A guide to the questions we ask before we write anything — and the same questions are worth asking about any health article you read, including ours.
Was it done in people?
A large amount of longevity research is done in mice, worms or cells. This work is genuinely valuable, and it is also where most exciting-sounding findings come from and where most of them stop. Mice are not small humans, and the majority of interventions that extend life in animals have never been shown to do anything comparable in people.
When the evidence is animal evidence, we say so in the sentence.
Does it show cause, or just company?
An observational study can show that two things occur together. It cannot show that one produced the other.
This matters more than any other single point in health reporting. People who sleep well also tend to exercise more, drink less and have more stable lives — so a finding that good sleepers live longer may be about sleep, or about everything that travels with it. Good studies adjust for these factors. No adjustment is complete.
We also watch for reverse causation, where the outcome causes the exposure rather than the other way round. Early disease can disturb sleep years before diagnosis, which makes short sleep look like a cause when it may be an early symptom.
How big is the effect, really?
Relative risk is the most abused number in health journalism. A doubling of a rare risk is still a rare risk.
Where we can, we give the underlying numbers and the confidence interval — the range within which the true value probably sits. When that range crosses the point of no effect, the result does not support a firm conclusion, and we say so rather than reporting the headline figure.
How many people, and which people?
A trial of thirty people is a starting point, not an answer. Large cohorts are more reliable, but their composition matters: a study drawn overwhelmingly from one population may not describe others well, and we note it when that is the case.
Who paid for it?
Funding does not make research wrong. It does shift the odds, and industry-funded studies produce results favourable to the funder more often than independent ones. We report the funding source when it is relevant.
Is it one study or many?
A single striking result is a hypothesis. Our confidence rises when independent groups, using different methods, arrive at the same place — and it should fall when they do not.
Where the field is genuinely divided, we describe the disagreement instead of picking the version that makes a cleaner story.