All Times EDT
Keywords: inference, effect estimation, time trend, non-concurrent control, platform trial, shared control
Platform trials aim at evaluating the efficacy of several experimental treatments within a single trial. The number of experimental arms is not prefixed, as arms may be added or removed as the trial progresses. Platform trials offer the possibility of comparing the efficacy of experimental arms using a shared control group. Compared to separate trials with an own control, this increases the statistical power and reduces the number of required patients. Shared controls in platform trials include concurrent and non-concurrent control data. Non-concurrent controls for a given experimental arm refer to data from patients allocated in the control arm before the arm enters the trial. The use of non-concurrent controls is appealing because it may improve the trial’s efficiency while decreasing the sample size. However, since the trial is perpetually open to adding arms and recruiting patients, the randomization occurs at different times. This lack of true randomization over time might introduce biases due to time trends. Moreover, platform trials allow the treatment given in the control group to be modified during the trial. If a treatment proves superior to the control, it can become the new control. This change of control arm reformulates the hypothesis to be tested and should be considered. In this talk, we review methods to incorporate non-concurrent control data in treatment-control comparisons allowing for time trends. We focus mainly on frequentist approaches that model the time-trend and on Bayesian strategies that limit the borrowing level depending on the heterogeneity between concurrent and non-concurrent controls. We examine the impact of time trends on the operating characteristics of treatment effect estimators for each method, considering both single-stage studies and studies with interim analyses. Finally, we discuss statistical inference for comparing treatment arms with changing control arms and discuss the impact on hypothesis tests and the estimands.