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 havebeen employed for similar purposes to benchmark
and inform the selection of the safety threshold. Such threshold analogues can be found in1: previous NI trials that used a similar efficacy endpoint as the safety endpoint of interest, and2 data used by agencies when making regulatory decisions, such as adding a black box warning to a drug label or withdrawing a drug from the market