GLP-1 drugs shrink existing sleep apnea. In 206,000 diabetics they did not prevent new cases.
The trial evidence for GLP-1 drugs in sleep apnea is about treating a problem that already exists. In SURMOUNT-OSA, tirzepatide ↗ cut the number of breathing interruptions per hour in patients who already had moderate-to-severe obstructive sleep apnea, enough to win a US label for that use. Obstructive sleep apnea is the disorder where the airway repeatedly collapses during sleep, and its severity is scored by how often breathing stops each hour. The open question was whether the same drugs, by driving weight down, could stop the disorder from ever starting. A new cohort study says the answer, in routine care, is no.
Writing in the American Journal of Respiratory and Critical Care Medicine ↗, a team led from Grenoble and McGill tracked 47,315 adults with type 2 diabetes who started a GLP-1 drug and 159,066 who started a DPP-4 inhibitor, a different diabetes pill that does not move weight. That weight-neutral comparison is the point of the design. If GLP-1 drugs prevent apnea by trimming fat, then the group losing weight should develop fewer new cases than the group holding steady. Everyone in the study was obese, had type 2 diabetes, and had no record of sleep apnea when they started. The researchers followed them for up to three years.
New diagnoses came at almost the same pace in both arms. There were 612 new cases among GLP-1 users and 1,197 among the DPP-4 group, which works out to 5.8 versus 5.4 cases per 1,000 person-years. A person-year is one patient followed for one year, so the two rates are close to identical. The hazard ratio was 1.07, with a confidence interval running from 0.93 to 1.23. A hazard ratio near 1.0 means no difference in risk, and because the interval comfortably crosses 1.0, the study cannot claim GLP-1 use lowered, or raised, the chance of a new apnea diagnosis. The result held across weight categories, across sexes, across specific drugs, and through several sensitivity checks.
This is the design epidemiologists reach for when they want to avoid the trap that makes healthy-user drugs look protective. It compares new starters of one drug against new starters of another, rather than against people taking nothing, so the two groups resemble each other in the reasons they were prescribed a diabetes medicine at all. Salvatore Suissa and Laurent Azoulay, two of the authors, have spent years documenting how weaker designs manufacture false benefits. Their fingerprints here are a reason to take the null seriously rather than explain it away.
There are honest reasons the finding might understate a real effect. The database spans 2007 to 2023, and much of that GLP-1 use predates the potent weight-loss agents like semaglutide ↗ and tirzepatide, so the average patient here may have lost less weight than a patient starting one of today's drugs would. The endpoint is a clinical diagnosis of sleep apnea, not a sleep-lab measurement, so a case only counts when a doctor orders the test and records the result. If GLP-1 patients feel better and get referred for sleep studies less often, some real apnea would go uncounted in their arm. And three years catches fast prevention, not slow prevention that might take a decade to show.
Still, the split between the trial story and the cohort story is the useful takeaway. Shrinking an apnea that is already there, which the randomized trial ↗ showed, is a different claim from keeping one from forming, which this study could not show. Losing weight remains the standard advice for people who have sleep apnea. The idea that a GLP-1 receptor ↗ drug doubles as apnea prevention for the millions of obese diabetics who do not yet have it now has a large, carefully built study pointing the other way.