Conceptual problems and questionable evidence in defining non-inferiority margins - a reply to Flückiger et al.

From 19 meta-analyses comparing active bona fide treatments, Flückiger et al. (2026) recently calculated a between-group effect size (SMD) of -0.17 for primary outcomes and of -0.16 for secondary outcomes. They suggested using these effect sizes as empirically derived non-inferiority margins. However, Flückiger et al.'s approach is conceptually problematic and lacks empirical support. Without a rationale, empirically found differences between active treatments cannot per se be regarded as non-inferiority margins.