You went in because your shoulder hurt. You came out with an X-ray, a word, and a piece of paper.
The word was arthritis. The paper had a surgeon's name on it. Somewhere in between, someone said "you could try physical therapy first," and nobody explained what that actually meant.
So now you are sitting at your kitchen table with a question that sounds simple and is not. Is physical therapy worth doing? Or is it just a few months of exercises before you end up having the replacement anyway?
You deserve a straight answer to that. Here it is.
We are a physical therapy clinic. So we are going to be more careful with this answer, not less.
For shoulder arthritis specifically, the direct trial evidence for non-surgical physical therapy does not exist yet. Not "is weak." Does not exist. A systematic review published in 2026 searched the literature through June 2025, screened 582 articles, and found four randomized trials. All four studied people who had already had surgery. Its conclusion is one sentence: "No published RCTs have evaluated physiotherapist-led interventions for patients with GHOA undergoing non-surgical care."
The professional guidelines still recommend trying it. Both the surgeons' guideline and the physical therapists' guideline say a course of physical therapy may help. But both say it as expert opinion, with no trial evidence behind it, not as proof.
The shoulder replacement evidence, by contrast, is strong. Large registries and long surgical series show that anatomic total shoulder replacement relieves pain, restores a lot of motion, and lasts a long time in most people.
So here is the honest version. Physical therapy for shoulder arthritis is a reasonable first step, recommended by the people who would perform your surgery, but it is not a proven alternative to a replacement. It is a fair, low-risk trial you can run before you commit to an operation you cannot undo.
The rest of this article is about how to run that trial fairly, and how to know what it told you.
Your shoulder has two joints that can wear out. The big ball-and-socket joint is the glenohumeral joint. The smaller one on top, where your collarbone meets the tip of your shoulder blade, is the acromioclavicular joint, usually shortened to the AC joint.
Arthritis here means the smooth cartilage that caps the ends of those bones wears away. The American Academy of Orthopaedic Surgeons describes it as a condition that "destroys the smooth outer covering (articular cartilage) of bone," and says it "usually affects people over the age of 50." AAOS also notes it is more common in the AC joint, though AC arthritis often causes fewer symptoms.
Where it hurts is a clue. AAOS puts glenohumeral pain "centered in the side or back of the shoulder," and AC joint pain "focused on the top of the shoulder." Grinding or clicking is common. So is night pain, and so is losing range of motion.
A Japanese research team X-rayed both shoulders of 541 residents of a single village, aged 40 to 89, and graded the images. They found shoulder osteoarthritis in 17.4 percent of them.
Age mattered a great deal. In residents 65 and older, the rate was 20.3 percent. Under 65, it was 11.1 percent.
Two limits matter. This was one village in one country. And it counted joints that looked arthritic on film, not people who were in pain. Those are not the same group, which is a theme you will meet again below.
Not every stiff, aching shoulder after 50 is arthritis. Rotator cuff problems, frozen shoulder and a pinched nerve in the neck all overlap, and each is treated differently.
If that is not settled yet, start with our guide to why you cannot lift your arm overhead after 50 and its two-minute self-test. If the word you keep hearing is "frozen," read our frozen shoulder guide. This article assumes arthritis is confirmed.
This is the part most articles skip. It is also the part that answers your question.
The AAOS published its clinical practice guideline on shoulder arthritis in 2020. On physical therapy before surgery, it says this:
"In the absence of reliable evidence, it is the opinion of the work group that physical therapy may benefit select patients with glenohumeral joint osteoarthritis."
Its strength rating is Consensus, the lowest rung.
In guideline language, that means the expert panel looked for studies, did not find usable ones, and wrote down their own opinion so that clinicians are not left with nothing. It is honest, and it is genuinely useful. It is not proof.
The same guideline names a gap it wants research to fill: whether physical therapy and other non-surgical care work as an alternative to joint replacement. The surgeons themselves are saying nobody has answered your exact question.
In 2023, the American Physical Therapy Association put out its own guideline. Its advice on care without surgery reads:
"In the absence of high or moderate quality evidence, the opinion of the GDG based on clinical expertise is that physical therapist services may benefit patients with GHOA who have not undergone TSA."
Evidence Quality: insufficient. Recommendation Strength: best practice. Included studies: zero.
Read that last line again. Our own profession's guideline, recommending our own profession's services, is built on zero included studies.
