What we cover
Four things make it onto this page. First, changes to clinical guidance — a screening age moving, an activity target being restated, a vaccine schedule updated. Those change what a reasonable person should do, which makes them the most useful health news there is. Second, regulatory decisions: an approval, a clearance, a label change, a safety communication. Third, large or unusually well-conducted studies, particularly randomised trials and rigorous systematic reviews, which occasionally shift the picture on a question people actually ask us about.
Fourth, and more quietly, we cover the corrections: the retraction, the failed replication, the widely repeated claim that turned out to rest on almost nothing. That category rarely trends, but it is often the most valuable, because bad health information is stubborn and needs to be met more than once. What we deliberately do not cover is the daily churn of single observational studies about one food or one supplement, which generate headlines out of proportion to what they can support.
How we choose a story
The first question is whether anything would change. If a reader could not do anything differently, and their understanding of their own health would not shift, the story usually does not survive that question. The second is whether the finding is strong enough to carry the weight a headline would put on it. A randomised trial with thousands of participants and a hard outcome such as death, stroke or hospital admission is a different animal from a questionnaire study of a few hundred people reporting how they felt.
The third is whether we can explain it honestly in plain language without either inflating it or flattening it into uselessness. Some genuinely important findings are difficult to convey — competing risks, absolute benefit in a low-risk group, uncertainty that is wide but not infinite — and we would rather take the space to do that properly than publish something crisp and misleading. Where a story mainly matters to people with a specific condition, we say so in the first paragraph rather than implying it applies to everyone.
Our standard for reporting a finding
Before we report that something works, we want to know four things and we will tell you all four: what kind of study it was, how many people were in it, how long they were followed, and what it was compared against. A result with no comparator is not a result. A result in mice is a result in mice. An improvement in a laboratory measurement is not the same as an improvement in how long or how well someone lives.
We report absolute numbers alongside relative ones. A treatment that cuts risk by a third sounds identical whether it moves three people in a hundred to two, or thirty in a hundred to twenty, and those are very different propositions. Where a study used a surrogate endpoint — a biomarker standing in for the outcome anyone actually cares about — we name it. Where the funders had a commercial interest in the result, we say that too. None of these things automatically invalidates a study; all of them change how much weight it can bear.
How we handle uncertainty
Most interesting health questions do not have a settled answer, and pretending otherwise is the most common failure in health journalism. When the evidence is genuinely mixed, we say it is mixed and describe the shape of the disagreement: whether the trials point in different directions, whether the observational data and the randomised data conflict, whether the effect is real but small enough that reasonable people weigh it differently.
We also try to be clear about what kind of uncertainty is involved. Not knowing whether a treatment works at all is different from knowing it works and not knowing for whom it works best, which is different again from knowing it works and not knowing what it does over twenty years. Readers can handle that distinction, and it usually matters more to a decision than the headline finding does. Where our own view has changed, the page says so and carries the date.
What we will not print
We do not use the word cure unless a treatment cures something. We do not describe early-stage research as a breakthrough. We do not run a single study as though it overturns a body of evidence, and we do not run fear as a hook — a risk without its size attached is not information, it is an emotional prompt. We do not accept payment to cover a product, and our advertising policy and affiliate disclosure set out how commercial relationships are handled.
We also avoid the false balance that treats a fringe claim as one side of a live debate. Where scientific consensus is strong — on vaccine safety, on tobacco, on the basic direction of evidence for blood pressure treatment — we report the consensus and describe genuine open questions within it, rather than staging an argument that does not exist in the literature. Our full approach is in the editorial policy and corrections policy.
Reading the news yourself
The most useful habit is to go one step upstream. Most health headlines are written from a press release, and the press release is written from an abstract. The abstract usually names the design, the number of participants and the comparison, which is often enough to tell you whether the headline is fair. Registries and guideline bodies are even better starting points, because they show a claim in the context of everything else known about the question.
Three habits do most of the work. Check what kind of study it was before you react to the finding. Check whether the numbers quoted are absolute or relative, because a halved risk means nothing until you know what it was halved from. And check whether anyone has found the same thing twice, since a single unreplicated result is a starting point rather than a conclusion. Our Research and Insights hub is a longer guide to all of it: how study designs rank, what a confidence interval is telling you, why associated with and causes are not interchangeable, and how to spot a surrogate endpoint standing in for the outcome you actually care about. If you read one thing before your next health headline, read that.