Fig. 5
From: On model-based time trend adjustments in platform trials with non-concurrent controls
Type 1 error rate and power of rejecting H02 for binary endpoints in the presence of step-wise trends (for λ0=λ2 and varying λ1) with respect to the response rate in the treatment arm 1 in the second period and according to the model used (see “Methods” section) for scenarios presented in “Design” section and described in Tables 2 and 3. Vertical lines refer to those situations when time trends are equal across arms on odds ratio (OR), risk difference (RD), and relative risk (RR) scales. ALLTC-step refers to models using all treatment data and control and adjusting for time by a step function (see (1)), ALLTCI refers to models using all treatment data and control with interaction between time and treatment arm and adjusting for time by a step function (see (2)), and pooled and separate approaches refer to logistic regression models without adjusting for time comparing treatment 2 to control using concurrent and non-concurrent control data, and concurrent control data only, respectively. Lines for ALLTCI-Step and the separate approach are overlapping