The GLA:D Program: What It Is, and What the Evidence Says

If you have arthritis in a knee or a hip in Canada, the sequence is familiar enough to be a joke. Somebody tells you to lose weight and stay active. Nobody says what active means, or how you are supposed to do it on a joint that hurts when you cross a parking lot. An X-ray gets ordered. A year or two later you are on a list for a new joint, and the list is long — in the most recent national figures, 69% of hip replacements and 63% of knee replacements were done inside the 182-day benchmark, which is another way of saying that roughly a third of people waited longer than half a year after the decision was made.

GLA:D is the thing that is meant to go in that gap, and it is now running at more than four hundred sites across the country. It is also, unusually for anything in this field, a program that has been studied hard enough to argue about — including by people who think the numbers on its own website are flattering. Both of those are worth knowing before you phone a clinic.

What the acronym is hiding

GLA:D stands for Good Life with osteoArthritis: Denmark, which is a strained acronym for a plain idea. It began in Denmark in 2013, and the paper describing why is refreshingly blunt about the problem it was built to solve. Every clinical guideline in the world already said the same thing: for hip and knee osteoarthritis, education and exercise come first, before drugs and long before surgery. And only 36% of Danish patients with osteoarthritis were getting that care.

So GLA:D is not a treatment. It is a delivery system for a treatment that everybody already agreed on and almost nobody was receiving. That distinction explains most of what is odd about it — why it has a trademark, why clinicians have to be certified to run it, and why the whole thing is wrapped around a database.

There are three parts, and all three are mandatory. Clinicians take a two-day course on a fixed protocol. Patients get a fixed program: education sessions, then twelve supervised exercise sessions. And every participant goes into a registry, measured before the program, at three months and at a year. The registry is not an afterthought — it is the reason we know anything about how the program performs outside a trial, and it is also, as we will get to, the reason its published results have to be read carefully.

It arrived in Canada in 2017, licensed to the University Health Network in Toronto, and it has spread further here than anywhere outside Denmark.

What actually happens

Before anything else there is an assessment. A trained therapist checks that what you have is osteoarthritis in a hip or knee and that it is the sort likely to respond, then takes a set of measurements: pain on a 0 to 10 scale, a 40-metre walk test, how many times you can stand up from a chair in 30 seconds, and a quality-of-life questionnaire. Those same measurements get repeated later, which is more than most treatments in this area bother to do.

Then two education sessions, 45 minutes each. What osteoarthritis is and is not, what is actually happening in the joint, why loading it is safe, what the exercises are for. One of the two is co-delivered by somebody who has osteoarthritis and has done the program. That is a structural choice rather than a nice touch: the single hardest thing to accept about this diagnosis is that the joint you have been protecting needs to be worked, and that lands differently from a person who has been where you are than from a clinician who has not.

Then twelve exercise sessions, twice a week for six weeks, an hour at a time, in a small group. Each hour has the same shape — ten minutes warming up on a stationary bike, a circuit of four stations covering core stability, postural alignment, leg strength and functional movements at two or three sets of 10 to 15 repetitions, then ten minutes of walking drills and stretching to finish.

The word the program uses for the middle part is neuromuscular, and it is not marketing. The level you are given is set by how well you control the joint, not by how much weight you can move. A knee that collapses inwards on a step-down gets an easier variation until it does not, and the point of the whole circuit is the quality of that movement rather than the load. This is why the program can be run on people whose knees hurt: nobody is being asked to grind out a heavy set.

Pain during the sessions is expected and has a ceiling. Participants are taught to work in a range they would call 0 to 5 out of 10, during the session and afterwards. Above that, the exercise gets regressed. It is a small piece of the education and probably the most useful single thing in it, because the alternative most people have been running on for years is to stop at the first twinge and never find out what the joint can do.

Who is actually in it

A profile of everyone who went through GLA:D Canada between 2017 and 2022 gives an unusually clear picture of who this program is reaching. There were 15,193 registrations — 11,228 knees and 3,965 hips. The average participant was 66 years old, and about three-quarters were women. Average pain was 5 out of 10. Around two-thirds were taking something for it. And before starting, 30.6% of the knee participants and 35.3% of the hip participants said they wanted joint surgery.

Geographically it is lopsided. Ontario supplied around 60% of participants and Alberta around a quarter, British Columbia about 7%, and Quebec is essentially absent from the registry. By the 2023 annual report the program counted 402 sites across nine provinces and two territories, and 2,015 clinicians trained, three-quarters of them physiotherapists.

That profile is worth holding onto while you read the results, because it is the population the results describe. Moderate pain, mid-sixties, mostly women, mostly carrying extra weight, mostly already medicated, a third of them already contemplating a surgeon. If your joint is much worse than that, or much better, the numbers below were not measured on you.

