For decades obstructive sleep apnoea had no approved drug treatment at all. The options were a machine, a mouthpiece, surgery, or weight loss pursued by other means. That changed when one drug in this class was approved specifically to treat it — and because the approval is recent and singular, the evidence behind it is small enough to read in full.
Reading it in full is worth doing, because the two trials say several things that the coverage of them does not.
Labels and trial results below were read on 13 September 2026 from DailyMed, the National Library of Medicine's index of current FDA labelling, and from the trial registration. Nothing here is medical advice, and nothing here is a dose or a schedule; our editorial standards explain why.
Exactly one drug holds the indication
Zepbound (tirzepatide) is indicated to treat moderate to severe obstructive sleep apnoea in adults with obesity. That is the whole list.
No semaglutide product has an apnoea indication — not Wegovy, not Ozempic. Neither does Mounjaro, which contains the identical molecule to Zepbound under a diabetes brand. The indication attached to the brand whose trials were run in apnoea, not to the molecule, which is the ordinary way approvals work and a frequent source of confusion when two brands share an ingredient. We set out the brand-by-brand position on Zepbound vs Wegovy and Mounjaro vs Ozempic.
What the two trials were
The indication rests on a master protocol (NCT05412004) containing two separate randomised, double-blind, placebo-controlled trials of 52 weeks, enrolling 469 adults in total.
| Study 5 | Study 6 | |
|---|---|---|
| Participants | 234 | 235 |
| Positive airway pressure therapy | Unable or unwilling to use it | Already using it, and continued |
| Mean age | 48 | 52 |
| Male | 67% | 72% |
| Mean baseline apnoea-hypopnea index | 51.5 events/hour | 49.5 events/hour |
| Severe apnoea at baseline (≥30 events/hour) | 63.1% | 68.2% |
Two features of that design decide how far the results travel.
Everyone had obesity, and nobody had type 2 diabetes. The label states plainly that patients with type 2 diabetes were excluded from both trials. Obstructive sleep apnoea and type 2 diabetes are among the most commonly co-occurring conditions in this population — which means the pivotal evidence for the first drug approved for apnoea excludes one of the largest groups likely to be offered it. The exclusion is defensible as trial design; it is also a real limit on the evidence, and it is not mentioned in general coverage of the approval.
This was a severe population. Entry required at least 15 events an hour, and in practice the average was around 50, with roughly two-thirds classified severe. Mild apnoea was not studied at all.
What happened at week 52
| At week 52 | Study 5 placebo | Study 5 drug | Study 6 placebo | Study 6 drug |
|---|---|---|---|---|
| Change in apnoea-hypopnea index (events/hour) | −5.3 | −25.3 | −5.5 | −29.3 |
| Difference vs placebo | — | −20.0 | — | −23.8 |
| Percentage change | −3.0% | −50.7% | −2.5% | −58.7% |
| Achieving ≥50% reduction | 19.0% | 61.2% | 23.3% | 72.4% |
| Remission or mild, non-symptomatic | 15.9% | 42.2% | 14.3% | 50.2% |
"Remission or mild, non-symptomatic" is the label's composite: an apnoea-hypopnea index below 5, or between 5 and 14 with an Epworth Sleepiness Scale score of 10 or lower. Every drug-versus-placebo difference above was statistically significant with control for multiplicity.
Two secondary findings sit alongside them: participants on the drug had greater reductions in systolic blood pressure and in high-sensitivity C-reactive protein than those on placebo, and the reduction in apnoea events held irrespective of age, sex, ethnicity, baseline body-mass index and baseline apnoea severity.
The finding that runs against the intuition
The bigger effect was in the group already using a machine.
Study 6 — participants on positive airway pressure throughout — recorded a 58.7% reduction in events and 72.4% achieving at least a halving. Study 5, where participants were using nothing, recorded 50.7% and 61.2%. The drug did more, not less, on top of existing therapy.
This needs its caveat stated as loudly as the finding. These were two separate trials in two different populations, not one trial randomising people to drug-alone or drug-plus-machine. The Study 6 cohort was four years older on average and had a marginally lower baseline event count; nothing in the design licenses a direct subtraction. The honest reading is that the result gives no support at all to the intuition that the drug is an alternative for the machine, and some support to the idea that the two act on different parts of the problem.
Half of those treated still had sleep apnoea
This is the number a reader is least likely to be given and most likely to want.
At the better of the two results, 50.2% of treated participants finished at an apnoea-hypopnea index below 5, or 5 to 14 with low daytime sleepiness. In the other trial, 42.2%. The rest — the majority in one trial, half in the other — remained above that threshold after a year of treatment.
Consider where they started. A mean of about 50 events an hour, reduced by 25 to 29, leaves an average in the low twenties: still moderate apnoea by the standard classification. A 50% reduction is a large clinical effect and it is not the same thing as resolution, and a severe case halved is frequently still a case.
That is not a criticism of the drug. It is the difference between the question the trials answered — does this reduce apnoea severity against placebo — and the question most people bring to the subject, which is whether it makes the condition go away.
What this page does not answer
It does not say whether anyone should take this drug, change an existing therapy, or stop one, and it contains no dose and no schedule. Whether a machine can be reduced or set aside is a clinical question that these trials were not built to answer, and one where the evidence points away from the assumption rather than toward it.
It also does not price the treatment. Self-pay figures for the brand holding the indication are on our Zepbound cost page with the date each was read, and what a plan will pay turns on the indication a prescription is written under — the mechanism set out on our insurance coverage page. The apnoea indication matters there for a specific reason: it is an approved use that is not weight loss, which is exactly the distinction plan exclusions are built on.
Sources
- DailyMed, ZEPBOUND (tirzepatide) injection, solution / ZEPBOUND KwikPen, Eli Lilly and Company. SPL version 40, published 2 September 2026 — read 13 September 2026. Indication; section 14.2, "Obstructive Sleep Apnea Studies in Adults with Obesity", for the design, baseline characteristics, exclusion of type 2 diabetes and all week-52 results quoted above.
- ClinicalTrials.gov, SURMOUNT-OSA master protocol (NCT05412004) — read 13 September 2026.
- DailyMed, WEGOVY (semaglutide) injection / tablet and OZEMPIC (semaglutide) injection, Novo Nordisk Pharmaceutical Industries, LP — read 13 September 2026, to confirm that neither carries an obstructive sleep apnoea indication.
- DailyMed, MOUNJARO (tirzepatide) injection, Eli Lilly and Company — read 13 September 2026, same purpose.
Indications in this category widen between label versions. Every claim above is tied to a named label version and the date it was read.