This site and its domain are for sale. Email the owner if you want to talk about it.

compoundglp.com

What FDA Inspectors Write Down at Compounding Pharmacies

Inspectors mostly write up process, monitoring and labels, while the GLP-1 warning letters increasingly read like a review of websites.

A clipboard with a blank inspection form beside a sealed injectable vial on a stainless steel table in a clean room.

Today the most common thing FDA inspectors write down at outsourcing facilities (the 503B kind of compounder) is a complaint about how they make the drugs. A review of FDA Form 483 data presented at the FDA Compounding Quality Center of Excellence’s 2024 conference found production and process control deficiencies in close to 29 percent of all observations issued to outsourcing facilities. Facility control systems came second at roughly 13 percent, environmental and personnel monitoring gaps at about 11 percent, and packaging and labeling defects at roughly 9 percent.

So if you typed fda 483 compounding into a search box, that is the short answer. A Form 483 is the list of observations an inspector leaves with a facility at the end of an inspection, and at compounders the list is mostly about process, monitoring and labels. The longer answer is that a 483 is the opening move rather than the verdict. In the GLP-1 business, a striking share of the FDA’s written output has also been about words on websites rather than particles in vials.

What an FDA 483 at a compounding pharmacy is, and what it isn’t

The sensible reaction is: of course. Compounders make sterile injectable drugs, and inspectors check whether they make them cleanly. So when the FDA cracks down on compounded semaglutide and tirzepatide, it is presumably because it found dirty rooms. That’s a reasonable guess. Some of it is even right.

But the mechanism is a bit odder. A 483 works like the form the rental car agent fills out when you return the car. She walks around it, notes the scratch on the bumper (and the faint smell of french fries), and hands you a copy. That is not a judgment that you are a bad driver. It is a list of things she saw, and what happens next depends a lot on what you say about the scratch.

What you say is where it goes wrong. BioSpace ran a headline about the FDA’s worry over contaminating drugs “with filth” (a phrase that does a lot of work in a headline). It reported that Empower Clinic Services’ response to the FDA’s Form 483 observations about insanitary conditions failed to satisfy the agency, which noted “gaps and discrepancies” in the paperwork submitted. So the inspector writes down what she saw, the facility writes back about what it fixed, and the agency grades the second document. In some sense the thing being inspected is the homework about the homework.

What inspectors write down most often

Those percentages (29, 13, 11 and 9, adding up to about 62) are for outsourcing facilities generally, not GLP-1 makers specifically. The facts I have do not break out the remaining third or so. They also don’t describe the corner pharmacy. 503A compounding pharmacies are regulated primarily by state boards of pharmacy rather than the FDA, and they are not required to register with the FDA or follow the Current Good Manufacturing Practices that drug manufacturers must meet. The federal 483 tally is basically a portrait of the 503Bs.

Translated out of inspection language, the categories are: the process for making the drug, the building it is made in, the checks on the air and the people, and the label. Which is the whole drug.

The more vivid catalog shows up in warning letters. On September 18, 2026, the FDA sent a warning letter to a Houston compounding pharmacy citing, among other things, sterile compounding failures including inadequate smoke studies and rewritten media fill records. A smoke study is what it sounds like: you release visible smoke in the clean work area and film it, to show that air flows away from the open vial rather than toward it. A media fill is a dress rehearsal (nutrient broth standing in for the drug) to see whether anything grows. Both exist to produce a record, and that’s fine. Records are how a regulator who is not in the room every day can know anything at all. “Hey, can I see the smoke study video?” is something an investigator says, I’d guess, at most sterile inspections.

A rewritten record is a different problem. The whole point of a media fill is that the broth either grew something or it didn’t, and the record is supposed to be the one part of the process that has no opinion.

Why the prescriptions themselves got written up

The same Houston letter cited something less physical: pretextual significant-difference determinations and boilerplate prescribing records. In plain words: you generally may not make a copy of a drug people can buy. You may make something for a patient who needs it to be a bit different. And the paperwork for a great many patients said, in the same sentence each time, that they needed it to be a bit different. The official name for this is a “pretextual significant-difference determination,” and the prescribing setup that produced it was, let’s say, very accommodating.

