GLP-1 statistics (2026): how many people use them, what it costs them, and why people stop

How many people take GLP-1 medications?

In KFF's poll, 12% of US adults said they currently take a GLP-1 drug for any reason, and 18% said they have ever taken one.

How common is prescription drug use overall?

For context, CDC FastStats reports prescription drug use across all medicines, from the National Health and Nutrition Examination Survey (2017 to March 2020).

• 49.9% of people used at least one prescription drug in the past 30 days.
• 24.7% used three or more.
• 13.5% used five or more.

How affordable are GLP-1s for people who take them?

In the same KFF poll, 56% of current GLP-1 users said the drugs were difficult to afford, including 55% of users with health insurance.

How many people stop, and how fast?

A JAMA Network Open study by Rodriguez and colleagues followed a US health-record cohort of 125,474 adults who started a GLP-1 medicine between 2018 and 2023.

• One-year discontinuation: 64.8% of adults without type 2 diabetes, compared with 46.5% of adults with type 2 diabetes.
• One-year reinitiation among people who stopped: 36.3% without type 2 diabetes, compared with 47.3% with type 2 diabetes.

A 2026 national survey in the Journal of Managed Care and Specialty Pharmacy (DiStefano and colleagues) asked 440 adults who had stopped a GLP-1 medicine, in an Ipsos survey fielded in June 2025. Of them, 54.6% stopped within 6 months of starting, and 79.7% within 12 months.

Why do people stop?

In the JMCP 2026 survey of people who stopped, the most frequently cited reasons were cost (36.1%), side effects (33.3%) and no insurance coverage (28.2%). Other reasons were also reported.

In the KFF poll, among adults who had ever taken a GLP-1 drug, 14% said they had stopped because of the cost and 13% because of side effects.

If you are thinking about stopping, keep a clear record of what you notice: see what to write down for your clinician and how to bring notes to your appointment. Talk to your prescriber before changing anything.

Shortage and compounding: the regulatory numbers

Semaglutide injection products were in shortage since 2022. FDA announced the shortage resolved on February 21, 2025. It gave state-licensed 503A pharmacies until April 22, 2025 and 503B outsourcing facilities until May 22, 2025 before it would act against compounded copies of the approved product.

Tirzepatide: FDA's May 1, 2026 Federal Register notice says FDA-approved tirzepatide products are not on the FDA drug shortage list.

On April 30, 2026, FDA proposed not including semaglutide, tirzepatide and liraglutide on the 503B bulks list. The comment period was extended to July 30, 2026 (Federal Register, June 26, 2026). No final decision had been published as of September 30, 2026. This is still a proposal: check FDA for the final decision.

FDA says compounded drugs are not FDA-approved, and the agency does not review them for safety, effectiveness or quality before they are marketed. As of May 31, 2026, FDA had received 990 adverse-event reports for compounded semaglutide and more than 730 for compounded tirzepatide.

Across all medicines, ASHP counted 227 active US drug shortages in the second quarter of 2026, compared with the all-time high of 323 in the first quarter of 2024.

How much GLP-1 research is published? (our dataset)

We counted records indexed in Europe PMC for each publication year with the query below, run on 30 September 2026. These are records indexed, not individual studies, and one study can appear in more than one record.

("GLP-1 receptor agonist" OR "GLP-1 receptor agonists") AND PUB_YEAR:<year>
Run: 2026-09-30, Europe PMC REST search API

• 2005: 30 records indexed
• 2015: 425 records indexed
• 2020: 924 records indexed
• 2024: 2,503 records indexed
• 2025: 5,189 records indexed

Side effects and safety: what trials, regulators and studies report

These numbers come from the FDA labels, published trials, regulators and studies. Each row says who was counted and when. For the exact wording on when to call a clinician, see when to call: the official wording.

• Placebo rates aren’t zero. The gap between the drug and placebo is what matters.
• Both labels say trials ran under widely varying conditions, so rates can’t be directly compared between drugs.
• Spontaneous reports, like the MHRA Yellow Card counts, can’t show how often something happens or prove cause.
• Observational studies, like the Swedish cohort, show associations, not cause and effect.
• The Reddit study is a preprint, not peer-reviewed, and counts what people chose to post.
• The pooled STEP analysis has Novo Nordisk employee authors, and SURMOUNT-1 was supported by Eli Lilly.
• The food-noise survey below is maker-funded: Novo Nordisk employees are among its authors, and Novo Nordisk paid the panel company that ran it.

Food noise: what the first numbers show

A 2025 paper (Dhurandhar EJ, et al. Nutrition & Diabetes 2025;15:30, published 8 July 2025) defines food noise as "Persistent thoughts about food that are perceived by the individual as being unwanted and/or dysphoric and may cause harm to the individual, including social, mental, or physical problems." It says the term "appears to have originated from patient anecdotes", and that "very little is known about food noise thus far, and rigorous experimental tests are warranted." The paper's acknowledgements say the work was funded by Ro, a telehealth company, and some authors report industry ties.

The INFORM survey (Arnaut T, et al. Adv Ther 2026;43(8):3649-3659) asked 550 US adults using injectable semaglutide to fill in the five-item Food Noise Questionnaire, scored out of 20. The median score was 13 for the recalled "before" and 6 now. Agreement with food-noise statements fell from 47–63% to 15–20%.

An earlier trial (Friedrichsen M, et al. Diabetes Obes Metab 2021;23(3):754-762, doi:10.1111/dom.14280) in 72 adults over 20 weeks found less hunger and fewer and weaker food cravings than placebo. It measured appetite, not "food noise", and Novo Nordisk sponsored it.

No one can promise a specific effect.

How we compiled these numbers

We used primary sources only: KFF, FDA labels and communications, peer-reviewed journals, a regulator (MHRA), NIDDK, CDC and ASHP, plus one preprint, which is labelled as one. Every number links to the page it comes from. The side-effect, safety and food-noise sources were read on 1 October 2026 and the rest on 30 September 2026. Industry funding and author ties are stated where the source discloses them. We excluded vendor and telehealth statistics, prices, and weight-outcome figures.