Inference in the proposed framework employs the confidence
interval (CI) of some parameter that compares
the frequency of outcomes in patients exposed to the
new drug versus an active comparator. Here, we focus on the risk ratio (RR), but the risk difference (RD), rate ratio and rate difference could also be used. To illustrate,
we present six scenarios in Figure 1 where larger RRs indicate that the new drug is associated with a higher rate of the adverse outcome of interest versus the comparator. The dashed vertical line represents the hypothetical safety threshold (Δ). We describe approaches to developing this threshold below. The solid vertical line represents the null threshold (RR=1), and the horizontal lines represent the CIs