990 and 730: How to Read the FDA's Compounded GLP-1 Adverse Event File
The reporting totals are real, the undercount is real, and the causation claim people make from them usually is not. A careful look at what the numbers support.
The short version
- As of May 31, 2026, the FDA had received 990 adverse event reports associated with compounded semaglutide and more than 730 associated with compounded tirzepatide.
- The FDA cautions that these reports do not establish causation.
- The agency also says the totals are likely undercounts, because many state-licensed pharmacies are not required to report adverse events.
- Separately, America's Poison Centers managed 22,966 GLP-1 exposure cases from 2019 through 2025, with call volume rising nearly 1,500% over that period.
Adverse event numbers get used badly in both directions. One side treats every report as a proven injury. The other dismisses the whole dataset as noise. Both readings are wrong, and the second is more dangerous, because the structural problem with this particular dataset runs in the direction of underreporting.
What the system actually is
The FDA's adverse event reporting system is passive surveillance. Someone — a patient, a clinician, a manufacturer — voluntarily submits a report describing something that happened while a drug was being used. No one verifies that the drug caused it. The agency states this plainly: the reports do not establish causation.
That is not a weakness so much as a design choice. Passive surveillance is built to catch signals, not to prove them. A cluster of similar reports tells regulators where to look next.
It cannot tell you the drug caused the event. It cannot tell you how common the event is, because there is no denominator — nobody knows exactly how many people took the product. And it cannot be compared cleanly against a different drug with a different reporting culture.
Why the undercount matters more here
For FDA-approved drugs, manufacturers face mandatory reporting requirements. The compounded market does not work that way. The FDA has specifically noted that its totals are likely undercounts because many state-licensed pharmacies are not required to report adverse events at all.
So the comparison people reach for — compounded reports versus brand-name reports — is structurally unfair to the compounded side, in the sense that it flatters it. One market has mandatory reporting; the other largely does not. When the loosely monitored market still generates over 1,700 combined reports, the signal is worth taking seriously.
The poison center data is the more interesting file
America's Poison Centers reports managing 22,966 GLP-1 agonist exposure cases from 2019 through 2025, with calls increasing nearly 1,500% across that period. Most involved accidental therapeutic errors with FDA-approved drugs.
But one pattern is specific to the compounded market: 10-fold dosing mistakes caused by confusion over syringe measurement units. That is not a pharmacology problem. It is a packaging and instruction problem, and it is largely a consequence of dispensing multidose vials that require the patient to draw their own dose.
Compare providers on dispensing format, not just price
Every provider below is a paid partner. Pricing changes frequently — confirm current cost on the provider's own page before you enrol.
Semaglutide injection through a licensed US telehealth network. Injectable only — no oral formulation offered.
See Embody pricing Paid partner linkMarketplace for cash-pay appointments with clinicians who prescribe FDA-approved brand-name medication only — Rybelsus, the oral Wegovy pill, and Foundayo. No compounded product.
See Sesame Care pricing Paid partner linkTelehealth intake with both injectable and oral pathways in the questionnaire flow.
See Care Bare Rx pricing Paid partner linkHow to use this data for your own decision
Do not use it to decide whether GLP-1s work
Efficacy questions are answered by controlled trials, not by voluntary reports. The trial literature on semaglutide and tirzepatide is large and consistent.
Do use it to evaluate dispensing format
The dosing-error signal points at a specific, fixable thing. Prefilled pens or single-dose presentations remove the unit-conversion step entirely. If you are choosing between providers, ask how the medication arrives before you compare monthly prices.
Do use it to weight product provenance
The FDA has warned that some compounded products carry false label information, including pharmacy names that either do not exist or did not make the drug. A report volume that keeps climbing in a market with that documented labelling problem is a reason to care about who actually compounded your medication.
Contact your prescriber first. You can also submit a report to the FDA's MedWatch program directly as a patient. Include the compounding pharmacy name, lot number, and full formulation including any additives — reports missing the additive information are much harder for the agency to act on.
What honest interpretation looks like
The numbers do not show that compounded GLP-1s are dangerous. They show that a large, loosely monitored market has produced a rising volume of reports, that a specific and avoidable error type keeps recurring, and that the true totals are almost certainly higher than what is recorded. That is a real signal, and it is an argument for better dispensing practice and better provenance — not for panic.
Sources
- FDA adverse event report totals for compounded semaglutide and tirzepatide as of May 31, 2026.
- America's Poison Centers, GLP-1 agonist exposure case data 2019 through 2025.
- Drug Discovery Trends reporting on FDA adverse event logs for compounded GLP-1s.
- FDA guidance on the limitations of passive adverse event surveillance.
- FDA warnings regarding label accuracy in compounded GLP-1 products.