Background: Various types of supraglottic airway devices (SGAs) have been developed for children. To compare the clinical performance of SGAs in children, we conducted Bayesian network meta-analyses, which combined direct and indirect evidence to estimate the difference in the effect of and probabilities of ranking of SGAs.
Methods: We searched MEDLINE, the Cochrane database, Embase, the Web of Science, and clinicaltrials.gov. RCTs comparing SGAs in children who underwent general anaesthesia were included. The primary endpoints were oropharyngeal leak pressure (OLP), insertion success/failure at first attempt, and blood staining on device. The study quality was assessed using the Cochrane risk of bias tool. The pooled estimates of mean difference (MD) or odds ratio (OR) with corresponding 95% credible interval (CrI) were obtained using the Markov chain Monte Carlo method. We performed sensitivity analyses after restricting to high quality RCTs. 
Results: In total, 57 RCTs with 5035 participants were identified, including 13 types of SGAs (Fig 1). Laryngeal tube (LT), i-gel, Cobra, and Proseal LMA had a statistically significant difference on OLP than did LMA Classic (MD [95% CrI] = 6.9 [2.8, 11], 4.1 [2.3, 5.9], 4.1 [0.6, 7.5], 3.6 [2.1, 5.1], respectively). The effect of LT could not be confirmed by the sensitivity analysis because of lack of high quality RCTs for LT. Insertion failure at first attempt was similar in all devices. I-gel was the only device that significantly reduced the risk of blood staining on device compared to LMA Classic (OR [95%CrI] = 0.42 [0.19, 0.88]) (Fig 2). 
Conclusions: Considering beneficial and harmful effects, i-gel may be the best device. Although LT showed clinically meaningful difference on OLP compared to LMA Classic, the effect of LT should be examined in future trials with high methodological quality.

A comparison of the clinical performance of various types of supraglottic airway devices in children: a Bayesian network meta-analysis