By Priya Nandakumar
A single scroll through a social feed can surface a dozen health claims before breakfast: a supplement that supposedly melts fat overnight, a study “proving” coffee cures headaches, a doctor on video warning that a common food is secretly toxic. Some of these claims trace back to real research. Many do not. Most people never click through to a source, and even those who do rarely have training in how to judge whether a study supports the headline built on top of it.
The good news is that separating a solid health claim from a shaky one does not require a medical degree. It requires a short, repeatable process built on the same handful of questions that researchers, journal editors, and health journalists ask before they trust a finding. This guide walks through that process step by step, with a target of under ten minutes per claim, plus the red flags that let you reject the weakest claims almost instantly and worked examples that show the checklist in action.
Why rapid evaluation matters
Health misinformation spreads differently than other kinds of misinformation. A 2018 MIT study on the spread of true and false news on Twitter, led by Soroush Vosoughi and colleagues and published in Science, found that false stories were 70 percent more likely to be retweeted than true ones, and that this effect was strongest for stories in the political and health-adjacent categories. Claims that promise a quick fix, name a villain ingredient, or contradict official guidance travel faster precisely because they are more emotionally engaging than a hedge-filled scientific summary.
At the same time, not every questionable-sounding claim is wrong, and not every claim wrapped in scientific language is right. Early observational findings sometimes hold up. Headlines from prestigious outlets sometimes overstate a single small trial. The only reliable way through this is a consistent evaluation habit rather than a gut check based on how confident the source sounds.
The ten-minute evaluation framework
The checklist below is built around six questions. Each one takes roughly one to two minutes once you have practiced it a few times. Working through all six in order, rather than jumping straight to “does this sound plausible,” keeps you from being swayed by a claim that only passes one or two of the tests.
Step 1: Trace the claim to its original source (2 minutes)
Before judging the claim, find out where it actually comes from. A headline, an influencer post, or a supplement ad is not a source; it is a retelling, often several retellings removed from the underlying research. Search for the name of the study, the lead author, or the journal it supposedly appeared in. Media literacy researchers Mike Caulfield and Sam Wineburg, formerly of the Stanford History Education Group, popularized a technique called lateral reading: instead of scrutinizing the page in front of you, open new tabs and see what independent sources say about the claim and the outlet reporting it. If a claim cannot be traced past a wellness blog or a press release, treat it as unverified rather than false, but do not treat it as established either.
Step 2: Identify the study type and where it sits on the evidence hierarchy (2 minutes)
Not all research carries the same weight. Evidence-based medicine organizes study designs into a rough hierarchy based on how well each design controls for bias and chance. The GRADE framework, developed by the international GRADE Working Group and used by the Cochrane Collaboration, the World Health Organization, and hundreds of clinical guideline panels, rates the certainty of a body of evidence as high, moderate, low, or very low based partly on study design and partly on how consistent, precise, and direct the findings are.
A quick way to gauge study type from a news article: look for the words “randomized,” “controlled,” “double-blind,” “cohort,” “observational,” or “in mice” / “in vitro.” Articles built on animal or lab-dish research routinely describe results as though they apply directly to people, which they frequently do not.
Step 3: Check the sample size and duration (1 minute)
A study with 18 participants followed for two weeks cannot tell you much about long-term safety or an effect size you should expect in the general population. Small samples produce noisy, unstable results: a handful of participants responding unusually well or poorly can swing the reported average substantially. As a rough gut check, treat findings from fewer than roughly 50 to 100 participants as preliminary, and treat findings from a study lasting only a few weeks as uninformative about anything described as a long-term or chronic-disease outcome.
Step 4: Look for conflicts of interest and funding sources (2 minutes)
Funding does not automatically invalidate a study, but it is a variable worth knowing. In the United States, the Physician Payments Sunshine Act, part of the 2010 Affordable Care Act, requires drug and device manufacturers to report payments made to physicians and teaching hospitals, and that data is published in the Centers for Medicare and Medicaid Services’ Open Payments database. Most reputable journals also require authors to disclose funding sources and financial ties in a conflict-of-interest statement, usually located at the end of the article. A supplement study funded entirely by the company that sells the supplement, with no independent replication, deserves more scrutiny than one funded by a public research agency.
Step 5: Check whether the finding has been replicated (2 minutes)
One study is a data point, not a conclusion. The Open Science Collaboration’s 2015 project, published in Science, attempted to replicate 100 psychology studies and found that only about 36 percent produced a statistically significant result matching the original, compared with 97 percent of the original studies reporting a significant effect. Medicine has documented similar patterns. Search for the claim alongside terms like “replication” or “meta-analysis,” or check whether a systematic review has since pooled it with other studies. A single splashy finding that has stood alone for years without follow-up work is a caution sign, not a confirmation.
