Journal

FDA Went Back to the Citations. Most Did Not Say What Was Claimed

The 1 May 2026 notice is, among other things, a citation audit. It is worth reading for that alone, because the failure pattern it documents is the same one that fills the consumer side of this market.

Why a regulatory notice is worth reading as evidence practice

We spend most of our time on this site doing one narrow thing: opening the source behind a number and checking whether it says what the page citing it claims. It is unglamorous and it is most of the work.

The FDA notice proposing to exclude semaglutide, tirzepatide and liraglutide from the 503B Bulks List is the same exercise conducted at scale, with the documents on the record and the reasoning written down. Whatever you think of the outcome, the method is legible in a way that very little in this field is. Below are the specific instances, each taken from the notice itself.

A prevalence statistic with no source behind it

One nominator argued that some patients are “hyper-responders” who need a lower dose than is commercially available, stating that an estimated 5–15% of the population are hyper-responders and 15% non-responders.

FDA noted the nominator provided no reference supporting what “has been estimated.” The paper the nominator did cite, Wilding et al. 2021, does not mention hyper-responders or the 5–15% statistic at all.

FDA went further and read what Wilding actually reported: more patients on semaglutide than placebo discontinued after gastrointestinal events, and nausea occurred primarily during dose escalation — but most gastrointestinal events were mild to moderate, transient, and resolved without permanent discontinuation. A study describing transient escalation-phase side effects does not establish a population of patients for whom the approved strengths are unsuitable.

A clinical argument resting on a newspaper article

The Outsourcing Facilities Association submitted that microdosing trends have been widely reported and allow patients otherwise intolerant of standard doses to continue treatment. The citation offered was an article in The New York Times.

FDA’s response was that the article contains no data or information supporting the proposition that the approved product could not be used to obtain lower doses, or that it is otherwise medically unsuitable — and that the clinical need analysis does not turn on general “trends” without identifying patients for whom the approved drug would be unsuitable.

The agency then pointed at the shelf. Semaglutide is approved in lower concentrations — 0.25 mg/0.5 mL, 0.5 mg/0.5 mL and 1 mg/0.5 mL among them — and the approved labelling already instructs prescribers to consider delaying dose escalation by four weeks where a dose is not tolerated. The nominations did not explain why those existing options could not be used.

We have a standing page on the microdosing evidence gap, and this is the same gap seen from the regulator’s side of the desk.

A patient preference the study did not report

A nominator argued that approved semaglutide injections for type 2 diabetes contain propylene glycol, which had caused increased injection-site irritation, and stated that 95% of patients preferred the formulation without it. The source given was Snitker et al. 2022.

FDA read the paper and reported that it does not contain that finding. The authors concluded that the injection-site experience with the formulation lacking propylene glycol was almost indistinguishable from the comparator multidose pen formulation, with either product associated with no or very mild injection-site pain.

The agency then checked its own adverse-event data. A search of the FDA Adverse Event Reporting System identified numerous reports of injection-site reactions associated with semaglutide, occurring in users of both Ozempic, which contains propylene glycol, and Wegovy, which does not. Of the reports identified, one referenced propylene glycol.

And there is a simpler answer sitting behind the argument: Wegovy and Wegovy HD are approved formulations that do not contain propylene glycol. Even if the excipient were unsuitable for some patients, the nomination offered no basis to conclude those patients could not use an approved product that omits it.

A dose argument overtaken by its own trial

The nominations advanced a general hypothesis that higher doses show a superior effect, citing among other things a registered trial, NCT05486065.

That trial has since been published. In it, 8 mg and 16 mg doses of semaglutide were compared against 2 mg and placebo. On the treatment policy estimand, the study did not demonstrate a statistically significant improvement in glycaemic control at either 8 mg or 16 mg compared with 2 mg. For body weight, the 16 mg dose showed a statistically significant decrease against 2 mg; the 8 mg dose did not. Adverse events and discontinuations due to adverse events were more frequent in both higher-dose groups.

FDA’s note is dry and worth quoting in substance: the results do not support the nominator’s argument. This is what it looks like when a claim is filed on anticipation and the evidence arrives afterwards pointing the other way.

The agency made a related structural point elsewhere in the notice. Obtaining a better response at a higher dose does not mean the approved strengths are ineffective or medically unsuitable. Better is not the same as necessary, and the statutory test asks the second question.

An argument that skipped over the obvious alternative

A nominator proposed sublingual and buccal semaglutide, arguing that swallowing the tablet can be daunting and citing dysphagia affecting up to a third of older adults.

FDA did not dispute the premise. It said it had no reason to disagree that difficulty swallowing is a well-documented issue, or that compounded drugs can serve an important need for patients who cannot swallow tablets. The failure was elsewhere: the nominators referred to oral routes generally without identifying any unsuitability in the approved semaglutide products, and never explained why the injectable products would be medically unsuitable for a patient who cannot swallow a tablet.

A similar gap ran through the 50 mg oral tablet proposal. The trials cited for high-dose oral semaglutide used a new formulation developed specifically to enhance bioavailability, with excipients omitted relative to the approved tablet; the PIONEER PLUS authors noted it was not possible to assess whether efficacy and tolerability were affected by that reformulation. Oral semaglutide’s absolute bioavailability is approximately 0.4–1% and 1–2% depending on formulation, and the approved products are co-formulated with an absorption enhancer. The nomination said nothing about the formulation of the compounded product, and did not assert it would overcome that problem.

The pattern, and what it does not prove

Five arguments, one failure mode. In each case the claim was plausible, widely repeated, and not supported by the document offered for it. Nobody had to be acting in bad faith for this to happen — it is what occurs when a figure circulates long enough that citing it feels like sourcing it.

Two things this does not establish. It does not establish that compounded GLP-1s have not helped patients; many people have had a good experience on them and that is not in dispute here. And it does not establish that no valid clinical need argument exists — only that these particular nominations, on this record, did not make one.

The uncomfortable symmetry is that the same audit run against the consumer side of this market — the ranking pages, the price roundups, the assistant-generated summaries — produces the same result, which is the entire reason this site exists. Our casebook is a record of the checks that failed.

Frequently asked questions