Evidence
How to judge diet research and health claims
"Coffee adds years to your life." "Eggs are as bad as smoking." Diet headlines flip every few months. A few simple questions will help you tell a solid finding from a shaky one.
Ask what kind of study it was, how many people took part and for how long, what it was compared with, whether the risk is given as a relative or absolute number, and who paid for it. One study rarely settles anything. Look for what reviews of many studies say.
Types of study, from weakest to strongest
- Anecdotes and testimonials. "I lost 20 lb on this shake." It may be true for that person, but it tells you nothing about what usually happens.
- Lab and animal studies. Useful for working out how something might work in the body, but mice and cells in a dish are not people.
- Observational studies. Researchers follow large groups and look for links, such as people who eat more nuts having fewer heart attacks. They can't prove cause, because nut eaters may also smoke less or exercise more.
- Randomized controlled trials. People are randomly put on one diet or another and compared. This is the best way to test cause and effect, but diet trials are often short and people don't always stick to their assigned diet.
- Systematic reviews and meta-analyses. These pool the results of many studies on the same question. They are the strongest evidence, but only as good as the studies they include.
Relative vs absolute risk
This is the trick behind many scary headlines. Here is a made-up example to show how it works.
Say 2 in every 1,000 people who never eat a certain food get a disease, and 3 in every 1,000 people who eat it daily get it. The headline says "Food raises risk by 50%." That is the relative risk, and it is technically true. But the absolute risk went from 0.2% to 0.3%, an extra 1 person in 1,000. Both numbers matter, and a fair report gives you both.
Size, length and funding
- How many people? A trial of 20 people can easily find a result by chance. Hundreds or thousands are more convincing.
- How long? Most diets work for a few weeks. What matters is whether the results hold at one or two years, and many studies stop long before that.
- Compared with what? "Keto beat the standard diet" means little if the standard diet was a poor one.
- What was measured? A small drop in a blood marker is not the same as fewer heart attacks.
- Who paid? Industry funding doesn't make a study wrong, but a juice study paid for by a juice company deserves a closer look. Funding is usually listed at the end of the paper.
Headline red flags
- Words like "miracle", "cure", "toxic" or "melts fat".
- A big claim from a single new study.
- A percentage with no starting number.
- Results from mice reported as if they apply to you.
- A study linked to someone selling a product, supplement or plan.
- Advice that everyone should cut out a whole food group.
For more, see how to spot a fad diet and are online diet plans safe?
Who writes this and why
OnlineDiets is written by its founder, Greg Parsons. Greg is not a dietitian or doctor, and he says so openly. The guides lean on mainstream sources such as the NHS, the US National Institutes of Health, the Harvard Nutrition Source, Mayo Clinic, the WHO and the American Heart Association, rather than single studies or personal opinion. Where the evidence is mixed or short-term, the guides say so.
You can read more in the editorial policy and how guides are reviewed. Nothing here replaces advice from your own doctor; see the medical disclaimer. Spotted a mistake? Get in touch.