Research · 11 min read
Contraindication, Warning, Reported: How a Label Sorts What Is Known
Three sections of an approved label hold three different standards of proof, and the regulation says what each one requires. One demands a demonstrated hazard. One explicitly does not require causation to be established. One says its own contents cannot be counted.
Key takeaways
- The label's sections carry different standards of proof, and 21 CFR 201.57 states each one.
- A contraindication must be a known hazard and not a theoretical possibility, and a label with none must print the word None.
- All six of these labels carry two contraindications, and the rarely quoted second one reaches any of the excipients in that particular product.
- A warning is added as soon as there is reasonable evidence of a causal association, and the regulation says a causal relationship need not have been definitely established.
- The same nine warnings appear on all six labels in five different orders; the regulation sets no ordering rule for that section, though it does set one for adverse reaction listings.
- One label alone carries a heart rate warning, and the never-share-a-device warning appears only on labels with an injectable presentation.
- The postmarketing section names hazards that appear nowhere else, including hair loss on three labels, and states in its first sentence that its contents cannot be turned into a frequency or a proof of cause.
Answer first: the sections are graded, and the grades are written down
An approved label looks like one document with a lot of warnings in it. It is closer to four documents stacked in descending order of certainty, and the federal regulation that sets its content, 21 CFR 201.57, defines the bar for each.
A contraindication has the highest bar. The regulation says known hazards and not theoretical possibilities must be listed, and that if no contraindications are known the section must state the word None.
A warning has a deliberately lower one. The same regulation says labeling must be revised to include a warning about a clinically significant hazard as soon as there is reasonable evidence of a causal association. The same sentence adds that a causal relationship need not have been definitely established.
An adverse reaction in the trials section is a measured rate under a stated threshold, which is covered in a separate article on what those percentages count. And a postmarketing entry is a report, which every one of these labels says cannot be turned into a frequency or a proof of cause.
Reading the label as one flat list of scary things collapses all four into each other. The gradient is the information.
A contraindication has to be demonstrated
The regulation's language here is unusually plain. The contraindications section must describe any situations in which the drug should not be used because the risk of use clearly outweighs any possible therapeutic benefit. It names the kind of situation it has in mind. That is a patient who, because of age, sex, concomitant therapy, disease state or other condition, has a substantial risk of being harmed, and for whom no potential benefit makes that risk acceptable.
Then the sentence that gives the section its weight. Known hazards and not theoretical possibilities must be listed. The regulation gives its own illustration: if severe hypersensitivity to the drug has not been demonstrated, it should not be listed as a contraindication.
And a rule that tells you how to read a short section. If no contraindications are known, the section must state the word None. So an empty contraindications section is not a gap in the document. It is an affirmative statement, and its absence on a label means the opposite.
That is the standard behind the difference in tone people notice between section 4 and section 5. Section 4 is a rule. Section 5 is a description of hazards, and it is allowed to describe hazards that have not been proven to be caused by the drug.
There are two contraindications on these labels, not one
This is the part that gets lost, because the first one is famous. Every one of the six approved labels examined here lists two contraindications, and the second gets almost no attention anywhere.
The first is the thyroid one: a personal or family history of medullary thyroid carcinoma, or the inherited endocrine condition abbreviated MEN 2. That contraindication, the boxed warning behind it and the animal evidence underneath both are the subject of a separate article.
The second is a prior serious hypersensitivity reaction to the active ingredient or to any of the excipients in the product. Each label states it in its own words, and each adds that serious hypersensitivity reactions, including anaphylaxis and angioedema, have been reported. Every one of the six cross-references its own hypersensitivity warning, and two of them also point to the postmarketing section.
The phrase worth slowing down on is any of the excipients. A contraindication that reaches the inactive ingredients is a contraindication attached to a formulation rather than to a molecule. Two products containing the same active ingredient do not necessarily contain the same excipients, and a compounded preparation's excipients are set by whoever compounded it. The patient-facing Medication Guides make the same point in plain words, telling readers to see the ingredient list printed at the end.
None of that is a reason to conclude anything about a specific product. It is a reason to treat a past reaction to a medicine as a question with a formulation in it. That is a question for a prescriber and the dispensing pharmacy rather than for a comparison page.
