PCOS is one of the commonest reasons women look at these drugs, and what the evidence actually supports is set out in nobody has priced it for PCOS. This study measures something different: how quickly prescribing moved [1].
Within a year of a PCOS diagnosis, the share of women starting one of these drugs went from 0.14% in 2018 to 5.97% in 2024. Calendar year alone was strongly associated with initiation after adjustment (OR 1.64, 95% CI 1.56–1.73), and the pattern held after excluding women with diabetes.
What the study does not contain is any outcome at all. No weight, no menstrual cycles, no ovulation, no fertility, no androgen levels. It counts prescriptions. A rising line here describes what clinicians decided, and reading it as evidence that the drugs work for PCOS would be inferring a result from a decision — a distinction this desk has had to draw before in perinatal prescribing rose thirtyfold.
The BMI figures show prescribing tracking weight closely: a median of 31.2 among users against 26.3 for women on metformin alone and 22.1 for the untreated. That is broadly what you would want to see, and it also means the three groups are incomparable on anything else. Users also carried a greater comorbidity burden.
Two boundaries limit how far this reads across. It is a private healthcare network in Poland, so these are women paying privately or privately insured in one country — a market, not a population. And prescription records are not dispensing records: the authors call for studies using dispensing, persistence and outcome data, which is the same gap that makes the persistence figures in two thirds stop within a year and the Danish youth cohort worth reading alongside any adoption curve.