Comparative efficacy and safety of exercise modalities in knee osteoarthritis: systematic review and network meta-analysis

Yan L, Li D, Xing D, Fan Z, Du G, Jiu J · BMJ

The largest and best-graded comparison of exercise types in this disease, and it puts aerobic exercise first. Against control on moderate certainty, aerobic exercise improved pain by a standardised mean difference of 1.10 at four weeks and 1.19 at twelve, function by 1.78 at twelve weeks, gait performance by 0.85 and quality of life by 1.53. Strengthening exercise improved function by 0.86 at twelve weeks, also on moderate certainty, and mixed exercise by 1.07. Aerobic exercise ranked highest across outcomes, with a mean surface under the cumulative ranking curve of 0.72, and no clear difference emerged between types of aerobic exercise in a post hoc comparison. On safety, 40 trials supplied usable data: risk ratios against control ran from 0.61 for mind-body exercise to 2.33 for aerobic exercise, every interval crossed one, and the reviewers read that as imprecision rather than a difference.

Population

217 randomised trials and 15,684 people with symptomatic knee osteoarthritis, searched to August 2024, assessed with RoB 2 and GRADE at four, twelve and twenty-four weeks

Intervention

Aerobic, flexibility, mind-body, neuromotor, strengthening or mixed exercise, with no restriction on dose, duration or intensity

Comparator

Non-exercise control, and the other exercise modalities

Limitations

A network meta-analysis compares interventions that were mostly never randomised against each other, so the ranking is a modelled probability rather than a measured gap. Only 40 of the 217 trials, 18 percent, reported adverse events at all, which is why the safety intervals are too wide to separate anything. Long-term estimates rest on far fewer trials than the short and mid-term ones and mostly carry low certainty. Exercise trials cannot blind their participants.