Optimal resistance training strategies for knee osteoarthritis symptom relief: a systematic review and network meta-analysis

Yang Y, Wang J, Wang T, Yuan Y, Qiu B, Hu W · BMC Musculoskeletal Disorders

Every type of resistance training beat control on pain: all seven, across 45 trials and 3,357 participants, with standardised mean differences from 1.35 for high-speed training down to 0.56 for concentric-eccentric isotonic. High-speed training, meaning concentric contractions performed at maximum speed, ranked first on all three symptom outcomes: pain 1.35 (95% credible interval 0.81 to 1.89), stiffness 1.26 (0.70 to 1.81) and function 1.70 (0.96 to 2.45), and it beat every other type on stiffness in the pairwise comparisons. The dose analysis found the optimum for pain and function at a moderate intensity of 43 to 47 percent of one-repetition maximum, sustained for 35 to 37 weeks at 610 to 640 repetitions a week; stiffness responded better to more repetitions over a shorter programme, 1,200 a week for 12 weeks.

Population

46 randomised trials and 3,463 people with knee osteoarthritis, searched to April 2025, analysed in a Bayesian network

Intervention

Seven types of resistance training, including high-speed, isokinetic, isometric and concentric-eccentric isotonic training

Comparator

Non-exercise control, and the other resistance training types

Limitations

The dose figures come from modelling across trials rather than from any trial that randomised people to different doses, so they are the shape of an association between study-level characteristics and effect size. Stiffness rests on 26 trials and 1,350 participants, function on 35 and 2,411, so the type ranking is better supported for pain than for the rest. Credible intervals on the leading estimates are wide, and no trial in a network like this can blind its participants.