Step 6: Ask whether the claim is biologically plausible and proportionate (1 minute)
Finally, weigh the claim against basic plausibility and against the size of the effect being promised. A claim that a specific food “boosts metabolism by 300 percent” or “detoxifies” the liver through a mechanism no textbook describes should raise questions regardless of how the study is dressed up. Extraordinary claims, such as a single food curing a chronic disease or a supplement outperforming a class of prescription drugs in days, need correspondingly strong, well-replicated evidence, not one small trial with an eye-catching press release.
Red flags that let you reject a claim faster
Some claims fail so many checks at once that you do not need the full ten minutes. The callout box below collects the signals that, in combination, are strong predictors of a low-quality claim.
Red Flags Checklist: Low-Quality Health Claims Usually Show Several of These
- Uses words like “miracle,” “secret,” “cure,” “breakthrough,” or “doctors don’t want you to know”
- No named study, journal, or researcher — only “a new study shows” or “scientists say”
- Only source is a press release, product page, or single influencer, with no independent coverage
- Effect size sounds implausibly large for a short intervention (dramatic weight loss, disease reversal in days)
- Study was in animals, cells, or a tiny human sample, but the headline implies a proven effect in people
- No mention of a control group or comparison condition
- The only people quoted defending the claim are financially tied to the product
- No sign the finding has been replicated, reviewed, or cited by other researchers
- Claim asks you to distrust an entire field of medicine rather than question one specific finding
If a claim trips three or more of these boxes, it is reasonable to set it aside without spending the full ten minutes on formal evaluation.
Worked examples
Example 1: “New study finds cinnamon lowers blood sugar as effectively as medication”
Tracing the claim leads to a small trial with around 25 to 30 participants, run over 40 days, measuring fasting blood glucose in people with type 2 diabetes. That places it at level 2 on the evidence hierarchy, a randomized trial, but with a small sample and short duration, so step 3 flags it as preliminary. No mention of comparison against an actual diabetes medication appears in the original paper, only in the headline built on top of it. Searching for follow-up work turns up a mixed picture across several small trials, with a 2013 Cochrane-style review noting inconsistent effect sizes. Verdict: a real but modest, unreplicated-at-scale finding, seriously overstated by the headline comparing it to medication.
Example 2: “Doctor reveals the one vitamin big pharma doesn’t want you to take”
This one collects red flags immediately: no named study, a framing that asks you to distrust an entire industry, and a product link in the same post. There is no traceable source beyond the video itself. Under the red-flags checklist, this can be dismissed in under a minute without a full six-step review, though dismissing the marketing does not mean the vitamin itself is useless, only that this particular claim about it is unsupported.
Example 3: “Large cohort study links regular walking to lower dementia risk”
Tracing the source leads to a cohort study following tens of thousands of adults over several years, published in a peer-reviewed journal with a clear methods section. That places it at level 3 on the hierarchy, an association, not a randomized proof of causation. The study discloses public research funding, and similar associations appear in several independent cohorts, satisfying the replication check. Plausibility is high, since physical activity’s cardiovascular and metabolic benefits are well established through other lines of evidence. Verdict: a reasonably solid association worth taking seriously, described honestly as correlational rather than as proof that walking prevents dementia.
Common mistakes people make when weighing evidence
Even people who know the checklist trip over a few recurring errors. Naming them in advance makes them easier to catch in the moment.
Treating peer review as a guarantee of truth. Peer review filters out some obvious flaws before publication, but it does not test whether results replicate, and it does not catch every methodological problem. Plenty of peer-reviewed papers have later been corrected, retracted, or simply failed to hold up under further scrutiny. Peer review is a floor, not a ceiling.
Confusing statistical significance with practical importance. A large study can detect a real but tiny effect, one so small it makes no meaningful difference to a person’s health, and still report it as statistically significant. Before acting on a finding, ask not just whether the effect is real but whether it is big enough to matter.
Rejecting an entire field because of one bad actor. A single retracted paper, a single scandal involving a researcher who fabricated data, or a single industry-funded study with weak methods does not mean an entire area of medicine is compromised. Discrediting one flawed piece of evidence is different from discrediting the broader body of work built from many independent groups.
Assuming correlation implies causation. Cohort studies can show that people who do X also tend to have outcome Y, but they cannot, on their own, rule out that a third factor explains both. Income, education, baseline health status, and access to care all tend to correlate with many lifestyle behaviors, which is why researchers spend so much effort on statistical adjustment and why randomized trials remain valuable when they are ethical and feasible.
Anchoring on the first version of a claim you saw. Once a claim has been repeated across several posts, it starts to feel familiar, and familiarity is easily mistaken for credibility. Psychologists call this the illusory truth effect: repeated exposure to a statement increases the likelihood people rate it as true, independent of whether it actually is. Running the six-step check again, even on a claim you have seen many times before, guards against this.
