A permutation-based funnel plot for subgroup analysis

Abstract This article presents an extension to the Subgroup Explorer, an interactive R Shiny app used for explorative post-hoc subgroup screening in clinical trials. The original method calculates point estimates for all subgroups in a clinical trial and visualizes them in a scatter plot. The proposed extension introduces confidence intervals for the treatment effect estimates, obtained through the inversion of permutation tests. The confidence intervals are constructed for a larger number of point estimates, allowing for a twodimensional confidence region along the funnel plot. The method's performance is evaluated on synthetic clinical trial data.