Shoulder Arthritis: What Physical Therapy Can and Cannot Promise

The glenohumeral joint is the ball and socket of the shoulder: the round head of the humerus sitting against the shallow glenoid of the shoulder blade. Both surfaces are covered in smooth articular cartilage, and the joint is held together less by bony fit than by the capsule, the labrum and the four rotator cuff muscles, which press the ball into the socket every time the arm moves. In the mobility-stability model the shoulder is a stability joint that has been given an enormous range of motion, and it relies on the cuff and the shoulder blade muscles to control it.

Glenohumeral osteoarthritis is what happens when that cartilage thins and wears, the joint space narrows, and the bone underneath remodels, often with bone spurs (osteophytes) at the edges [2]. The result is pain with movement, stiffness, clicking or grinding, and a steady loss of rotation. Physical therapy is one of the most commonly offered treatments. A 2026 systematic review set out to measure how well it works and found almost nothing to measure: four randomized trials, all after surgery, and none of non-surgical care [1]. That gap is the subject of this post.

What Is Happening

Primary glenohumeral osteoarthritis is a wear process in the joint surfaces. Cartilage has no blood supply of its own, so once it thins it does not rebuild in any meaningful way. As the surfaces lose their cushioning, bone moves closer to bone, the capsule tightens, and osteophytes grow around the rim of the humeral head. The AAOS, the main US professional body for orthopedic surgeons, describes the typical picture as shoulder pain that increases with movement, clicking, reduced range of motion and, often, pain at night [2].

Rotation suffers first in most people. Reaching behind the back, fastening a seat belt, or turning the hand outward with the elbow at the side all depend on the humeral head gliding and spinning freely in the socket, and a stiff capsule plus a rough surface limits exactly that.

The shoulder blade then compensates. When the glenohumeral joint stops moving well, the scapula tips and rotates more to get the hand where it needs to go [3]. That is a sensible short-term workaround, and it is also why neck and upper back aches often accompany shoulder arthritis. It is the mobility-stability model in reverse: a joint that should move has stiffened, and the structures around it are doing extra work.

Shoulder arthritis is not a single condition. Primary osteoarthritis usually leaves the rotator cuff intact. Rotator cuff arthropathy is arthritis that develops after a large, long-standing cuff tear, and it behaves differently, because the muscles that center the ball are no longer doing their job. The 2026 review included both [1]. If you have been told you have a cuff tear as well, the rotator cuff tear post covers that side of the picture.

X-ray findings and symptoms do not always match. Some people have marked joint-space narrowing and modest pain, and others the reverse. “Bone on bone” on a report describes the X-ray, not how much your shoulder can still do.

What the Evidence Supports

The evidence verdict for physical therapy in glenohumeral osteoarthritis is short. Garba and colleagues searched five databases up to June 2025 for randomized trials of physical therapist-led care for adults aged 45 and over with glenohumeral osteoarthritis or rotator cuff arthropathy [1]. Of 582 articles identified, four trials were eligible. All four tested care after surgery. None was placebo-controlled.

The 2026 evidence reviewA common treatment with almost no trials behind it

• 582Articles screened for physical therapy in shoulder arthritis
• 4Randomized trials eligible, all after surgery
• 0Trials of non-surgical physical therapy
• 0Trials with a placebo control

The one finding the authors could rate was low-certainty evidence that starting postoperative physical therapy early, within the first 6 weeks, made no difference to outcomes at any time point compared with starting after 6 weeks [1]. Everything else in the review was rated very low certainty. The authors concluded that the evidence base is “extremely limited and of poor quality” and that no published randomized trial has evaluated physical therapist-led non-surgical care for glenohumeral osteoarthritis.

My read is that this is a gap, not a negative result. Nobody has shown that physical therapy fails for shoulder arthritis. Nobody has properly tested whether it works. Those are different statements, and the second is the honest one.

So where does the advice to try exercise come from? From guidelines and expert reviews that reason from what is known about the shoulder rather than from shoulder-arthritis trials. The AAOS plain language summary of its guideline says exercise or physical therapy “may” help reduce pain and restore function [2]. A 2023 expert review describes the goals of physical therapy as reducing pain, increasing range of motion and protecting the joint, recommends rotator cuff strengthening for that last goal, and suggests therapy may help pain with movement more than pain at rest [3]. That is a reasonable clinical rationale. It is not the same as trial evidence, and it does not support claiming that any one program is proven best.

What Physical Therapy Can and Cannot Promise

What it can reasonably offer. A structured trial of exercise is low risk, inexpensive and keeps your options open. It can help you keep the range you still have, keep the rotator cuff and scapular muscles working so the joint is controlled through that range, and find which movements aggravate the shoulder and which do not. A good therapist will also help with practical modifications: how to reach a high shelf, how to sleep, how to keep lifting without provoking a flare.

