Reviews of a prescription medicine usually consist of anecdotes. This one is built from three public datasets — the posted results of its withdrawal trial, its approved label, and the FDA adverse event reports filed about it — read on 2026-09-10. The third of those contains something that, as far as we can find, nobody covering this drug has reported.
This page describes what those records contain. It does not recommend a medicine or a dose to anyone; that is a decision for a prescriber, as our editorial standards set out.
For where this molecule's results sit against semaglutide's as a spread rather than a headline — including the 9.1% of patients on the highest dose who did not reach a 5% reduction — see the trial outcome distributions. The brand this review mostly concerns, its two indications and the label changes FDA has approved since 2023, is described on our Zepbound brand page.
What it achieved in the trial that measured longest
SURMOUNT-4 (NCT04660643) is the most informative single source on tirzepatide's sustained effect, because it ran 88 weeks and posted its results to the registry. Participants took open-label tirzepatide for 36 weeks to a maximum tolerated dose of 10 mg or 15 mg, then continued for a further 52. Mean starting weight was about 107.8 kg.
At week 88, participants who continued were 26.0% below their starting weight. The proportions reaching each threshold from baseline:
| Weight reduction from baseline | Continued tirzepatide | Randomised off the drug at week 36 |
|---|---|---|
| At least 5% | 98.49% | 69.00% |
| At least 10% | 93.98% | 44.38% |
| At least 15% | 87.05% | 24.01% |
| At least 20% | 72.59% | 11.55% |
Nearly three quarters of continuing participants lost a fifth or more of their body weight. That is a large effect by any standard applied to weight management, and it is measured rather than claimed.
One caveat about that right-hand column, because it is routinely misread. It is labelled placebo in the trial, and it is not a drug-naive placebo group. Everyone in it had already taken tirzepatide openly for 36 weeks before being switched. So 11.55% reaching 20% loss at week 88 is not "what happens without the drug" — it is what remained a year after stopping the drug. A genuine never-treated comparison would be lower. Any page presenting that column as a placebo response rate is comparing the wrong two things.
The other honest limit: trial populations are supervised, screened, provided with the medicine, and consist of people who stayed enrolled for 88 weeks. Ordinary prescribing has none of those properties. The other SURMOUNT trials, with the percentages converted into kilograms, are on our Zepbound results record.
What the label counts against it
From the pooled placebo-controlled trials in the Zepbound label, effective 2026-08-28:
- Nausea 25%, 29% and 28% at 5, 10 and 15 mg, against 8% on placebo
- Diarrhoea 19%, 21% and 23%, against 8%
- Constipation 17%, 14% and 11%, against 5% — the one reaction that falls as the dose rises
- Vomiting 8%, 11% and 13%, against 2%
- Hair loss 5%, 4% and 5%, against 1%, and 7.1% among female participants against 0.5% among male
The same label also carries the class boxed warning about thyroid C-cell tumours, and what sits underneath that box differs by molecule more than the box does. One entry in the same label is more notable for how small it is. Cholelithiasis was reported in 1.1% of treated patients against 1% on placebo — effectively no excess, where the two leading semaglutide and liraglutide products both report a clear one. The gallbladder numbers run opposite to the weight loss across this class, and this label is the reason.
And the figure that puts those in proportion: 4.8%, 6.3% and 6.7% permanently discontinued because of adverse reactions, against 3.4% on placebo, most of them in the first few months. A quarter of participants reported nausea; around one in fifteen left because of any adverse reaction at all.
Three of those lines have their own pages here, because each turns out to be more interesting than its number: the side effect tables and what they hide, why hair loss is counted in this label and not in Mounjaro's, and what the label does not say about lean tissue. The weight brand's whole table, with the placebo rate subtracted and the contraception warning most readers miss, is on our Zepbound side effects page; why the diabetes brand prints roughly half the nausea rate at the same doses is set out in Mounjaro vs Zepbound.
The thing nobody counts
FDA's adverse event reporting system collects voluntary reports from patients, clinicians and manufacturers. Queried on 2026-09-10 for tirzepatide, it returns 160,663 reports. The most frequently reported term is this:
| Reported term | Reports |
|---|---|
| Incorrect dose administered | 29,672 |
| Nausea | 17,092 |
| Injection site pain | 13,906 |
| Diarrhoea | 10,006 |
| Extra dose administered | 9,608 |
| Off-label use | 8,941 |
| Vomiting | 8,267 |
| Drug ineffective | 6,965 |
| Constipation | 6,502 |
| Injection site haemorrhage | 5,281 |
The most reported thing about this medicine is not something it did. It is people saying they got the dose wrong — more often than nausea and diarrhoea combined. Add the neighbouring terms — extra dose administered 9,608, accidental underdose 4,609, product dose omission issue 4,474 — and 48,363 of the 160,663 reports concern dosing rather than an effect of the drug. Close to one report in three. Partial doses measured outside the labelled pen steps are the same error class, and they are what the microdosing market is built on.
