More structured exercise each week was linked to better motor scores in Parkinson disease

Design

Bayesian network meta-analysis and meta-regression of randomised trials

Population

173 randomised trials in Parkinson disease

Primary outcome

UPDRS III, balance, walking speed, Timed Up and Go

Effect

Mind–body exercise g 0.65 (CrI 0.42–0.87); +0.12 SMD on UPDRS III per +250 MET-min/week

A Bayesian network meta-analysis combined 173 randomised trials of exercise in Parkinson disease, comparing types of exercise and looking at whether more planned exercise per week was associated with larger gains in motor scores, balance, walking speed and the Timed Up and Go test.

Mind–body exercise, such as tai chi and yoga, had the largest effect on motor scores (UPDRS III) against no exercise (g = 0.65, 95% CrI 0.42–0.87). Each extra 250 MET-minutes per week, roughly an hour of brisk walking, was associated with better motor scores, balance and Timed Up and Go, but not walking speed. Confidence in the evidence was rated low or very low, and prediction intervals for every exercise type crossed zero.

The practical message is less about which exercise than about doing it, regularly and in some volume. No single best type or dose was established, and the dose link comes from comparing trials rather than from randomising doses. For a patient who asks what to do, recommending a structured activity they will keep up, including tai chi or yoga, is consistent with this evidence.

• Recommend regular structured exercise to every patient with Parkinson disease from diagnosis.
• Consider tai chi or yoga, which had the largest effect on motor scores here, alongside walking or strength work.
• Encourage more weekly exercise where tolerated, since more was associated with better motor scores and balance.
• Refer to physiotherapy for a programme where balance or falls are a concern.

Why it matters

It gives a patient a concrete reason to exercise more, not just to exercise.

Don't overread it

The dose association is between trials, not within them, and confidence was low.

The statistics, in plain English

A standardised effect of 0.65 is moderate to large. But all prediction intervals crossed zero, meaning a future trial could find no benefit for any single type, and the link with dose comes from comparing different trials rather than from testing doses directly.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.