GLP Watchdog

SURMOUNT-OSA: Tirzepatide for Sleep Apnea

Last verified November 2026 · Phase 3 · Finished; headline results published June 2024 · NCT05412004 ↗

By Ruth Calder · Enforcement Editor
Editorially reviewed. Not medical advice · How we verify contentLast reviewed

Obstructive sleep apnea is strongly driven by weight, so a drug capable of taking off a fifth of someone's weight ought to help. What SURMOUNT-OSA added was the measurement: apnea-hypopnea index scored by overnight polysomnography, the same instrument a sleep clinic uses, rather than a symptom questionnaire. The size of the change is what makes it clinically meaningful. Moving someone from severe to mild on that scale is the difference between a condition that needs a machine every night and one that may not.

Enrollment
469
Duration
52 weeks of randomized treatment, including a dose-climbing period of up to 20 weeks, plus 4 weeks of safety follow-up
Drug
Tirzepatide (Zepbound)
Population
Adults 18 and over with a BMI of 30 or above and moderate-to-severe obstructive sleep apnea confirmed by an overnight sleep study (at least 15 breathing interruptions an hour). Type 2 diabetes was an exclusion. The trial ran as two separate cohorts: one of people who could not or would not use a breathing machine, one of people already established on one for at least three months. Across both, the average person was stopping breathing about 51 times an hour and weighed around 115 kg at a BMI near 39. Roughly 70% were men, the reverse of every weight trial on this site.

The Zepbound drug page covers the drug itself: its dosing, cost and side effects.

Primary endpoint

Breathing interruptions per hour of sleep after a year, measured in a sleep lab

Treatment arm

Cohort 1 (without PAP): −25.3 events/hour; Cohort 2 (with PAP): −29.3 events/hour

Comparator

Cohort 1: −5.3 events/hour; Cohort 2: −5.5 events/hour

Treatment difference: Estimated treatment difference vs placebo: cohort 1 −20.0 events/hour (95% CI −25.8 to −14.2); cohort 2 −23.8 events/hour (95% CI −29.6 to −17.9); P<0.001 for both

Starting from about 51 events an hour, people came down by 25 to 29 while placebo moved about 5. That moved the average participant from severe apnea into the moderate range, and many went further, to mild or none, measured by the same overnight study a sleep clinic would run rather than by a questionnaire.

Secondary endpoints

SURMOUNT-OSA secondary endpoints, treatment against comparator. Last verified November 2026.
EndpointTreatmentComparatorDifference
Cut their breathing interruptions by at least half

Six in ten of the no-machine group and nearly three quarters of the machine-using group halved their events, the usual bar for calling a treatment a success in this condition.

Cohort 1: 61.2%; Cohort 2: 72.4%Cohort 1: 19.0%; Cohort 2: 23.3%Relative risk 3.3 (95% CI 2.1 to 5.1) in cohort 1 and 3.1 (95% CI 2.1 to 4.5) in cohort 2; P<0.001 for both
Met the trial's definition of the apnea being gone

Roughly four in ten without a machine and half with one, against about one in seven on placebo. The approval itself rests on the primary endpoint above; this row shows how many people ended up with little or no apnea.

Cohort 1: 42.2%; Cohort 2: 50.2%Cohort 1: 15.9%; Cohort 2: 14.3%Relative risk 2.9 (95% CI 1.8 to 4.8) in cohort 1 and 3.3 (95% CI 2.0 to 5.4) in cohort 2; P<0.001 for both
How much body weight came off

18 to 20%, in line with what the same drug produces in weight trials. That raises a question this trial cannot answer: whether the breathing improved because of the drug or simply because of the weight.

Cohort 1: −17.7%; Cohort 2: −19.6%Cohort 1: −1.6%; Cohort 2: −2.3%Cohort 1: estimated treatment difference −16.1 percentage points (95% CI −18.0 to −14.2); cohort 2: −17.3 percentage points (95% CI −19.3 to −15.3); P<0.001 for both
How much oxygen the body lost overnight

From a starting level of roughly 130 to 155, this fell far more on the drug than on placebo, and it may be the most important row on the page. Hypoxic burden combines how deep and how long the oxygen drops are, and it predicts death from cardiovascular causes better than the event count everyone quotes.

Cohort 1: −95.2; Cohort 2: −103.0 (%·min/hour)Cohort 1: −25.1; Cohort 2: −41.7Estimated treatment differences −70.1 (95% CI −90.9 to −49.3) in cohort 1 and −61.3 (95% CI −84.7 to −37.9) in cohort 2; P<0.001 for both
How much sleep problems interfered with daily life, self-reported

About twice the improvement seen on placebo. Numbers on a sleep study only matter if the days get better too.

