation has not held up.
The operation built on that explanation — arthroscopic subacromial decompression — has been tested against placebo surgery and followed for a decade. It performed no better than a fake operation, and no better than exercise. [1]
Key points
- Decompression surgery is no better than placebo surgery at ten years. Mean difference in pain at rest -1.5 points (95% CI -8.6 to 5.6) and on arm activity -3.2 (95% CI -13.0 to 6.5), with 87% of participants followed up. [1]
- It is no better than exercise therapy either, in the same trial over the same decade. [1]
- A BMJ guideline panel makes a strong recommendation against the surgery, concluding that almost all informed patients would choose to avoid it — no benefit, some harm, and burdensome. [2]
- Frozen shoulder may be more common after the operation. [2]
- Exercise reduces how many people end up having surgery. Over ten years, 35% of a specific-exercise group chose surgery against 65% of a control exercise group. [4]
- Rotator cuff tears accumulate with age regardless. In that trial 55% had a tear at ten years against 28% at baseline — and 51% had one in the opposite, largely asymptomatic shoulder against 3% at baseline. [4]
- What should replace the surgery is genuinely unsettled. The guideline panel says so explicitly. [2]
What is actually happening
The traditional account runs like this: the rotator cuff tendons pass through a narrow space beneath the acromion, the bony roof of the shoulder; with overhead use, or with a hooked acromial shape, they are compressed against it; that compression irritates the tendon and the bursa, and produces pain on lifting the arm.
It is an appealing mechanical story, and it generated a matching operation: remove some of the bone, widen the space, relieve the pinch. If the story were right, the operation should work.
It does not. In the FIMPACT trial, patients were randomised to decompression surgery, to placebo surgery — an arthroscopy with no decompression performed — or to exercise therapy. At ten years there were no differences between decompression and placebo on either primary pain outcome, and none on any secondary outcome or in adverse events. [1] Comparing decompression against exercise gave the same answer. [1] The five-year report of the same trial had already found this; the ten-year data confirm it did not change with time. [3]
When an operation aimed squarely at a proposed mechanism performs identically to a sham version of itself, the most economical conclusion is that the mechanism was not the problem. Current thinking treats this as tendon-related shoulder pain with several contributing factors — load, capacity, tendon health, movement control, and sensitivity — rather than a structure being pinched.
A related finding from a separate ten-year study makes the point about structure even more sharply. Rotator cuff tears became substantially more common over the decade — 55% at ten years against 28% at baseline — but they became just as common in the other shoulder, rising from 3% to 51%, and tear progression did not differ between operated and non-operated patients. [4] Tears accumulate with age in shoulders that are not complaining.
What it feels like
- Pain on the outer shoulder and upper arm, often described as a band rather than a point.
- Worse reaching overhead, out to the side, or behind the back — putting on a shirt, reaching a high shelf.
- A painful arc partway through lifting the arm, sometimes easier at the very top.
- Night pain, particularly lying on that side, and difficulty finding a comfortable position.
- Weakness that is often pain-related rather than a true loss of power.
If the shoulder is stiff in every direction, including when someone else moves it for you, this is probably not the right page — see frozen shoulder.
How it is diagnosed
By history and examination: where the pain is, what provokes it, how the shoulder moves and how strong it is through range. The various named impingement tests provoke the painful structures but do not reliably identify which structure is responsible, and a positive test does not confirm that anything is being pinched.
Assessment establishes what the shoulder can and cannot do and how it responds to load. That is more useful than deciding which structure is being compressed, because the compression model did not survive being tested.
Imaging deserves particular caution here. Scans frequently show cuff tears, bursal change and acromial shape variations in people with no pain at all — and in the ten-year study, half of the largely asymptomatic opposite shoulders had a tear. [4] A scan finding is therefore not, on its own, an explanation for your symptoms. Imaging is useful to answer a specific question, not to name the pain.
How physiotherapy and rehabilitation help
With the surgical route closed off by the evidence, exercise-based rehabilitation is the mainstay by default as much as by demonstration. It is worth being honest about that distinction: exercise looks good partly because it performs as well as the alternatives, and the alternatives performed no better than placebo.
The strongest single argument for it comes from a ten-year follow-up in which patients received either a specific exercise programme or a control exercise programme, with surgery available to anyone who wanted it. By ten years, 35% of the specific exercise group had chosen surgery against 65% of the control group. All patients improved substantially on the Constant-Murley score, and those who had not been operated on scored 11 points higher than those who had (95% CI 4 to 18). [4] The authors' reading is that specific exercise reduced the need for surgery, and that decompression still gave satisfactory results in those who did not respond to exercise.
