Fourteen RCTs were identified (n=2,600, range 26 to 761): three neonatal studies, four paediatric studies and seven adult studies. Nine of the 14 RCTs had allocation concealment. Seven studies used an intention-to-treat analysis. Twelve RCTs were blinded. Six studies performed a cuff-leak test. Eight studies described criteria for extubation and five reported criteria for reintubation. The overall quality level of the evidence was considered to be high. All except one study reported at least one episode of reintubation. There was no evidence of publication bias.
There were significantly lower rates of reintubation with the use of corticosteroids compared to controls (OR 0.56, 95% CI 0.41 to 0.77, I2=38%, absolute NNT=35 events prevented per 1,000 extubations; 13 studies).
There were no significant differences between subgroups, although studies with a treatment duration of more than 12 hours had a greater reduction in reintubation rates (OR 0.41; six studies) compared with studies of a shorter duration (OR 0.79; seven studies) (p=0.09).
There were significantly lower rates of laryngeal oedema with the use of corticosteroids compared to controls (OR 0.36, 95% CI 0.27 to 0.49, I2=71%, absolute NNT=84 per 1,000; nine studies), but with significant heterogeneity.
Results for mortality and adverse events were available for very few studies. A summary was reported.