[Dr. Jared Powell] (0:11 - 21:55)
Hello, it's Jared.
Today, we contend with the question, is the dose of exercise important for musculoskeletal pain? Is there a preferred number of sets, reps, frequency, intensity, duration, a veritable sweet spot that if we could just identify it, would result in better outcomes for our patients? The answer, as best we can tell across multiple conditions and multiple studies is no.
There does not appear to be a reliable dose response relationship between exercise prescription parameters and clinical outcomes for individuals with musculoskeletal pain, not for low back pain or shoulder pain or knee osteoarthritis to name a few conditions. The dose doesn't seem to make the poison or the cure. I think this is one of the more interestinsg findings in our field because it necessitates a deeper follow-up question.
Why would we expect a dose response relationship in the first place? And what does it mean that we can't find one? So let's start with the word itself, dose.
It's a pharmacological term. When a physician prescribes a medication, dose is absolutely vital. Too little and the drug doesn't reach therapeutic levels.
Too much and you get toxicity. There's a curve, what we call a biological gradient, where the relationship between how much you give and what happens is fairly predictable, as predictable as we can get in biology anyway. It's predictable, it's measurable, and often linear within a therapeutic window.
Exercise science has borrowed or stolen this language decades ago and for good reason. In the strength and conditioning world, for example, dose does matter. If you want to get stronger, there are well-established principles such as progressive overload, volume, intensity, frequency.
The recent ACSM guidelines or the American College of Sports Medicine guidelines position stand recently released a massive umbrella review. And they confirmed that variables like load and volume reliably influence outcomes like strength and hypertrophy in healthy adults. The best and most efficient way to get strong is to lift heavy on average.
In this circumstance, the intensity factor of the dose equation warrants serious consideration. But the metaphor seems to break down when it comes to pain. Strength and hypertrophy are not the same outcome as the experience of pain.
When we recommend exercise for someone with shoulder pain or low back pain or knee osteoarthritis, the primary outcome we care about is often pain reduction and or functional improvement. That's the outcome that patients care about as well. And the mechanisms through which exercise influences pain are almost certainly different from the mechanisms through which exercise influences muscle cross-sectional area or the neurological change that in part leads to enhanced strength.
So this is the error at the heart of the dose debate. We've smuggled in a framework from exercise science and pharmacology that works pretty nicely for tissue adaptation and various biomotor tasks and assumed it applies equally to pain. But pain is not a tissue output.
It's an experience and a subjective and private experience at that. It's influenced biology, yes, but also by context, by expectation, fear, self-efficacy, the therapeutic relationship, sleep, stress and beliefs about the body. None of those things have a linear dose response curve with sets and reps.
So let's briefly look at the evidence across three musculoskeletal conditions, starting with knee osteoarthritis. The most comprehensive data we have comes from a recent systematic review and meta-analysis by Lawford in 2024.
They included 100 trials, over 9,000 participants that separated exercise programs into those with a higher and lower compliance with the ACSM prescription guidelines. They looked at resistance exercise, aerobic exercise and combined programs separately. The result was the same across all three.
No difference in pain or function between higher and lower compliance programs. They also looked at individual dosage variables such as frequency, intensity and duration, and none of them independently influenced outcomes either. The authors then concluded that exercise dosage may not be an important factor for driving symptom improvement in knee osteoarthritis.
Moving on to rotatic up tendinopathy, a systematic review found only three trials that directly compared higher versus lower exercise dose. The evidence was low to very low certainty and conflicting. So please consider this when listening to the next couple of sentences.
One trial suggested higher load and volume might improve function in individuals with rotatic up tendinopathy, but not pain. Another study found no difference between higher and lower load at six weeks. The evidence base was way too thin to draw any firm conclusions about dose, but what was there didn't support a clear dose response relationship.
Moving on to low back pain, a Bayesian network meta-analysis of 82 trials over 5,000 participants tried something I think more sophisticated. Rather than splitting trials into crude binary categories of high or low dose, the authors modeled dose as a continuous variable using met minutes or metabolic equivalent minutes per week. And they did find a dose response relationship, but it wasn't linear, it was U-shaped.
Pain improved as exercise dose increased up to a threshold, roughly 920 met minutes per week. And then the benefit plateaued and began to decline. So what we're seeing is too little didn't help much, but too much didn't help either.
The sweet spot was actually in the middle. Now, isn't that a more interesting finding than the simple more is better conclusion or slogan? A U-shaped curve is not what you predict from a pharmacological dose model.
In pharmacology, you generally see a sigmoid curve, a threshold, then a steep rise, and then a plateau. You don't see the effect start to reverse at higher doses unless you're approaching toxicity. A U-shaped curve suggests something else is going on.
It suggests there might be a threshold. You need to do enough to have an effect, but beyond that threshold, additional dose or just doing more doesn't add benefit and perhaps may even subtract from it. This seems to me much more consistent with a model where the active ingredient may not be the mechanical stimulus, but something that accompanies the stimulus, things like engagement, routine, confidence, moving without fear.