What it cannot promise. It cannot rebuild cartilage, remove osteophytes or reverse what the X-ray shows. It cannot promise a particular amount of pain relief, because no trial has measured that for non-surgical care. And no one can honestly tell you that a specific named protocol is the evidence-based one for shoulder osteoarthritis, because that evidence does not yet exist.

Medication and injections. The AAOS notes that over-the-counter acetaminophen and NSAIDs may improve short-term pain or function, and that a corticosteroid injection may reduce pain and improve function for the short term [2]. Hyaluronic acid injections are where the evidence argues with itself. The AAOS states that strong evidence supports no benefit [2]. A later meta-analysis of 15 trials and 1,023 people reported pain improvement with hyaluronic acid, but found probable signs of randomization bias in 46.7% of the included studies [4]. A 2025 review of injection studies with at least six months of follow-up, covering 1,125 patients, described the benefit of hyaluronic acid as consistent but modest [5]. The fair summary is that the question is not settled.

Surgery. When non-surgical care does not provide adequate pain relief, the AAOS describes total shoulder replacement, or a reverse replacement when the rotator cuff is not working, as an option [2]. This post does not argue for or against surgery. It is a decision for you and a surgeon, based on how much the shoulder limits you. If you do have surgery, the 2026 review found no outcome difference between starting postoperative therapy early or after 6 weeks, at low certainty [1]. Follow your surgeon’s protocol.

Exercises to Try

These are not a proven program for shoulder arthritis. They are low-risk movements that match the goals described above. Work in a range that is uncomfortable at most, not sharply painful, and stop any exercise that leaves the shoulder clearly worse the next day. If your shoulder is in a flare, do only the first two.

Pendulum swings. Lean forward with your good hand on a table and let the affected arm hang straight down. Rock your body gently so the arm swings in small circles, then forward and back. The shoulder muscles should stay relaxed. One to two minutes, two or three times a day. This moves the joint without asking the cuff to lift the arm.

Supported flexion with a stick. Lie on your back holding a broom handle in both hands, shoulder-width apart. Use your good arm to lift the stick, and the affected arm with it, overhead as far as is comfortable. Hold for five seconds and lower slowly. Ten repetitions. Lying down steadies the shoulder blade, so the movement comes from the glenohumeral joint rather than from shrugging. Common fault: arching the lower back to get more height.

External rotation with a band. Stand with a light resistance band anchored at elbow height. Bend the elbow to 90 degrees and tuck it against your side, with a rolled towel between elbow and ribs. Rotate the forearm outward against the band, then return slowly. Two or three sets of 10 to 15. You should feel the back of the shoulder working. If the elbow drifts away from your side, the deltoid is taking over. The rotator cuff exercise post explains how to progress this.

Scapular setting. Sitting or standing tall, draw the shoulder blades gently back and down, as if sliding them into your back pockets, and hold for five to ten seconds without shrugging or arching. Ten repetitions. This works the muscles that position the socket under the ball.

Wall slide. Face a wall with your forearms on it, elbows bent. Slide the forearms upward as far as is comfortable, keeping the shoulders down, then slide back. Ten repetitions. Stop below the point where grinding becomes painful.

Skip or modify any of these if they reproduce sharp pain, if you have had recent shoulder surgery and your surgeon has given different instructions, or if your shoulder has recently dislocated. The shoulder pain relief post covers other causes worth ruling out if your pattern does not fit arthritis.

The following is an invented scenario. Someone in their late sixties with an X-ray report of glenohumeral osteoarthritis starts with pendulums and stick flexion, adds band rotation after two weeks, and keeps a simple note of which daily tasks hurt. After eight weeks the X-ray is unchanged, but reaching the top cupboard hurts less and night pain has eased. Another person follows the same plan and notices little change. Both outcomes are possible, and nobody yet knows how often each happens.

When to See a Doctor

See a doctor or physical therapist if:

• Pain or loss of movement is new and has not been assessed, so that other causes such as a rotator cuff tear or frozen shoulder can be considered (the frozen shoulder post covers how it differs)
• The shoulder is hot, red and swollen, or you feel feverish, which can signal infection and needs prompt assessment
• Pain began after a fall or injury, or the arm suddenly cannot be lifted
• You have numbness, tingling or weakness in the arm or hand
• Pain at night or with daily tasks is significantly limiting your life despite a reasonable trial of non-surgical care

When to see a doctor. If joint pain follows a significant injury, is accompanied by swelling, giving way, locking, or loss of range of motion, or if you have neurological symptoms (numbness, tingling, weakness), see a doctor or physical therapist. They can assess whether imaging or specialist input is appropriate.