Everyone prescribes exercise for knee osteoarthritis and almost nobody specifies the dose. This systematic review with dose-response meta-analysis pooled 20 randomised trials of isolated isotonic lower limb resistance training — no other exercise modality mixed in — covering 791 people with knee osteoarthritis, and used restricted cubic spline models to locate where benefit peaks.
The peaks differ by outcome, which is the finding. Pain reduction was greatest at around 6,000 total repetitions (mean difference -3.52, 95% CI -4.08 to -2.96). Functional improvement peaked much earlier, at around 2,600 repetitions of moderate intensity (-16.66, -18.80 to -14.51). Strength gains maxed out earliest, at around 1,600 repetitions (standardised mean difference 0.12, 0.01 to 0.22). On intensity: very light to moderate produced the best pain relief, moderate the best function, and moderate to vigorous similar strength gains. Centre-based programmes outperformed home-based ones, though both worked.
The relationships were non-linear, which means more is not simply better. A patient grinding through a heavy programme may be past the point where pain relief improves, and one doing very light work may be getting good pain relief without much function or strength. Moderate intensity comes out best across the outcomes patients care about most.
How to use it: translate the numbers into something prescribable. Roughly 6,000 total repetitions is achievable over about 12 weeks at three sessions a week — around 170 repetitions per session, which is three or four exercises at three sets of 12 to 15. Set moderate intensity, meaning a weight the patient could manage perhaps 12 to 15 times before form fails, not a weight that hurts. Where a supervised programme is available, use it, since centre-based delivery performed better.
The certainty is low to very low by GRADE, and 791 patients across 20 trials is thin for a dose-response analysis. Treat these numbers as a reasonable target to prescribe against, not as a validated dose.
- Prescribe a specific dose rather than 'do some exercises' — around 6,000 total repetitions over about 12 weeks.
- Set moderate intensity: a load the patient could manage 12 to 15 times, not one that provokes pain.
- Expect function and strength to plateau earlier than pain relief; keep going for the pain benefit.
- Refer to a supervised centre-based programme where available; it outperformed home-based delivery.
- Certainty is low to very low — these are targets to prescribe against, not validated thresholds.
The statistics, in plain English
Restricted cubic splines fit a flexible curve to the relationship between dose and outcome rather than forcing a straight line, which is what allows a peak to be identified. The catch is that a curve estimated from 20 trials with 791 patients in total is not tightly determined, especially at the extremes where few studies contribute — so the figure of 6,000 repetitions is the top of a fairly flat and imprecisely located curve, not a threshold with a cliff either side. GRADE certainty of low to very low reflects small trials, unblinded exercise interventions where participants know their allocation, and inconsistency between studies.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for orthopaedics, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free