The background is the copy rule. In 2018, Pharmacy Times has noted, the FDA published guidance to 503A pharmacies on what counts as unauthorized compounding of a copy of an approved product. Compounding a copy is ordinarily prohibited, but the shortage exception permitted GLP-1 compounding while semaglutide and tirzepatide were on the FDA drug shortage list. Then the shortages ended. For semaglutide, the FDA’s February 21, 2025 declaratory order gave 503A pharmacies until April 22, 2025, and 503B facilities until May 22, 2025, to stop compounding under the exception. According to Legal 500, the agency said the delay was meant “to allow patients a reasonable amount of time to transfer their prescriptions.” (The tirzepatide version, with its court ruling and moving deadlines, lives in another article.)

Once the exception is gone, the significantly-different door is the only door, and the only door gets a lot of traffic. Here is a made-up example of where that ends. A pharmacy’s intake form adds a field: “Clinical reason this patient requires a non-commercial formulation.” The field autofills. Ten thousand patients in a row require it for the same reason, typed identically (including the typo). Eventually the typo is load-bearing, since fixing it would technically be a change in clinical documentation. An inspector reads the stack and writes down, in the flattest prose available, “boilerplate.”

Why so many GLP-1 warning letters are about advertising

Now the part the objection doesn’t predict. On September 9, 2025, the FDA sent more than 50 warning letters to GLP-1 compounders and online sellers citing misbranding under sections 502(a) and 502(bb) of the Federal Food, Drug, and Cosmetic Act. Section 502(bb) is a compounding-specific provision prohibiting false or misleading advertising or promotion of a compounded drug. Another trade report counted more than 55 warning letters to online sellers on September 16, 2025. Whether that is the same wave counted differently or a second one, I don’t know, and the sources don’t settle it.

The letters name the telehealth platform itself, not just the pharmacy, as the party responsible for misbranding. The reasoning, set out with the gravity it deserves, runs as follows. First premise: the FDA’s position is that compounding facilities, including pharmacies and outsourcing facilities, are not FDA-approved or FDA-licensed entities. Second premise: a website says or suggests that they are. Conclusion: the website is misbranding a drug, and whoever runs the website is the one doing it. The proof fits on an index card, and delivering it still took more than 50 letters, I guess.

It kept going. In March 2026, the FDA issued 30 warning letters to telehealth companies offering GLP-1 products with misleading claims. In April 2026, as ArentFox Schiff described it, a second round targeted companies marketing unapproved GLP-1, dual GLP-1/GIP and triple agonist products, “many labeled ‘research use only.’” Research use only is the pharmaceutical cousin of the old commercial that opened with “I’m not a doctor, but I play one on TV.” Everyone understands the disclaimer. The agency, judging by the letters, was not persuaded that the disclaimer was the part customers were reading.

I write about pharmacy paperwork for a living, which mostly means reading letters addressed to other people, and the striking thing about this pile is how little of it is about chemistry. The 483s describe clean rooms and airflow. The warning letters increasingly describe landing pages, checkout flows, or whatever the sales funnel looked like that quarter, sort of thing.

Which makes the website the most thoroughly inspected room in the business.

Sources

  1. cGMP Expectations for Outsourcing Facilities
  2. FDA slams GLP-1 compounder over risk of contaminating drugs ‘with filth’ - BioSpace
  3. Compounded Semaglutide vs Brand Name Comparison: What's Actually in Your Injection?
  4. Compounded GLP-1 Marketing After the Shortage Ended
  5. FDA and Novo Nordisk Warned of GLP-1 Telehealth Compounding Takedown. What’s Next? | Pharmacy Times
  6. Semaglutide Salt Forms: Why the Base Matters
  7. Compounded GLP-1s: Why doctors worry and the FDA is cracking down
  8. The FDA Removes Semaglutide from the Drug Shortage List
  9. FDA's Semaglutide Shortage Resolution: Legal Implications And Risks For Compounding Pharmacies | Legal 500
  10. FDA Warning Letters to Telehealth Companies Selling Compounded Drugs
  11. The Shot Heard ‘Round the Pharmacy’: FDA Takes Aim at GLP-1 Compounders | ArentFox Schiff

Compound GLP explains the law and published research. It is not medical or legal advice, and it never recommends a dose, a pharmacy or a product. Talk to a clinician before you start, stop or change any medicine.

Read next