What Actually Needs a Reference in Pharma Promotional Materials?

Every promotional piece eventually reaches the same moment. Marketing writes a confident line, "the best-tolerated option in its class," and a medical or regulatory reviewer writes back two words: prove it.

That exchange is the heart of MLR review, and it runs on a single principle. In pharmaceutical promotion, the burden of proof sits with you. A regulator does not have to show your claim is wrong; you have to show it is right, with evidence, before the piece goes out. So the practical question every writer faces is: which parts of my copy actually need a reference?

Here is the working answer.

The default: if it is a claim, it needs substantiation

Start from the assumption that every claim needs support. Under the FDA's rules on prescription drug promotion (and the equivalent codes elsewhere), promotional claims must be truthful, not misleading, fairly balanced, and backed by adequate evidence, usually "substantial evidence" from adequate and well-controlled studies. And a "claim" is broader than most people assume. It covers anything you assert about:

• Efficacy: response rates, time to onset, durability, symptom improvement.
• Safety and tolerability: adverse-event rates, "well tolerated," discontinuation rates.
• Mechanism of action: how the product works, particularly anything that goes beyond the label's wording.
• Comparisons: "more effective than," "faster than," "first-line," "preferred." (The highest bar of all, covered below.)
• Numbers: every statistic, percentage, p-value, and patient count carries an implied source.
• Economic and quality-of-life claims: cost savings, adherence, patient-reported outcomes.

And it is not only words. A graph, a chart, a bolded figure, or an image that implies a benefit is a claim too, and it needs the same substantiation as a sentence.

What usually does not need a reference

A smaller set of statements can stand on their own:

• On-label statements taken directly from the prescribing information. Here the approved label is the reference. If you state the indication, the dosing, or a safety statement exactly as the PI does, you are citing the PI by definition (and you should still point to it).
• Company facts that are not claims: an address, a congress date, the existence of a product.
• Genuinely common disease-state background. Even here reviewers often ask for a citation, because "everyone knows that" is exactly where misstatements hide.

The rule of thumb: if a reasonable reader could ask "says who?", it needs a reference.

A reference is not the same as substantiation

This is where most pieces come unstuck. Attaching a citation is not the same as supporting the claim, and reviewers know the difference. The reference has to actually say what your claim says, for the population and the context you are using it in. Three failures show up over and over:

• The source does not support the specific claim. You cite a paper, but the sentence you need sits in the abstract or the introduction rather than the results, or the study measured something adjacent to what you wrote. Citing a secondary source instead of the primary data is the classic version.
• The claim has drifted from the source. A previously approved line quietly picks up a "unique" or a "first-line," and the reference underneath it no longer covers the stronger statement. The specifics matter just as much: the exact patient number, the comparator arm, and the endpoint all have to match the study you are citing, not a rounded-off memory of it.
• The claim contradicts the label. This is the one that surprises teams. A claim can be perfectly referenced to a real study and still fail, because it conflicts with the approved prescribing information, for instance by implying use in a patient group the label does not cover, or a benefit the label specifically qualifies. The label wins. A useful check does not just say "no evidence found"; it tells you the claim contradicts the PI, which is exactly the reason a reviewer needs in order to send it back.

One caveat: all of this is about promotional material. In genuinely non-promotional, unbranded settings, an independent expert's congress talk or true scientific exchange, the rules differ. An off-label point that a KOL makes as their own scientific opinion can sometimes be tolerated there, because it is the expert's own view rather than the company's claim. It is a grey area, and a contested one. But the moment a piece is promotional, the approved label is the ceiling.

The traps that quietly raise the bar

A few ordinary-looking claims need far more substantiation than teams plan for:

• Comparative and superlative claims. "Effective" needs evidence; "more effective than" a named competitor needs head-to-head evidence, a much higher bar and a frequent cause of rejected promotional claims. Superlatives ("best," "#1," "the only") are comparative claims in disguise, and they draw the same scrutiny.
• The same claim in a different region. A claim that is fully substantiated against the US label can be non-compliant against the EU or UK one, because the approved indication and wording differ. The reference that works for one market may contradict another market's label, so "which PI?" is part of the question "does this need a reference?".
• The same claim for a different audience. The bar shifts with who is reading. A claim aimed at patients should, as a rule, stay tightly aligned with the approved label. A claim aimed at prescribers can reach further, supported by a peer-reviewed publication that the label does not spell out, as long as it complements the label rather than contradicting it. So the same evidence can be enough for a prescriber audience and too much for a patient one.

Getting it right the first time

The teams that clear review fastest do not guess. They make substantiation explicit while they draft, not after:

• Anchor every claim to the exact supporting line of its source, so a reviewer can verify it in seconds instead of hunting. (Formatting that source cleanly helps; our free AMA citation generator turns a PubMed record into a clean citation in one paste.)
• Check each claim against the right region's label, not just against the literature.
• Re-read the claim against what the source actually says, word for word, before you submit.

This is what PharmaText.ai automates. Our Precision Traceability engine ties every claim to specific coordinates in your source PDFs as you write, and flags the claim that has no support, or that contradicts the label, before the piece ever reaches MLR. The "prove it" happens at the desk, not in the review meeting.

Related: start with what MLR review is and how it works; then see how unsubstantiated claims blow up the timeline in why materials get sent back 10+ times and why your MLR process is broken before it starts.