In overweight and obese people with moderate-to-severe sleep apnea, GLP-1 drugs knocked about 15 breathing interruptions an hour off the count a sleep lab measures. The daytime grogginess those same patients feel barely moved.
That split is the useful finding in a network meta-analysis published August 17 in Frontiers in Endocrinology ↗. A group at the First Affiliated Hospital of Henan University of Chinese Medicine pooled 15 studies, 13 randomized trials and 2 case-control studies, covering 1,877 people, to line up three obesity treatments against each other for sleep apnea: GLP-1 receptor agonists, SGLT2 inhibitors, and aerobic exercise.
The number that moved and the one that didn't
Obstructive sleep apnea is the airway collapsing over and over during sleep, and it is measured by the apnea-hypopnea index, the number of times per hour breathing stops or goes shallow. GLP-1 drugs, the injected peptides that include semaglutide and the GIP/GLP-1 dual agonist tirzepatide, came out ahead of a dummy injection on that count by a wide margin: 15.28 fewer events per hour (the plausible range ran from about 8 to 22 fewer). They also trimmed body mass index by 1.78 points and nudged average overnight blood-oxygen up by 0.4 of a percentage point.
Then there is the Epworth Sleepiness Scale, the questionnaire that asks how likely you are to doze off reading, watching television, or sitting in traffic. This is the symptom that sends people to a sleep clinic in the first place. GLP-1 drugs improved it by 0.20 points. That is statistically real, but the smallest change a patient would actually notice on the scale is about 2 points, ten times larger. So the machine's count dropped hard and the feeling of being tired all day did not. On sleepiness, plain aerobic exercise, not either drug, ranked first.
What to make of it
The honest headline is that GLP-1 drugs are the strongest of the three at cutting the apnea count, which is worth something. Fewer breathing interruptions is the mechanism by which sleep apnea damages the heart and metabolism over years, even when a patient does not feel dramatically different in the morning. That is also the exact basis on which the FDA cleared tirzepatide, marketed as Zepbound, as the first drug for sleep apnea in December 2024 ↗: the pivotal SURMOUNT-OSA trials were built on the apnea count, not on how sleepy people felt.
The caution is that this is a stack of mostly small studies, and the authors say so plainly. Not one of their comparisons earned a high-certainty rating under the GRADE-based framework they used. The evidence was moderate only for the effect on BMI, low for the apnea index and oxygen and sleepiness, and very low for everything else, dragged down by bias within studies, imprecision, and suspected selective reporting. The head-to-head comparisons between the three treatments were largely non-significant and rested on indirect math rather than trials that actually pitted them against each other. The authors call their ranking hypothesis-generating, not a basis for prescribing.
For a general reader deciding whether a weekly shot will fix their sleep, the takeaway is narrower than the marketing. GLP-1 drugs reliably shrink the number a sleep study counts, and that number matters for long-run heart risk. Whether they make you feel less exhausted at your desk is a separate question this analysis could not answer, and on that measure the cheapest option on the list, moving your body, came out on top. The result is a reminder that in obesity medicine the surrogate a drug is approved on and the symptom a patient came in with are not always the same thing, a gap that also showed up when semaglutide lowered a dementia risk score without moving the disease itself ↗.