Causal Inference in Clinical Trials and Observational Study

This course offers an overview of statistical foundations for causal inference and introduces new analytic methods for causal inference in both randomized controlled trials and observational study. The covered topics include outcome regression, propensity score, doubly robust estimation, and instrumental variables for estimating the average treatment effect, their extensions to estimating the heterogeneous treatment effect which is useful for developing precision medicine, marginal structure modeling for adjusting time-varying confounding, and sensitivity analysis for evaluating the effect of unmeasured confounding. This course also offers study design issues such as the choice of estimand and adaptive randomization. The course is a continuation of the BIODS 248A on clinical trial design and mainly focuses on making causal inferences via observational study including real world data. However, BIODS 248A is not required for this course, which is self-contained. Prerequisites: Working knowledge of statistical inference, probability theory, and R.