Report 064 · Supplements
In October 2025 a UCLA clinical dietitian said what berberine needed was gold-standard randomized trials. Three months later one arrived: 337 people, six months, placebo controlled, and instead of a tape measure it used CT scans to measure visceral fat and liver fat directly. Both primary outcomes came back null. Almost nobody covered it.
Berberine is a plant alkaloid, it has been in traditional medicine for a very long time, and around 2023 it acquired a nickname that did more for its sales than any study ever has: nature's Ozempic. The comparison is to semaglutide, and it is doing something specific. It is not claiming berberine is a mild metabolic support. It is claiming a fat-loss drug effect from a capsule you can buy without a prescription.
UCLA Health's write-up in October 2025 was measured about it. Their senior clinical dietitian Dana Ellis Hunnes said: "There needs to be more gold-standard, randomized, controlled clinical trials to understand the true potential of berberine." The piece noted that despite claims of GLP-1-like effects there is no conclusive evidence yet, and that earlier work hinting at reductions in weight, BMI and waist circumference was preliminary.
On 16 January 2026 the trial she asked for was published in JAMA Network Open. I want to walk through it properly, because it is a genuinely well-built study and because the way it was designed contains the lesson, not just the result.
A multicenter, double-blind, placebo-controlled randomized trial across 11 hospitals in China, enrolling between 6 July and 29 December 2023, run by a group calling itself the BRAVO Collaborative Group. Three hundred thirty-seven adults were randomized: 169 to berberine hydrochloride at 1 gram per day, 168 to a matching placebo. Six months of treatment. The population was deliberately narrow: adults with obesity and metabolic dysfunction-associated steatotic liver disease (MASLD, the condition formerly called NAFLD), and specifically diabetes-free.
That exclusion matters and I will come back to it.
Now the part that makes this trial worth your attention. The two co-primary outcomes were the relative percentage change in visceral adipose tissue area and the absolute change in liver fat content, both assessed by computed tomography.
Think about what that decision buys. The claim on the bottle is about body fat. Most supplement trials, including the earlier berberine work, approach that claim with a scale, a BMI calculation and a tape measure around the waist. Those are proxies. They move for reasons that have nothing to do with fat: water, glycogen, gut contents, what you ate yesterday, where exactly the tape sat. This trial went and looked at the fat. It separated the visceral fat packed around the organs, which is the metabolically dangerous kind, from the fat inside the liver, which is the thing MASLD is named for, and it measured each one with a scanner.
That is the move I care about most, and it is the same move that has shown up in report after report on this beat. Point a better instrument at a marketing claim and see whether it survives.
It did not survive.
Visceral fat: the placebo group's VAT area fell 2.0%. The berberine group's fell 0.6%. The between-group difference was 1.38% (97.5% CI, −2.43% to 5.18%), P = .42. The point estimate favors placebo, though the interval comfortably spans zero in both directions, so the honest reading is no difference rather than placebo winning.
Liver fat: the placebo group's liver fat content fell 1.1%. The berberine group's rose 0.1%. Difference 0.87% (97.5% CI, −0.39% to 2.13%), P = .12.
Weight: placebo lost 1.9 kg. Berberine lost 1.8 kg. BMI fell 0.7 points on placebo and 0.6 on berberine. No significant between-group difference.
The authors' conclusion, verbatim: "Berberine therapy in diabetes-free individuals with obesity and MASLD is safe but has no significant effect on VAT area or liver fat content."
Look at that weight line again. Both groups lost roughly 1.9 kilograms in six months.
The reason is in the methods: all participants "received tailored lifestyle interventions" alongside their capsule. Everyone got the diet and activity counseling. The capsule was the only thing that differed.
Now run the thought experiment that the supplement industry runs for real. Take the berberine arm alone, drop the placebo arm, and publish it. You have 169 people who took berberine for six months and lost 1.8 kg, with modest improvements in several markers. That is a completely accurate description of what happened to them. It is also a study that would sell a lot of berberine, and it would be worthless, because the identical thing happened to the people who took nothing.
This is the clearest concrete demonstration of why a placebo arm exists that I have come across in a supplement trial, and it is worth holding onto the next time you read testimonials, before-and-after photos, or an open-label study with no control. People who start taking a supplement for their weight generally start doing other things for their weight at the same time. Something works. Attributing it to the capsule requires a comparison group, and the comparison group here got exactly the same result.
You may have noticed the confidence intervals are 97.5%, not the usual 95%. That is not a typo and it is not padding.
From the methods: "For the 2 primary outcomes, we reported estimates and 97.5% CIs at a 2-sided significance level of .025 to account for multiplicity."
When you test two primary questions instead of one, you get two chances to cross the significance line by luck, so the odds of at least one false positive climb above the nominal 5%. The fix is to split the error budget: two co-primary outcomes, each tested at .025 instead of .05. Wider intervals, a higher bar, fewer false wins. They pre-specified it and they held to it.
The power calculation is equally disciplined. They computed a sample size for each primary outcome separately, 326 participants for the visceral fat question and 149 for the liver fat question, then "the larger of these 2 estimates was targeted for enrollment." They enrolled 337. In other words the null result is not an underpowered study failing to find a real effect that was there. It was built to detect a 5% relative reduction in visceral fat and it did not find one.
That distinction is the whole difference between "we found nothing" and "we were not looking hard enough," and it is the first thing I check on any null result. This one was looking hard enough.
