"Is a GLP-1 safe" is a question about a person, and a drug label cannot answer questions about people. What a label does is name the groups for whom the drug is ruled out entirely, list the harms that have been counted or reported, and say what should be watched. Those are narrower answers, and they are checkable, which is what this page reports.
Two other pages carry the adjacent questions and this one does not repeat them: the counted adverse reactions from the approval trials are on the side-effects hub, and what the longest randomised trials recorded over years is on long-term side effects. This page is about who the labels exclude, and how uniform the class actually is.
Nothing here is advice, and it is not a clearance. A decision about any of these drugs belongs with a prescriber who knows a person's history, as our editorial standards set out.
The absolute bar is two lines long
A contraindication is the strongest thing a label says: not a caution, not a monitoring instruction, but a statement that the drug should not be used. Across eight current labels, the list is remarkably short and remarkably consistent.
| Product | Contraindications |
|---|---|
| Wegovy | MTC/MEN 2 history · serious hypersensitivity |
| Ozempic injection | MTC/MEN 2 history · serious hypersensitivity |
| Oral semaglutide (Rybelsus, Ozempic tablets) | MTC/MEN 2 history · serious hypersensitivity |
| Mounjaro | MTC/MEN 2 history · serious hypersensitivity |
| Zepbound | MTC/MEN 2 history · serious hypersensitivity |
| Trulicity | MTC/MEN 2 history · serious hypersensitivity |
| Saxenda | MTC/MEN 2 history · serious hypersensitivity |
| Foundayo (orforglipron) | MTC/MEN 2 history · serious hypersensitivity |
| Byetta | serious hypersensitivity · history of drug-induced immune-mediated thrombocytopenia from exenatide products |
MTC is medullary thyroid carcinoma, an uncommon cancer of the thyroid's C cells. MEN 2 is Multiple Endocrine Neoplasia syndrome type 2, an inherited condition that strongly predisposes to it. The bar covers a personal or family history, which is why it is a question a prescriber asks about relatives and not only about the person in front of them.
Everything else the labels say about who might be harmed sits in the warnings section — graded, conditional, and framed as something to monitor rather than something that rules the drug out.
The one product that breaks the pattern, in both directions
Byetta is the outlier twice over, and both halves are worth stating plainly.
It has no boxed warning and no thyroid contraindication. Every other label in the set opens with the boxed warning about rodent thyroid C-cell tumours. Byetta's does not, and medullary thyroid carcinoma appears nowhere in its contraindications.
It has a contraindication nobody else has. A history of drug-induced immune-mediated thrombocytopenia from exenatide products rules the drug out, and the label states that serious bleeding from it "may be fatal". No other label in the class carries this.
The documents do not explain either difference. One observation is available from the labels themselves and is offered as an observation rather than an explanation: exenatide is not a modified human GLP-1. Its label describes a 39-amino acid peptide whose sequence only partially overlaps the hormone's — a different molecule that fits the same receptor, as set out on is GLP-1 a peptide. Whether that is why its safety documentation diverges is not something the record states.
The practical consequence is the point of this section: "is a GLP-1 safe" cannot be answered for GLP-1s as a group, because the class's own defining warning is absent from one member and its most serious individual warning belongs to that member alone.
And one product carries a warning its own data did not produce
Foundayo, the non-peptide approved in July 2026, carries the thyroid C-cell boxed warning. The text inside it is unusual enough to quote in substance: orforglipron is not pharmacologically active in rats or mice and did not produce tumours in rodents. The warning is kept because the rodent effect is considered a GLP-1 receptor-dependent one, orforglipron is active at the human receptor, and the human relevance of the rodent finding has not been determined.
So the boxed warning follows the receptor rather than the molecule or its animal data. That is a fact about how class labelling works, and it cuts both ways for a reader: the warning is not evidence that this drug caused anything, and its presence is not an oversight.
Where the real length is: the warnings
The contraindication lists are two lines. The warnings sections are not. Zepbound's runs to eight items, and the set is broadly representative of the class:
- Severe gastrointestinal reactions — and a population line worth noticing: the drug is not recommended in patients with severe gastroparesis. That is a group excluded in the warnings section rather than in the contraindications, which is a weaker instrument than an outright bar.
