Each health claim gets one of four grades. Strong means several good randomised trials, or a review of them, agree. Moderate means trials point the same way but are small or few. Weak means small, short or mixed studies. None means no sound human study supports the claim, or the good studies found no effect.
On this page
What do the four grades mean?
| Grade | What it means | What would move it up |
|---|---|---|
| Strong | Several good randomised trials, or a solid review of them, agree | Already the top grade |
| Moderate | Trials point the same way but are small or few, or have some risk of bias | Larger, longer or independent trials |
| Weak | Small or short studies, physiology studies, or results that disagree with each other | Consistent results from well-run trials |
| None | No sound human study supports the claim, or the good studies found no effect | Any well-run human trial with a positive result |
The dots beside each grade fill up with it: three for strong, two for moderate, one for weak and none for no good evidence.
A grade tells you how sure we can be, not how big the effect is. A strong grade can go with a small effect, and a weak grade doesn’t mean something doesn’t work, only that it hasn’t been tested well.
Why is breathing research hard to grade?
Three reasons come up again and again:
- You can’t hide a breathing exercise. People always know whether they are doing one, so trials can’t be properly blinded, and expectations may explain part of the effect. Good trials compare against another active treatment, such as asthma education or a different breathing pattern, rather than nothing.
- Exercises are often tested as part of a package. Pursed-lip breathing is usually taught within pulmonary rehabilitation, and breathing retraining within CBT for panic. It is hard to say how much each part contributes.
- Many trials are small and short. A trial of 30 people over four weeks can find a large effect by chance.
So an exercise that physiotherapists have taught for decades can still carry a weak grade if it has never been tested on its own. We explain why on each page.
What do we look at in each study?
- Was there a fair comparison? Ideally an active control group that got the same attention.
- What was measured? Symptoms and quality of life matter most to people; lung function and heart rate variability are useful but are not the same as feeling better.
- How many people, and for how long?
- Who paid for it, and who ran it? We note when the trial was funded or run by the people who sell the method or device.
- Has it been repeated? One trial, however good, is a starting point.
First published 4 November 2026. Spotted an error? Tell us and we'll correct it. Read our editorial policy.