A team at NYU Langone used Cosmos, a dataset built from Epic health systems that covers more than 300 million patients, to count how often GLP-1 drugs are prescribed to adults who do not have an FDA-approved reason to take them.[1] The answer: rarely as a share of everyone, but far more often than a few years ago, and in a group that looks different from the patients the drugs were tested in.
What the study counted
The researchers looked for adults prescribed liraglutide, semaglutide or tirzepatide between January 2021 and December 2025 with no record of an approved indication. They defined that as no type 2 diabetes, and no obesity as the drugs’ labels define it: a BMI of 30 or more, or 27 to 30 together with a weight-related condition such as high blood pressure, high cholesterol, heart disease, heart failure, sleep apnea or fatty liver disease (MASH).[1]
They counted by active ingredient, not brand name, because people without an approved indication sometimes get the diabetes-branded version for weight loss. To avoid counting someone who qualified at a higher weight and then lost it, they excluded anyone without a recorded weight at least six months before the first prescription.[1]
The numbers
- More than 1.1 million adults out of about 92.4 million without an approved indication were prescribed a GLP-1 over the five years (1.2%).[1]
- The annual rate rose from 0.11% in 2021 to 1.54% in 2025, about 14 times higher (the authors round it to 15-fold). In 2025 alone that was 844,508 people in the dataset.[1]
- 85.5% were women, against 54.2% of people who were not prescribed one. In 2025, 2.2% of women without an approved indication got a GLP-1 prescription, compared with 0.58% of men.[1]
- 35.1% had a BMI in the normal range. Their median BMI was 25.9, compared with 24.5 in people not prescribed one.[1]
- A history of eating disorders was six times more common: 1.8% against 0.3%.[1]
- They were better off: 78.0% had private insurance (against 47.8%), and they were more likely to live in the least socially vulnerable areas.[1]
The authors also ran a stricter version that left out anyone who had ever had a BMI in the obesity range outside pregnancy. Prescribing in that group still rose about 16-fold, from 0.03% to 0.5%, with the same patterns.[1]
Why it matters
The drugs’ benefits, the authors write, “have largely been studied in populations with greater comorbidity burden,” and the balance of risk and benefit for people without an approved indication “is unknown and merits additional study.”[1] The trials behind Wegovy and Zepbound enrolled people with obesity or overweight plus a related condition. Nobody has tested what these drugs do over years in people at a normal weight.
The eating-disorder finding is the one the authors flag most. Most of those diagnoses were coded as “unspecified,” so the study cannot say which disorders were involved, and the data cannot show why each prescription was written. For what is known about GLP-1s and disordered eating, see our evidence review on GLP-1s and binge or disordered eating.
What the study cannot see
The authors list the limits plainly. A missing diagnosis in a health record is not proof the person had no indication, because BMI and conditions are sometimes not recorded. And compounded semaglutide and tirzepatide, along with any prescription written outside Epic systems, were not counted, so the real numbers are likely higher.[1] That gap matters: many of the cash-pay telehealth services we review sell compounded versions through their own platforms, outside the health records this study used.
If you are weighing a GLP-1 without diabetes or obesity, our reviews of off-label Ozempic use without diabetes and GLP-1 microdosing cover what the evidence does and does not show. The label criteria themselves are covered in what disqualifies you from semaglutide.