You need enough exercise to activate these things to get the ball rolling, but you don't need more and more exercise to get more and more of them. There appears to be a ceiling and pushing past that ceiling might introduce other problems such as fatigue, pain, flare-ups, an exercise burden, and ultimately reduced adherence to the exercises that actually undermine the benefit. So if the dose-response relationship keeps not showing up in research, why do we keep looking for it?
I think there are a few reasons. The first is that it would make our jobs a hell of a lot easier if there were an optimal dose of exercise, we could say, right, three sets of 15, three times a week for 12 weeks, off you go. And no, that was the right answer.
Clinical practice becomes a lot more straightforward. The output follows the input in a predictable way. The second is professional identity.
Musculoskeletal clinicians like myself and perhaps like you have built a significant part of their professional identity around exercise prescription. We're seen as movement experts and quite rightly so. We're the ones who know how to design and dose a program.
If dose doesn't matter for the outcome we're most commonly trying to influence, which is pain, that's an awkward finding. The third is a mechanistic assumption that runs deep in many clinicians and patients. The dose-response expectation assumes that exercise reduces pain primarily through a biomechanical stimulus, stronger muscles, better motor control, or more stable joints, for example.
Under that model, dose should be important because biomechanical adaptations are dose dependent. More load, more adaptation, less pain. But what if the biomechanical stimulus isn't the main driver of pain relief?
What if the mechanisms that matter are things like improved pain self-efficacy, reduced fear of movement, the therapeutic relationship, the experience of doing something active and feeling capable, the contextual reassurance embedded in a supervised exercise program. None of those things scale linearly with simple sets and reps. You don't get twice as much pain self-efficacy by doing twice as many exercises.
In fact, I wrote about this in a recent British Journal of Sports Medicine editorial in 2026 at the start of this year, arguing that clinicians should reconsider the narrative that patients need to get stronger to feel better. The causal evidence that we have, the studies that try to actually identify the causal pathway between exercise and pain relief, doesn't support strength as a primary mediator. What keeps showing up instead are psychological variables, pain catastrophizing, kinesiophobia, pain self-efficacy, and these may not respond to simple slogans like just keep loading it and more, more, more.
So if you know me, hopefully you know I like to look at things from a number of different perspectives, and this includes steel manning an argument that potentially falsifies my own. So let's play devil's advocate on the argument I'm making in this podcast for a bit of fun. The argument goes something like this.
The reason we don't see a dose response relationship in the exercise literature is because we chronically under load our patients. We've been so cautious, so protective, and so afraid of flaring people up that the exercise programs tested in most clinical trials barely qualify as a noteworthy stimulus. In effect, what we've shown is that two sub-therapeutic doses produce similar results, which is exactly what you'd expect.
This is a good argument and I'm sympathetic to it. It's true that many exercise programs in the musculoskeletal pain literature are by strength and conditioning standards remarkably conservative. Low loads, limited progression, short durations, high rep ranges that wouldn't challenge a novice trainee.
If you come from a background in sports science or strength and conditioning, looking at the typical rotator cuff tendinopathy trial is a bewildering experience. The idea that three sets of 15 with a yellow TheraBand followed by five sets of wall crawls three times a week for eight weeks represents a significant bioactive loading stimulus is, to put it mildly, debatable. So maybe we've never properly tested high dose against low dose in a way that reflects what a real loading stimulus looks like.
Again, I think this critique has partial validity. We probably are too conservative with load in some clinical population. And there's a broader cultural tendency in the physical therapies to err on the side of caution that may not always serve the patient.
The idea that we should load tissues more confidently is, in many contexts, sound advice. And it's something I tend to follow in my own clinical practice. But I think this critique also has limits.
And here's where I'm gonna start to push back. First is the empirical problem. We do have some trials that have tested properly high doses against lower doses, and the results are mixed at best.
In rotator cuff tendinopathy, a well-designed randomized controlled trial found that adding a large dose of resistance exercise to usual care provided no additional benefit. Another found no difference between a comprehensive supervised heavy loading program and a simple home program. In knee osteoarthritis, several head-to-head trials have compared proper high-intensity and low-intensity resistance exercise and have found no meaningful difference in pain and function outcomes.
If underloading were the primary explanation for this lack of effect, you'd expect some effect when someone actually increases the load. This effect hasn't appeared yet. Second is the conceptual problem.
The underloading argument still assumes the biomechanical model, that the active ingredient is the mechanical stimulus and we just haven't given enough of it. But if the mechanisms of pain relief are predominantly nonspecific and contextual, then giving more mechanical stimulus won't necessarily produce more pain relief, no matter how high you push the load or the sets or the reps. You could design the most aggressive, progressive overload program imaginable and still not see a difference in pain outcomes if the thing that's actually reducing pain is the therapeutic relationship or the patient's sense of agency or the reduction of fear that comes from doing something active.
Third is the clinical reality. Not every patient wants to be loaded heavily. Not every patient can be loaded heavily.
