No using data from off-label dosages to support favorable claims

AI-verifiedblockerBinding lawRaised as a question — a person decides

Rule

Favorable data derived from patients treated at dosages different from those recommended may not be used to support a drug claim: for a drug subject to section 505 of the act, the comparator is the dosages recommended in approved or permitted labeling; for other drugs, the comparator is the dosages recommended in labeling and generally recognized as safe and effective. Citation of studies that include some patients treated at different dosages is acceptable if those patients' results are not used.

Check

Determine whether the advertised drug is subject to section 505 of the act. For a section 505 drug, compare the dosages used in cited supporting studies against the approved or permitted labeling. For a non-505 drug, compare the dosages used against the labeling dosing that is generally recognized as safe and effective. In either case, confirm that favorable data relied upon in the ad derives only from on-label dosing; citation of a study that includes some off-label-dosed patients is acceptable only if their results are not used.

Escalate when

Whether specific data points relied upon derive from off-label dosing, or which labeling comparator applies to the drug, requires study-level review.

Where this comes from

21 CFR 202.1(e)(6)(xvii). The highlighted words are quoted exactly; a script confirms they appear in the source before the rule can be published.

…xistence of a real clinical difference. (xvi) Uses statements or representations that a drug differs from or does not contain a named drug or category of drugs, or that it has a greater potency per unit of weight, in a way that suggests falsely or misleadingly or without substantial evidence or substantial clinical experience that the advertised drug is safer or more effective than such other drug or drugs. (xvii) Uses data favorable to a drug derived from patients treated with dosages different from those recommended in approved or permitted labeling if the drug advertised is subject to section 505 of the act, or, in the case of other drugs, if the dosages employed were different from those recommended in the labeling and generally recognized as safe and effective. This provision is not intended to prevent citation of reports of studies that include some patients treated with dosages different from those authorized, if the results in such patients are not used. (xviii) Uses headline, subheadline, or pictorial or other graphic matter in a way that is misleading. (xix) Represents or suggests that drug dosages properly recommended for use in the treatment of certain classes of…

Open the original at these words on ecfr.gov ↗Read our full copy →

Document: 21 CFR 202.1
Version: 2026-09-16 — the link above is pinned to this date, so it shows the text as it stood then
Retrieved: 2026-09-21 from www.ecfr.gov
Fingerprint: a542640a872b3f2702d549e8…

Applies when

Naturepromotional

How this rule was checked

Citation proven by script and passed an independent verifier running on a different model. Not yet confirmed by a human.

Drafted
claude-code:sonnet · 2026-09-21
Independent verification — revise
claude-code:opus · 2026-09-21
Reviewer's reasons
- The quote is from 21 CFR 202.1(e)(6)(xvii), an operative list of things that make a prescription-drug advertisement false, lacking in fair balance, or otherwise misleading. It is an operative prohibition, not an example, definition, or preamble, so the rule statement and check do follow from it.
- The rule is narrower than the source, not broader. Source (e)(6)(xvii) reaches two populations: drugs subject to section 505 of the act, AND 'in the case of other drugs, if the dosages employed were different from those recommended in the labeling and generally recognized as safe and effective.' The rule as written covers only the section 505 branch and, by naming that limit explicitly, implies off-label-dose favorable data is unproblematic for non-505 drugs. That is an under-scoped applies_when relative to the cited provision and needs fixing before it is shown as AI-verified.
- The parenthetical carve-out in the rule statement faithfully reproduces the source's final sentence: 'This provision is not intended to prevent citation of reports of studies that include some patients treated with dosages different from those authorized, if the results in such patients are not used.' Good.
- Atomic: one requirement (do not use favorable data derived from off-label dosing) with its own statutory carve-out; the carve-out is not an independently checkable second requirement.
- determinism 'judgment' is correct: establishing which favorable data points derive from patients dosed off-label requires reading the underlying studies against approved labeling, not a literal match.
- authority 'binding' is correct: 21 CFR 202.1 is a regulation.
- Severity 'blocker' is defensible: (e)(6) items are stated as things that make the ad false/misleading and thus misbranding under section 502(n) per paragraph (k), rather than merely 'may be'.
- Locator '21 CFR 202.1(e)(6)(xvii)' matches the quoted text exactly.
Revised by the drafter
claude-code:sonnet · 2026-09-21
Independent verification — pass
claude-code:opus · 2026-09-21
Reviewer's reasons
- The quote is 202.1(e)(6)(xvii) and the rule tracks its two comparators precisely: approved or permitted labeling dosages for a drug subject to section 505, and labeling dosages generally recognized as safe and effective for other drugs. That is a conditional within one requirement, not a bundle.
- The rule carries over the express proviso — citation of studies including some patients dosed differently is permissible if those patients' results are not used — which is the limitation most likely to be dropped and turned into an overstatement. It is not dropped.
- Because (e)(6) is conclusive, framing this as 'may not be used' rather than 'may be misleading' is faithful to the source.
- 'Judgment' is correct: identifying which favorable data points derive from off-label dosing requires reading the underlying studies, not a literal match in the piece; the escalate condition says so.
- Authority 'binding' is correct (codified regulation) and 'blocker' is defensible for a (e)(6) item. applies_when is limited to promotional material, consistent with a rule about advertisements, and adds no limit the source does not have.