Frequently, and the clearest measurement comes from one diagnosis. Among women prescribed an incretin drug after a polycystic ovary syndrome diagnosis in a Polish private network, only 11.5% had coded type 2 diabetes [1]. The remaining 88.5% were being treated for weight and cardiometabolic reasons outside the classical glycemic indication, which the authors state explicitly. What the evidence actually supports in that condition is separate, and is set out in whether a GLP-1 helps PCOS.
How fast one indication moved
The network tracked women diagnosed with PCOS between 2018 and 2024 and whether they started an incretin drug within a year. That share went from 0.14% to 5.97%, and calendar year alone remained strongly associated with initiation after adjustment, at an odds ratio of 1.64, 95% CI 1.56 to 1.73. The pattern held after excluding women with diabetes.
Prescribing tracked weight closely. Median body mass index was 31.2 among users, against 26.3 for women on metformin alone and 22.1 for the untreated, and users carried a greater comorbidity burden as well. Three quarters of them, 75.5%, were also taking metformin, so for most this was an addition rather than a replacement. That matters to anyone pricing a regimen rather than a single drug.
The same shape in two more populations
A US retrospective cohort tracked semaglutide and tirzepatide prescriptions in the year either side of delivery from 2019 to 2024 [2]. In the year before delivery the rate went from 0.2 to 6.4 per 1,000 deliveries, and in the year after from 0.3 to 14.6 per 1,000, which is about 1.5% of all deliveries. Change-point analysis placed the acceleration at June 2022 for the predelivery window and March 2021 for the postdelivery one.
One measurement detail changes what the predelivery figure means. The window covers the whole year before delivery, including months before anybody knew a pregnancy existed. A fill in that period is not evidence of a clinician knowingly prescribing to a pregnant patient. The postdelivery figure carries no such ambiguity and it is the larger of the two.
A Danish nationwide registry followed everyone aged 12 to 24 who filled one of these prescriptions between 2018 and 2025 [3]. Use rose more than fiftyfold, reaching 418 new users per 100,000 by 2025, which is under half a percent of the age group. Among 18 to 24-year-olds it went from 13 to 686 per 100,000, and among 12 to 17-year-olds from 1.7 to 72.
What off-label actually means here
It does not mean unlawful or improper. Prescribing outside an approved indication is ordinary medical practice and is how a great deal of useful treatment reaches people, and what it changes is who has tested the thing being done. An approval carries trial evidence in the population named on the label, and a prescription outside it carries whatever evidence exists elsewhere, which in these three populations is thin.
It also changes who pays. Coverage follows the indication, so the same molecule at the same dose meets a different bill depending on which line of a form is ticked, as set out in what insurance actually covers. That is a large part of why the cash market exists at all. Why so many people who qualify on paper still never start is counted in why a prescription is hard to get.
What a buyer should take from a prescribing curve
Two things, and neither is about efficacy. The first is that a rapid rise in prescribing is not a rise in evidence, and the authors of all three studies say so in their own words rather than in this desk’s. The second is that the people writing these prescriptions increasingly are not specialists: 78% of the Danish youth treatments were started in general practice.
A telehealth intake sits further along that same line again, with no prior relationship and no record behind it. What separates a seller that screens properly is the subject of the three gates before the price. How long anybody lasts once they start is tracked in how long people stay on a GLP-1.