- Design
- Population-based active-comparator, new-user cohort study
- Population
- 206,381 adults with type 2 diabetes and obesity, no prior OSA
- Primary outcome
- Incident clinical diagnosis of obstructive sleep apnoea
- Effect
- Hazard ratio 1.07 (95% CI 0.93-1.23) for GLP-1 agonist vs DPP-4 inhibitor
GLP-1 receptor agonists reduce the severity of established obstructive sleep apnoea in trials, and it is tempting to assume their weight loss would also prevent it. This large UK cohort tested that directly.
Using an active-comparator, new-user design, it compared 47,315 people with type 2 diabetes and obesity starting a GLP-1 agonist against 159,066 starting a weight-neutral DPP-4 inhibitor, excluding anyone with prior sleep apnoea. Over up to three years, incident OSA diagnoses were essentially the same: 5.8 versus 5.4 per 1,000 person-years, a hazard ratio of 1.07 with a confidence interval spanning no effect. The result held across body-mass-index strata and sensitivity analyses.
The practical message is not to prescribe or expect GLP-1 agonists to prevent sleep apnoea in diabetes. They remain useful for weight and glycaemic goals, and for reducing severity in diagnosed OSA, but preventing new OSA is not among their demonstrated benefits.
- UK cohort of 47,315 GLP-1 agonist starters versus 159,066 DPP-4 inhibitor starters.
- All had type 2 diabetes and obesity, with no prior sleep apnoea.
- Incident OSA was similar: 5.8 versus 5.4 per 1,000 person-years.
- The hazard ratio was 1.07 (95% CI 0.93-1.23), consistent with no effect.
- Results held across body-mass-index strata and sensitivity analyses.
Why it matters
It punctures the assumption that GLP-1-driven weight loss will prevent sleep apnoea at a population level.
Don't overread it
This measured incident clinical OSA diagnoses, not sleep-study severity; it does not contradict trials showing GLP-1 agonists reduce severity in people who already have OSA.
The statistics, in plain English
A hazard ratio of 1.07 with a confidence interval from 0.93 to 1.23 straddles 1.0, so no protective effect was detected. The active comparator (a weight-neutral drug) makes this a fair test of the weight-loss hypothesis.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for pulmonology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free