SURMOUNT-OSA: baseline factors

A new post hoc analysis of the two SURMOUNT-OSA trials asks a narrower question than the original trial report: did changes in sleep-apnea measures with tirzepatide look materially different according to participants' starting demographic factors and baseline measurements? The authors describe improvements across the baseline groups examined, but the analysis is exploratory. It can describe patterns within the trial data. It cannot determine which individual will respond, or replace a clinician's assessment of obstructive sleep apnea.^[1]

The practical reading is therefore cautious. The original randomized trials established a comparison with placebo in adults with obesity and moderate-to-severe obstructive sleep apnea. This later paper examines whether the reported changes were visible across selected starting characteristics. It is useful context for readers following the tirzepatide evidence base, not a patient-selection rule or a plan for changing prescribed care.^[1]

What the underlying trials studied

SURMOUNT-OSA combined two Phase 3, randomized, double-blind, placebo-controlled trials that followed adults with obesity and moderate-to-severe obstructive sleep apnea for 52 weeks. One trial enrolled people who were not using positive airway pressure, commonly called PAP. The other enrolled people using PAP at baseline. The registered programme enrolled 469 participants and is listed as completed on ClinicalTrials.gov.^[3]

The primary endpoint in the 2024 report was change in the apnea-hypopnea index, or AHI, a count of breathing interruptions per hour of sleep. At 52 weeks, the estimated difference versus placebo was 20.0 fewer events per hour in the trial without PAP at baseline and 23.8 fewer events per hour in the trial with PAP at baseline. The original report also measured body weight, hypoxic burden, blood pressure, and patient-reported sleep outcomes.^[2]

What the new post hoc analysis adds

The August 2026 paper re-examined the trial data across specified demographic groups and baseline measurements. It reports that participants assigned to tirzepatide generally had larger improvements than placebo across the subgroups examined. Reported AHI reductions in the tirzepatide groups ranged from 27.7 to 34.1 events per hour across age groups, from 19.8 to 32.6 across sex groups, and from 12.1 to 52.2 across baseline AHI categories.^[1]

Those ranges should not be used as a ranking of which type of person benefits most. They are descriptive values from subgroup slices with different starting measurements. A larger numerical reduction can partly reflect a higher baseline AHI and more room for change. The paper itself calls the analyses hypothesis-generating, which is the appropriate limit on the conclusion.^[1]

AHI is one measure, not the whole clinical picture

AHI is central to the trial, but it is not the only outcome a sleep clinician considers. The original SURMOUNT-OSA report included hypoxic burden, body weight, systolic blood pressure, and sleep-related patient-reported measures. The newer analysis also describes improvements in body weight, systolic blood pressure, and sleep-apnea-specific hypoxic burden across the baseline groups it examined.^[1]

That does not mean every measure changes by the same amount for every participant, or that an AHI result can settle a treatment decision by itself. Symptoms, sleep testing, existing PAP use, other conditions, adverse effects, and the approved label all remain relevant. For an overview of the medicine and its established evidence, see our tirzepatide reference page.

Why baseline characteristics are tempting to overread

Baseline characteristics are attractive because they seem to promise a simple answer to a difficult question: who is likely to benefit? This paper does not provide that answer. It does not report a validated prediction tool, and its subgroup findings do not establish that age, sex, body size, starting AHI, or neck circumference should be used alone to decide whether a person should receive a medicine.^[1]

The distinction matters in sleep apnea, where the trial populations were specific. Participants had obesity and moderate-to-severe disease. The 2024 trials also had separate PAP-related entry groups, so their results should not be recast as a universal instruction to stop, continue, or alter PAP. Readers comparing related incretin medicines should also avoid assuming that findings for tirzepatide automatically apply to semaglutide or another product.^[2]

What we do not yet know

The post hoc results do not show how to predict an individual response before treatment, how durable changes remain beyond the trial period, or how care should be coordinated for people with different PAP histories. They also cannot establish causation for every observed subgroup pattern. Future pre-specified research would be needed to test whether a baseline characteristic truly changes the relative effect of treatment.^[1]

Regulatory status also varies by jurisdiction. In the United States, readers can consult our educational page on tirzepatide regulation, while medicine-specific questions should go to the clinician or pharmacist managing the prescription. This article explains research findings and does not provide medical advice.

How to read this result

The new paper supports a modest conclusion: within SURMOUNT-OSA, improvements in OSA-related measures associated with tirzepatide were reported across several baseline-characteristic groups. That broad pattern is reassuring as trial context, but it is not evidence that a particular age, sex, BMI, AHI, or neck measurement identifies the right treatment for an individual. The source study was funded by Eli Lilly and included company employees among its authors, another reason to keep the descriptive, post hoc framing visible.^[1]

For related background, our tirzepatide guide explains the molecule, while the original trial report remains the best place to inspect the randomized comparison. For related evidence context, see the tirzepatide trial summary and our tirzepatide safety overview. Neither source replaces individualized assessment for obstructive sleep apnea.^[2]