The sentence sits on every seller page here: compounded drugs are not FDA-approved, and are not reviewed by the FDA for safety, efficacy or quality before they are dispensed. Most people read it as a warning label. It is closer to a receipt for a procedure that never happened, and the procedure has a definition in statute — five findings, one of which is about a factory rather than a molecule. Knowing which five changes what the sentence is worth, and it changes what to ask the company in the intake questions worth failing on.
An approval attaches to a product, never to a molecule
Ozempic was approved in December 2017 under new drug application 209637. Wegovy was approved in June 2021 under application 215256. Both are semaglutide, and the approvals belong to those two products — that applicant, that manufacturing process, those strengths, that labeling.[1] Nothing about either approval extends to a different preparation of the same active ingredient made somewhere else.
This is why the substitution most sellers invite is the wrong one. Semaglutide has an enormous trial record behind it, and a compounded vial of semaglutide inherits none of that record, because the record was built on products whose identity was fixed by an approval. The distance between the molecule and the container is the subject of the comparison with the branded products.
The five findings that have to come out right
Section 505(d) of the Federal Food, Drug, and Cosmetic Act is written backwards: it lists the grounds on which the agency must refuse an application. The investigations must include adequate tests, by all methods reasonably applicable, to show whether the drug is safe under the conditions in the proposed labeling. The results of those tests must actually show it is safe. The methods, facilities and controls used for manufacture, processing and packing must be adequate to preserve the drug’s identity, strength, quality and purity. There must be enough information in front of the agency to decide at all. And there must not be a lack of substantial evidence that the drug does what its labeling claims. Two further grounds in the same subsection cover the patent information an application must contain and labeling that is false or misleading in any particular.[2]
Two of those are worth reading twice. The third is not about the drug; it is about the plant, the process and the controls, which is why an approval decision involves inspectors as well as reviewers. And the fifth carries its own definition in the same subsection: substantial evidence means evidence consisting of adequate and well-controlled investigations, including clinical investigations, by experts qualified by scientific training and experience, on the basis of which those experts could fairly and responsibly conclude that the drug has the claimed effect.[2]
Three sections of the act, switched off
Compounding does not fail that review. It sits outside it, by a provision Congress wrote for the purpose. Section 503A states that three sections of the act do not apply to a drug compounded for an identified individual patient on receipt of a valid prescription, provided a list of conditions is met: section 355, the approval requirement; section 352(f)(1), the requirement that labeling bear adequate directions for use; and section 351(a)(2)(B), the requirement that the drug be made in conformity with current good manufacturing practice.[3]
So “not FDA-approved” is one of three exemptions, and it is the one that gets printed. The other two are at least as consequential to somebody holding a vial: a compounded preparation is not required to carry the directions-for-use labeling an approved drug carries, and it is not made under the manufacturing rulebook an approved drug is made under. Which of those last two applies depends entirely on the category of compounder, and that split is set out in the article on 503A and 503B.
What “not reviewed for quality” looks like in a laboratory
The phrase describes a missing check rather than a discovered problem, which makes it easy to dismiss. One published analysis gives it content. Researchers compared 16 injectable semaglutide, 8 oral semaglutide and 2 injectable liraglutide follow-on products against the corresponding originator products, using chromatography with ultraviolet and mass spectrometry detection, inductively coupled plasma methods, nuclear magnetic resonance, dissolution testing and fibrillation assays.[4]
The injectable follow-on substances and products carried new impurities and impurity patterns, including high molecular weight proteins, trace metals, anions, counterions and residual solvents. Several commercial oral follow-on products contained a markedly lower quantity of semaglutide than the label claimed. Neoepitopes were identified, indicating potential immunogenicity, and the liraglutide follow-on products showed increased fibrillation and reduced physical stability.[4]
Now the part that has to travel with those findings. Every author on that paper was employed by Novo Nordisk, which manufactures the originator products the follow-ons were measured against. The authors state their own limit plainly: the impact of these differences on efficacy and safety outcomes remains unknown and should be investigated by clinical studies.[4] The most detailed comparative chemistry in print was run by the party with the largest commercial interest in the answer, and it still declines to claim harm. Both halves of that are the finding.
Five things the phrase does not mean
It does not mean illegal. Compounding under section 503A is a lawful activity with statutory conditions attached, and a pharmacy meeting them is operating inside the law rather than around it. It does not mean untested, in the sense that the active ingredient has no evidence behind it; it means the evidence attaches to somebody else’s product. It does not mean a defect was found, because no one looked.
Nor does the converse hold. An approval is not a certificate of safety: the agency describes it as a determination that the benefits of a drug outweigh the risks for the intended use.[5] And an approval covers the labeled use only, so a dose or a schedule outside the label is outside the thing that was reviewed, whichever product it is dispensed from — a point that matters most to the protocols described in the microdosing article.
Registered, cleared and approved are three different words
The agency is explicit about the first of them. Mere registration of an establishment or listing of a drug does not denote approval of the establishment or of the drug, nor does it mean that a product may be legally marketed. The same page states that the agency does not approve health care providers, including physician offices, and that it does not approve compounded drugs at all: by statute, compounders do not have to prove to the agency that their products meet the standards a drug manufacturer must meet.[5]
Those distinctions do not survive contact with the market. A cross-sectional study of websites advertising compounded GLP-1 products for weight loss in Colorado identified 93 business websites covering 188 physical locations. Semaglutide was advertised by 92 of the 93. 41 of the 93 referred to FDA approval when describing compounded products, and 5 described them as “generic.”[6] Neither word is available for a compounded preparation, which makes close to half of that sample misdescribing the one fact a buyer most needs.
The reporting duty runs the same direction
There is a second asymmetry, and it distorts the record rather than the product. Federal law does not require state-licensed pharmacies that are not outsourcing facilities to submit adverse events to the agency, so the FDA states directly that adverse events from compounded versions of these drugs are likely underreported.[7] The counts the agency does publish are set out in the article on the contents of the vial.
The consequence is narrow and worth holding. A quiet safety record in this corner of the market is partly the absence of an obligation to speak, which is not the same as the absence of events. Comparing a compounded product’s reported event rate against a branded product’s is comparing a voluntary channel with a mandatory one.
Not obvious to clinicians either
A 2026 survey distributed to 523 clinicians and pharmacists across a multi-site health system returned 99 eligible responses. Fifty-one percent were aware of the 2024 American Diabetes Association guidance statement on compounded GLP-1 and dual GIP/GLP-1 receptor agonists, and a majority reported low confidence in the safety of these agents and had not incorporated them into practice, citing safety, efficacy, legality and knowledge gaps.[8]
That is the shape of the problem. The phrase on the label is not a specialist term that consumers alone find opaque; half of the professionals surveyed had not seen the governing guidance. Anyone expecting the regulatory status to be explained accurately at checkout is expecting something the market has not been delivering.
What remains checkable
Three facts survive all of this and can be verified before a card is charged: which pharmacy prepared the medication, which statutory section it operates under, and what the label on the container actually says. The first two are the subject of the supplier questions, and what the sellers here publish about price and dose sits on the semaglutide board.
None of this argues for or against a compounded prescription. It argues that the sentence means something specific: no application, therefore no findings, therefore no inspection of the line that made this vial. A reader who treats that as a verdict is overreading it, and a seller who treats it as boilerplate is underreading it by the same distance.