Can balance exercises reduce falls in older adults?

Trials in the US, Canada and Australia show why structured support and the person’s situation matter more than a generic exercise list.

Prepared by PoshanSense · AI-assisted educational writing
Our sources and editorial approach · Not clinically reviewed

Complementary human studies · Published 2014, 2018, 2019

A stumble can make someone wary of the stairs, the pavement or even getting up from a chair. It is understandable to look for an exercise that restores confidence. The research supports a more careful question: what kind of programme fits this person, with what support?

Balance is part of the answer, not the whole assessment

Exercise programmes can practise strength, stepping, weight transfer and controlled movement. But a fall is not simply proof that someone needs to stand on one leg more often. The trials enrolled defined groups and used designed programmes; they did not replace individual assessment with a universal challenge.

That matters particularly after a recent fall, illness or hospital stay. Being able to do an exercise and being safer in everyday life are related but different outcomes.

What did the US trial show?

The 2018 Oregon trial recorded 152 falls in the therapeutic Tai Ji Quan group, 218 with multimodal exercise and 363 with stretching. The Tai Ji programme had a lower fall rate than either comparison over six months.

It was a specifically adapted, taught intervention. Participants were screened for ability to exercise safely, and the programme progressed through supported learning. This does not establish that any generic class, a short video or an unsupervised balance challenge produces the same result.

A Canadian study adds a home-based example

In a 2019 Canadian trial, 345 older adults were randomized after a fall; 344 remained after one withdrew data permission. Home strength-and-balance retraining supported by a physiotherapist reduced the rate of subsequent falls compared with usual geriatric care. But the proportion falling at least once was similar, and fracture reduction was not established.

That distinction stops “fewer falls” becoming “everyone avoided falling.” The study received a correction adding omitted conflict disclosures, including the lead author’s cofounder role in a brain-health company. It did not revise the reported fall result.

The counterexample is important

A 2014 Australian trial tested home exercise in 340 older people recently discharged from hospital. Performance-based mobility improved, but the exercise group had more falls over the following year: 177 versus 123, with a higher estimated fall rate.

The authors discussed several possible explanations, but did not establish a single mechanism. This is not a reason to tell all older people to avoid movement. It is a reason not to assume that any programme proven useful elsewhere transfers unchanged to a recently hospitalised population.

What should someone do with this evidence?

After a fall or hospital stay, seek assessment and a plan appropriate to current health, mobility and surroundings. Ask whether a suggested programme specifically targets falls, who teaches and adjusts it, and what support is available. The evidence favours those practical details over a list of apparently simple exercises.

Do not practise an unstable or unsupported challenge solely because it appears in a health article. This article explains trial results; it is not an exercise prescription, a fall-risk assessment or a replacement for rehabilitation.

The US study was supported by the National Institute on Aging. Its lead author disclosed ownership of the company receiving licensing fees for the Tai Ji programme; the Australian study by the NHMRC, with no competing interests declared. The Canadian trial reported CIHR funding and corrected its author disclosures. The studies broaden the geographical picture, but they do not establish the same benefit in every home, health system or population. Educational article; not clinically reviewed.

Published 11 October 2026 · Sources checked 11 October 2026. Suggest a correction.