In Reply We thank Dr van Besien and colleagues for their comments regarding our Original Investigation.1 They identified 3 areas for further clarification. First, they commented on the forced introduction in the statistical models of the main variable (ie, the main effect) and its potential to obscure the effects of recognized covariates. We included the main effect in each of the outcomes because our hypothesis testing was on the main effect. In addition, because the analysis was based on registry data, we adjusted for the other significant covariates to test the hypothesis. We also tested interactions between the main effect and the selected covariates and found no significant interactions. Furthermore, we conducted propensity score matching to validate the results. All of these confirmed that there was no confounder with the main effect. Because there was no interaction between the main effect and the other covariates, we believe that obscuring the selected covariate effects by including the main effect in each outcome should be minimal.