And maybe not every patient needs to be. The person in front of you might be a 65-year-old with a rotator cuff problem and moderate fear of movement whose primary goal is to hang out the washing without wincing in pain. This person also happens to loathe exercise.
Telling that person they need, the operative word being need, to load progressively to 85% of their one rep max is not patient-centered care. It's the clinician's framework imposed on the patient's life. Sure, we could all benefit from some extra strength.
And what an absolute gift you could give someone, especially as we age. But just make sure it's what the patient wants. And by all means, you can emphasize that a bit of strength training may indeed be helpful for their general health status.
But saying it is the only way to reduce their pain is not in alignment with the evidence. So yes, we should probably be less afraid of load than we've historically been. But the idea that we just haven't loaded enough, that the dose-response relationship is there, we just need to push harder, is, I think, an auxiliary hypothesis that tries to rescue the model from falsification rather than engaging truthfully with the possibility that the model itself might need revising.
Before we go any further, let's talk about what the term dose actually means from an exercise perspective. What is a dose of exercise? So in pharmacology, the dose is precise.
You know exactly how many milligrams of the active ingredient are in the tablet. You also know the bioavailability. You know the half-life.
You can measure the serum levels after taking the medication. The concept of dose rests on the assumption that there is an identifiable active ingredient that you can control how much of it reaches the target. Now, if we apply this to exercise, what is the active ingredient?
Is it the mechanical load on the tendon? Is it the metabolic stress on the muscle? Is it the neurological stimulus, the cardiovascular demand, or the sense of accomplishment?
Even the interaction with the therapist, the patient's belief that they're doing something helpful. I could keep going all day here. All of these things happen simultaneously during a single set of shoulder abduction exercises.
And we are not sure which one or which combination is responsible for any given patient's improvement. So when we say higher dose exercise, what are we actually saying we're giving more of? More load, more volume, more time in the clinic, more attention, more of the active ingredient, which we haven't identified.
This is a fundamental problem. The dose response framework presupposes that we know what the active ingredient is and can titrate it. We can't.
We don't even know what it is with any certainty for most musculoskeletal pain conditions. And until we do, optimizing dose is optimizing a variable we haven't defined for a mechanism we haven't identified in a system that almost certainly doesn't behave like a simple input-output pharmacological model. I want to emphasize something here because I can feel the objection forming in you learned listeners.
I can already feel the knives coming for me. What the evidence suggests is that for the outcome of pain and function in musculoskeletal conditions, the specific prescription parameters such as sets, reps, load frequency, or even contraction type don't seem to be the primary drivers of improvement. That's different from saying exercise is unimportant.
Exercise clearly helps, and I'm bullish on exercise as a treatment for a broad spectrum of musculoskeletal pain conditions. The question is, what about exercise helps? And the answer is probably not the thing we've been obsessing over.
What might be more important is whether the person does the exercise consistently, whether they feel safe doing it, and also whether the program fits into their life and is in alignment with their goals and their preferences. What also may be important is whether the clinician explains it in a way that makes sense and doesn't frighten them, and whether the therapeutic relationship supports engagement over weeks and months. And most importantly, I think, whether the exercise experience builds confidence rather than reinforcing the belief that their body is fragile.
These are contextual and relational variables, and they're hard to put in a typical dosage equation. They're also much harder to standardize in a clinical trial, which might be part of why the exercise dose literature keeps coming up empty. We've been measuring the things that are easy to measure rather than the things that move the needle.
There's a useful heuristic that I proposed in a paper called, Is Exercise the Right Treatment for Rotated Calf-Related Shoulder Pain? The specificity of exercise type or intensity and contraction probably matters little for reducing pain, but it might matter more for enhancing function and performance. If someone needs to return to overhead sport or heavy manual labor, then yes, the biomechanical specificity of their exercise program should be considered in fine detail.
So where does this leave us? There probably isn't an optimal dose of exercise for musculoskeletal pain. That's freeing.
It means working with the person in front of you, rather than mindlessly chasing a guideline that may not be relevant to the outcome you're trying to achieve. I think the narrative needs updating too. You need to get stronger to get better is not well supported by the mediation evidence that we have.
Instead, exercise can help, and the type and amount that works best will depend on what you can tolerate, what you can enjoy, and what fits into your life is a more honest and probably more effective way of explaining the possible beneficial effects of exercise. I'll leave you with this. Exercise is not a tightly controlled and formulated drug.
So maybe it's time we stop treating exercise prescription as family logical medicine. Thanks for listening. I'm Dr. Jared Powell, and this has been another episode of the Shoulder Physio podcast. I'll see you next time. The Shoulder Physio podcast would like to acknowledge that this episode was recorded from the lands of the Tirribalang people. I also acknowledge the traditional custodians of the lands on which each of you are living, learning and working from every day.
I pay my respects to elders past, present and emerging and celebrate the diversity of Aboriginal and Torres Strait Islander peoples and their ongoing cultures and connections to the lands and waters of Australia.
