The certainty column
A systematic review published this year did something most coverage of GLP-1s and PMOS (PCOS) does not: it put a confidence rating next to every single finding. Eleven trials survived the screen. Most of them did not test the drugs you are actually being offered.
Before anything else
This is a description of what one body of published evidence does and does not establish. It is not a reason to start anything, and it is not a reason to stop anything you are currently taking. Those are conversations for a clinician who knows your history.
There is a habit in health writing of reporting the number and skipping the footnote. A trial finds that something went down by some amount, the amount becomes the headline, and the part where the researchers say we are not very sure about this gets left in the paper.
Systematic reviews that use GRADE do not let you do that. GRADE is a formal rating for how much confidence the evidence deserves — high, moderate, low, or very low — and it sits in a column next to each result. A review of GLP-1 receptor agonists in PMOS (PCOS), published in the European Journal of Endocrinology in February, filled that column in for every outcome. It is worth reading the column rather than the numbers.
What survived the screen
The authors started with 9,654 records and read 365 in full. Twenty-nine articles were relevant. Twelve of those were then thrown out — five for an unacceptably high risk of bias, seven over unresolved research-integrity concerns. What remained was 17 articles reporting 11 randomised controlled trials.
Eleven trials is not a large literature for a condition affecting more than 170 million women. But the more useful detail is which trials they were. Five used exenatide. Four used liraglutide. One used beinaglutide, one dulaglutide. Six ran for twelve weeks. The longest ran for thirty-two.
The GLP-1 evidence base in PMOS is built almost entirely on molecules that are no longer the ones being prescribed, over durations of twelve to thirty-two weeks.
Read that list again and notice what is missing. Semaglutide — the molecule that accounts for a great deal of the current conversation, and appears on the intake form at nearly every service we review — does not appear in the pooled trials at all. Tirzepatide is absent for a different and less damning reason: it is a dual GIP/GLP-1 agonist, so a review restricted to GLP-1 receptor agonists would exclude it by definition. Either way, the practical position is the same. The two drugs most women are actually offered are not the drugs this evidence is made of. Every included study also enrolled only women with overweight or obesity, which means none of it speaks to the substantial group of women with PMOS who are not.
What the ratings say
On weight, the review found a real effect. As an add-on, GLP-1 treatment reduced BMI by a mean difference of 1.38 kg/m² against control, with a confidence interval running from 0.38 to 2.39. Against placebo specifically, the reduction was 2.18 kg/m². The certainty rating on that headline figure: low.
And in a sensitivity analysis — the version where the studies at high risk of bias are removed — the point estimate actually grew slightly, to 1.67, but the interval widened and the result was no longer statistically significant. That is not a scandal. It is what a small, uneven evidence base looks like when you stress it.
Everything downstream is thinner. Fasting glucose, fasting insulin and HOMA-IR each came back showing no difference between groups — but at very low certainty, which is why the authors do not report those as null results. Their abstract says something more honest and more frustrating: the evidence was insufficient to draw a conclusion at all. Waist-to-hip ratio: no difference, very low. Hirsutism: one study, insufficient. Menstrual regularity: an advantage for GLP-1s, reported too heterogeneously to pool, very low. The only outcome that reaches moderate certainty anywhere in the review's results table is LDL cholesterol — where the finding is that nothing happened. (The abstract summarises even that one more conservatively, which tells you how little in here anybody wants to stand on.)
Then there is the gap that we find hardest to read past. On quality of life, on mental health, and on cost-effectiveness, the review finds no studies. Not weak studies. None.
Why this doesn't read as a verdict
"Uncertain" is not a synonym for "doesn't work." Low-certainty evidence of benefit is not evidence of absence, and the reviewers' own conclusion is careful in exactly this way: GLP-1s are associated with modest short-term weight loss in this population, consistent with what they do in obesity generally, and the benefits beyond that remain unestablished rather than disproven. Their closing line asks for longer, better-designed trials, which is a request, not a warning.
It does mean something specific for how a service should talk to you. A platform that describes a GLP-1 as a treatment for PMOS is describing something the trial literature has not shown. A platform that describes it as a weight and metabolic intervention that may help downstream, alongside the things that are better established for insulin dysfunction, is describing something the evidence supports — modestly, over short horizons, in women with overweight or obesity.
No GLP-1 is FDA-approved specifically for PMOS. That sentence appears at the bottom of every page we publish, and the certainty column is the reason it is not boilerplate.
The version of this you can use
You are not going to litigate GRADE ratings in a fifteen-minute intake. But there is a question that carries the same content: what is this expected to change for me, and over what timeframe?
An answer that names weight, and metabolic markers, and a horizon, and admits that the reproductive and psychological benefits are hoped for rather than demonstrated, is an honest answer from someone who has read the same literature. An answer that promises your cycles back, your androgens down and your PMOS resolved is selling past the evidence — and the fact that it might still turn out to be true for you does not make it a thing anyone can currently promise.
We built the Care Standard to reward the first kind of answer. The screening schedule is the other half of it: what gets measured, and how often, so that "did this change anything for me" stops being a feeling and becomes a number you own.
The honest disclaimer. This is general information from the women who write Selene — not medical advice, not a diagnosis, and not a recommendation about any medication. Effect sizes and certainty ratings described here are drawn from one published systematic review and are not a substitute for a clinician deciding what is right for you. No GLP-1 is FDA-approved specifically for PMOS (PCOS), and GLP-1 medications are not a treatment for endometriosis. Selene is editorial: we do not prescribe, sell, or dispense anything.
Sources. Forslund M, Wändell P, Forsberg L, Österberg M, Dagerhamn J, Wernersson E, Kärrman Fredriksson M, Ringborg A, Lindén Hirschberg A, "GLP-1 receptor agonist treatment in women with polycystic ovary syndrome — a systematic review and meta-analysis," European Journal of Endocrinology, 194(3), pages S25–S39, published online 17 February 2026 (doi:10.1093/ejendo/lvag033); GRADE certainty ratings, effect estimates, study counts and the absence of quality-of-life, mental-health and cost-effectiveness data are as reported in that review. Prevalence figure (more than 170 million women worldwide) from Teede HJ et al., "Polyendocrine metabolic ovarian syndrome, the new name for polycystic ovary syndrome: a multistep global consensus process," The Lancet, published online 12 May 2026.