Daniel Fletcher
Why a headline is never the whole story
A health headline has one job: to make you read the article. It has to be short, striking and clear, which means it often flattens something complicated into a single confident claim. "Two cups of coffee a day cuts stroke risk" sounds like advice. The study behind it might have been a questionnaire filled in by 400 people once, ten years ago.
That doesn't mean the news is wrong, or that the researchers did a bad job. It means the gap between what a study showed and what a headline claims can be very wide. The skill worth learning is not scepticism for its own sake, but a quick, calm way of working out how much a story should actually change what you do on a Tuesday morning.
Start by identifying the type of study
Study design is the single most useful clue to how much weight a finding deserves. As a rough hierarchy, from strongest to weakest:
- Systematic reviews and meta-analyses pool the results of many studies. They are usually the most reliable guide, though they can only be as good as the studies they include.
- Randomised controlled trials assign people randomly to a treatment or a comparison group. Done well, these are the best single studies for showing cause and effect.
- Observational studies (cohort, case-control, cross-sectional) follow or compare people as they already are. They can show a link between two things, but not that one caused the other.
- Laboratory and animal studies happen in cells or mice. They are often the earliest stage of research, and many promising results never translate to humans.
- Case reports describe one person or a handful. Useful for spotting rare problems, useless for judging what most people should do.
If a headline is built on cells in a dish, it belongs in the "interesting, watch this space" pile, not in your shopping list.
Ask how many people, and how much they resemble you
Sample size matters, though bigger isn't automatically better. A well-run trial of 600 people can be more informative than a sprawling survey of 200,000 with sloppy measurements. What matters is whether the study was large enough, and long enough, to detect a real effect rather than random noise.
The more important question is often who took part. A trial of fit 30-year-olds tells you very little about a 70-year-old taking three medicines for type 2 diabetes. Look for age, sex, ethnicity, existing health conditions and medication use. If the participants are nothing like you, the result may still be interesting, but it isn't personal guidance. Also check how many people dropped out, and how long the study ran — three weeks is a short time in a human life.
Watch what was actually measured
Headlines often report a change in a marker rather than an outcome you'd notice. A trial might lower a blood reading by a few points, which then gets translated into "cuts heart attacks" — a claim the study never tested. Likewise, food studies often rely on people remembering what they ate, which is notoriously unreliable.
Be alert, too, to language such as "linked to", "associated with" or "may". These words are honest signals from the researchers that they found a pattern, not proof. Confounding is the usual culprit: people who do one healthy thing often do several others, and teasing apart the effects is genuinely difficult.
Relative versus absolute risk: the sleight of hand to know
This is the trick that frightens people most, and it is worth understanding properly. Suppose a headline says a habit raises your risk of a condition by 50%. That sounds alarming. But the meaning depends entirely on where you started.
- If the baseline risk was 2 in 10,000 over ten years, a 50% rise takes it to 3 in 10,000. That's an extra one person in 10,000.
- If the baseline was 20 in 100, a 50% rise takes it to 30 in 100. That's ten extra people in every hundred.
Same percentage, wildly different real-world meaning. Whenever you see a relative risk, hunt for the absolute numbers underneath, the time period involved, and who the risk applies to. If the article never gives them, treat the figure as incomplete.
A quick checklist before you change your habits
Run through this in five minutes, and you'll make better decisions than most of the people sharing the story.
- Can you find the original study, and was it reviewed by other scientists?
- Was it in humans, and do those humans resemble you?
- How large, and how long, was it?
- What was measured — a marker, or something that actually matters to you?
- What's the absolute effect, over what time frame?
- Does it fit with what the wider body of evidence already says?
- What would changing cost you — money, time, sleep, enjoyment?
Most single studies, however exciting, are one brick in a wall. Sensible health advice tends to be unglamorous and slow to change: move more, eat mostly plants, sleep enough, limit alcohol, don't smoke, keep up with screenings. If a headline contradicts all of that on the basis of one small trial, patience is a perfectly reasonable response. And if a story genuinely matters to your health, your GP, practice nurse or pharmacist can help you put it in context.
