Fig. 2From: Handling an uncertain control group event risk in non-inferiority trials: non-inferiority frontiers and the power-stabilising transformationType I error (top) and power (bottom) of ‘do not modify margin’ and ‘modify margin’ procedures, using the risk difference (left) or risk ratio (right) scale. Data are generated according to the base-case scenario of Table 1 for varying values of control event riskBack to article page