Exercise Therapy for Knee Osteoarthritis: A Systematic Review and Network Meta-analysis

Mo L, Jiang B, Mei T, Zhou D · Orthopaedic Journal of Sports Medicine

Every exercise type except aquatic exercise beat control on WOMAC pain, with weighted mean differences from -4.20 to -4.45 for cycling, resistance training and traditional exercise and -0.57 for yoga. Ranked by the cumulative ranking curve, stationary cycling came first for WOMAC pain (80.8%) and for the six-minute walk (76.1%), yoga first for stiffness (90.6%) and function (77.4%), and aquatic exercise first on the visual analogue pain scale (77.2%). Most included trials failed to blind participants or researchers, which the authors report as a high risk of performance bias.

Population

Thirty-nine randomised trials covering 2,646 participants with knee osteoarthritis

Intervention

Aquatic exercise, stationary cycling, resistance training, traditional Chinese exercise, or yoga

Comparator

Non-exercising controls, and each exercise type against the others in a network

Limitations

Level 3 by its own grading, and the ranking probabilities that make it quotable are the weakest thing in it: a surface-under-the-curve percentage orders interventions without saying whether the gaps between them are real. Most of the 39 trials could not blind participants, so performance bias runs through the whole network. Exercise types are pooled as categories rather than as doses, so it answers which kind rather than how much.