What a programme involves
| Component | What it is for | Evidence |
|---|---|---|
| Progressive loading of the rotator cuff the core of the programme, advanced over months |
Rebuilding the tolerance of the cuff tendons to the load of reaching, lifting and working overhead. | Supported By ten years, 35% of the specific exercise group had chosen surgery against 65% of the control group [4] |
| Scapular stabilisation exercise | Improving the control of the shoulder blade that the cuff works against. | Supported, low certainty Added to general exercise it reduced pain (mean difference -0.8 cm) and improved function and abduction range, on low to very low certainty evidence [5] |
| Feedback added to those exercises | Coaching the movement more precisely, on the assumption that accuracy matters. | No added benefit Adding feedback made no significant difference [5] |
| Thoracic spine manual therapy | Short-term relief of pain and stiffness to make the exercise possible. | Supported, short term Reduces pain and disability and improves range of motion in the short term [6]. Short term is the operative phrase |
| Load management and activity modification | Temporarily reducing the overhead and end-range demands that provoke it. | Not separately tested Standard practice; not isolated in the cited trials |
| Subacromial decompression surgery | Surgically enlarging the space beneath the acromion. | Not recommended Strong guideline recommendation against [2]; no better than placebo surgery at ten years [1] |
| Electrotherapy ultrasound, TENS, shockwave, laser |
Commonly offered for shoulder pain. | Not tested here None of the sources cited on this page evaluated them, so this page makes no claim either way |
An amber badge means the sources cited on this page did not test that component — not that it fails. It is a statement about the evidence available here, not a verdict.
What a course of treatment looks like
Assessment first, to establish what the shoulder can do and what provokes it, and to identify the overhead or loaded tasks — at work, in sport, at home — that the programme has to get you back to.
Then progressive loading of the cuff and the shoulder blade, made harder as the shoulder tolerates it. The trials that support this ran their programmes over months, not weeks, and the follow-up that matters was measured in years. [4] Manual therapy to the thoracic spine is reasonable early, to make the exercise possible; it is not the treatment. [6] Appointments are for progressing load and checking technique, and most of the work happens between them.
One finding deserves to be said plainly, because it is unusual for a page to be able to say it. In the ten-year follow-up everyone improved substantially — including the group who had neither the specific exercise programme nor surgery. [4] Time is doing a great deal of the work here. That is a reason for patience rather than for an operation, and a reason to be sceptical of anyone attributing a slow recovery entirely to their own intervention.
Where this leaves the decision about surgery
The BMJ guideline panel reviewed the same evidence and made a strong recommendation against subacromial decompression, judging that surgery did not provide important improvements in pain, function or quality of life compared with placebo surgery or other options, that frozen shoulder may be more common after it, and that almost all informed patients would choose to avoid it. [2]
The panel then said something that is easy to skip past and worth quoting the sense of: there is substantial uncertainty about what the best alternative treatment is. [2] Ruling out an operation is not the same as knowing what to do instead. This page can tell you with confidence what does not work; it cannot tell you with the same confidence which exercise programme is best.
What the evidence supports — and what it does not
Supported
- Avoiding subacromial decompression surgery — a strong guideline recommendation against it. [2]
- Exercise as first-line management, which reduced the proportion of people going on to surgery from 65% to 35% over ten years. [4]
- Scapular stabilisation added to general exercise — improved pain and function, on low to very low certainty evidence. [5]
- Thoracic manual therapy for short-term relief. [6]
- Expecting improvement over time — all groups improved markedly across ten years. [4]
Not supported
- The impingement model itself, as an explanation that justifies surgery — the operation matched placebo at five years and at ten. [1][3]
- Decompression surgery, over placebo or over exercise, on any outcome measured. [1]
- Treating a scan finding as the cause. Half of the asymptomatic opposite shoulders had a cuff tear at ten years. [4]
- Adding feedback devices to scapular exercise — no significant difference against the same exercises without feedback. [5]
- Confidence about which rehabilitation programme is best. The guideline panel states the uncertainty plainly. [2]
How certain is this?
Evidence grade: High for the negative finding, low for the positive one. That split is the honest summary of this whole topic.
What we know well is what does not work. A placebo-controlled surgical trial is among the hardest studies in medicine to run and among the most informative when it succeeds. FIMPACT randomised patients to real surgery, sham surgery and exercise, retained 87% of them for a decade, and found no differences. [1] A guideline panel working to GRADE standards reached a strong recommendation against the operation on that basis. [2] This is about as settled as surgical evidence gets.
What we know poorly is what to do instead. The scapular stabilisation meta-analysis rates its own evidence low to very low certainty and calls for higher-quality trials. [5] The exercise follow-up study is informative but modest — 83 patients at ten years — and was not designed to compare exercise against no treatment at all. [4] The guideline panel's admission of substantial uncertainty about alternatives is the most accurate statement available. [2]
So: be confident when someone recommends the surgery, and appropriately humble about the specifics of the rehabilitation being offered instead.