Applying the framework on social media specifically
Short-form video and image posts strip away most of the context a full article would normally include: no methods section, no disclosure statement, often no named study at all. That makes steps 1 and 4 of the framework harder to complete directly from the post itself. A practical workaround is to search the claim in a few extra words, such as the ingredient or intervention name plus “study” or “randomized trial,” rather than relying on whatever caption accompanies the video. If a creator cites a real journal, searching the journal name alongside the topic will usually surface the actual paper or a summary of it within a minute or two.
It also helps to notice the account’s business model. A creator selling the exact product mentioned in the claim has a direct financial incentive that a claim from an independent educator, clinician, or science journalist does not share. That does not make the claim automatically false, but it does mean conflicts of interest deserve extra weight when the person making the claim and the person profiting from it are the same person.
When to bring a claim to a clinician instead of evaluating it alone
The framework in this guide is designed for evaluating the credibility of a general claim, not for making individual medical decisions. Some situations call for a conversation with a doctor or pharmacist rather than a solo search session: when a claim involves stopping or changing a prescribed medication, when it involves a child, pregnancy, or an existing chronic condition, or when the intervention carries a real risk of harm, such as a high-dose supplement or an unregulated product. In those cases, the ten-minute framework is still useful preparation. Arriving at an appointment with the original study, the sample size, and the funding source already identified turns a vague question into a specific, answerable one.
Building the habit
The six-step framework works best as a habit rather than a one-time checklist you consult only for claims that already seem suspicious. Applying it consistently, including to claims that confirm what you already believe, is what separates genuine evaluation from confirmation bias wearing a lab coat. Health journalists at outlets that once ran structured review projects, such as the now-archived HealthNewsReview.org, used a similar approach for over a decade: checking whether an article quantified benefits and harms, disclosed conflicts of interest, compared the intervention to existing alternatives, and avoided disease-mongering language. Running the same checks yourself, in a few focused minutes, gives you a private version of that editorial process for anything that crosses your feed.
None of this requires becoming a skeptic of medicine as a whole. It requires treating each new claim as a small research question: where did this come from, what kind of study produced it, how many people were involved, who paid for it, has anyone else found the same thing, and does the size of the claimed effect match the strength of the evidence behind it. Ten minutes spent on those six questions will not catch everything, but it will catch most of what matters.
Frequently asked questions
Is it really possible to evaluate a health claim in under ten minutes?
For most claims circulating online, yes. The red-flags checklist alone can eliminate the weakest claims in under a minute, and the full six-step framework of source, study type, sample size, conflicts of interest, replication, and plausibility typically takes eight to ten minutes once you have practiced it a handful of times.
What if I can’t find the original study at all?
Treat the claim as unverified rather than assuming it is either true or false. A claim with no traceable source has already failed step one, and it is reasonable to withhold judgment or simply set it aside until a primary source surfaces.
Does industry funding automatically mean a study is wrong?
No. Funding source is one input, not a verdict. A well-designed, adequately powered, independently replicated study funded by an interested party is still meaningful evidence, though it deserves closer scrutiny of the methods and a check for whether independent researchers have found similar results.
Are observational studies useless if they can’t prove causation?
Not at all. Cohort and case-control studies generate much of what is known about long-term risk factors, precisely because randomized trials are often impractical or unethical for questions like smoking or diet over decades. The key is describing them accurately as associations rather than proof of cause and effect.
How do I evaluate a claim when I don’t have medical training?
The framework in this guide was designed for exactly that situation. None of the six steps require interpreting statistics or reading a full paper; they require reading an abstract, checking a disclosure statement, and doing a couple of searches to see whether other sources or researchers back up the claim.
What’s the single fastest check if I only have one minute?
Run the claim against the red-flags checklist. A claim promising a dramatic effect, naming no study, and citing only a product page or single influencer can usually be set aside immediately, without a full evaluation.
References
- GRADE Working Group. “GRADE guidelines” and certainty-of-evidence framework. gradeworkinggroup.org
- Cochrane Collaboration. “About Cochrane Reviews” and systematic review methodology. cochrane.org
- Open Science Collaboration. “Estimating the reproducibility of psychological science.” Science, 2015. science.org/doi/10.1126/science.aac4716
- Ioannidis, J.P.A. “Why Most Published Research Findings Are False.” PLOS Medicine, 2005. journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.0020124
- Vosoughi, S., Roy, D., Aral, S. “The spread of true and false news online.” Science, 2018. science.org/doi/10.1126/science.aap9559
- Centers for Medicare and Medicaid Services. Open Payments database (Physician Payments Sunshine Act data). openpaymentsdata.cms.gov
- Caulfield, M. and Wineburg, S. Stanford History Education Group, lateral reading research. sheg.stanford.edu
- HealthNewsReview.org (archived). Ten-criteria review methodology for health journalism, 2006-2018.


























