Time Partitioning in Target Trial Emulation

Abstract

In target trial emulation, time partitioning enables researchers to handle time-varying confounders and immortal time bias with appropriate methods. Based on two clinical scenarios, this study aimed to explore issues related to time partitioning and to provide guidance for trial emulation. After formalizing the research question within the framework of structural causal models, we show how a given time partitioning may be too fine or too coarse depending on the clinical context. When the partitioning is too fine, the dimensionality of the model is unnecessarily high. When the partitioning is too coarse, the resulting causal structure may hinder effect estimation. We also show that cloning-censoring-weighting may not be valid when treatment influences outcome within study periods, and we confirm this through simulations. In conclusion, we provide practical guidance for actively specifying an appropriate time partitioning in trial emulation, rather than using the available data resolution as a default.

Figures (7)

• Figure 1: Causal graph for the first three periods of for the first clinical scenario. $X_1,X_2,X_3$: Treatment. $Y_1,Y_2,Y_3$: Vital status. $C$: Baseline confounders. $A$ and $B$: Unobserved variables.
• Figure 2: Ancestral Multi-World Networks (AMWN) for testing conditional independence \ref{['eq:cain']} in simplified versions of the clinical scenarios. The variables of interest are represented in blue, the conditioning set is represented with rectangles, and biderected dashed arrows represent hidden confounders (emerging from the AMWN procedure). In red, we highlight an open path between the variables of interest in the second clinical scenario.
• Figure 3: Simulation results assessing the bias of nonparametric cloning-censoring-weighting (CCW) and of a nonparametric maximum likelihood estimator of the estimand. Comparison of "always treat" versus "never treat" in simplified 3-period scenarios. 1,000 simulated datasets of 1,000 patients each. CI: Confidence interval.
• Figure 4: Causal graph for the first three periods of the second clinical scenario. $X_1,X_2,X_3$: Treatment. $Y_1,Y_2,Y_3$: Vital status. $C$: Baseline confounders. $A$ and $B$: Unobserved variables.
• Figure 5: Cyclic causal graph. $X_1,X_2,X_3$: Treatment. $Y_1,Y_2,Y_3$: Vital status. $C$: Baseline confounders. $A$ and $B$: Unobserved variables.