Unsubstantiated hacks, cures, and quick fixes for a diagnosed condition are warning signs, not shortcuts. A New York Times report on health misinformation quotes Dr. Agarwal urging that such claims be validated before they are acted on. When the claim is about a packaged food, the nutrition label is a faster check than the front of the box.

Framing tells are just as quick to catch. Watch for all-or-nothing language, a single personal story standing in for evidence, the claim that a mainstream institution is hiding the truth, and posts with no author, no date, or no link to an original source. Each substitutes emotion or borrowed authority for evidence you can verify yourself. Emotional language like 'miracle' or 'breakthrough' often signals a weak claim. A fifth tell, this one about delivery rather than framing: algorithmic delivery. A study of debunking on TikTok found that the platform's For You feed often pushes videos containing false or unvetted medical claims. A recommendation is not an endorsement. Algorithms optimize for engagement, not accuracy, so a sensational claim can reach you because it gets clicks, not because it is true.

Most flags surface in a ten-second scan, and the work splits into setup you do once and checks you repeat on every new claim.

Setup, once:

  • Mute or unfollow accounts that habitually use the patterns above.
  • Bookmark two or three fact-checking sites so verification is one click away.

Repeat, on each new claim:

  • Scan for the flags before reading any further.
  • Check the date first. Recirculated old research is a common source of false alarm.

Before judging a claim, identify who produced it. Ask what credentials the author holds, not what title the page implies, and whether they would gain financially or professionally if you believed them. A conflict of interest does not make a claim false, but it raises how much independent support the claim needs. A named author with a traceable institution and a publication date guarantees nothing, but it means someone can be asked to answer for the claim. A supplement seller that publishes a study of its own product has a financial stake in the result, whether or not the study is well run. A 'nutritionist' title may not require a license, while a registered dietitian does.

Disinformation also travels through political elites, traditional media, and trusted community members. Familiarity of the messenger is not evidence about the message.

Separate a site's health content from its business model. A page that sells the product it recommends is advertising, however clinical the layout. A sponsored post labeled as research is still a sponsored post.

Check whether the page links to the primary research rather than another news article. A preprint posted before peer review is a draft; citing it as settled evidence skips peer review. A claim that names no study is harder to verify and easier to overstate. For a publication's track record, look for a dated corrections page or a visible errata note rather than a general claim of accuracy. Ethical lapses should be brought to newsroom leaders, a reminder that accountability processes matter.

The strength of a claim should match the study behind it. Observational studies track existing habits and outcomes; randomized controlled trials assign participants to groups and compare. A randomized trial is the gold standard for testing whether a treatment works, but it must be well designed and large enough to detect an effect. Observational findings can suggest a link but cannot establish causation. Randomized trials exist because self-selected habits carry confounders that no data volume removes. People who already take a supplement, for example, may also sleep longer, move more, or see a doctor more often, and those differences rather than the supplement can explain a better outcome. An observational study might find that coffee drinkers live longer, but that could be because coffee drinkers are wealthier and have better healthcare.

Correlation is not causation; two things may correlate due to a shared cause or the study population. A headline that converts a correlation into a cause has overreached.

Check for consensus rather than a single study. Ask whether major health organizations such as the WHO, the CDC, or the NIH describe the same conclusion. A lone finding that contradicts broad agreement is a question to investigate, not a settled reversal.

Read the original study when you can, or at least its abstract and limitations section, rather than the article about the article. A study of ten people and a study of ten thousand carry different weight, and the abstract usually states which you are reading. Guides that separate real scientific findings from misleading science coverage are useful for unfamiliar parts. Check the confidence interval or p-value if you can; a wide interval means less certainty.

Bring questions you cannot resolve to a clinician who knows your situation. A doctor can weigh a claim against your history; this is general education, not medical advice. Your health screening checklist is a better starting point for testing decisions than anything you read in a feed. Before your appointment, write down the claim and its source so you can ask about it. A clinician can tell you whether a claim applies to your specific condition, age, or medications. If the claim is about a treatment, ask whether it has been tested in a clinical trial.

Most claims will not resolve into certainty. The routine sorts them into ones you can act on and ones to keep at arm's length until better evidence appears. Pick one recurring health topic you follow, check its next three headlines against this checklist, and note which sources survive. That short list becomes your default reading on the subject.