Testing for Mechanistic Interactions in Long-Term Follow-Up Studies
Fig 2
Type I error rates under the null hypothesis of no mechanistic interaction: interaction contrast (B) and type I error rates (C) for proportional hazards (A), interaction contrast (E) and type I error rates (F) for non-proportional hazards (D), and interaction contrast (H) and type I error rates (I) for crossover hazards (G).
Bold solid lines in A, D and G are hazard rates for people with exposure profile of X = 1 and Z = 1, thin solid lines, those for X = 1 and Z = 0, bold dotted lines, those for X = 0 and Z = 1, and bold broken lines, those for X = 0 and Z = 0. Bold solid lines in C, F and I are for MIT (mechanistic interaction test), bold dotted lines, for PRISM (peril ratio index of synergy based on multiplicativity), bold broken lines, for RERI (relative excess risk due to interaction), and the horizontal thin solid lines, the nominal α level of 0.05. The thin broken line in C is for the interaction test in the Cox model.