How to Read a Weight-Loss Study Headline in 5 Steps

Key takeaways

• Of 76 highly cited animal studies, 37% were confirmed in human trials, 18% contradicted and 45% never tested in people.^[3]
• Relative and absolute numbers differ: in SELECT, heart events fell from 8.0% to 6.5%, a 20% relative but 1.5-point absolute reduction.^[9]
• Study length matters: one year after stopping semaglutide, STEP 1 participants had regained about two-thirds of the weight they lost.^[7]
• The comparison group usually loses weight too; the useful number is the difference, such as 6.4% vs 4.4% for Plenity vs a sugar capsule.^[8]
• Industry-sponsored drug and device studies report favorable results more often (risk ratio 1.27), so check who paid.^[12]

In short: Before you believe a weight-loss headline, ask five questions: who was studied (people or mice, and people like you?), how many, for how long, compared with what, and who paid. Then look for the absolute numbers behind any percentage. A drug that “cuts heart risk by 20%” lowered the share of people with a heart attack, stroke or heart death from 8.0% to 6.5% in its trial: real and useful, but smaller than the headline sounds.

Weight-loss news moves fast, and much of it starts as a press release, a single small study or an experiment in mice. This guide walks through five checks, each with a real example from the research we use on InstaTuck. It is the same method behind our editorial standards, and it works for any health story. You can find more of our research explainers in the InstaTuck research library and our Learn hub.

The National Institutes of Health suggests a very similar checklist for reading health news: was it tested in people, were there enough of them, were they similar to you, was the study long enough, is it the first study to find this, and did someone with a financial interest pay for it. The CONSORT 2025 statement, the international standard for reporting randomized trials, asks researchers to report the same details in a 30-item checklist, so good papers make these answers easy to find.

Step 1: Who was studied: mice, or people like you?

Many exciting headlines come from animal or lab studies. They are an important first step, but they often do not carry over to people. When researchers followed 76 highly cited animal studies published in top journals, 37% were later confirmed in human randomized trials, 18% were contradicted, and 45% had never been tested in people at all.

• Confirmed in human trials37%37%
• Contradicted in human trials18%18%
• Never tested in people45%45%

What happened to 76 highly cited animal studies when tested in people

Item	Value
Confirmed in human trials	37%
Contradicted in human trials	18%
Never tested in people	45%

Studies from leading journals, 1980 to 2000. Shares of the 76 studies.

Source: Hackam DG, Redelmeier DA. JAMA 2006 (checked on October 7, 2026)

Brown fat is a good weight-loss example. Small mammals rely on it to make heat, so it looks powerful in rodent studies. In adults, direct measurement found that brown fat burned only about 15 to 25 extra calories a day during mild cold, even in people with plenty of it. Our article on brown fat, cold plunges and saunas tells that story in full.

Even in human studies, check whether the volunteers resemble you. A trial in young men may not apply to women after menopause, and results in people with type 2 diabetes are often different: in the trials of the newest medicines, people with diabetes usually lost less weight than people without it. The “who was studied” section of a paper, often called “participants” or “eligibility”, tells you.

Step 2: How many people took part?

Small studies can find big effects by chance, and those effects often shrink or vanish when the work is repeated with more people. Spot reduction is a clear example. A small 2023 trial in 16 men reported extra trunk-fat loss with long sessions of ab exercises. But a 2022 pooled analysis of 13 studies with 1,158 people found no spot-reduction effect at all. Until the small result is repeated, the bigger body of evidence is the safer guide; our page on whether you can spot-reduce fat explains both.

Size also decides what a study can detect. A trial of a few hundred people can show average weight change, but rare side effects, such as a problem that affects one person in a thousand, only appear in very large trials or after a treatment is widely used. That is one reason the FDA keeps watching medicines and devices after approval. As a rough rule, be most cautious with a single study of under 100 people and most confident in results repeated across several large trials.

Step 3: How long did it last?

Weight usually comes off fastest in the first months and the real test is what happens next. A diet study of 8 weeks cannot tell you whether people stayed on it for a year. Even strong results can change once treatment ends. In the STEP 1 trial, adults on semaglutide 2.4 mg (Wegovy) lost 17.3% of their weight over 68 weeks in the group followed afterward. One year after stopping the medicine and the lifestyle program, they had regained about two-thirds of it and were 5.6% below where they started.

17.3%Week 68, end of treatment

5.6%Week 120, one year after stopping

STEP 1 extension: average weight change on and after semaglutide

Item	Value
Week 68, end of treatment	17.3%
Week 120, one year after stopping	5.6%

Percent of starting body weight lost, compared with the start of the trial. Exploratory analysis of a subset of participants.

