Condition guide · Shoulder
Rotator Cuff Pain
The commonest cause of shoulder pain, and the subject of one of the most instructive trials in orthopaedics — in which the standard operation was compared against a pretend one.
In plain words. Four muscles wrap around the ball of the shoulder and their tendons blend into a cuff that holds it centred while the big muscles move the arm. When those tendons are irritated, overloaded or degenerate, the shoulder hurts — typically when you lift the arm out to the side, and typically at night when you roll onto it.
For decades the leading explanation was that a bony spur was pinching the tendon, and the leading treatment was surgery to shave the bone away. Then somebody ran the obvious experiment: they compared that operation with an identical operation in which the surgeon did nothing. The results are the reason this page exists.
Key points
- Subacromial decompression is no better than placebo surgery — high-certainty evidence of no improvement in pain, function or quality of life up to one year. [1]
- In the placebo-controlled trial, real surgery and pretend surgery finished 1.3 Oxford points apart (95% CI −3.9 to 1.3, p=0.31) — a difference of nothing. [2]
- Both surgical groups did slightly beat no treatment, but by less than the trial's own 4.5-point threshold — and since the placebo operation did it too, the benefit is not coming from the bone removal. [2]
- Surgery is not risk-free. Serious adverse events within 30 days of shoulder arthroscopy ran at about 0.5–0.6% in registry data, including deep infection, pulmonary embolism, nerve injury and death. [1]
- The people in CSAW had already failed a non-operative programme including exercise and at least one steroid injection — so this is evidence about persistent pain, not first-line care. [2]
- Exercise-based rehabilitation remains the first-line treatment, and the surgical evidence gives no reason to hurry past it.
What is actually happening
The rotator cuff is four muscles — supraspinatus, infraspinatus, teres minor and subscapularis — running from the shoulder blade to the head of the humerus. Their job is not power but control: they hold the ball centred in its shallow socket while the larger muscles move the arm. The supraspinatus tendon passes through a narrow gap between the head of the humerus below and the acromion above, cushioned by a bursa.
The subacromial space, and the operation that targets it
That theory — "impingement" — is intuitive, and it dominated shoulder practice for thirty years. The difficulty is that when the bone removal was isolated and tested against a sham operation, it added nothing. [1][2] Whatever is producing the pain, shaving the acromion is not what fixes it.
The current understanding is less tidy: tendon tissue that has been overloaded relative to what it can tolerate, in a system where strength, control of the shoulder blade, and load history all matter. That is a less satisfying story than a spur on an X-ray, but it is the one that survives testing.
What it feels like
- Pain on the outer shoulder, sometimes referred down towards the upper arm but rarely past the elbow.
- Worst reaching up, out to the side, or behind — a painful arc partway through lifting the arm.
- Night pain, particularly lying on that side. This is one of the most characteristic features.
- Difficulty with overhead work, hanging washing, or reaching into a high cupboard.
- Weakness that may be genuine loss of strength, or simply the arm giving way because it hurts — distinguishing the two is part of the examination.
How it is diagnosed
Clinically, by pattern of pain and by testing the cuff muscles individually against resistance while watching how the shoulder blade behaves. The main job is to separate ordinary cuff-related pain from a full-thickness tear, from a frozen shoulder, and from pain referred from the neck.
Note the deliberate exclusion in the trials cited here: CSAW required intact rotator cuff tendons and excluded full-thickness tears, and the Cochrane review likewise excluded them. [1][2] Nothing on this page tells you what to do about a complete tear — that is a different problem with a different evidence base.
What treatment involves
A progressive loading programme for the cuff and the muscles controlling the shoulder blade, advanced over months rather than weeks, alongside a sensible plan for the activities that provoke it. Tendon tissue adapts to load; the skill is choosing a starting dose the shoulder accepts and building from there.
A corticosteroid injection is sometimes used to reduce pain enough to make that loading possible. It is worth being clear about what that is for: creating a window in which you can do the work, not repairing the tendon.
Everyone in the CSAW trial had already completed a non-operative programme including exercise and at least one steroid injection before being randomised. [2] The findings therefore apply to people whose pain has persisted despite proper first-line treatment — and even in that group, the operation did not beat placebo.
What the evidence supports — and what it does not
CSAW: real surgery, placebo surgery, and no surgery at six months
Supported
- Exercise-based rehabilitation as first-line treatment — it is what everyone in these trials received before anything else was considered, and the surgical alternative did not beat a sham. [1][2]
- Time and continued loading — even the untreated group in CSAW improved over six months, from a starting point of persistent pain. [2]
- Being clear about what a steroid injection does — reduce pain to permit rehabilitation, rather than repair tissue.
