There has been considerable research interest in the effects of exercise on depression over the past three decades and many systematic reviews have reported moderate to large effect sizes, with the standardised mean difference for the most recent Cochrane review being −0.82 (95% confidence interval −1.12 to −0.51).1 2 3 A new linked trial (TREAtment of Depression with physical activity (TREAD); doi:10.1136/bmj.e2758) adds to this evidence base.4

At first glance reviews suggest that exercise is effective in the treatment of depression. However, most trials included in systematic reviews recruited small numbers of patients, had a short follow-up, and did not adequately conceal randomisation or recruited non-clinical community volunteers (or both). Volunteers are more likely to be motivated to exercise and may be less severely depressed than people identified in clinical settings. Subgroup analyses that included only the higher quality trials in the Cochrane review reduced the effect size to −0.42 (−0.88 to 0.03),1 casting doubt on the main finding.

In 2009 the UK National Institute for Health and Clinical Excellence recommended that people …