The numbers everyone quotes

The headline result comes from a pooled analysis of 28,370 participants across Denmark, Canada and Australia. Immediately after the program, average pain intensity had fallen 26 to 33%, walking speed had improved 8 to 12%, chair stands 18 to 30%, and quality of life 12 to 26%. Roughly half of participants met the threshold for a clinically meaningful improvement in pain. The pattern held across three countries with quite different health systems, which is a real finding in itself.

GLA:D Canada’s own reported results are in the same territory: 55.8% of hip and 58.3% of knee participants improved on pain, 51% and 56.4% on function, and 57.6% and 64.4% on quality of life.

The most informative account of what that feels like is not a number. Researchers interviewed 30 Canadian participants, over half of them rural, and what they described was specific and small and exactly the sort of thing that does not show up on a pain scale. Stairs. Getting off the toilet. One person had gone hiking that summer, having not managed it the summer before.

The same study is the reason to trust the rest of it: 23% of the people interviewed said they got nothing out of it. Their reasons were consistent — the arthritis was too advanced, the exercises hurt more than they helped, the thing they had hoped for did not happen. “I was too far gone for the GLA:D program,” one of them said. Their collective suggestion was that people should be sent earlier, which is probably the single most actionable sentence in the whole literature.

Why those numbers are weaker than they look

Here is the problem, and it is not a small one. Almost everything in the section above comes from a registry, and a registry has no control group. Everybody in it did the program. Nobody in it did nothing, or did something else, or was told to come back in eight weeks.

At least three different things would produce exactly the graph the registry produces.

Regression to the mean. Osteoarthritis pain fluctuates in waves, and people sign up for programs during the bad ones. Measure a group of people on the worst month they have had in a year and measure them again eight weeks later, and they will look better, because the average week is better than the worst week. This is not cynicism; it is arithmetic, and it contaminates every before-and-after measurement ever taken in pain care.

Attention. The program is fourteen supervised appointments over a couple of months with a professional who is interested in your knee, in a room with other people who have the same problem. Nobody has ever managed to design a treatment for pain in which that part contributes nothing.

The exercise itself. Which is the part everyone assumes is doing the work.

A single-arm registry cannot separate those three, and a 2025 paper in the Journal of Rheumatology made the point about GLA:D directly and unsparingly: studies of this design are not built to support causal claims, and the finding that people with milder symptoms improve most is exactly what you would expect from natural fluctuation rather than from a program working best on them.

The trial that put a number on it

There is one study that gets around all of this, and it is the most interesting thing in the field.

The DISCO trial randomised 206 Danish adults over 50 with symptomatic, X-ray-confirmed knee osteoarthritis into one of two arms. Half did the eight-week exercise and education program. The other half received four injections of sterile saline into the knee over the same eight weeks — salt water, openly described as an inert placebo, delivered by a needle into the joint.

At nine weeks, pain had improved by 10.0 points in the exercise group and 7.3 points in the saline group, on a 0 to 100 scale. The difference was 2.7 points, with a confidence interval running from −0.6 to 6.0, and the trial’s formal test for equivalence passed comfortably. A year later the two groups were still indistinguishable.

There are two honest readings of that, and the argument between them is the most useful thing you can know about this program.

The deflationary reading is the obvious one. Six weeks of supervised exercise did not beat four squirts of salt water, so whatever is producing the improvement is mostly not the exercise.

The other reading, argued in a companion editorial in the same journal, is that an injection into a joint is not a fair definition of nothing. Injected placebos are among the most powerful in medicine — considerably stronger than a placebo pill — and matching one is not the same as failing. The editorial also points out what the two arms did not have in common afterwards: the exercise group needed fewer additional treatments and were more likely to still be exercising on their own. A syringe does not leave you with a routine.

Zoom out one more level and the picture stays honest but stays modest. The 2024 Cochrane review of exercise for knee osteoarthritis pooled 139 trials and 12,468 people. Against no treatment, exercise improved pain by around 13 points on a 0 to 100 scale and function by a similar margin — at low to moderate certainty, with benefits the reviewers described as of uncertain clinical importance, and with 94% of the included trials at unclear or high risk of bias on the domains that matter most. The reason is structural and unfixable: you cannot blind a person to whether they spent six weeks in a gym.

So the fair summary is this. Exercise for hip and knee osteoarthritis is the most strongly recommended treatment in the field, and its measured effect is real, modest, and less precisely known than the confidence of the recommendations suggests. GLA:D is a well-organised, consistently delivered, properly measured way of receiving it. The registry percentages are the ceiling of what it does, not the estimate.

That is still a better hand than most things sold for arthritis pain. It is not the hand the brochures deal.