What to expect
Slow improvement, measured in months rather than weeks, and considerable variation between people. In the ten-year study everyone improved substantially, by an average of 37 points on the Constant-Murley score. [4] That is genuinely encouraging, and it happened over years rather than over a course of treatment.
Expect the shoulder to be irritable in the early phase, and expect progress to be judged by what you can reach and lift rather than by whether it aches. If you are not improving with a properly progressed programme, that is a reason to reassess the diagnosis rather than to escalate straight to surgery — particularly given that decompression matched a sham operation. [1]
Common questions
My surgeon has recommended decompression surgery. Should I have it?
That is a conversation to have with them, armed with the evidence. A BMJ guideline panel makes a strong recommendation against the operation, having concluded that almost all informed patients would decline it. [2] The trial behind that recommendation compared the surgery against a placebo operation and found no difference at ten years, and no difference against exercise either. [1] Reasonable questions to ask are what the surgery is expected to achieve given those results, and what has already been tried.
Is "impingement" a real diagnosis?
The pain is real. The explanation the name carries — that a structure is being pinched and needs decompressing — has not survived testing. Many clinicians now prefer terms like subacromial pain syndrome or rotator cuff-related shoulder pain, which describe where the problem is without asserting a mechanism the evidence does not support.
My scan shows a rotator cuff tear. Does that change things?
Less than you would expect. In a ten-year follow-up, 51% of the largely asymptomatic opposite shoulders had a tear by the end, against 3% at baseline. [4] Tears accumulate with age in shoulders that are not painful. A tear found on a scan is not automatically the cause of your symptoms. Large and massive tears are a different discussion — see rotator cuff pain.
Which exercises should I be doing?
Progressive loading of the cuff and shoulder blade muscles is the mainstay. Adding scapular stabilisation exercise to general exercise improved pain and function, although on low to very low certainty evidence. [5] The honest answer is that the specific programme matters less than doing one and progressing it, and the guideline panel is explicit that the best alternative to surgery remains uncertain. [2]
Do I need a device or feedback machine for the scapular exercises?
No. The meta-analysis found no significant difference in pain or functional recovery between scapular stabilisation exercise with feedback and the same exercise without it. [5]
How long before it settles?
Months, and sometimes longer. The reassurance from the ten-year data is that substantial improvement is the norm across all groups. [4] The caution is that it is slow, and that a shoulder which is not improving deserves reassessment rather than either persistence or an operation the evidence does not support.
References
- Kanto K, Bäck M, Ibounig T, et al. Arthroscopic subacromial decompression versus placebo surgery for subacromial pain syndrome: 10 year follow-up of the FIMPACT randomised, placebo surgery controlled trial. BMJ. 2025 Dec 2;391:e086201. doi:10.1136/bmj-2025-086201 PMID 41330610 Randomised placebo-controlled trial, 10-year follow-up
- Vandvik PO, Lähdeoja T, Ardern C, et al. Subacromial decompression surgery for adults with shoulder pain: a clinical practice guideline. BMJ. 2019 Feb 6;364:l294. doi:10.1136/bmj.l294 PMID 30728120 Clinical practice guideline
- Paavola M, Kanto K, Ranstam J, et al. Subacromial decompression versus diagnostic arthroscopy for shoulder impingement: a 5-year follow-up of a randomised, placebo surgery controlled clinical trial. British Journal of Sports Medicine. 2021 Jan;55(2):99–107. doi:10.1136/bjsports-2020-102216 PMID 33020137 Randomised controlled trial, 5-year follow-up
- Petersson AH, Björnsson Hallgren HC, Adolfsson LE, et al. No need for subacromial decompression in responders to specific exercise treatment: a 10-year follow-up of a randomized controlled trial. Journal of Shoulder and Elbow Surgery. 2025 Jun;34(6):e477–e487. doi:10.1016/j.jse.2024.10.027 PMID 39716615 Randomised controlled trial, 10-year follow-up
- Brasileiro A, Sousa C, Schindler I, et al. Scapular stabilization exercise on pain and functional recovery in people with shoulder impingement syndrome: a systematic review and meta-analysis. The Physician and Sportsmedicine. 2025 Jun;53(3):189–196. doi:10.1080/00913847.2025.2470115 PMID 39983700 Systematic review and meta-analysis
- Yu S, Chen S, Yang Z, et al. Effectiveness of Thoracic Spine Manual Therapy in Treating Subacromial Impingement Syndrome: A Systematic Review and Meta-analysis. Archives of Physical Medicine and Rehabilitation. 2025 Dec;106(12):1886–1898. doi:10.1016/j.apmr.2025.07.008 PMID 40712864 Systematic review and meta-analysis
About this guide
If you need assessment
This page explains. It does not diagnose.
Shoulder pain has several causes that look similar and are managed differently, so it is worth having assessed rather than self-labelled. Physiotherapist India publishes information and takes no bookings. Ask a qualified physiotherapist to examine your own case before acting on anything written here.