Fig. 4From: How unmeasured confounding in a competing risks setting can affect treatment effect estimates in observational studiesResults from Scenario 2 Legend Back to article page