Most people research health questions before or instead of seeking professional advice, and the quality of what they find varies enormously.
The source hierarchy
National health services, professional bodies and established medical charities generally maintain accurate, reviewed content.
Which is the sensible starting point and is frequently less prominent in search results than commercial content.
Academic sources are authoritative and are frequently inaccessible to non-specialists in both language and availability.
Commercial content
Content produced to sell something has an interest in a particular conclusion.
Which does not make it false and does mean the emphasis will be selective.
Identifying who benefits from you believing something is the single most useful check.
Study reporting
Media coverage systematically overstates findings, and press releases from institutions contribute substantially to this.
Which has been documented in studies of the pathway from paper to headline.
The useful questions are how many participants, was it in humans, was there a comparison group, and what exactly was measured.
Relative and absolute risk
A doubling of a small risk is still a small risk.
Which is why relative figures make findings sound more dramatic than absolute ones.
Asking what the absolute change is converts an alarming headline into a usable number.
Individual accounts
Compelling and unreliable as evidence, since they cannot distinguish treatment effect from natural course or coincidence.
Which is why they feature heavily in marketing for treatments without evidence.
They are genuinely valuable for understanding what an experience is like, which is a different question.
Specific warning signs
Claims of effectiveness across unrelated conditions.
Explanations invoking suppression by established medicine.
Requests for payment before information.
And discouragement from consulting a doctor, which is the most serious of them.
Search behaviour
Searching symptoms produces results weighted toward serious explanations.
Which increases anxiety without improving accuracy, and the phenomenon is well documented.
Searching a condition after diagnosis is generally more useful than searching symptoms before one.
Using it well
Reading beforehand to prepare questions makes consultations better.
Which clinicians generally welcome, whatever the stereotype suggests.
Bringing what you found and asking what they make of it is a reasonable way to use it.
Guidelines
National clinical guidelines are publicly available in most countries and are more readable than people expect.
Which makes them a useful source for what treatment should be offered for a condition.
Patient versions are frequently published alongside the clinical ones.
Preprints
Research published before peer review, which became widely known during recent public health events.
Which speeds dissemination and removes a quality filter.
Coverage of preprints as established findings is a specific and growing problem in health reporting.
Systematic reviews
Assessments of the whole body of evidence on a question rather than of a single study.
Which is the most reliable form of evidence for most treatment questions.
Several organisations produce these with plain-language summaries, and they are freely available.
Conflicts of interest
Declared in academic publication and frequently absent from popular coverage.
Which means checking the original source can reveal funding relationships that the coverage did not mention.
Funding does not invalidate research and is information worth having.
Generated content
Automatically produced health content has expanded rapidly, and quality is inconsistent.
Which makes source attribution more important rather than less.
Content with no identifiable author, institution or review date warrants more caution than it used to.
When to stop reading
Research is useful for preparing a question and is not a substitute for assessment.
Anyone worried about a symptom should get it assessed rather than continuing to search.
Social platforms
Health content on social platforms is ranked by engagement rather than accuracy.
Which systematically favours strong claims over cautious ones.
Accounts identifying professional credentials are worth checking, since claimed credentials are frequently unverifiable and sometimes false.
Comment sections and forums
Patient forums provide genuinely valuable experiential information.
Which is different from clinical information, and the two are easily conflated.
What a treatment feels like is a question other patients answer better than clinicians do.
Language and translation
Reliable health information is unevenly available in languages other than the dominant one.
Which pushes people toward less reliable sources.
Major health services publish translated material, and it is frequently poorly signposted.
A workable routine
Start with a health service or professional body source, check the date, note who produced it, and take any single study cautiously.
Which takes little longer than the alternative and produces far better information.
Bringing what you found to an appointment, and asking what they make of it, is a good use of it.
When it becomes unhelpful
Repeated searching about the same worry generally increases anxiety without adding information.
Which is a signal to seek assessment rather than to keep reading.
Talking to clinicians
Saying what you read and where is more useful than presenting a conclusion.
Which gives them something specific to address.