How to read a health study in 60 seconds
Four questions that tell you how much to trust a headline. You can use them on any study, including the ones on this site.
The short version
- Ask what kind of study it was. A randomized trial beats an observational study for proving cause.
- Ask who was studied. Mice, young athletes, and 70-year-olds with high blood pressure are not the same.
- Ask how big the change was, not just whether there was one.
1. What kind of study was it?
- Randomized controlled trial: people are randomly put into groups, one gets the thing, one doesn’t. The best way to show cause and effect.
- Observational (cohort) study: researchers track what people already do and what happens to them. Great for spotting links over many years, but it can’t prove the habit caused the result. That’s why we say “linked to.”
- Meta-analysis: combines many studies. Strong when the studies inside it are good.
- Animal or cell study: useful for ideas, but often doesn’t carry over to people. We label these clearly.
2. Who was studied?
A result in 36 people for six weeks is a promising start. A result in 50,000 people over 20 years is harder to ignore. Also check whether the people look like you: age, sex, health conditions.
3. How big was the change?
“Lowered risk by 50%” sounds huge. But if the risk went from 2 in 1,000 to 1 in 1,000, the real-world change is small. When a study gives the actual numbers, we report them.
4. Who paid for it?
Industry funding doesn’t make a study wrong, but it is worth knowing. Our articles mention it when it applies, like the blueberry trial partly funded by a blueberry growers’ group.
One more thing: watch for a single study
One study is a clue. Several studies pointing the same way, using different methods, is evidence. Our Research Revisited section looks at how older findings held up over time.
Educational information, not medical advice. Read our medical disclaimer.