Skip to main content
. 2022 Jan 5;31(2):300–314. doi: 10.1177/09622802211065158

Figure 1.

Figure 1.

The approximate mediation structure of the hypothetical RCT where treatment initiation is randomized. (Top) The total effect of early treatment initiation may be divided into the mediated effect on early delivery and the direct effect on the outcome. U indicates unmeasured common causes of early delivery and the outcome. (Bottom) If we are also interested in later treatment effects, early delivery is both a mediator and a variable that prohibits future treatment initiation.