The control that makes it a finding
A single number from this database means very little. So the same query was run on semaglutide, in the same class, in the same database, on the same day — the molecule whose own trial and reporting record we read separately.
| Tirzepatide | Semaglutide | |
|---|---|---|
| Most reported term | Incorrect dose administered (29,672) | Nausea (10,674) |
| Incorrect dose administered | 29,672 | 880 |
| Total reports | 160,663 | 73,001 |
| Dosing reports as share of total | 18.5% | 1.2% |
| Reports from consumers | 91.4% | 70.3% |
Semaglutide's list looks like a drug label: nausea, off-label use, vomiting, diarrhoea, decreased appetite, constipation. Incorrect dose administered does not appear in its top twelve. Thirty-four times fewer such reports in absolute terms, and fifteen times fewer as a share.
What it means, and what it does not
Two readings survive the data, and honesty requires both.
The first is about the product. Tirzepatide is dispensed in several presentations — single-dose pens, multi-dose KwikPens and vials — across eight strengths, on a stepped escalation schedule, and a substantial share of its users are self-injecting at home for the first time. That is a lot of surface for a mistake. If mis-dosing were genuinely more common with this product than with semaglutide, this is the pattern it would produce.
The second is about the pipeline. These are not measurements; they are reports, and reports have to reach the database. Manufacturers submit the majority of what FAERS holds and differ substantially in how they classify a patient phone call about a device or a schedule. Tirzepatide's reports are 91.4% consumer-sourced against semaglutide's 70.3% — a real difference, and one that points the same way.
But the reporter-mix difference cannot carry the result. A gap of 91% against 70% in consumer share does not produce a 34-fold gap in a single term. Something beyond who is filing is contributing, and this database cannot say what.
What can be stated firmly is the third fact, and it is the reassuring one: of the 29,672 incorrect-dose reports, 29,331 were classified non-serious and 340 serious, and 28,793 came from consumers. Whatever is driving the volume, the overwhelming majority of these events were recorded as having no serious outcome.
None of this is an incidence rate. FAERS has no denominator, its reports are unverified, and a report is not a finding of causation. A person cannot use these numbers to work out a probability. What the numbers do show is that the commonest thing filed about tirzepatide belongs to a category most coverage of it never mentions — and that on the drug it is usually compared with, the same category barely registers.
Set against the labels, that finding gets stranger rather than clearer: not one of the seven approved GLP-1 labels defines what an overdose is, while the reports carrying the word are accidental over intentional by roughly 129 to 1.
What this review will not give you
A score. This site does not assign numbers to medicines, because a score implies a ranking across patients whose circumstances differ in ways no published source captures, and because a number invites citation long after the evidence underneath it has moved.
What the record supports instead is a summary a reader can check: a large and well-measured effect that depends on continuing, since randomised withdrawal produced 14.8% regain in a year; a common but rarely treatment-ending burden of gastrointestinal side effects; a counted hair loss reaction weighted heavily toward women; an acknowledged but unquantified loss of lean tissue; and a reporting record dominated by people getting the dose wrong. Every one of those is dated and sourced below, which is the most a page like this should claim.
On the last of those, the mechanism is now documented separately from the count. Tirzepatide's approved presentations span a six-fold concentration range at a single fixed volume so that no dose is ever calculated by hand, which is exactly the step a compounded vial reintroduces.
Sources
- SURMOUNT-4, NCT04660643, ClinicalTrials.gov posted results, read 2026-09-10 — week 88 change from baseline, threshold responder rates, trial design and baseline weight.
- FDA label for Zepbound (tirzepatide) injection, effective 2026-08-28, via the openFDA drug label endpoint, read 2026-09-10 — adverse reaction table, discontinuation rates, hair loss subsection.
- openFDA drug adverse event endpoint, queried 2026-09-10 — report counts by reaction term, by reporter qualification and by seriousness, for tirzepatide and for semaglutide as a control.