Glenn Wellmann DPT

If your knee has arthritis and you only have the energy for one kind of exercise, which kind should it be? A 2025 network meta-analysis in The BMJ pooled 217 randomised trials to answer that, and the winner is not the one most people expect.

The study, briefly

The paper is Yan, Li, Xu and colleagues, “Comparative efficacy and safety of exercise modalities in knee osteoarthritis: systematic review and network meta-analysis,” BMJ 2025;391:e085242. It is a systematic review with a network meta-analysis, prospectively registered (PROSPERO CRD42023469762), with databases searched from inception through August 2024.

The scale is what makes it worth your attention: 217 randomised controlled trials and 15,684 participants, published between 1990 and 2024. Everyone included was an adult with symptomatic knee osteoarthritis, diagnosed clinically, on X-ray, or on MRI. Trials were excluded if participants had already had knee surgery or also had hip arthritis, so this is a fairly clean picture of the ordinary arthritic knee.

The researchers sorted every exercise program into buckets: aerobic exercise (walking, cycling, swimming), strengthening, flexibility, neuromotor or balance work, mind-body work such as tai chi and yoga, mixed programs, and control. Then they compared outcomes at roughly 4 weeks (short term), 12 weeks (mid-term), and 24 weeks (long term). A network meta-analysis lets you rank options that were never tested directly against each other, by routing the comparisons through the shared control groups.

What they found

Aerobic exercise won, and it was not close. For pain at 12 weeks, aerobic exercise produced a standardised mean difference of -1.19 (95% CI -1.59 to -0.79) versus control, with moderate certainty evidence. In plain terms, a standardised mean difference near 1.0 is a large effect, the kind of change a person actually notices in daily life, not a statistical curiosity.

• Pain: aerobic exercise led at both time points, -1.10 (95% CI -1.68 to -0.52) short term and -1.19 (95% CI -1.59 to -0.79) at 12 weeks, both moderate certainty.
• Function: aerobic exercise also led at 12 weeks (SMD 1.78, 95% CI 1.05 to 2.51) and finished on top of every outcome ranking, with a mean SUCRA of 0.72. SUCRA is just a probability score for how often an option lands near the top of the rankings.
• Strength work still mattered: strengthening improved 12-week function (SMD 0.86, 95% CI 0.53 to 1.18) and mixed programs did better still (SMD 1.07, 95% CI 0.68 to 1.46). Mixed programs were the ones that still held a benefit at 24 weeks (function SMD 0.56, 95% CI 0.26 to 0.86).
• Safety: no exercise type produced more adverse events than doing nothing. Aerobic exercise carried a risk ratio of 2.33, but the confidence interval ran from 0.73 to 7.14, which crosses 1 and therefore tells you nothing definitive.

That last bullet is the one I wish more people heard. The fear that moving an arthritic knee grinds it down faster did not show up as extra harm across 217 trials.

The catch

Here is where I stay honest with you. Most of the included trials compared exercise against a control group, not head to head against a different style of exercise. Only 63% of studies (137 of 217) even had a control arm, and very few pitted aerobic work directly against strength work. So a good chunk of this ranking is built on indirect comparison, which is legitimate statistics but weaker than a direct trial.

The authors also found significant funnel plot asymmetry for pain (P=0.002 at 4 weeks, P less than 0.001 at 12 weeks), which is a signal that small studies with flattering results may be over-represented in the literature. Exercise dose and adherence were reported inconsistently, so we cannot say how much walking or cycling produced these numbers. And durability is a real gap: only 12 studies, 5.5% of the total, followed anyone past a year.

Read this as strong evidence that aerobic exercise deserves top billing, and weaker evidence that it is definitively superior to a well built strength program for your particular knee.

What this means for you

In the clinic, the most common thing I hear from someone with knee arthritis is a version of “I stopped walking because it hurts.” This study is a good argument for reversing that decision carefully rather than permanently.

1. Pick an aerobic activity your knee tolerates today, not the one you did at 30. Brisk walking, a stationary bike with the seat set high enough that your knee never bends past comfortable, or pool walking and swimming. Tolerance is the filter, not tradition.
2. Aim for consistency over intensity. The trials in this analysis mostly ran 8 to 12 weeks of regular sessions. Three to five sessions a week of something you can repeat beats two heroic efforts followed by a flare.
3. Use the 24 hour rule. Some ache during and shortly after activity is expected with arthritis. If pain is clearly worse the next morning and stays worse, you overshot the dose. Back off about 20% and rebuild.
4. Keep strength in the program. Mixed programs were the ones that still showed benefit at 24 weeks, which matches what I see: cardio buys you comfort, strength protects the gains. Two days a week of leg work is enough to matter.
5. Do not wait for the knee to feel normal first. Nothing in this data suggests you need to be pain free before you start moving.

If you are wondering whether walking alone is enough to cover the strength side, I wrote about that specific question in is walking enough to build strength. Short answer: it is excellent for your joints and your pain, and it is not a substitute for loading your legs. If hiking is your preferred aerobic option, hiking and joint health covers how to build the mileage without paying for it later.

Where I would start with you

The hard part is rarely knowing that exercise helps. It is knowing which aerobic option your knee will tolerate this month and what dose to start at, then progressing without flaring. That is the whole point of a structured home exercise program: a plan matched to your knee, written down, with clear rules for when to add and when to hold. Bring me your knee and your calendar, and we will build something you will actually repeat.

References

• Yan L, Li D, Xu D, et al. Comparative efficacy and safety of exercise modalities in knee osteoarthritis: systematic review and network meta-analysis. BMJ 2025;391:e085242. doi:10.1136/bmj-2025-085242. https://www.bmj.com/content/391/bmj-2025-085242