How to Tell Good Health Information from Bad Online (2026)

The fastest way to tell good health information from bad online is to check four things before you believe anything: who wrote it and whether they are actually qualified on that exact topic, who medically reviewed it, what evidence it cites, and when it was published or last reviewed. Add a fifth habit that settles most arguments: find the original source behind the post or headline. None of that requires a medical degree, and the whole check takes about a minute once you know what you are looking for.

That matters because health misinformation is unusually convincing. It borrows real medical vocabulary, it names conditions you have personally Googled at 2am, and it arrives in a format built to make you forward it. The failure is rarely a completely fake claim; more often it is a real study stretched past what it actually found, or a true statement from a real study paired with a conclusion it does not support.

Below is a repeatable six-step process you can run on any health page, plus the mistakes that trip up almost everyone, including people who read carefully for a living.

What You Need

You need a device with a browser, the original claim in front of you, and about a minute. That is genuinely the whole list.

Three things help more than the rest. First, the link itself, not a screenshot of it. Screenshots strip out the byline, the date and the source, which are the three things you most need. If someone sends you a screenshot, ask for the link before you judge it.

Second, the ability to open the page and scroll to the bottom, where reputable health sites put the byline, the last-reviewed date, the funding disclosure and the references. If the page ends with a signup box and no references, you already have your answer.

Third, somewhere to park the claim while you check it. A note on your phone is enough. Researching a claim while still half-believing it is how people end up defending something they have not actually verified.

You do not need a subscription to a journal database, a research affiliation, or a paid medical app. Public sources like MedlinePlus, the NIH, the CDC, the WHO, the National Center for Complementary and Integrative Health and Cochrane are free, and they are enough for the overwhelming majority of questions people search at home.

Step-by-Step: How to Tell Good Health Information from Bad Online

A 60-second routine you can run on any claim

Before the six checks, run this. Pause for ten seconds before you react, because the urgency you feel is part of the design. Read the whole page rather than the headline and the first paragraph. Check who is speaking. Check the date. Then find one more independent source that is not a copy of the first.

If a claim is about your own treatment, a medication you take, a pregnancy, or a child, stop here and run this routine in the presence of a clinician rather than alone. The rest of this guide is for information, not decisions.

A quick reference for checking a health source

Four indicators carry most of the weight. Here is the compact version, and each one has a fuller section below.

• Authority. A named author with credentials relevant to this specific topic, plus a separate medical reviewer. Warning sign: no byline, anonymous quotes, or a quoted expert whose specialty does not match the topic.
• Evidence. Cites studies, guidelines or systematic reviews you can open. Warning sign: no references at all, or every claim traces back to one paper.
• Currency. A publication date and a visible last-reviewed date. Warning sign: undated, or written once years ago with no review stamp.
• Transparency. Funding, sponsorship and conflict-of-interest statements. Warning sign: supplement sales, affiliate links, or a page that is entirely a sales pitch.

1. Identify the Exact Health Claim

Pull the single central claim out of the page and write it in plain words, then strip away everything around it. Opinion, prediction, advertising, personal story and speculation all borrow the same clothes as fact, and they are much easier to spot once they are separated.

Vague claims cannot be checked, which is exactly why they are used. “Science has finally cracked the code on weight loss” is not a claim you can evaluate; it is a promise dressed as one. “A specific extract from a specific plant reduced a measured outcome in a specific group of people over a specific period” can be searched, and it can turn out to be false in a way you would never notice.

Watch for the modifiers. Words like could, may, associated with and suggests each do different work, and pages that swap them for will, prevents and reverses are usually doing it on purpose.

2. Check Who Created the Information

Look for a byline, then check whether the credentials match the subject. A nurse writing about breastfeeding is exactly the right author for a breastfeeding question. A physiologist writing about childhood sleep disorders is not, however impressive the title sounds.

Then look for a separate reviewer. High-quality health sites name an author and a medical reviewer who is not the same person, and they say what that reviewer’s qualifications are. This does not guarantee accuracy, but the absence of one tells you nobody is accountable for the clinical claims.

Here is what the domain suffix can and cannot tell you:

• .gov usually means a US government health agency such as the CDC, NIH or NIA. The catch: still written for professionals, so a topic page can be dense, and guidance changes over time.
• .edu usually means a university or academic medical centre. The catch: a page can be a student project or a research abstract aimed at other researchers, not readers.
• .org usually means a non-profit, professional body or patient organisation. The catch: anyone can register one, and some are advocacy groups or funded by sponsors.
• .com means a commercial site, which includes many excellent health publishers and many bad ones. The catch: the suffix tells you nothing about the editorial process on its own.

Read the About page and the editorial standards if they exist. A site that publishes how it picks topics, who reviews them and how it corrects errors is behaving like a newsroom, and that behaviour is the single most useful signal on the whole page.

3. Look for Evidence, Not Just a Quotation or Statistic

Numbers are the most persuasive part of most bad health pages, so check where they came from. A statistic with no source, no sample size and no comparison group is a decoration, not evidence. So is a quotation from a doctor whose specialty you have not checked.

Evidence strength is roughly ordered, and it is worth knowing the order. Systematic reviews and meta-analyses, which pool many studies and weigh them together, sit at the top. Below them are randomized controlled trials, where participants are assigned to treatments or placebo by chance. Below those sit observational studies, which find links without proving cause. Laboratory and animal work sits lower still, and a single personal story sits at the bottom.