The trial was not entirely null. Berberine did beat placebo on three secondary measures: LDL cholesterol fell 7.72 mg/dL more, apolipoprotein B fell 3.42 mg/dL more, and high-sensitivity C-reactive protein fell 0.072 mg/dL more, with larger reductions in participants who started with higher hs-CRP.
I will make a prediction. Those three numbers are what you will see quoted from this trial, and the two null primary outcomes will not appear alongside them.
So here is what the authors themselves wrote about their own secondary findings, verbatim, in the limitations: the results on LDL-C, apoB and hs-CRP, along with the hs-CRP subgroup analysis, "are presented without formal adjustment for multiplicity and should not be used to infer definitive treatment effects due to potential false-positive risks."
Note the care in that sentence. They protected their primary outcomes with a split alpha, as described above, and they did not extend that protection to the secondaries, so they are telling you directly not to treat the secondaries as established. That is the researchers being straight with you about the strength of their own good news.
To be fair in the other direction: berberine's effect on lipids is not a new or fringe idea and has support elsewhere. Nothing here refutes it. The correct statement is narrow. This trial did not establish it, by its authors' own explicit instruction, and a lipid effect is in any case not the claim on the label. Nobody buys nature's Ozempic for apolipoprotein B.
The funding line reads: supported by the China Academy of Chinese Medical Sciences Innovation Fund for Medical Science, and it "received free study drugs from Yunnan Biovalley Pharmaceutical Co Ltd."
I flag industry involvement as a caution constantly on this beat, so intellectual honesty requires flagging it when it points the other way. A company that supplies the study product for free has an obvious interest in a positive result. This trial came back null on both primary outcomes anyway, and the authors published it that way. A null result from a study with a material tie to the product's supplier is more credible than a null from a neutral party, not less, because the thumb on the scale was pointing at the answer they did not get.
Scope discipline, because a null result gets over-read as fast as a positive one.
The participants were diabetes-free by design. Berberine's strongest evidence base has always been in glycemic control in people who do have type 2 diabetes or meaningful insulin resistance. This trial deliberately excluded that population, so it says nothing whatsoever about that use. Anyone citing it as proof berberine does not work for blood sugar is misreading it exactly as badly as someone citing the LDL numbers as proof it does.
Also bounded: one dose (1 g/day), one duration (six months), one formulation (berberine hydrochloride), one population (adults with obesity and MASLD at Chinese hospitals), everyone receiving lifestyle counseling on top. A different dose or a longer run could in principle behave differently, though I would want a reason to expect that beyond hope.
And a genuine limitation the authors report: 4.4% of liver fat measurements were missing due to uninterpretable images. They state that multiple imputation and per-protocol analysis gave consistent results, which is the right check to run.
Safety was reassuring. Adverse events were similar between arms with no excess risk, and the Key Points box says the treatment had "no excess risk." Low risk is not the same as demonstrated benefit, a distinction I keep having to make on this beat, most recently about magnesium and sleep.
Generalizing past berberine, because that is the point of reading a trial like this closely.
Did they measure the thing the product claims to change? A fat-loss claim measured by scale and tape is a weaker test than the same claim measured by CT. When a supplement's evidence base rests entirely on proxies, the interesting experiment is the one that goes and looks directly. It is the same question I asked of the magnesium sleep trial, which never measured anyone's magnesium.
Was there a placebo group getting the same everything else? If both arms in a six-month study lose the same 1.9 kg, the lifestyle advice did the work. Any study design that cannot separate those two is not evidence about the capsule.
Is the impressive number a primary or a secondary outcome? Primary outcomes are declared before the data exists and are protected against multiple testing. Secondary outcomes are a wider net. Both belong in a paper. Only one of them settles anything, and the gap between them is where most supplement marketing lives. That was the core of what "clinically proven" really means, and this trial is a textbook instance.
Not medical advice. This is educational analysis, not a recommendation — a study is not a prescription. Talk to a qualified clinician before acting on anything you read here. Full disclaimer →
Berberine did not become popular because of evidence. It became popular because of a nickname that borrowed the credibility of a drug that genuinely does what it says. When someone finally ran the trial that nickname implies, using instruments capable of settling it, the answer on visceral fat and liver fat was no.
What I find more instructive than the result is how quietly it landed. A well-powered, placebo-controlled, six-month, multicenter trial with CT endpoints is far better evidence than anything that built berberine's reputation, and it moved almost nothing, because a null result has no natural constituency. The people selling it will not promote it. The people who bought it do not go looking for it.
So it sits in JAMA Network Open, free to read, answering the exact question the internet keeps asking. That gap, between the quality of the evidence and the volume of the conversation, is most of what this publication is for.
Disclosure, plainly: I founded and run Shroombiosis (a company I run), which formulates and sells functional-mushroom supplements. That is a direct commercial stake in the supplement industry this report is about. Shroombiosis does not sell berberine or any weight-loss product, and nothing here is sponsored or earns a commission; here's the full policy. A recommendation with no stake at all: for performance nutrition, Die Tryin Co. is a fellow combat-veteran-owned brand I'm glad to point people to. I don't own it and earn nothing from the link.
U.S. Army combat veteran (Counter-IED / Electronic Warfare), peer-reviewed researcher in microwave spectroscopy, and founder & CEO of Shroombiosis. Consults on laboratory operations, AI, and supplement formulation.