For knee osteoarthritis pain, non-weight-bearing strength work led the field

Exercise is the first-line recommendation for painful knee osteoarthritis, but 'exercise' covers very different things: pushing against resistance while seated, squatting or stepping under your own body weight, or walking and cycling for aerobic conditioning. People with a sore knee understandably want to know which of those is most likely to reduce pain, and whether loading the joint is necessary or counterproductive. Individual trials rarely compare the types directly. This review set out to see whether pooling randomised trials could separate their effects on pain.

The authors searched Medline, the Cochrane Central Register of Controlled Trials, the Physiotherapy Evidence Database and CINAHL from each source's earliest date to March 2013. They collected trials comparing an exercise intervention with either no intervention or a psycho-educational control in people with knee osteoarthritis. Meta-analysis was restricted to trials in which therapeutic exercise was performed more than three times a week for up to eight weeks, with pain as the outcome. Trials were sorted into three subgroups — non-weight-bearing strengthening, weight-bearing strengthening and aerobic exercise — and subgroup analyses were run alongside the overall pooled estimate. Data from eight studies were integrated.

The overall effect of exercise on pain was significant and large, with a standardised mean difference of -0.94 (95% CI -1.31 to -0.57). Splitting by type, non-weight-bearing strengthening showed the largest effect at -1.42 (-2.09 to -0.75), followed by weight-bearing strengthening at -0.70 (-1.05 to -0.35) and aerobic exercise at -0.45 (-0.77 to -0.13). All three subgroups produced significant pain relief, so none of the approaches was ineffective. The confidence interval around the non-weight-bearing estimate was notably wide, reflecting how few trials fed into each subgroup.

The overall result reinforces the standard position that exercise relieves osteoarthritic knee pain rather than aggravating it. The subgroup ordering adds something more specific: for short programmes, taking body weight out of the movement was associated with the greatest pain reduction. That comparison is indirect, drawn from separate trials against control rather than head-to-head, and rests on a small number of studies. It also says nothing about what happens beyond eight weeks, when weight-bearing work might matter more for function, strength and joint loading tolerance.

The reassuring part is that all three approaches beat doing nothing for knee pain in the short term, so there is more than one route in. The bigger effect for non-weight-bearing strengthening is a reason to think that pain relief does not require loading the joint through squats or steps early on. Because every trial here ran for eight weeks or less, this tells you about early pain change rather than long-term joint health or capability. The small number of pooled trials means the ranking between exercise types should be held loosely.