The same guideline adds that "no one specific intervention performed by a physical therapist is superior to another for patients with GHOA." Anyone who tells you their particular technique is the proven one for an arthritic shoulder is going beyond what the guideline supports.
Worth noting: the guideline's strongest advice on treatment is about care after a shoulder replacement, a sling combined with progressive exercise. The evidence in this field is mostly about rehab after surgery, not instead of it.
We would be leaving something out if we stopped there.
In 2024, researchers studied a national insurance claims database covering 2,710,463 patients with a diagnosis of shoulder arthritis between 2010 and 2021. They compared people who had physical therapy within a year of diagnosis against people who did not, and counted who went on to a total shoulder replacement.
The physical therapy group had more replacements, not fewer. At five years, among people who had not had a steroid injection, 0.78 percent of the physical therapy group had a replacement versus 0.37 percent of the no-therapy group. The authors' conclusion was blunt:
"Individuals who had physical therapy as a part of their nonsurgical treatment did not have any decrease in the probability of requiring TSA."
Three things have to be said about that, and all three are true at once.
First, it does not mean therapy caused surgery. This is a records study, not an experiment. Nobody was randomly assigned. People with worse shoulders are more likely to be sent to physical therapy and more likely to end up in an operating room. The therapy may simply be a marker of a worse shoulder.
Second, the authors themselves list the limits. The data comes from billing codes never designed for research. Coding rules changed over the years. And they state plainly that their design left them unable to evaluate severity, pain, physical activity, or quality of life. They call their own result a "narrow perspective."
Third, and most usefully: this study measured surgery, not how people felt. It cannot tell you whether the therapy group slept better, reached higher, or hurt less. It only counted operations.
So the honest lesson is not "physical therapy does not work." It is narrower and more useful than that. Do not start physical therapy in order to avoid a replacement. Start it to feel and function better, and to make the surgical decision from a better position. If someone promises you therapy will keep you out of surgery, they are promising something no study supports.
There is one more thing worth knowing, because it is genuinely good news.
A Nordic research team is running a randomized trial called ProAct. It enrolls people with shoulder arthritis who are already eligible for a replacement. Each person is assigned at random to the operation with standard rehabilitation, or to a 12-week exercise program. The published protocol says the trial needs 102 patients, and the main measurement is a shoulder score at 12 months.
The trial registry lists it as completed, with data collection finishing in May 2026. The results were not published as of the writing of this article. When they land, the honest answer in this section may change, and we will update it.
Given all of the above, that is a fair challenge. Here is the case, with its limits attached.
The UK's National Institute for Health and Care Excellence publishes a guideline on osteoarthritis, NG226, which covers people over 16 and all affected joints. It is not shoulder specific, and that scope matters.
It names the core treatments for osteoarthritis as "therapeutic exercise and weight management (if appropriate), along with information and support."
It also sets an expectation you should hear before you start. NICE tells clinicians to "advise people with osteoarthritis that joint pain may increase when they start therapeutic exercise," while noting that sticking with an exercise plan over the long term increases the benefit.
That is borrowed evidence, mostly from hips and knees. Borrowed evidence is weaker than direct evidence. It is also the reason both guidelines above still land on "try it."
The APTA guideline describes the clinical picture of shoulder arthritis as "a global reduction in range of motion (ROM) with the greatest loss in passive external rotation with the arm at the side," confirmed on X-ray.
A sensible plan works on that picture directly:
A Danish team ran a 12-week program in 18 people, average age 69, all already eligible for a total shoulder replacement. Three sessions a week, one supervised and two at home with elastic bands, progressed according to how much pain each session produced.
Nobody had an adverse event. Attendance at supervised sessions averaged 89 percent. Shoulder scores improved, active flexion rose by about 17 degrees, and the number of people whose pain disturbed their sleep fell from 17 out of 18 to 9 out of 18.
Now the honest part. This was a single-arm feasibility study, 15 of the 18 had shoulder arthritis and 3 had a different diagnosis, and the authors wrote that without a control group "it is not possible to distinguish between effects that might have occurred due to the natural cause of the condition, regression to the mean, placebo, or the exercise intervention."
What it does establish is worth having. A structured, progressive shoulder program is safe and doable in exactly the population that is deciding about surgery. Nobody got worse.