Both cohorts pooled: −7.5 points (improvement)Both cohorts pooled: −3.6 pointsEstimated treatment difference −3.9 points (95% CI −5.7 to −2.2); P<0.001
Inflammation in the blood

Fell about twice as far as on placebo in the no-machine cohort and several times as far in the machine-using cohort. The weight and cardiovascular trials report the same pattern.

Cohort 1: −1.4 mg/L; Cohort 2: −1.4 mg/LCohort 1: −0.7 mg/L; Cohort 2: −0.3 mg/LEstimated treatment differences −0.7 mg/L (95% CI −1.2 to −0.2; P=0.004) in cohort 1 and −1.0 mg/L (95% CI −1.6 to −0.5; P<0.001) in cohort 2
Top blood-pressure number, at week 48

Down about 8 to 10 points on the drug, which is a lot, though placebo also fell in the machine-using cohort and the gap there was smaller. Untreated apnea is a leading cause of blood pressure that will not respond to drugs, so this result matters clinically in its own right.

Cohort 1: −9.5 mmHg; Cohort 2: −7.6 mmHgCohort 1: −1.8 mmHg; Cohort 2: −3.9 mmHgEstimated treatment differences −7.6 mmHg (95% CI −10.5 to −4.8; P<0.001) in cohort 1 and −3.7 mmHg (95% CI −6.8 to −0.7; P=0.02) in cohort 2

Adverse events

SURMOUNT-OSA adverse events, rate on treatment against rate on comparator. Last verified November 2026.
EventTreatment rateComparator rate
Diarrhea

The most frequent stomach complaint in both cohorts, mostly mild to moderate, and concentrated in the first 20 weeks or so while the dose climbed.

Cohort 1: 26.3%; Cohort 2: 21.8%Cohort 1: 12.5%; Cohort 2: 8.8%
Nausea

Second most common, generally passing, and worst during dose increases.

Cohort 1: 25.4%; Cohort 2: 21.8%Cohort 1: 10.0%; Cohort 2: 5.3%
Vomiting

Around 9 to 18%, several times the placebo rate.

Cohort 1: 17.5%; Cohort 2: 9.2%Cohort 1: 4.2%; Cohort 2: 0.9%
Constipation

About 15% in both cohorts.

Cohort 1: 15.8%; Cohort 2: 15.1%Cohort 1: 2.5%; Cohort 2: 4.4%
Reaction where the injection went in

Uncommon, and consistent with what other tirzepatide trials report.

Cohort 1: 7.0%; Cohort 2: 5.0%Cohort 1: 0.8%; Cohort 2: 0.0%
COVID-19

Listed because the trial ran through the pandemic. Rates were slightly lower on the drug than on placebo, so this is background noise rather than a drug effect.

Cohort 1: 5.3%; Cohort 2: 6.7%Cohort 1: 8.3%; Cohort 2: 9.6%
Serious adverse events of any kind

Similar overall: a little higher on the drug in one cohort and lower in the other. No deaths occurred in either trial.

Cohort 1: 7.9%; Cohort 2: 5.9%Cohort 1: 5.8%; Cohort 2: 10.5%
Stopped the drug for good over a side effect

Under 5% on the drug in both cohorts. On placebo it was 1.7% in one cohort and 7.0% in the other, so there was no consistent excess.

Cohort 1: 4.4%; Cohort 2: 3.4%Cohort 1: 1.7%; Cohort 2: 7.0%

Clinical significance

This trial matters commercially as much as clinically, because a diagnosable condition with an objective test is far easier to get covered than obesity alone. For readers who have been denied coverage for weight, an OSA diagnosis and this evidence may be the route that works. The result belongs to tirzepatide at weight-management doses in people with obesity; it does not establish anything for sleep apnea without obesity, and it does not transfer to compounded preparations, which were not studied here.

Who sells tirzepatide, and for how much

This trial tested Tirzepatide (Zepbound). Across the 734 sellers on this register, 453 publish a standing monthly cash price for compounded tirzepatide, from $69 a month, with a median of $228.28.

⛔ None of those sellers was in this trial. A compounded preparation is not the product studied here: no agency has reviewed that specific preparation for safety, effectiveness or manufacturing quality, and its concentration is set by whichever pharmacy filled it. The result above is the strongest evidence available for tirzepatide and it was generated on something else. Every seller, with its price record.