Not supported
- Arthroscopic subacromial decompression — high-certainty evidence of no improvement in pain, function or quality of life versus placebo surgery up to one year. Pain at one year was 2.9 after placebo and 0.26 better after surgery (95% CI 0.84 better to 0.33 worse). [1]
- The "impingement" rationale — if pain came from the bone pressing on the tendon, removing the bone should have outperformed an operation that removed nothing. It did not. [1][2]
- Global success as a reason to operate — 655 per 1,000 after placebo versus 708 per 1,000 after surgery, risk ratio 1.08 (95% CI 0.93 to 1.27). [1]
- Treating shoulder arthroscopy as trivially safe — serious adverse events within 30 days ran at 0.5–0.6% in registry data. Small, but not zero, for a procedure with no demonstrated benefit over placebo. [1]
How certain is this?
Evidence grade: High, for the negative finding. This is one of the few places on this site where the evidence is genuinely strong, and it is strong in the direction of not doing something. The Cochrane review rated the placebo comparison high-certainty for pain, function and quality of life, and both placebo-controlled trials were at low risk of bias for that comparison. [1] Placebo-controlled surgical trials are rare and difficult; when one exists, it settles far more than a stack of observational studies.
The limits are worth stating. The evidence excludes full-thickness rotator cuff tears entirely. [1][2] The trials studied people whose symptoms had persisted for at least three months despite non-operative care, so they cannot tell you about the first few weeks. And the Cochrane review's non-placebo comparisons — surgery versus exercise — came from trials at high risk of bias, which is why the placebo comparison carries the weight. [1]
Note also what is not established: exactly which exercise programme is best, or for how long. The strength of the evidence here is about what to avoid, not a precise prescription for what to do.
What to expect
Rotator cuff pain is usually a matter of months. In CSAW, all three groups — including the one that received no treatment at all — improved between randomisation and six months. [2] That is worth holding onto when progress feels slow: the natural course of this problem is toward improvement.
Night pain is often the last symptom to settle, and it is common for the shoulder to remain sensitive to a sudden increase in overhead work for some time after everyday pain has resolved. Building tolerance back deliberately is more reliable than avoiding the activity until it feels ready.
When shoulder pain needs urgent assessment
Seek immediate medical attention if you have:
- Shoulder pain with chest pain, breathlessness, sweating or nausea — this can be cardiac and needs emergency care.
- Sudden inability to lift the arm at all after a fall or a wrench — this can indicate an acute full-thickness tear, which is time-sensitive.
- Obvious deformity, or the shoulder feeling out of joint after an injury.
- A hot, swollen shoulder with fever or feeling generally unwell.
- Numbness, pins and needles, or weakness spreading down the arm.
- Unexplained weight loss, a history of cancer, or severe unrelenting night pain that does not vary with position.
This list is not exhaustive. If you are worried about a symptom, seek professional advice.
Common questions
My surgeon has recommended a decompression. Should I have it?
That is a decision to make with your surgeon, but make it knowing the evidence. A Cochrane review rated as high-certainty that subacromial decompression provides no improvement in pain, function or quality of life compared with placebo surgery up to one year [1], and in the trial that compared them directly the difference was −1.3 Oxford points, well within chance. [2] It is fair to ask what specifically in your case is expected to change. Note this evidence excludes full-thickness tears.
But people do get better after the operation.
They do — and that is exactly why the placebo comparison matters. People also got better after an operation in which the surgeon deliberately removed nothing, and after no operation at all. [2] Improvement following a treatment is not evidence that the treatment caused it, which is the whole reason sham-controlled surgical trials are worth their considerable difficulty.
My scan shows a spur and impingement. Doesn't that prove it?
It shows the anatomy that the impingement theory was built on. What the trials tested was whether removing that anatomy helps, and the answer was no better than pretending to. [1][2] As elsewhere on this site, a scan finding is a description, not automatically an explanation.
Should I rest it?
Not completely. Tendon tissue responds to graded load, and the first-line treatment in all of this research was an exercise programme, not immobilisation. Avoiding the movements that hurt for a few days is reasonable; avoiding them for months tends to leave you weaker and no less sore.
How long before it settles?
Usually months. The trial groups were followed at six months and a year, and improvement continued across that period in every group, including no treatment. [2] Anyone offering a shorter guaranteed timeline is going beyond the evidence.
References
- Karjalainen TV, Jain NB, Page CM, et al. Subacromial decompression surgery for rotator cuff disease. Cochrane Database of Systematic Reviews. 2019 Jan 17;1(1):CD005619. doi:10.1002/14651858.CD005619.pub3 PMID 30707445 Systematic review
- Beard DJ, Rees JL, Cook JA, et al. Arthroscopic subacromial decompression for subacromial shoulder pain (CSAW): a multicentre, pragmatic, parallel group, placebo-controlled, three-group, randomised surgical trial. Lancet. 2018 Jan 27;391(10118):329–338. doi:10.1016/S0140-6736(17)32457-1 PMID 29169668 Placebo-controlled randomised trial
About this guide
If you need assessment
This page explains. It does not diagnose.
Separating cuff-related pain from a full-thickness tear, a frozen shoulder or pain referred from the neck requires examination — and the evidence on this page deliberately excludes full-thickness tears. Physiotherapist India publishes information and takes no bookings. Ask a qualified physiotherapist to examine your own case before acting on anything written here.