Australian researchers set out to pool the randomised trials of berberine in polyendocrine metabolic ovarian syndrome, or PMOS, the condition previously called polycystic ovary syndrome. Earlier reviews had presented the supplement as a match for metformin. Every eligible trial, four of them covering 452 women, was rated at high risk of compromised integrity, so none was pooled. The review, published on 6 October in Advances in Therapy, concludes that no reliable conclusion about berberine in PMOS can be drawn from the current evidence.
How the trials were checked
The team searched MEDLINE, EMBASE, the Cochrane Library and CINAHL up to October 2025 for randomised controlled trials of berberine in women with the condition, and registered the review in advance. Before any data were combined, each eligible trial was assessed independently with two published tools: the Research Integrity in Guidelines and Evidence Synthesis (RIGID) framework and the Trustworthiness in Randomised Controlled Trials (TRACT) checklist. The rule, set beforehand, was that a trial rated at moderate or high risk of compromised integrity would be left out of the pooled analysis, and that its authors would be contacted for clarification first.
All four trials were rated high risk. The concerns the reviewers list are an absence of trial registration, recruitment timelines they judged implausible, and possible duplication of data. None of the trial authors responded to requests for clarification. The four trials had reported improvements in insulin resistance, blood lipids and androgen markers, but the reviewers say those findings cannot be considered reliable.
This is not a finding that berberine does nothing. It is a finding that these trials cannot answer the question either way: once they are set aside, the reliable evidence base is empty.
What this changes
A conventional risk-of-bias assessment asks how well a trial was designed and reported, and generally takes for granted that the trial ran as described. Integrity tools ask an earlier question: are there signs that the data, the timeline or the registration do not add up? Applied here, they removed every trial behind the “as good as metformin” reading of the earlier reviews.
That matters for anyone weighing berberine against a prescribed drug. In the US, dietary supplements do not need FDA approval for effectiveness before they are sold, so published trials are usually the only evidence there is. The site’s guide to supplements marketed alongside GLP-1 drugs covers berberine’s wider evidence. Our News reports in September on a 93-person berberine combination trial, and last week on 34 lifestyle trials in PMOS, were based on different studies; this review does not assess either of them.
What is not yet known
The review covers berberine in PMOS only. It says nothing about berberine for weight loss or type 2 diabetes in general, where the trial base is different. Four trials is a small set, and integrity ratings involve judgement, although two separate frameworks were applied and the trial authors were given a chance to respond. We read the published abstract; we could not retrieve the full open-access text, so the individual trials are not named here. One author sits on the journal’s editorial board and, according to the disclosure, had no role in its peer review. A reliable answer would need new, prospectively registered trials.
Sources: Perera E, Mantzioris E, Moran L, et al. “Questionable Evidence for Berberine in Polyendocrine Metabolic Ovarian Syndrome: An Integrity-Based Reappraisal”, Advances in Therapy, published online 6 October 2026, doi:10.1007/s12325-026-03799-7.