A study followed 129 people over 65 with shoulder arthritis for three years while their physicians treated them without surgery. Treatments were mixed and chosen case by case, including anti-inflammatory medication, injections and education, with only some patients receiving physical therapy. So this is not a physical therapy trial, and there was no comparison group.
At three years, most patients scored better than at the start on pain and shoulder function. But the useful detail is the shape of the curve. Scores rose at three months, dipped at six and twelve months, then recovered.
The authors suggested conservative treatment "should be extended for longer than 12 months before the decision regarding shoulder arthroplasty is made."
Take that as a caution about timing, not proof of anything. If your shoulder is worse at month six than at month three, that is a known pattern, and not automatically a reason to give up.
An honest article has to include the treatments that are not worth your time or your money, including some that are sold enthusiastically.
Hyaluronic acid injections. The AAOS position is unusually firm: strong evidence supports that there is no benefit to hyaluronic acid for glenohumeral osteoarthritis. NICE goes further and says do not offer them for osteoarthritis at all.
There is a wrinkle worth knowing. A 2019 review and meta-analysis found real pain improvement after these injections, about 26 mm on a 100 mm scale at three months. But the comparison groups improved by similar amounts, and the authors concluded these improvements "may not be directly related to HA" and that "a significant placebo effect may be present." People do feel better after the needle. The contents do not appear to be why.
Injectable biologics. AAOS states that injectable biologics such as stem cells or platelet-rich plasma cannot be recommended for this condition. That is a consensus statement, meaning the supporting evidence is not there.
Electrotherapy of all kinds. NICE recommendation 1.3.9 says do not offer any of the following because there is insufficient evidence of benefit: "transcutaneous electrical nerve stimulation (TENS), ultrasound therapy, interferential therapy, laser therapy, pulsed short-wave therapy, neuromuscular electrical stimulation (NMES)." AAOS separately lists TENS among the treatments its work group "cannot recommend for or against", a consensus statement.
We use a device called the Neubie in our clinic. It is an FDA-cleared Class II device that uses direct current and is not a TENS unit. Its clearances cover things like neuromuscular re-education and chronic pain management.
It is not cleared for arthritis, and it is not the answer to a worn shoulder joint. We take the NICE electrotherapy recommendation seriously. It includes neuromuscular electrical stimulation, and we do not offer any electrical device as a treatment for shoulder arthritis.
Acupuncture, dry needling and glucosamine. NICE says do not offer acupuncture or dry needling to manage osteoarthritis, and do not offer glucosamine.
Braces, straps and supports. NICE 1.3.11 says do not routinely offer insoles, braces, tape, splints or supports unless three conditions are all met: there is joint instability or abnormal loading, and exercise is ineffective or unsuitable without the device, and the device is likely to improve movement and function. A shoulder strap is not a default.
Hands-on treatment as the main event. This one constrains our own profession. NICE says only consider manual therapy for people with hip or knee osteoarthritis, and only alongside therapeutic exercise. It adds that "there is not enough evidence to support its use alone for managing osteoarthritis." If a plan for your shoulder is mostly someone else's hands and a table, ask what the exercise component is.
The honest answer is that the shoulder-specific evidence is thin, and we will not dress it up.
The AAOS guideline says nothing at all about steroid shots for shoulder arthritis treated without surgery. NICE, covering all joints, says consider an injection when other medication is ineffective or unsuitable, or to support therapeutic exercise, and tells clinicians to "explain to the person that these only provide short-term relief (2 to 10 weeks)."
A small prospective study of 30 shoulders in 29 patients, average age 66, tracked people after a single image-guided injection and found meaningful improvement lasting up to around four months. There was no comparison group, so some of that may be the same placebo effect the hyaluronic acid review flagged.
A separate records review of 311 shoulders at one institution found that 37.3 percent went on to a replacement within three to eight years of the injection. That number is often quoted as if it means something about how well injections work. It does not. Those authors state that they collected no data on pain or function at all.
So: an injection may buy you a window of comfort. It is not a repair, and it is not a shortcut. The most defensible use of one is exactly what NICE describes, as support for doing the exercise.
This section exists because a good decision needs both sides argued properly.
NICE sets the referral rule plainly. Consider referral for joint replacement, and it names the shoulder by name, if joint symptoms such as pain, stiffness or reduced function "are substantially impacting their quality of life" and non-surgical management "is ineffective or unsuitable."
