Shoulder Arthritis and Replacement: Types, When Surgery Is Needed, and Which Operation

Shoulder replacement is done chiefly to relieve the pain of an arthritic joint, most often glenohumeral osteoarthritis, rheumatoid arthritis, avascular necrosis, or rotator cuff tear arthropathy, once painkillers, injections and physiotherapy no longer control it. Arthritis earns a replacement when the pain limits sleep and daily life and the changes on imaging match that pain, not on a scan finding alone¹.

For two years I treated my shoulder as a nuisance rather than a joint that had ground itself smooth, working through the painkillers and an injection that helped for a while and then did not, until the nights in a chair made the decision for me. What I could not find was the plain account of when arthritis actually earns an operation and which one it points to. This is that account, written from my own reverse replacement outward and checked by a consultant shoulder surgeon. If you want the whole picture first, start with the shoulder replacement overview.

What is shoulder arthritis?

Shoulder arthritis is damage to the surfaces of the ball-and-socket (glenohumeral) joint, and it comes in several forms: glenohumeral osteoarthritis (simple wear), rheumatoid arthritis (an inflammatory disease), post-traumatic arthritis after an old fracture, avascular necrosis where the bone of the humeral head dies, and rotator cuff tear arthropathy, where a long-standing cuff tear leads on to arthritis. These share a common end point of a painful, worn joint, but they behave differently and matter to the choice of operation².

The distinction that turned out to matter most for me was not the label on the arthritis but what it had done to the rotator cuff, the sleeve of tendons that lifts and steadies the arm. Osteoarthritis often leaves the cuff intact; cuff tear arthropathy, by definition, does not. That single fact shapes almost everything that follows.

When does arthritis earn a replacement?

Arthritis earns a replacement when the pain limits sleep and daily life, the changes on imaging match that pain, non-surgical measures no longer control it, and general health is fit for an anaesthetic. It is the pain and its grip on your life, not the appearance of a worn joint on a scan, that drives the decision to operate¹.

That order matters, because a shoulder can look dramatic on an X-ray and still be liveable, while another looks less severe and steals every night’s sleep. The honest test is whether the shoulder still responds to the things short of surgery, not how worn it looks on a scan.

What comes before surgery

Non-surgical measures come first and can control arthritic shoulder pain for years: painkillers, changes to how you use the arm, physiotherapy to keep the joint moving and strong, and corticosteroid injections into the joint. None of these reverses the arthritis, but they can settle pain and delay an operation, and a replacement is generally reserved for when they have genuinely stopped working².

My own injection bought me a good few months before it faded, and I understand now why a surgeon does not reach for the operation while gentler measures still help: a replacement lasts well but not forever, so there is a real case for not spending its years too early.

Which operation the arthritis points to

The rotator cuff, not the type of arthritis or your age, mainly decides the operation: an arthritic shoulder with a working cuff usually has an anatomic total replacement, while an arthritic shoulder with a torn, irreparable cuff (cuff tear arthropathy) usually has a reverse replacement, where the deltoid lifts the arm in place of the cuff. The reverse design exists precisely because an anatomic replacement fails early when the cuff is gone¹.

So osteoarthritis with an intact cuff points toward the total (anatomic) shoulder replacement, while cuff tear arthropathy points toward the reverse operation set out in rotator cuff arthropathy and reverse replacement. My own cuff was long past repair, so a standard total was never on the table and I had a reverse instead. Rheumatoid shoulders can go either way depending on the state of that cuff and the bone, which is the fork that quietly decides the whole operation.

What a replacement will and will not do for the arthritis

A replacement removes the arthritic surfaces of that joint and settles the pain, with roughly 90% to 95% of people reporting less pain and better function and most series describing around 90% good or excellent results, but it relieves arthritic pain rather than restoring a twenty-year-old shoulder. Overhead and rotational range improves rather than returns in full, and the joint should not be loaded like an undamaged one³.

It is also worth being clear that surgery treats the shoulder, not the disease behind it: rheumatoid arthritis or osteoarthritis can still affect other joints, and the arthritis is not cured, only removed from that joint. The honest boundaries are set out in what shoulder replacement will not fix. Set against that, most replacements last well, with around 90% still in place at 10 years, so a shoulder that has genuinely run out of other options usually rewards the operation⁴.

The arthritis that made the decision for me

The turning point for most arthritic shoulders is not a scan but the night, when the pain stops responding to painkillers and stops you sleeping on the affected side. That is the point at which a worn joint moves from a nuisance to a reason to operate, and it is a lived signal a scan cannot capture⁵.

Mine was a chair in the spare room, because lying flat was unbearable and the arm ached whatever I did with it. I have written that stretch honestly in the shoulder pain that stopped me sleeping. If your arthritis has reached that point and the injections and physiotherapy have stopped earning their keep, the question is no longer whether the joint is worn, but which operation the cuff points to and whether the rest of the picture fits, and that is a judgement for a surgeon examining you and your imaging in person.