7 Questions to Ask About Any Health Story You Read

1. What kind of study is it

Laboratory and animal studies suggest mechanisms and do not establish effects in people.

Observational studies find associations and cannot on their own establish cause, because the groups being compared differ in many ways.

Randomised controlled trials allocate participants deliberately and are much stronger for establishing cause.

Systematic reviews and meta-analyses combine studies and sit at the top of the hierarchy when done well.

A headline based on a mouse study and a headline based on a large trial are treated identically by the news and should not be by you.

2. Relative against absolute risk

A doubling of risk sounds alarming and means little without the starting figure.

If a risk rises from one in ten thousand to two in ten thousand, that is a hundred per cent increase and a one in ten thousand absolute change.

Reports quote the relative figure because it is larger. Look for the absolute numbers.

3. Size and duration

Small studies produce unstable results that frequently fail to replicate.

Short studies cannot measure long-term outcomes, so a finding about a marker is not a finding about health.

Surrogate outcomes — a blood measurement rather than an illness — are common and are not the same as the outcome you care about.

4. Who funded it

Funding does not invalidate research, and it is worth knowing.

Industry-funded studies of the funder's own product show more favourable results on average.

Declared conflicts of interest are usually published with the paper.

5. Practical habits

Find the original study rather than relying on the article, since abstracts are freely available.

Be sceptical of single studies that overturn established consensus; that occasionally happens and usually does not.

Ask whether the study population resembles you in age, sex and health status.

6. Where to look instead

National health service websites and public health bodies summarise evidence for a general audience and are updated as it changes.

Systematic review libraries publish plain-language summaries alongside the technical versions.

These are less exciting than news coverage and considerably more reliable.

7. Common phrases and what they conceal

Linked to and associated with signal an observational finding rather than a demonstrated cause.

Scientists say frequently means one research group said something in one paper.

Could and may indicate possibility rather than finding, and both survive into headlines where the study found nothing conclusive.

Final Thoughts

Health reporting compresses complicated research into a sentence, and the compression removes what you need to judge it. Check the study type and look for absolute rather than relative figures before changing anything.