Figure 4From: Meta-analysis of randomized phase II trials to inform subsequent phase III decisions Probability that the upper bound of the OR's 95% credible interval may be less than 1 in a new trial for ICH. Fixed variance (option 2) indicates the variance of Y i new is assumed to be a fixed value which is approximated by assuming a baseline risk of 0.01 in the control group, a particular sample size, and a treatment effect of 0.485 (summary OR estimate from model 2). Sampled variance (option 1) indicates the variance of Y i new is calculated for each sample of the estimation process, assuming a baseline risk of 0.01 in the control group, a particular sample size, and using the sampled ( θ i new ) (Equation 6).Back to article page