The short version
- Conservative care is not “do nothing”. It is a specific, supervised, progressive programme — and that is not what most people are given.
- The effect of exercise is real but modest. A large individual-participant analysis called it of questionable clinical importance beyond the short term.
- It still goes first, because nothing else in this space does better, and it is the only option with any prospect of changing the trajectory.
- People with worse pain and function benefit most — which is the opposite of how it is usually offered.
I have spent this section telling you what injections do and do not do. Ending on “try exercise instead” without applying the same standard to exercise would be the exact failure I have been criticising elsewhere.
So here is the evidence, on the same terms.
What does the evidence for exercise show?
Two large analyses, and they land in different places.
| Cochrane review, 54 studies | IPD meta-analysis, 4,241 participants | |
|---|---|---|
| Pain, short term | 12 points better than no exercise | 6.4 points better at 12 weeks |
| Function, short term | 10 points better | 4.5 points better |
| Sustained effect | 6 points at 2–6 months after stopping | 3.8 at 6 months; 3.4 at 12 months |
| Authors' framing | Short-term benefit, sustained 2–6 months | Small positive effect, “of questionable clinical importance” beyond the short term |
The second analysis is the more sobering and the more rigorous. Pooling individual participant data from 31 trials rather than published summaries, it found a real effect that shrinks with time and questioned whether it is clinically meaningful in the medium and long term.
That is not a reason to skip it. It is a reason to stop overselling it, which is what produces patients who conclude physio does not work.
Then why does it still go first?
Because everything it is competing against is worse, and the comparison is not close on the terms that matter.
- Corticosteroid produces a moderate effect at 1–2 weeks that is undetectable by 26 weeks. Exercise still shows a measurable effect at twelve months.
- Hyaluronic acid and PRP are more expensive, more contested, and in PRP's case failed its largest rigorous trial.
- Exercise has no meaningful downside, no cost per dose, no cartilage question, and no frequency ceiling.
- It is the only one that can change anything structural — strength, load distribution, body weight. The injections all leave the joint exactly as they found it.
Both the OARSI and ESCEO guidelines place education, exercise and weight management as core treatments for precisely this reason. Not because the effect is dramatic, but because it is durable, cheap, safe, and the only thing on the list that alters the underlying situation.
Who benefits most?
This is the most practically useful finding in the whole analysis, and it inverts the usual assumption.
People with higher baseline pain and worse physical function benefited more than those who started out better, with the evidence most certain in the short term. The authors suggested targeting exercise at people with higher levels of pain and disability.
That is the reverse of how it tends to be offered. The person whose knee is bad enough to be discussed for surgery is often the one told exercise will not help, and they are precisely the group the data says gains most.
Why people think it did not work
Almost nobody who tells me physio failed has actually done what the trials did. They were given a photocopied sheet, told to do it twice a day, and seen again in six weeks. That is not the intervention that was studied.
What was studied is supervised, progressive, and adjusted as it goes — someone watching, loading increased over time, and enough sessions to matter. The gap between that and a handout is the single biggest reason conservative care gets written off.
That is a clinical observation rather than a trial finding. But it is the pattern behind most of the people who arrive asking for an injection because “the other stuff did not work.”
What good actually looks like
- Supervised at the start. Not forever, but long enough for someone to correct what you are doing and progress it.
- Progressive. The load goes up. A programme that is identical in week eight to week one has stopped being treatment.
- Strength, not just stretching. Quadriceps and hip strength do the work for a knee.
- Long enough to judge. Twelve weeks is the timescale the trials used. Six weeks of half-effort is not a trial of anything.
- Weight, where it applies. It sits alongside exercise in the guidelines rather than instead of it.
- An injection is allowed to help. If pain is the reason you cannot do the programme, using an injection to open that window is the best argument for giving one.
That last point is where this whole section lands. Injections and conservative care are not rivals. The injection is worth giving when it makes the thing that actually works possible — and worth questioning when nothing follows it.
Sources
- Fransen M, McConnell S, Harmer AR, et al. Exercise for osteoarthritis of the knee: a Cochrane systematic review. British Journal of Sports Medicine 2015;49(24):1554–1557. 54 studies. Retrieved via PubMed.
- Holden MA, Hattle M, Runhaar J, et al. Moderators of the effect of therapeutic exercise for knee and hip osteoarthritis: a systematic review and individual participant data meta-analysis. The Lancet Rheumatology 2023;5(7):e386–e400. 31 trials, 4,241 participants. Retrieved via PubMed.
- Arden NK, Perry TA, Bannuru RR, et al. Non-surgical management of knee osteoarthritis: comparison of ESCEO and OARSI 2019 guidelines. Nature Reviews Rheumatology 2021;17(1):59–66. Retrieved via PubMed.
- Jüni P, Hari R, Rutjes AWS, et al. Intra-articular corticosteroid for knee osteoarthritis. Cochrane Database of Systematic Reviews 2015;10:CD005328. Retrieved via PubMed.