This review looked at many studies on sleep‑apnoea treatments and found that while CPAP is still the top way to cut the number of breathing pauses, the diabetes drug tirzepatide also cuts those events by a lot and has solid evidence behind it. Exercise boosts quality of life the most, and most other options help daytime sleepiness. However, long‑term safety and how well people stick with these treatments need more research.

Obstructive sleep apnoea (OSA) is a prevalent chronic condition that is associated with cardiometabolic and neurocognitive complications. While continuous positive airway pressure (CPAP) remains the first-line therapy, suboptimal adherence limits its effectiveness, highlighting the need to evaluate alternatives such as mandibular advancement devices (MADs), hypoglossal nerve stimulation (HNS), physical activity, different modalities of weight loss management including glucagon-like peptide-1 (GLP-1) agonists and combination therapies. We conducted an umbrella review to synthesise high-level evidence from meta-analyses of randomised controlled trials (RCTs) evaluating the efficacy, adherence, and safety of therapies used in patients with OSA. A comprehensive search was performed in PubMed, Embase, Web of Science, and the Cochrane Database of Systematic Reviews covering the period from January 1, 2017, to July 5, 2025. Eligible studies were meta-analyses published in English assessing interventions targeting key OSA outcomes, including changes in apnoea-hypopnoea index (AHI), Epworth Sleepiness Scale (ESS), quality of life (QoL), blood pressure (BP), treatment adherence, and safety. When multiple meta-analyses addressed the same intervention-outcome pair, the one including the highest number of RCTs was retained. Two reviewers independently screened studies and extracted data. Methodological quality was assessed using the AMSTAR 2 tool, and the certainty of evidence was evaluated using the GRADE framework. Meta-analyses published in languages other than English, those focusing on paediatric populations or interventions outside the scope of conventional OSA management, and meta-analyses that did not report any of the pre-specified outcomes/interventions of interest were excluded. The review protocol was registered in PROSPERO (CRD42023420729) and the Open Science Framework (https://osf.io/2jvsx). A total of 5571 meta-analyses were identified. Of these, 34 met the inclusion criteria, encompassing 230 RCTs and 36,353 participants (n = 26,058 [72.3%] male). GRADE assessment showed that 12 meta-analyses (35%) had evidence that was of low certainty, 23 (68%) provided moderate-certainty evidence, and only one (3%) provided evidence that was of high certainty. CPAP was the most effective treatment for reducing AHI (mean difference [MD] -30.7 events/h; standardised mean difference [SMD] -1.65, 95% confidence interval [CI] -1.87 to -1.43; low-certainty evidence), followed by GLP-1 receptor agonists (tirzepatide: MD -21.86 events/h; SMD -0.84, 95% CI -1.01 to -0.68; moderate-certainty evidence) and MADs (MD -11.91 events/h; SMD -0.73, 95% CI -14.25 to -9.75; low-certainty evidence). CPAP, wake stimulants, HNS, and myofunctional therapy significantly reduced daytime sleepiness (ESS score SMDs of -0.80 to -0.88; moderate-certainty evidence except for pitolisant and solriamfetol, which were supported by high-certainty evidence). Physical activity led to the greatest improvements in QoL (SMD 1.3, 95% CI 0.58 to 2.02; moderate-certainty evidence), while CPAP also showed modest benefits (SMD 0.16, 95% CI 0.11 to 0.21; critically low-certainty evidence). This umbrella review identified CPAP as the most effective intervention for reducing AHI and daytime sleepiness in patients with OSA, while physical activity yielded the greatest improvements in quality of life. Data on safety, long-term adherence, and combination therapies remain scarce, underscoring the need for more comparative and longitudinal research to support personalised treatment strategies. Data need to be interpreted in the context of several limitations, including those relating to the meta-analysis inclusion criteria and the quality of data in the meta-analyses themselves. None.