Source: Wilding JPH et al., Diabetes, Obesity and Metabolism 2022 (STEP 1 extension, 327 participants) (checked on October 7, 2026)

That does not mean the medicine failed; it shows that obesity is a long-term condition, which is why the study length matters so much. Our guide to life after stopping a GLP-1 covers what helps. When you read a headline, look for the length in weeks and ask whether anyone was followed after the treatment stopped.

Step 4: Compared with what, and by how much?

In most weight-loss trials, the comparison group loses weight too, because everyone gets diet and activity advice. So the number that matters is the difference between the groups. In the trial behind the Plenity capsule, people lost 6.4% of their body weight, which sounds solid, but people taking a look-alike capsule filled with cane sugar lost 4.4%. The device added about 2 percentage points. Our list of FDA-authorized weight-loss devices shows the same pattern across devices.

Then check whether a percentage is relative or absolute. In the SELECT trial of 17,604 adults with heart disease and overweight or obesity, a heart attack, stroke or heart-related death occurred in 8.0% of people on placebo and 6.5% of people on semaglutide over about three years. That is a 20% relative reduction (the hazard ratio was 0.80), the figure most headlines used. The absolute reduction was 1.5 percentage points: roughly 67 people would need treatment for about three years to prevent one such event, by our arithmetic. Both numbers are true; the absolute one tells you more about your own odds.

8%Placebo

6.5%Semaglutide 2.4 mg

SELECT trial: share of people with a heart attack, stroke or heart death

Item	Value
Placebo	8%
Semaglutide 2.4 mg	6.5%

Mean follow-up 39.8 months. Relative reduction 20% (hazard ratio 0.80); absolute difference 1.5 percentage points.

Source: Lincoff AM et al., New England Journal of Medicine 2023 (SELECT, 17,604 adults) (checked on October 7, 2026)

Watch for two different ways of counting in medicine trials, too. The “treatment-regimen” result counts everyone who was randomized, including people who stopped early; the “efficacy” result estimates what would have happened if everyone had stayed on the drug, and it is always higher. Company press releases often lead with the efficacy number. Our retatrutide results explainer shows how one trial produced 25.0% by one method and 28.3% by the other. And an average is not a promise: in any trial, some people lose much more than the average and some lose little.

Finally, ask what kind of study it is. In a randomized trial, chance decides who gets the treatment, so the groups start out alike. In an observational study, researchers compare people who already differ, such as those who walk more and those who walk less, and other differences can explain the result. Observational findings show links, not proof of cause. Our 10,000 steps article is built on large observational studies and says so.

Step 5: Who paid, and where was it published?

Most large medicine trials are paid for by the company that makes the drug; SELECT, for example, was funded by Novo Nordisk. That does not make a result wrong, because big trials are expensive and are registered and peer reviewed. But funding is worth knowing. A Cochrane review of 75 studies found that industry-sponsored drug and device studies reported favorable efficacy results more often than studies with other funding (risk ratio 1.27). We disclose funding when it matters; our fiber supplements comparison, for example, notes that a key psyllium review was written by authors from its manufacturer.

Where the claim appears matters as much. A peer-reviewed journal paper has been checked by outside experts. A company press release or a conference “topline” result has not, and the full paper may tell a more mixed story. A trial registered in advance on ClinicalTrials.gov lets anyone see what the researchers planned to measure, which makes it harder to cherry-pick results later. Supplement marketing is the extreme case: “clinically studied” can refer to a tiny, unregistered study of one ingredient at a different dose. Our guides on how to read a supplement label and “Nature’s Ozempic” supplements show what to look for.

How to find the study behind a headline

You do not need a science degree to check a source. A good news story names the journal, the lead author or the trial name. Search that name on PubMed, the free database run by the National Library of Medicine, and you will usually find the abstract, a one-paragraph summary with the methods and main numbers. For trials, search the trial name or its NCT number on ClinicalTrials.gov: the record shows how many people were planned, how long the trial runs, what the main outcome is, who sponsors it and whether results have been posted.

When you open an abstract, read it in this order: the methods (who, how many, how long, compared with what), then the results (look for numbers in both groups, not just the difference), then the conclusions, and finally the funding line, which often appears at the end. If a story gives no way to find the study at all, or links only to a product page, treat the claim as unverified. When we write about a study on InstaTuck, we link the paper or the official record in our sources so you can check it yourself.

Putting it together: one headline, five checks

Imagine the headline “New injection wipes out 25% of body weight”. Run the checks. Who: adults with obesity and without diabetes, in a large trial. How many: 2,339. How long: 80 weeks. Compared with what: placebo, which gave 3.9%, so the difference is about 21 points, counting everyone who started. Who paid and where: the maker, with results published in a major peer-reviewed journal. That is strong evidence for the average weight change. It still leaves questions a headline skips: side effects, what happens after stopping, how it compares with existing medicines, and whether the FDA approves it at all. For drugs still in trials, see our tracker of treatments awaiting FDA decisions and our monthly news roundup.