What it will not do

It will not change the joint. Nothing in the program regrows cartilage or reverses anything visible on an X-ray, and the education sessions say so plainly, which is to their credit.

It will not reliably cancel a joint replacement, and the way this gets sold is the part worth being careful about. Denmark’s first-year data showed meaningful drops in painkiller use — 19.2% among knee participants by three months — and a 9.4% reduction in sick leave over the following year. Those are good numbers. They are not the same as surgery avoided, and a third of Canadian participants walk in already wanting an operation.

The more useful framing is that GLA:D is a way of finding out whether you need surgery yet. Six weeks of properly supervised loading tells you something an X-ray cannot: how much of your pain and disability belongs to the joint and how much belongs to a leg that has been getting weaker for three years because the joint hurt. If the answer turns out to be that you need the operation anyway, you arrive at it stronger, which is not a consolation prize — it is most of what determines how the first three months afterwards go.

Getting into one in Canada

Start at the GLA:D Canada locations page, which lists participating sites with phone numbers. Phoning is the actual step, because the three things you need to know all vary by clinic and none of them are on the website.

• Whether you need a referral. Usually not — most clinics take self-referrals — but confirm rather than assume.
• What it costs. This is the big variable. The program is publicly funded at select clinics in every Alberta Health Services zone, and runs out of some hospital and university clinics elsewhere. Privately, Canadian clinics commonly land somewhere between $300 and $500 for the full program, sometimes quoted per session in the $35 range across the fourteen or so visits. Ask whether the initial assessment is billed on top, because it often is.
• When the next group starts. It runs in cohorts, so there may be a wait of weeks, and a clinic that can start you tomorrow may be running something that is not quite the program.

If you have extended health benefits through work or a pension, this is ordinarily billed as physiotherapy and comes out of that annual cap — check the cap first, because fourteen visits will eat a modest one whole, and if you also claim massage or orthotics against the same pot you want to know before you book. There is no provincial health plan that covers it as a physician visit.

There is also a virtual version, run as a live group over video with the therapist watching everyone on screen. Any phone, tablet or laptop with a camera will do. For people two hours from the nearest participating clinic — and remember that over half the participants in that Canadian interview study were rural — it is the difference between the program existing and not.

A few people are screened out, and it is better to know why in advance than to be turned away at the assessment. The program is for osteoarthritis specifically, so another cause for the hip or knee problem — a soft tissue injury, a fracture, an inflammatory arthritis — rules it out. So does having other symptoms more prominent than the joint: chronic widespread pain or fibromyalgia is named explicitly, not as a judgement but because the program was built and tested on a different problem. And the sessions run in English, so you need enough English to follow them or someone with you who can bridge it.

If your problem is a back rather than a hip or knee, GLA:D Back is a separate program built on the same model for persistent or recurrent back pain, run through the University of Alberta and now expanding nationally.

Giving it a fair test

Six weeks is a real commitment of time and usually of money, so it is worth setting it up to actually tell you something.

Write down your own baseline before the first session, in your own terms, alongside whatever the clinic measures. How many times you can stand up from a kitchen chair in thirty seconds. How far you get before you have to stop. The one thing you have given up that you most want back — the stairs, the garden, a particular walk. Those are what you will be comparing against at the end, and a memory of how bad it was beforehand is not a measurement.

Go to all twelve. This is the whole of the advice and it is where the program is most often wasted. Twice a week for six weeks is the dose that was studied, and the people who dip in and out are testing something else.

Use the pain ceiling honestly. Working at 4 out of 10 during a session is fine and expected; so is being sore that evening. What is not fine is being worse the next morning, repeatedly, and saying nothing because you do not want to be the one holding the group up. Tell the therapist — regressing an exercise is a normal part of the protocol, not a failure, and quietly pushing through is how a six-week program turns into a flare-up.

Fill in the three-month and twelve-month questionnaires when they arrive. They are how anybody knows whether this works, and the honest answer from someone it did nothing for is worth more to the next person than another satisfied one.

And treat the twelve sessions as the training, not the treatment. The program ends; the arthritis does not. What you should walk out with is a small set of movements you know how to do at the right level and a clear idea of what to do when it flares — and then something to keep the leg loaded afterwards, whether that is a walking programme you build up slowly, a pool, or the same circuit at home twice a week. The registry’s twelve-month numbers hold up better than you would expect for people who kept going, which is the least surprising finding in this entire article.

The strongest thing GLA:D has going for it is not the size of its effect. It is that somebody finally organised the delivery of the treatment everyone had been recommending and nobody had been giving.

If you want the wider picture of managing an arthritic joint day to day — medication, heat, pacing, the rest of it — our overview of arthritis pain management covers the ground around this program, and Arthritis Society Canada can point you at services in your province if there is no GLA:D site within reach of you.