A warning does not require proof of cause, and says so
The warnings and precautions section must describe clinically significant adverse reactions, including any that are potentially fatal or are serious even if infrequent, or that can be prevented or mitigated through appropriate use. It must also describe other potential safety hazards, and the regulation is explicit that this includes hazards expected for the pharmacological class, and hazards arising from drug interactions. It must state limitations in use imposed by them, and steps that should be taken if they occur.
Then the timing rule. Labeling must be revised to add a warning about a clinically significant hazard as soon as there is reasonable evidence of a causal association, and a causal relationship need not have been definitely established.
Two consequences follow, and both explain things people find confusing about these labels. First, the presence of a warning is not a finding that the drug causes the thing. It is a finding that there is reasonable evidence of an association and that a reader should know. Second, a hazard can be listed because it is expected for the class rather than because it was seen with that particular molecule. That is precisely why a newer product in this class can arrive carrying warnings inherited from its relatives.
The list is also alive. These labels carry a change log at the top, and across the six it records warnings being added and revised in recent months, and one being removed outright. A warning list is a snapshot of what was known when the version was issued, not a permanent property of the drug.
The numbered list is a list, and the numbers are not a ranking
Read the six labels side by side and the same nine warnings appear on every one of them. Risk of thyroid C-cell tumors. Acute pancreatitis. Severe gastrointestinal adverse reactions. Acute kidney injury due to volume depletion. Hypoglycemia. Hypersensitivity reactions. Diabetic retinopathy complications. Acute gallbladder disease. Pulmonary aspiration during general anesthesia or deep sedation.
Counting only those nine, the six labels put them in five different orders. The only two that match are the two semaglutide labels from the same sponsor, one an injection and one a tablet. Acute gallbladder disease sits third on one label and ninth on another. Severe gastrointestinal adverse reactions sits second on one and seventh on another. Diabetic retinopathy sits third on two of them and eighth on two others.
The regulation explains why that is possible. Paragraph (c)(6), which governs warnings and precautions, contains no ordering requirement at all — the words order and rank do not appear in it. Paragraph (c)(7)(ii), which governs adverse reaction listings, does. Reactions must be categorized by body system, by severity, or in decreasing order of frequency. Within a category they must be listed in decreasing order of frequency, or by decreasing severity if frequency cannot be reliably determined.
So the section that looks like a ranking is not one, and the section that looks like a jumble of percentages is ordered by rule. A page that reports the first two warnings on a label as the most important ones has read a numbering convention as a judgment.
Two entries that are not on every label
Beyond the shared nine, two warnings appear on some of these labels and not others, and both differences are informative rather than random.
One label carries a warning titled Heart Rate Increase that the other five do not. It is on the semaglutide weight-management label. It is also the only warning in the whole set that asks for a recurring measurement, which is the subject of a separate article on what these labels ask anyone to monitor.
The other is a warning against sharing an injection device between patients. It appears on the four labels that include an injectable presentation and on neither of the two that are tablets only. That is a warning about a physical object rather than about a molecule, which is a useful reminder that section 5 mixes hazards of the drug with hazards of the delivery.
The heading wording moves too. The diabetic retinopathy warning is scoped to patients with type 2 diabetes on three labels, to patients with a history of the condition on one, and left unscoped on two. Three of the six attach trial figures underneath it, and the other three state only the class observation about rapid improvement in glucose control. The heading is where a limiting clause lives, and it is the part a summary usually drops.
The section that tells you its own contents cannot be counted
Section 6.2 is called Postmarketing Experience, and it exists because the regulation requires a listing of adverse reactions identified from domestic and foreign spontaneous reports, kept separate from the reactions identified in clinical trials.
Every one of the six labels opens it with the same qualification. Because these reactions are reported voluntarily from a population of uncertain size, it is not always possible to reliably estimate their frequency or establish a causal relationship to drug exposure.