- Acute kidney injury due to volume depletion — the mechanism is dehydration from vomiting and diarrhoea, with renal function monitored in anyone reporting those.
- Acute gallbladder disease — reported in the trials; covered on gallbladder side effects.
- Acute pancreatitis — observed across GLP-1 receptor agonists, drug discontinued if suspected; covered on pancreatitis.
- Hypersensitivity reactions — anaphylaxis and angioedema, reported postmarketing.
- Hypoglycaemia — conditional, see below.
- Diabetic retinopathy complications in people with type 2 diabetes — monitored where there is a history.
- Pulmonary aspiration during general anaesthesia or deep sedation.
Trulicity's, Wegovy's and the others differ in ordering and in a few additions rather than in kind. Thyroid C-cell tumours appear at the head of each as a cross-reference to the boxed warning, and the thyroid question has its own page.
Hypoglycaemia is conditional, and the labels give a number
Low blood sugar is the harm people most often assume is inherent to these drugs, and the labels consistently frame it as a drug-interaction risk instead. The insulin effect is glucose-dependent — it stands down as blood sugar falls — so every label ties the hypoglycaemia warning to concomitant insulin or an insulin secretagogue such as a sulfonylurea.
The Zepbound label quantifies it. Among adults with type 2 diabetes in its trial, plasma glucose below 54 mg/dL was reported in 10.3% of tirzepatide-treated patients who were also taking a sulfonylurea, against 2.1% of those who were not. The mechanism behind that split is on how GLP-1 works.
The newest item is on every label at once
Pulmonary aspiration during general anaesthesia or deep sedation now appears on all eight labels read. These drugs delay gastric emptying, so a stomach assumed to be empty before a procedure may not be, and stomach contents can enter the lungs. Each label instructs that healthcare providers be informed of any planned surgery or procedure.
This is the warning most likely to matter to someone who feels entirely well, because it is triggered by an unrelated event — a routine operation, a colonoscopy, dental sedation. What should change and how far in advance is a decision for the anaesthetist and the prescriber; we cover the documentary record on GLP-1 and surgery.
One label writes down when to stop
Most safety documentation concerns harm. Saxenda's contains something different: a stopping rule tied to lack of benefit. For paediatric patients, the label instructs evaluating the change in BMI after 12 weeks on the maintenance dosage and discontinuing if BMI has not fallen by at least 1% from baseline, since sustained clinically meaningful weight loss is then unlikely.
That is a safety instrument of a kind rarely discussed — a rule against continuing exposure that is not producing the effect it was accepted for. What the outcome distributions actually look like is on GLP-1 before and after.
What the labels do not settle
- Whether the drug is safe for any particular person. Contraindications identify groups; a history, other medicines and other conditions are what decide an individual case, and only a prescriber holds those.
- Why Byetta's documentation differs from the rest of the class in both directions. The labels state the difference and not its reason.
- How often the postmarketing harms occur. Reactions reported voluntarily after approval come from a population of uncertain size, and every label says so; they cannot be turned into rates.
- Anything about pregnancy, breastfeeding or fertility. Those have their own page, GLP-1 and pregnancy, because a summary line is the wrong form for them.
- Anything about products sold under the GLP-1 name without a label. Patches, capsules and boosters are not in this analysis at all; there is no contraindication list to read, which is itself the finding on GLP-1 supplements.
How this page was built
Sections 4 (Contraindications), 5 (Warnings and Precautions) and the boxed warnings were read in full from eight current FDA labels on DailyMed on 2026-09-18, each with its effective date recorded: Wegovy (2026-06-18), Ozempic injection (2026-06-01), oral semaglutide covering Ozempic and Rybelsus tablets (2026-01-30), Mounjaro (2026-08-27), Zepbound (2026-08-28), Trulicity (2026-06-16), Byetta (2025-09-02), Saxenda (2026-02-25) and Foundayo (2026-07-29). The contraindication table records what each document lists, clause by clause; quoted phrases are the labels' own. Presence or absence of a boxed warning was checked in each document rather than inferred from the class.