FIGURE 3.
Illustration of the Reverse‐Bayes borrowing method. The data from the (non‐significant) REMAP‐CAP trial are combined with a mixture prior consisting of the (significant) RECOVERY trial data and a unit‐information prior (both estimates shown with 95% confidence interval). The resulting posterior medians with equal‐tailed 95% credible intervals are shown for a range of mixing weights. The Reverse‐Bayes mixing weight leads to the highlighted posterior with upper credible interval limit fixed at one [Colour figure can be viewed at wileyonlinelibrary.com]