The same five questions protect you from the other end of the market: before-and-after photos and testimonials skip every step at once. Our article on before-and-after photos explains why, and our list of weight-loss myths science has busted shows the method at work. When a study does seem to fit your situation, take it to your clinician; they can tell you whether it applies to you.

Questions to ask a professional

• Does this study include people like me in age, health and weight?
• What was the actual difference between the groups, in percentage points?
• Has this finding been repeated in other studies?
• Does this change anything about my current plan?

Frequently asked questions

What is the difference between relative and absolute risk?

Relative risk compares the groups as a ratio; absolute risk is the actual difference in percentage points. In the SELECT trial, heart events dropped from 8.0% to 6.5%: a 20% relative reduction, but a 1.5-point absolute one.

Should I trust a study done in mice?

Treat it as an early clue, not an answer. In one review of 76 highly cited animal studies, only 37% were later confirmed in human randomized trials. Wait for studies in people.

How big should a weight-loss study be?

There is no single cut-off, but be cautious with one small study, especially under 100 people. Results repeated across several large randomized trials are the most reliable.

Does company funding mean a study is wrong?

No. Most large drug trials are paid for by the maker, and many are registered and peer reviewed. Funding is still worth knowing: industry-funded studies report favorable results more often, so look for independent confirmation.

What is an estimand?

It is the question a trial result answers. The treatment-regimen estimand counts everyone randomized, including people who stopped; the efficacy estimand estimates the result if everyone had stayed on treatment, so it is higher.

References

1. Checklist for Understanding Health News Stories (Know the Science). National Center for Complementary and Integrative Health, NIH. (accessed October 7, 2026) Government page
2. Hopewell S, Chan AW, Collins GS, et al.. CONSORT 2025 statement: updated guideline for reporting randomised trials. BMJ, 2025. PMID 40228833 (accessed October 7, 2026) Guideline
3. Hackam DG, Redelmeier DA. Translation of research evidence from animals to humans. JAMA, 2006. PMID 17032985 (accessed October 7, 2026) Review
4. Muzik O, Mangner TJ, Leonard WR, Kumar A, Janisse J, Granneman JG. 15O PET measurement of blood flow and oxygen consumption in cold-activated human brown fat. Journal of Nuclear Medicine, 2013. PMID 23362317 (accessed October 7, 2026) Other
5. Ramirez-Campillo R, Andrade DC, Clemente FM, Afonso J, Pérez-Castilla A, Gentil P. A proposed model to test the hypothesis of exercise-induced localized fat reduction (spot reduction), including a systematic review with meta-analysis. Human Movement, 2022. doi:10.5114/hm.2022.110373 (accessed October 7, 2026) Meta-analysis
6. Brobakken MF, Krogsaeter I, Helgerud J, et al.. Abdominal aerobic endurance exercise reveals spot reduction exists: A randomized controlled trial. Physiological Reports, 2023. doi:10.14814/phy2.15853 · PMID 38010201 (accessed October 7, 2026) Randomized trial
7. Wilding JPH, Batterham RL, Davies M, et al.. Weight regain and cardiometabolic effects after withdrawal of semaglutide: The STEP 1 trial extension. Diabetes, Obesity and Metabolism, 2022. doi:10.1111/dom.14725 · PMID 35441470 (accessed October 7, 2026) Randomized trial
8. De Novo classification request for Plenity (DEN180060). U.S. Food and Drug Administration, 2019. (accessed October 7, 2026) Government page
9. Lincoff AM, Brown-Frandsen K, Colhoun HM, et al.. Semaglutide and Cardiovascular Outcomes in Obesity without Diabetes (SELECT). New England Journal of Medicine, 2023. doi:10.1056/NEJMoa2307563 · PMID 37952131 · NCT03574597 (accessed October 7, 2026) Randomized trial
10. Jastreboff AM, Kaplan LM, Davies MJ, et al.. Retatrutide, a Triple Hormone Receptor Agonist, for Treatment of Obesity (TRIUMPH-1). New England Journal of Medicine, 2026. doi:10.1056/NEJMoa2604169 · PMID 42814954 · NCT05929066 (accessed October 7, 2026) Randomized trial
11. Lilly’s triple agonist, retatrutide, delivered substantial weight loss and A1C reduction (BLA planned Q1 2027). Eli Lilly and Company (press release), 2026. (accessed October 7, 2026) News (reported facts only)
12. Lundh A, Lexchin J, Mintzes B, Schroll JB, Bero L. Industry sponsorship and research outcome. Cochrane Database of Systematic Reviews, 2017. PMID 28207928 (accessed October 7, 2026) Meta-analysis
13. ClinicalTrials.gov records for the candidates listed (API v2 queries). U.S. National Library of Medicine. (accessed October 7, 2026) Government page

Facts checked on October 7, 2026

Educational information, not medical advice. Talk with a qualified healthcare professional about your own situation. In an emergency call 911.