Headlines also tend to promote a study up the ladder. A laboratory finding in cells is reported as a treatment, an observational association is reported as a cause, and a trial in one small group is reported as a universal rule. Read the abstract and the methods of the original paper, not the summary someone wrote about it.

Correlation versus causation is the most common stretch. Two things moving together may share a common cause, or the link may be coincidence. A nutrient, a behaviour and an outcome can all shift at once because something else entirely changed, and the summary will not mention it.

4. Compare the Claim with Independent Health Guidance

Take the claim you isolated in step one and search for it alongside the current position from a body that has no product to sell you. MedlinePlus, the CDC, the WHO, the NIH, the NCCIH and Cochrane all publish plain-language guidance that is reviewed and dated, and a major clinical guideline is a reasonable place to check whether a popular claim ever made it in.

Be fair about what a mismatch means. Mainstream guidance changes as evidence develops, so a recommendation being new is not evidence that it is wrong. What matters is whether there is a plausible mechanism, whether human evidence exists, and whether the strength of the claim matches the strength of the evidence behind it.

Unsupported, exaggerated and too absolute are three different failure modes, and they need different responses. Unsupported means you can find no evidence either way, so treat it as an open question. Exaggerated means the direction is roughly right and the scale is wrong. Too absolute means a real effect is presented as universal, which is a red flag even when the underlying finding is solid.

5. Look for Funding, Bias, and Commercial Intent

Find out who pays. Supplement companies, supplement sellers, device companies and clinics all fund content about their own products, and that funding is often disclosed properly on a trustworthy page rather than hidden.

Financial interest does not make a claim false. It does mean the page has a reason to emphasise benefits and minimise risks, and a conflict of interest is exactly that: a reason to read more carefully, not a verdict.

The commercial tells are easy to spot. A purchase button inside the article. A countdown timer. Affiliate links. A page that lists risks only after several paragraphs of benefits. Language designed to frighten you into a decision now, or to promise certainty that no honest source can offer.

One content category deserves its own note. Fertility, pregnancy and parenting pages attract a heavy commercial load, because the market for supplements, courses and tests is large and the emotional stakes are high. A page recommending herbs to bring on labour, or a supplement that reverses a diagnosis, deserves exactly the same scrutiny as any other health claim, and more of it, because the consequences fall on someone who cannot consent to the risk.

6. Check the Full Context and Cross-Check It

Before you accept a claim, look for the five context items that are easiest to skip: which population was studied, what the study design was, how certain the result is, who funded it, and when it ran. A finding in one narrow group does not transfer to everyone, and an old result may have been overtaken.

Then check whether multiple independent sources agree. Independence is the word that matters. Ten websites repeating one press release is one source, not ten, and the giveaway is identical phrasing and matching numbers. Genuine agreement means different teams, different methods, similar conclusions.

Use a three-step ending for your confidence. If two or more independent credible sources and current guidance all line up, it is reasonable to treat the claim as reliable. If sources disagree or the only support is a single study, keep investigating or ask someone qualified. If the claim involves symptoms, medication changes, pregnancy or a child, take it to a clinician rather than resolving it yourself.

One more practical habit closes the loop. When a claim arrives as a post or a headline, find the original study or statement before you forward it. Searching the article title in quotes, or the claim plus the word study on PubMed, gets you to the primary source in under a minute, and it resolves most doubts right there.

Common Mistakes

Treating a polished website as authoritative. Design is a business decision, not a research finding. Fix: check the byline, the reviewer and the references before you look at anything else on the page.

Confusing popularity with evidence. A claim that has been repeated for twenty years is still unproven. Fix: search for the source, not the slogan.

Trusting testimonials and before-and-after posts. These show that something happened to someone, not that it will happen to you, and they are usually filtered to the successes. Fix: ask what the denominator is, and treat personal stories as a reason to research, not a reason to conclude.

Ignoring dates. Health guidance genuinely changes, and a page from several years ago may have been accurate then. Fix: look for a last-reviewed date, and if there is none, assume nobody has checked.

Reading one study out of context. The abstract in a press release can differ from the paper. Fix: open the original and read the methods, the sample size and the limitations section.

Taking disagreement as proof of a cover-up. Science arguing openly is science working. Fix: look at the substance of the disagreement, not the emotional charge of the people having it.

Stopping at the first page that feels right. The first result that matches what you already suspected is the one most likely to confirm it. Fix: deliberately look for a source that would change your mind.

Losing the original claim while you research. Fix: keep the claim, its source and its date written down together before you start. Once you have paraphrased it three times, you can no longer tell whether you checked the claim or a version of it, and that distinction is where most of the errors creep in.

Frequently Asked Questions

Conclusion

Start with the first action, every time: identify the exact claim, trace it back to whoever actually said it, and compare it with current independent guidance before you believe it, act on it or send it to someone you care about.

Used this way, online health information is genuinely useful. It helps you describe what is happening, arrive at an appointment better prepared, and notice when something does not add up. It is not a diagnosis, a treatment plan or a substitute for individual medical advice, and nothing on this page should change what you take, stop taking, or do for a child, a pregnancy or an ongoing condition without talking to a qualified clinician first.