@marblenode40
Active 3 months, 3 weeks ago
We compared approximated treatment effects with patient-level modification for confounders, IV with treatment choice as the instrument, and modified IV, with hospital included as a fixed effect when you look at the regression models. The usage patient-level adjustment resulted in biased estimates for the analyses that included unmeasured […] View