Systematic Review to Inform a World Health Organization (WHO) Clinical Practice Guideline: Benefits and Harms of Structured Exercise Programs for Chronic Primary Low Back Pain in Adults.

Only 13 trials and 1362 people, with one to three trials per exercise type, so no individual type could really be assessed. 11 of the 13 trials (85%) were rated at unclear risk of bias, with the concerns mainly about whether participants and assessors knew which group they were in. The positive finding on pain and function comes from pooling all exercise types together in an analysis run after the main one, not from a pre-planned single comparison. Most trials measured outcomes in the first two weeks after the programme; only 4 trials had 12-month data, so nothing here speaks to longer-term benefit. Almost no usable data on quality of life, psychological outcomes, work and social participation, medication use, falls, or treatment burden, so those questions remain open. The reviewers could not test whether results differ by sex, race/ethnicity, or by whether people also had leg pain, so it is unknown whether the findings hold across those groups. All trials came from high, upper-middle or lower-middle income countries; no low-income settings were represented. Grey literature was not searched, so unpublished negative trials could be missing, although the authors searched trial registries and contacted authors of unpublished trials. Trials at high risk of bias were excluded from the main analysis. Adding them back did not change the direction of the findings but lowered the certainty.

The World Health Organization and the University of Zurich are named as funders. The review was commissioned by the WHO to feed into its guideline on chronic primary low back pain, which was still under development when the paper was submitted, and the reviewers worked with the Cochrane team behind the earlier review on the same question that this one updates. No individual competing interests are stated in the text supplied.

Do not read the pooled result as evidence that any particular exercise type works. When the eight types were analysed separately the evidence was low or very low certainty and the review could not say whether aerobic exercise, Pilates, motor control training or any other single type made a difference. The moderate-certainty finding comes from lumping all the types together in an analysis the authors ran after the main one. Do not read the yoga figure as proof that yoga harms people either: it comes from one trial and the confidence interval runs from 1.50 to 441.85.