Fig. 2From: Interval-cohort designs and bias in the estimation of per-protocol effects: a simulation studyA causal diagram representing the underlying mechanism for protocol deviations in our studyBack to article page