Twenty-six RCTs were included in the review (1,502 participants, range 23 to 200). Trials were of similar quality. None of the trials was blinded. Nine trials described randomisation by random numbers table or computer generation. All participants in all studies received their allocated intervention. Only one study described allocation concealment.
When all studies were pooled, desflurane reduced the mean time to extubation relative to propofol by 21% (95% CI 4% to 36%; 17 RCTs, p=0.01) and reduced mean time to follow commands by 25% (95% CI 5% to 41%; 19 RCTs, p=0.008).
Desflurane reduced the standard deviation of time to extubation by 30% (95% CI 6% to 48%, p=0.008) and the standard deviation of time to follow commands by 40% (95% CI 26% to 52%, p<0.001). The authors noted that variability in time to recovery mattered clinically as it contributed to the incidence of prolonged extubation times.
Sensitivity analysis did not substantially affect the main findings. There was statistically significant heterogeneity (p<0.001) for each endpoint and this was not explained by the measured variables. The fail-safe test did not suggest publication bias.