It adds a line that matters for anyone over 50 who has been made to feel like a poor candidate. Do not exclude people from referral because of "age, sex or gender, smoking, comorbidities, overweight or obesity."
AAOS puts it more simply: your doctor may consider surgery "if your pain causes disability and is not relieved with nonsurgical treatment."
A series of 377 consecutive anatomic total shoulder replacements, all done for shoulder arthritis in patients aged 70 or older, reported results after an average of 3.3 years.
That is a well-performing operation. It is also a surgical series reviewed after the fact rather than a randomized trial, and the average follow-up was a little over three years, so it describes the early years rather than a lifetime.
Complications were not zero. There were 3 medical complications, 10 minor surgical complications and 5 major surgical ones. Five people (1.3 percent) developed a later rotator cuff tear.
A Nordic joint registry covering Denmark, Norway and Sweden tracked shoulder replacements performed between 2004 and 2013. At ten years, cumulative implant survival was 0.96 for anatomic total shoulder replacement (2,340 procedures), 0.93 for stemmed hemiarthroplasty and 0.85 for resurfacing hemiarthroplasty. For patients under 55, every figure was lower, with anatomic total shoulder replacement falling to 0.87.
One clarification, because this is easy to misread. Implant survival means the implant had not been replaced again. It is not a measure of how satisfied the patient was.
AAOS reaches the same point from the evidence side, saying with strong evidence that anatomic total shoulder replacement gives better function and pain relief in the short to mid term than a partial replacement (hemiarthroplasty).
The AAOS guideline is unusually specific about who does well, which is useful when you are thinking about timing.
That fourth point is the one people miss. If your shoulder still works reasonably well, you have less room to improve, which is an argument about timing rather than an argument against surgery.
That is the real shape of this decision. Shoulder replacement is a well-evidenced operation with good results. The question is not whether it works. The question is when, and whether you are in the best shape to have it. Our guide to joint replacement for arthritis goes deeper on the operation itself.
Most people do not give physical therapy a fair test. They go a few times, feel unsure, drift away, and are left with no information. Months later they are making the surgical decision on the same hope and fear they started with.
Here is how to run the trial so it tells you something either way.
A reasonable window is 8 to 12 weeks. That is a clinical convention, not a study result, and we will say so. It is long enough for strength to change and short enough that you are not drifting. The Danish program ran exactly 12 weeks, and the three-year cohort is a reminder that looking further out is also reasonable.
Not "feel better." Goals with an edge on them, chosen on day one:
Write them down. Note where you are today on each, in plain terms. "Cannot reach the shelf at all." "Wakes me twice most nights."
Not a vague intention to reassess. A date. On that date you sit down with your therapist and compare against the list you wrote in week one.
Our guide to telling whether your physical therapy is actually working covers how to judge a change that is real versus a change that is noise, and what to do at week two, four and six if nothing is moving.
This is the step almost nobody takes, and it is the one that makes the trial worth running.
One more thing. If your surgeon has already recommended an operation and given you a date, this is not a reason to postpone it. Talk to them first. A trial of conservative care is something you plan with your surgical team, not around them.
When you get to the surgeon's office, these are worth asking. Write the answers down.
That last question is often the most valuable one. Our guide to deciding between surgery and physical therapy covers the wider decision across knees, hips, backs and shoulders, including what the comparison trials in other joints have found.
Most shoulder arthritis is not urgent. A few things are. The following are drawn from a review of shoulder assessment published in the British Journal of General Practice, which lists red flags that mean urgent investigation or referral.
Call 911 or go to an emergency department if:
Get seen the same day if:
Get it checked soon if:
This is one place where the law is on your side. Under Oregon statute ORS 688.132, a physical therapist has a legal duty to refer you out immediately when your symptoms need diagnosis or treatment by a medical provider, when physical therapy is contraindicated, or when the problem is outside what a physical therapist can treat. If we see something that needs a physician, sending you to one is not a courtesy. It is our obligation.
You may be the one doing this research, not the one with the sore shoulder. That is common, and it is genuinely useful. Two things help most.
Help them keep the record. Memory is unreliable over eight weeks. A parent who wrote down "cannot reach the top shelf, wakes twice a night" in week one has something concrete to compare against in week ten. A parent who did not will simply say "about the same," which is often not true in either direction.
Watch for what they will not mention. Sleep is the big one. Night pain is a very common feature of shoulder arthritis, and older adults often do not report it because they assume poor sleep is just age. Ask directly. Ask whether they have quietly stopped reaching into the back of a cupboard, or lifting a grandchild.
