The phrase "GLP-1 supplement" describes a product that contains no GLP-1. That is not a gotcha — the sellers mostly say so themselves. The premise is indirect: that a botanical ingredient prompts the gut's own L cells to release more of the body's own hormone.
That premise is testable, and on the ingredient the category is built around, it has been tested. This page reports what those trials found, including a large one that found nothing.
Every figure below carries its source and the date we read it. Nothing here is advice about what anyone should use; that belongs with a clinician or a pharmacist, as our editorial standards set out.
Berberine: the biggest trial in the category, and it was negative
Berberine is the single ingredient common to almost every product in this market. It is in the patches, it is in the capsules, and it is the compound behind the "nature's Ozempic" phrasing that circulates online.
In January 2026, JAMA Network Open published the BRAVO trial (Lei et al, 9(1):e2554152, registered as NCT05647915). It is the kind of study this field almost never gets:
- 337 participants, diabetes-free, with obesity and metabolic dysfunction-associated steatotic liver disease
- 11 hospitals, enrolled between 6 July and 29 December 2023
- double-blind, randomised, placebo-controlled
- 1 g of oral berberine a day for six months
- adherence of about 90% in both arms
- primary outcomes measured by CT, not by scales or tape
The primary results:
| Outcome | Berberine vs placebo | 97.5% CI |
|---|---|---|
| Visceral adipose tissue area (relative change) | +1.4% | −2.4% to 5.2% |
| Liver fat content (absolute change) | +0.9% | −0.4% to 2.1% |
Both intervals cross zero. The authors' conclusion is unhedged: six months of berberine at 1 g a day "had an excellent safety profile but did not reduce VAT area or liver fat content."
What it did change should be reported too, because it is real and it is not nothing:
| Marker | Difference vs placebo | 95% CI |
|---|---|---|
| LDL cholesterol | −7.72 mg/dL | −13.13 to −1.93 |
| Apolipoprotein B | −3.42 mg/dL | −6.33 to −0.51 |
| High-sensitivity CRP | −0.072 mg/dL | −0.140 to −0.004 |
So berberine did something measurable to lipids and inflammation, in a trial large enough to detect it, and did not do the thing it is sold for.
The mechanism papers and the outcome trial disagree
Search PubMed for papers with both berberine and GLP-1 in the title and you get six records, as of 2026-09-16. They are genuine research and they support the mechanism: berberine metabolites stimulating GLP-1 secretion by relieving oxidative stress (Am J Chin Med, 2024); berberine acting on GLP-1 signalling in db/db mice (Front Pharmacol, 2023); berberine promoting GLP-1 production in intestinal L cells (Eur J Pharmacol, 2021); restoration of GLP-1 secretion in colon enterocytes (Nutr Diabetes, 2018).
Every one of them is a cell or animal study. None measures GLP-1 in people.
This is a common and instructive shape. A mechanism can be real in a dish and still not produce an outcome in a person, because the step from "this cell released more hormone" to "this human lost visceral fat" contains a great deal of biology. When the mechanism papers and the outcome trial point different ways, the outcome trial is the one that answers the question a reader is actually asking.
Akkermansia: the one positive randomised result, and what it actually measured
The strongest supplement evidence in this space is not about berberine and is not about losing weight.
In June 2026 Nature Medicine published a randomised controlled trial of pasteurised Akkermansia muciniphila MucT (Mount et al, 32(6):2107-2116, NCT05417360). Ninety adults with overweight or obesity first lost at least 8% of body weight on an 8-week low-energy diet, then spent 24 weeks on a healthy ad-libitum diet with daily supplementation or placebo.
| Over 24 weeks of maintenance | MucT | Placebo | P |
|---|---|---|---|
| Weight regained | 1.2 ± 0.7 kg | 3.2 ± 0.4 kg | 0.012 |
| Net loss from baseline, difference | 3.1 ± 0.7 kg greater | — | 0.009 |
No serious treatment-related adverse events were observed.
This is a real, registered, peer-reviewed positive result and it should be read for what it is. It is a trial about regain, not about loss. Every participant had already lost the weight, by dieting, before the supplement started. The finding is that people on the supplement put back about 2 kg less over the following six months.
Three limitations belong with it, and two of them come from the authors:
- The authors name the short intervention and the absence of a control group receiving modified strains lacking the active components as limitations.
- The trial was part-run by the company that sells the organism. Several authors are employees, a co-founder and chief technology officer, and a chief medical officer of The Akkermansia Company SA; another is a paid scientific advisor; several are named inventors on a related patent application. All of this is declared in the paper, which is how it should be, and all of it is relevant when weighing the result.
- The baseline Akkermansia abundance was associated with response, meaning the effect was not uniform across participants.
The legal line between a capsule and a patch
One thing separates the products on this page from the patch products sold alongside them, and it is not a matter of degree.
The Dietary Supplement Health and Education Act defines a dietary supplement as a product intended for ingestion. A capsule is ingested. On the analysis published by White, Tai and Nuzi in Annals of Pharmacotherapy in August 2026 (60(8):813-816), a supplement worn on the skin is therefore not a dietary supplement at all, and the whole transdermal category is unlawfully marketed independently of what it contains.
So the oral products on this page clear a bar the patch products do not. That is a statement about statutory categories, not about whether they work.
The same survey found that none of the 24 patch products and 1 gel it examined posted a certificate of analysis. That absence applies to much of the oral market too. Where a supplement does not publish one, nobody outside the company has verified that the panel matches the contents.
How to read a claim in this category
Three questions separate a supportable claim from an unsupportable one, and they can be applied to any product here without any specialist knowledge.
Is the evidence in people? A mechanism in intestinal cells is a reason to run a trial, not a result. Berberine has both, and they disagree.
What did the trial measure? "Reduced weight regain after a diet" and "caused weight loss" are different claims, and the one positive randomised result in this space is the first kind.
At what dose, by what route? The berberine trial used 1,000 mg a day, swallowed. A patch declaring 400 mg on an adhesive is not a smaller version of that study; it is a different question that has not been studied at all, since a search for "transdermal berberine" on ClinicalTrials.gov returns zero results.
Where this leaves the category
There is one large, well-conducted randomised trial of the ingredient these products are named for, and it did not find a body-composition effect. There is one positive randomised trial of a different ingredient, it concerns maintenance rather than loss, its effect is around 2 kg over six months, and it was run in part by the seller.
Against that, the approved medicines have trial results in the mid-teens to about twenty per cent of body weight, which we set against each other on tirzepatide vs semaglutide — but the more useful comparison for anyone weighing a supplement is not the effect size. It is that one column of this market has registered human trials with posted outcomes at all, and the other mostly does not.
The related pages: why the chemistry rules the patch category out; a dated brand-by-brand census on GLP-1 patch brands; and the hormone all of this is named after, including why its half-life is the number that matters, on natural GLP-1.