In the benchmark approach, we consider the case where data are not available for the active comparator versus placebo on the outcome of interest. In these situations, we propose using thresholds that have been employed for similar purposes to benchmark
and inform the selection of the safety threshold.