Evidence and research
How do I tell whether a study cited in wellness marketing is any good?
To judge a study cited in wellness marketing, check four things: the level of evidence, the sample size, the delivery route, and whether it appeared in a named peer-reviewed journal. A systematic review pools all studies on a question; a meta-analysis statistically combines their results. Both outrank a single trial.
The evidence hierarchy, top to bottom
- Systematic review and meta-analysis. A systematic review searches the literature to a defined method and summarises every eligible study. A meta-analysis goes further, pooling their results into a single effect estimate.
- Randomised controlled trial (RCT). Participants are randomly assigned to treatment or control. Double-blind and placebo-controlled designs are stronger again: neither participant nor assessor knows who got what.
- Prospective controlled or cohort study. Followed forward in time, without randomisation.
- Pilot, preliminary and observational studies. Useful for hypotheses, weak for proving effect.
- Mechanistic, in vitro and animal studies. Explain how something might work; they do not show it works in people.
Four questions to ask of any cited study
- What level is it? A paper describing itself as a mechanism review is not clinical evidence.
- How many participants? 422 across 13 RCTs carries weight. Twenty in a pilot does not.
- What exactly was delivered, and how? The route, the device and the dose all matter. For light, intensity has a name and a unit: irradiance, the power landing on a given area (milliwatts per square centimetre at a stated distance), and fluence, the total energy per square centimetre (joules).
- Where was it published, and when? A named peer-reviewed journal with a year is checkable, or a link that leads to one. A title with neither is not. Our own studies pages link on to the sources, so any entry can be followed to its paper.
Three warning signs
A percentage with no study attached cannot be verified. A study cited for a claim it did not test misuses a real paper. And two different figures for the same thing on one site means at least one is wrong.
Terms used