Where this summary comes from
Yu H, Zhong J, Li M, Chen S. Effects of exercise intervention on falls and balance function in older adults: a systematic review and meta-analysis. PeerJ, 2025. PMID 41122235.
The paper itself — doi.org/10.7717/peerj.20190
Study design: Systematic review and meta-analysis · 37 randomised controlled trials · adults over 60 without diagnosed disease
What the paper asked
Does exercise improve balance and reduce falls in older people — and at what dose?
What was found
- Falls were fewer: odds ratio 0.32 (95% CI 0.20 to 0.51).
- Balance improved: Berg Balance Scale g = 0.92, and Timed Up and Go g = −0.62.
- Confidence in moving improved too (Modified Fall Efficacy Scale, g = 1.01) — which is an outcome in itself for someone who walks less because they are afraid of falling.
- The better dose for balance: sessions of 30 minutes or less, three times a week, for 24 weeks or more. For confidence: sessions over 30 minutes, three times a week, for 12 to 23 weeks.
What does not follow from it
- ⚠️ An odds ratio of 0.32 is a large effect, and the abstract reports no certainty grading such as GRADE. Large effects in exercise meta-analyses usually grow larger as small trials enter.
- The subgroups come from splitting those same 37 trials. "30 minutes, three times a week, 24 weeks" is the best of a ranking, not a proven prescription.
- Participants had no diagnosed disease. This does not transfer to an older person with Parkinson's, a stroke, or vestibular vertigo.
What it means for someone here
Falls are the commonest reason older people lose their independence, and this paper says exercise works against them — with the caveat that the size of the effect should be read cautiously. What transfers directly is the dose: short sessions three times a week, but for six months. Balance here is assessed with the same tests this paper used, so that the before and after are comparable.
Related findings
This page summarises one study. It is not treatment advice, and no programme is written for a person from a single paper.