Frequently Asked Questions

On its primary endpoint — Breathing interruptions per hour of sleep after a year, measured in a sleep lab — the treatment arm recorded Cohort 1 (without PAP): −25.3 events/hour; Cohort 2 (with PAP): −29.3 events/hour against Cohort 1: −5.3 events/hour; Cohort 2: −5.5 events/hour for the comparator. Estimated treatment difference vs placebo: cohort 1 −20.0 events/hour (95% CI −25.8 to −14.2); cohort 2 −23.8 events/hour (95% CI −29.6 to −17.9); P<0.001 for both. Starting from about 51 events an hour, people came down by 25 to 29 while placebo moved about 5. That moved the average participant from severe apnea into the moderate range, and many went further, to mild or none, measured by the same overnight study a sleep clinic would run rather than by a questionnaire.
469 participants, over 52 weeks of randomized treatment, including a dose-climbing period of up to 20 weeks, plus 4 weeks of safety follow-up. It was a Phase 3 trial, finished; headline results published June 2024. Registered as NCT05412004, which you can look up yourself rather than taking anyone's summary for it.
Adults 18 and over with a BMI of 30 or above and moderate-to-severe obstructive sleep apnea confirmed by an overnight sleep study (at least 15 breathing interruptions an hour). Type 2 diabetes was an exclusion. The trial ran as two separate cohorts: one of people who could not or would not use a breathing machine, one of people already established on one for at least three months. Across both, the average person was stopping breathing about 51 times an hour and weighed around 115 kg at a BMI near 39. Roughly 70% were men, the reverse of every weight trial on this site. ⚠ A trial result describes the population it was measured in. If you differ from it materially — in age, in baseline weight, in whether you have diabetes — the number is context rather than a prediction.
The most commonly reported were diarrhea (Cohort 1: 26.3%; Cohort 2: 21.8% against Cohort 1: 12.5%; Cohort 2: 8.8% on comparator), nausea (Cohort 1: 25.4%; Cohort 2: 21.8% against Cohort 1: 10.0%; Cohort 2: 5.3% on comparator), vomiting (Cohort 1: 17.5%; Cohort 2: 9.2% against Cohort 1: 4.2%; Cohort 2: 0.9% on comparator). The most frequent stomach complaint in both cohorts, mostly mild to moderate, and concentrated in the first 20 weeks or so while the dose climbed. Rates come from this trial's population under its own dose-escalation schedule; a faster or slower titration changes what people experience.
Secondary endpoints included cut their breathing interruptions by at least half (Cohort 1: 61.2%; Cohort 2: 72.4% against Cohort 1: 19.0%; Cohort 2: 23.3%); met the trial's definition of the apnea being gone (Cohort 1: 42.2%; Cohort 2: 50.2% against Cohort 1: 15.9%; Cohort 2: 14.3%); how much body weight came off (Cohort 1: −17.7%; Cohort 2: −19.6% against Cohort 1: −1.6%; Cohort 2: −2.3%). ⚠ Secondary endpoints are weaker evidence than the primary one — a trial is powered for its primary outcome, and the rest are read as supporting rather than as findings in their own right.
No, and this matters because the number gets used that way. SURMOUNT-OSA tested Tirzepatide (Zepbound) — the FDA-approved product, made by its manufacturer under Good Manufacturing Practice, at the doses and schedule the protocol specified. A compounded preparation of tirzepatide is a different product: no agency has reviewed that specific preparation for safety, effectiveness or manufacturing quality, and its concentration is set by whichever pharmacy filled it. ⛔ A seller quoting this trial's result beside a compounded price is quoting evidence that was generated on something they are not selling. That is not the same as saying compounded tirzepatide does not work — it is saying nobody measured it here.
It cannot tell you that. A trial reports an average across a defined population under a fixed protocol, with monitoring and adherence support most people do not get. It is the best evidence available and it is still a distribution rather than a promise — some participants lost far more than the mean and some lost nothing. What it is genuinely good for is comparing options against each other, which is what the rest of this section is for.

References

  1. 1.Malhotra A, Grunstein RR, Fietze I, et al.; SURMOUNT-OSA Investigators. Tirzepatide for the Treatment of Obstructive Sleep Apnea and Obesity. N Engl J Med. 2024. PMID: 38912654.
  2. 2.Eli Lilly and Company. Obstructive Sleep Apnea Master Protocol GPIF: A Study of Tirzepatide (LY3298176) in Participants With Obstructive Sleep Apnea (SURMOUNT-OSA, NCT05412004). ClinicalTrials.gov. 2024. https://clinicaltrials.gov/study/NCT05412004
  3. 3.U.S. Food and Drug Administration. FDA Approves First Medication for Obstructive Sleep Apnea (Zepbound, tirzepatide). FDA Press Announcement, December 20, 2024. 2024. https://www.fda.gov/news-events/press-announcements/fda-approves-first-medication-obstructive-sleep-apnea
  4. 4.Jastreboff AM, Aronne LJ, Ahmad NN, et al.; SURMOUNT-1 Investigators. Tirzepatide Once Weekly for the Treatment of Obesity. N Engl J Med. 2022. PMID: 35658024.

Related trial deep-dives