Surgery or exercise for rotator cuff shoulder pain: what the evidence shows

The question almost everyone asks first

If your shoulder aches when you reach overhead, catches when you lift, or wakes you when you roll onto it at night, sooner or later a version of the same question arrives: do I need surgery? It's a fair thing to ask. A tear or "impingement" sounds mechanical, and mechanical problems sound like they need fixing. For a long time, one of the most common shoulder operations — shaving bone and soft tissue to make more room under the shoulder blade — was offered on exactly that logic. Then researchers did something unusual: they tested that operation against a fake version of itself, and against a course of exercise. The results reshaped how the honest answer is given. This piece lays out what those trials actually found, how confident researchers are, and — just as importantly — which situations they don't cover. We're not here to talk you into or out of an operation. We're here to show you the evidence so you can have a sharper conversation with a regulated professional.

What the shoulder problem actually is

The rotator cuff is a group of four muscles whose tendons wrap around the top of the arm bone and let you lift and rotate the shoulder [1]. Most cuff problems are not sudden injuries. According to the American Academy of Orthopaedic Surgeons, the majority of tears come from a slow wearing-down of the tendon over years, more common after age 40, linked to repetitive overhead use and the natural decline in the tendon's blood supply [1]. This degenerative, gradual kind of shoulder pain — often called subacromial pain syndrome, impingement, or rotator cuff disease — is what most middle-aged patients with an achy shoulder have.

That is different from an acute, traumatic tear: a full-thickness tear caused by a fall, a heavy lift with a jerk, or a shoulder dislocation, often in a younger person [1][6]. The distinction matters more than almost anything else here, because — as you'll see — the big trials that questioned surgery studied the degenerative, atraumatic group, not the traumatic one. Keeping those two apart is the key to reading the evidence correctly.

Symptoms overlap across both: pain at rest and at night (especially lying on that side), weakness turning or lifting the arm, and sometimes a crackling sensation with movement [1]. Because the surface presentation looks similar, the decision about surgery rests less on how much it hurts and more on why the tendon is damaged and what the shoulder is being asked to do.

What the trials found

Two landmark trials tested the standard "make more room" operation — arthroscopic subacromial decompression — in people with longstanding, non-traumatic shoulder pain. Both included a placebo-surgery arm: patients who were taken to the operating room and given only a diagnostic look inside the joint, with no decompression, and who did not know which they'd received. That design is what makes the findings hard to dismiss.

The CSAW trial (Beard and colleagues, Lancet, 2018) randomised 313 UK patients into three groups: real decompression, placebo (arthroscopy only), or active monitoring with no surgery [2]. Six months on, shoulder scores (Oxford Shoulder Score, higher is better) were 32.7 with decompression, 34.2 with placebo surgery, and 29.4 with no surgery at all [2]. Read those numbers slowly. Real surgery did not beat the placebo operation — the difference was −1.3 points (95% CI −3.9 to 1.3), statistically indistinguishable from zero [2]. Both surgical groups edged out no-treatment by a couple of points, but the gap fell short of the 4.5-point difference the researchers had set in advance as the smallest change worth calling clinically important [2]. The authors concluded the results "question the value of this operation" for these patients [2].

The FIMPACT trial (Paavola and colleagues, BMJ, 2018) went a step further and added an exercise arm [3]. Its 210 Finnish patients were assigned to decompression, placebo surgery, or a structured exercise programme. At two years, on a 0–100 pain scale, decompression was no better than placebo surgery — the difference in pain during arm activity was −9.0 points (95% CI −18.1 to 0.2), and at rest −4.6 (−11.3 to 2.1), neither reaching the 15-point mark the trial defined as clinically meaningful [3]. Surgery did edge out exercise on the raw numbers (pain on activity −12.0, 95% CI −20.9 to −3.2), but again that gap did not cross the 15-point threshold for a difference a patient would actually feel, and it came at the cost, burden, and risk of an operation [3]. The authors' verdict was blunt: the surgery "provided no benefit over diagnostic arthroscopy," and the findings "do not support" performing it for this condition [3].

Pulling the trials together, a 2019 Cochrane systematic review (Karjalainen and colleagues) pooled eight trials and 1,062 patients [4]. Against placebo surgery it found decompression produced a pain difference of about a quarter-point on a 0–10 scale (0.26 better, 95% CI 0.84 better to 0.33 worse) — and rated this high-certainty evidence of no clinically important benefit for pain, function, or quality of life [4]. High-certainty is Cochrane's strongest grade; it means further research is unlikely to overturn the finding. The reviewers concluded the data "do not support the use of subacromial decompression in the treatment of rotator cuff disease" [4].

The honest headline, then, is not that one option "wins." It is that for many people with degenerative, non-traumatic subacromial pain, this surgery performs no better than a placebo operation, and produces outcomes broadly comparable to structured exercise — with exercise avoiding the costs and risks of the theatre. That's a statement about comparability, not a ranking of one therapy as superior.

Who surgery may still suit

None of the above means surgery has no place. It means the decompression operation, for atraumatic subacromial pain, has a weak evidence base — a narrower claim than "shoulder surgery doesn't work."

The 2019 BMJ Rapid Recommendation (Vandvik and colleagues), an independent guideline panel including patients, weighed exactly this evidence and issued a strong recommendation against decompression surgery for adults with atraumatic shoulder pain lasting more than three months [5]. Crucially, that guidance was written for the degenerative group and does not address acute, traumatic, or full-thickness rotator cuff tears [5] — a boundary the panel drew deliberately.

For those tears, the clinical picture differs. A 2021 review of traumatic rotator cuff tears notes they typically stem from higher-energy injuries, more often produce large full-thickness tears, and disproportionately affect people under 40 — a group treated as a distinct population, in whom timely surgical repair is more commonly considered because delays are associated with tendon retraction and muscle wasting [6]. Separately, AAOS describes surgery being weighed when pain persists despite months of non-surgical care, when tears are large, or when there is significant weakness or loss of function [1]. Whether any of that applies to a given shoulder is a judgment for a surgeon or specialist after examining the joint and imaging — not something a study average can decide from a distance.

Limitations worth holding in mind

• "Surgery" here means one specific operation. These trials tested arthroscopic subacromial decompression, not tendon repair of a torn cuff. Evidence about decompression does not transfer to repairing an acute tear.
• The populations were atraumatic and chronic. People with traumatic full-thickness tears, or with other shoulder diagnoses, were generally not who these trials enrolled [3][5].
• You cannot fully blind exercise, but you can blind surgery — and they did. The placebo-surgery design is a real strength; still, exercise programmes, dose, and adherence varied between trials, so the exercise comparison carries lower certainty than the placebo-surgery comparison [3][4].
• Comparable is not identical. Group averages hide individual variation; some people improve markedly with one path and little with another, and trials can't tell you in advance which you'll be.

What this means for you

A fair summary: for longstanding, non-traumatic subacromial shoulder pain, the best current evidence shows that the decompression operation is no better than a placebo operation, and gives outcomes comparable to a course of structured exercise — while an independent guideline panel recommends against that surgery for this group [2][3][4][5]. At the same time, acute traumatic tears are a genuinely different situation where timely surgical repair is more often considered, on the judgment of a specialist [1][5][6]. None of this is a promise about your shoulder, and none of it settles which path is right for you.

The right next step is a conversation with a licensed, regulated professional — your physician, a physiotherapist, or an orthopaedic surgeon — who can examine your shoulder, look at whether the problem is degenerative or a traumatic tear, and help you weigh exercise, other non-surgical care, and surgery against your own goals and history. That individualised assessment is the one thing the research averages cannot give you.