That single sentence is the most useful thing on the page, because of what is listed underneath it. Across these six labels the postmarketing sections name acute pancreatitis including necrotizing and, on some, hemorrhagic pancreatitis, sometimes resulting in death. They name ileus, intestinal obstruction, and severe constipation including fecal impaction. They name anaphylaxis and angioedema, and on some labels rash and urticaria. They name acute kidney injury, or acute renal failure or worsening of chronic renal failure sometimes requiring hemodialysis. They name pulmonary aspiration in patients undergoing elective procedures requiring general anesthesia or deep sedation.
Three of the six also list alopecia — hair loss — under skin and subcutaneous tissue. It appears nowhere else on those labels. Two of the six list nervous-system entries: headache and dysesthesia on one, and dizziness, dysesthesia, dysgeusia and headache on the other. So a symptom that many people report and that no trial table shows can still be somewhere in the document, in the one section that carries no numbers.
One further wrinkle sits in plain sight on the newest of these labels. Its postmarketing section describes reactions reported during post-approval use of GLP-1 receptor agonists as a class, rather than of that product. A newly approved drug has no postmarketing experience of its own yet, and the label says so by naming the class instead.
Reading the four tiers in order
Start at section 4 and read both entries. A contraindication is a rule that had to be demonstrated before it could be printed, and there are two of them here rather than the one everybody quotes.
Then read section 5 as an unordered set. Note which warnings are shared across products, which are attached to a device rather than a drug, and read the heading itself for the population it names before reading the paragraph.
Then read section 6.1 as measurement, within one label only, against its own placebo column and its own stated inclusion threshold. Then read 6.2 as report — a list of things that happened in people taking the drug, with no denominator and no established cause.
What that produces is neither reassurance nor alarm. It is a document where the certainty of each claim is declared by the section it sits in. Almost every misreading in circulation comes from moving a fact up a tier it did not earn.
Sources
- 21 CFR 201.57 — Specific requirements on content and format of labeling for human prescription drug and biological products described in § 201.56(b)(1)Paragraph (c)(5) for the contraindications standard, including the requirement that the section describe situations where the risk of use clearly outweighs any possible therapeutic benefit, that known hazards and not theoretical possibilities must be listed with severe hypersensitivity given as the illustration, and that a section with no known contraindications must state None. Paragraph (c)(6)(i) for the content of warnings and precautions, including hazards expected for the pharmacological class, and for the revision rule that a warning is added as soon as there is reasonable evidence of a causal association with a causal relationship not needing to be definitely established. The measured absence of any ordering requirement in paragraph (c)(6). Paragraph (c)(7) for the definition of an adverse reaction and the requirement of some basis to believe there is a causal relationship, paragraph (c)(7)(ii) for the categorization and decreasing-frequency ordering rules, and paragraph (c)(7)(i)(B) for the requirement that postmarketing reactions be listed from domestic and foreign spontaneous reports and kept separate from the clinical trial listing.
- WEGOVY (semaglutide) injection and tablets — full prescribing informationThe two contraindications, including a prior serious hypersensitivity reaction to semaglutide or to any of the excipients in either presentation with anaphylaxis and angioedema named; the eleven numbered warnings and their order; the Heart Rate Increase warning found on no other label in this set; the never-share warning; the retinopathy heading scoped to patients with type 2 diabetes and the trial figures beneath it; the change log recording recent warning additions, a revision and one removal; and the postmarketing section with its uncertain-population sentence and its listing of necrotizing pancreatitis, ileus, intestinal obstruction, severe constipation including fecal impaction, anaphylaxis, angioedema, rash, urticaria, pulmonary aspiration and acute kidney injury.
- ZEPBOUND (tirzepatide) injection — full prescribing informationThe two contraindications with the hypersensitivity entry cross-referencing both the hypersensitivity warning and the postmarketing section; the ten numbered warnings and their order, with severe gastrointestinal adverse reactions second and acute gallbladder disease fourth; the never-share warning; the retinopathy heading scoped to patients with type 2 diabetes mellitus with no trial figures beneath it; the change log recording a warning removal and a heading revision; and the postmarketing listing including hemorrhagic and necrotizing pancreatitis, ileus, intestinal obstruction, fecal impaction, anaphylaxis, angioedema and renal failure sometimes requiring hemodialysis.
