Fig. 1From: A novel method for controlling unobserved confounding using double confoundersCausal diagram for X (treatment) and Y (outcome) with C1, C2 (observed confounders) and U (other unobserved confounders)Back to article page