Short-term efficacy of minimally invasive treatments for adult obstructive sleep apnea: A systematic review and network meta-analysis of randomized controlled trials

Many treatments have been proposed for adult obstructive sleep apnea (OSA), but no comprehensive comparison of all interventions has been performed. We aimed to compare and rank the effectiveness of all minimally invasive treatments for adult OSA in a systematic review and network meta-analysis. Literature was searched within Ovid MedLine, EMBASE Classic+Embase, Cochrane library, and Cochrane Database of Systematic Reviews from inception to Aug 9th, 2016 for randomized controlled trials comparing minimally invasive treatments for adult OSA. The outcomes were the changes in apnea-hypopnea index (AHI) and Epworth sleepiness scale (ESS). Frequentist approach to network meta-analysis was used and treatment hierarchy was summarized according to the surfaces under the cumulative ranking curves. Eighty-nine randomized controlled trials comprising 6346 adult OSA participants and comparing 18 different interventions were included. In comparison with no treatment, positive airway pressure (PAP) was most effective in reducing AHI (23.28 [weighted mean difference]; 95% confidence interval: 19.20–27.35). PAP was ranked first followed by mandibular advancement device (MAD) in reducing AHI. Exercise was ranked first followed by cervico-mandibular support collar in reducing ESS. Considering the effectiveness in reducing both AHI and ESS, PAP was ranked the best, followed by MAD and positional therapy, while lifestyle modification alone was the least effective intervention. Interventions that are highly effective in reducing objective laboratory-derived AHI do not demonstrate equivalent effectiveness in improving patients' subjective sleepiness. Future improvement of the interventions is necessary to simultaneously improve both objective and subjective outcomes. Keywords: Network meta-analysis, Randomized controlled trial, Obstructive sleep apnea

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.