You can join a virtual visit from your own home, even from another state. Many families find that the easiest way to hear the same information at the same time.
HWY Physical Therapy is built around adults over 50. Dr. Raj Pusuluri, PT, DPT, sees patients at both of our clinics, and he will tell you honestly when a shoulder needs a surgeon rather than a therapist.
A first visit is a full evaluation. We measure your shoulder motion in both directions, test the strength around the joint, go through your imaging and your history, and ask what you have actually stopped doing because of this shoulder. Then we set the goals and the review date together, in writing.
A typical plan runs one or two visits a week at first, tapering as your home program takes over. Most of the work happens at home between visits, which is exactly how the Danish program described earlier was built. If you would rather start without driving anywhere, our Wellness Video Call is a virtual visit you can do from your kitchen table.
We are direct-pay, so no insurance is required and no referral is needed to start. For more, see a full walkthrough of a first physical therapy visit after 50 and a realistic guide to how many sessions people usually need.
No. Nothing in this article is a reason to cancel or postpone an operation your surgeon has recommended.
If you want to try conservative care first, that is a conversation to have with your surgeon, before you change anything. They know your images, your other health conditions and their own waiting list. A planned trial of physical therapy agreed with your surgical team is a reasonable thing. Quietly skipping a surgery date is not.
Increased soreness when you start is expected. NICE advises clinicians to tell people that "joint pain may increase when they start therapeutic exercise." In the Danish 12-week program, exercises were progressed by pain response, aiming to keep pain at 5 or below out of 10, and no adverse events occurred in 18 patients.
Pain that keeps climbing week after week, or new night pain, is different. That is a reason to change the plan, not to push through it.
Give it 8 to 12 weeks with written goals and a fixed review date. That window is a clinical convention rather than a study finding.
For context, the Danish program ran 12 weeks. And in the three-year cohort of people over 65 treated without surgery, scores rose at three months, dipped at six and twelve months, then recovered, which is why those authors suggested extending conservative treatment beyond twelve months before settling the arthroplasty question.
It is a window, not a shortcut. NICE tells clinicians to explain that intra-articular corticosteroid injections "only provide short-term relief (2 to 10 weeks)." A small shoulder-specific study of 30 shoulders found improvement lasting up to around four months, with no comparison group.
The AAOS shoulder guideline makes no recommendation on them either way. The most defensible reason to have one is to make it possible to do the exercise.
No study shows that a sensible, step-by-step exercise plan speeds up shoulder arthritis. NICE names therapeutic exercise as a core treatment for osteoarthritis, alongside weight management where relevant, and notes that long-term adherence increases the benefit.
Be clear about what exercise does and does not do, though. There is no cure for arthritis of the shoulder, in the AAOS's own words. Exercise does not regrow cartilage. It changes how well the muscles around the joint support it, and how much you can do with the shoulder you have.
Nobody can promise you that, and we will not.
The largest analysis available, covering more than 2.7 million patients, found that people who had physical therapy did not have a lower probability of eventually needing a total shoulder replacement. That study could not measure pain or function, and could not account for how bad each shoulder was to begin with, so it does not prove therapy is useless. It does mean avoiding surgery is the wrong reason to start.
Start because you want your sleep back, your reach back, and a clear answer about your own shoulder within a couple of months.
If you have been told you have shoulder arthritis and you are stuck between "try therapy" and "book the surgery," a proper evaluation gives you something better than a guess. You will leave with measured range of motion, written goals and a review date, so that in eight weeks you will know what your shoulder actually did.
Book a Wellness Video Call or call (971) 202-1979. Direct-pay, no insurance required, and no referral needed to get started.
HWY Physical Therapy
North Salem 2615 Portland Rd NE, Salem, OR 97301 (at Center 50+) Walk-ins welcome
South Salem HWY Physical Therapy Clinic South Salem 180 Ramsgate Square S, Salem, OR 97302 By appointment only
Phone (both locations): (971) 202-1979 Hours: Monday to Friday, 8:00 AM to 5:00 PM
This article is general information about a health condition. It is not a diagnosis and it is not medical advice for your situation. It is not a recommendation to decline, delay or cancel any treatment or surgery your doctor has recommended. Please talk to your own clinician about your own shoulder.