- MOUNJARO (tirzepatide) injection — full prescribing informationThe two contraindications; the ten numbered warnings and their order, with acute gallbladder disease eighth; the never-share warning; the retinopathy heading scoped to patients with a history of diabetic retinopathy; and the postmarketing listing, which names alopecia alongside the shared gastrointestinal, hypersensitivity, pulmonary and renal entries.
- OZEMPIC (semaglutide) injection — full prescribing informationThe two contraindications; the ten numbered warnings, with diabetic retinopathy complications third and the never-share warning fourth rather than last; the unscoped retinopathy heading; and the postmarketing listing including cholecystitis and cholecystectomy, dysesthesia, headache and alopecia.
- RYBELSUS (semaglutide) tablets and OZEMPIC (semaglutide) tablets — full prescribing informationThe two contraindications; the nine numbered warnings and the absence of any never-share warning on a tablet-only label; the shared-warning order matching the injection label from the same sponsor; and the postmarketing listing including cholecystitis and cholelithiasis requiring cholecystectomy, dizziness, dysesthesia, dysgeusia, headache and alopecia.
- FOUNDAYO (orforglipron) tablets — full prescribing informationThe two contraindications, with the hypersensitivity entry stating that serious reactions have been reported with GLP-1 receptor agonists rather than with this product; the nine numbered warnings, their order, and the absence of a never-share warning; the retinopathy heading scoped to patients with type 2 diabetes together with the statement that the drug has not been studied in patients with retinopathy or macular edema requiring acute treatment; and the postmarketing section describing reactions reported during post-approval use of GLP-1 receptor agonists as a class.
Frequently asked questions
What is the difference between a contraindication and a warning?
The standard of evidence, and the regulation states both. A contraindication describes a situation where the risk of use clearly outweighs any possible benefit, and known hazards rather than theoretical possibilities must be listed. If none are known, the section must state the word None. A warning is added as soon as there is reasonable evidence of a causal association with the drug, and the same sentence says a causal relationship need not have been definitely established. A contraindication is a rule. A warning is a hazard you are being told about.
How many contraindications do these labels actually have?
Two, on all six examined here. The first is a personal or family history of medullary thyroid carcinoma or the endocrine condition abbreviated MEN 2. The second is a prior serious hypersensitivity reaction to the active ingredient or to any of the excipients in that product, with each label adding that serious reactions including anaphylaxis and angioedema have been reported. The second one is rarely quoted and reaches the inactive ingredients, which means it is attached to a formulation rather than only to a molecule.
Are the warnings listed in order of importance?
No. The regulation governing the warnings and precautions section sets no ordering requirement — the words order and rank do not appear in it. The regulation governing adverse reaction listings does set one. Across these six labels, the same nine warnings appear on all of them in five different orders, with acute gallbladder disease third on one label and ninth on another. Reading the numbering as a ranking is reading a convention as a judgment.
Does a warning mean the drug causes that problem?
Not by itself, and the regulation is explicit about it. Labeling must be revised to add a warning as soon as there is reasonable evidence of a causal association, and a causal relationship need not have been definitely established. The section is also required to cover hazards expected for the pharmacological class. That is why a newly approved drug in an established class can arrive carrying warnings that come from its relatives rather than from its own trials.
Why is a side effect I experienced not in the trial tables?
It may be in a different section. Section 6.2, Postmarketing Experience, lists reactions identified from spontaneous reports rather than from trials, and it is required to be kept separate from the trial listing. Across these six labels that section names things the tables do not, including ileus, intestinal obstruction, severe constipation with fecal impaction, anaphylaxis and angioedema, and on three of them hair loss. Every one of those labels opens the section by saying that the reports come from a population of uncertain size, so frequency and cause cannot reliably be established from them.
Why does the retinopathy warning read differently on different labels?
Because the heading carries a scope and the scopes differ. Three of these labels scope it to patients with type 2 diabetes, one to patients with a history of the condition, and two leave the heading unscoped. Three of the six attach trial figures beneath it and the other three state only the observation that rapid improvement in glucose control has been associated with temporary worsening. The limiting clause in a heading is usually the first thing a summary drops, and on this warning it changes who the paragraph is about.