Figure 3.

Trial sequential analysis excluding the trial of Pryor. Meta-analysis of the trials by Greif [13], Belda [12] and Mayzler [14], excluding the trial of Pryor [15] with a required information size of 1304 (APIS, a priori information size) based on an a priori relative risk reduction (RRR) of 33% and a type I error risk of 5% and a power of 80%. The cumulative z-curve constructed for a fixed-effect model as heterogeneity is 0% crosses both the traditional boundary (P = 0.05) after the first trial and the trial sequential monitoring boundary during the second trial. So there may be evidence for an effect of at least 33% RRR in a cumulative meta-analysis of trials investigating a high oxygen fraction when the Pryor trial is excluded when adjusting for repeated analyses of accumulating data.