1. to define causal interventions and actions, draw and interpret causal diagrams, and apply the rules of causal diagrams to distinguish causal from non-causal statistical associations.
2. to decide which biostatistical/epidemiological methods must be used in different situations to derive causal effect parameters.
3. to use causal diagrams to estimate the direction of bias in "non-causal" models.
4. to understand the conceptual approaches of causal modeling
5. to know how these methods are applied in big real world data and pragmatic trials 

Published cardiovascular, oncology, HIV, nutrition and obstetrics examples will be used to:

• Adjust for non-adherence in randomized clinical trials, where both intention-to-treat and naïve per protocol analyses may fail to yield the true causal intervention effect
• Assess the “fallibility of estimating direct effects” (i.e., adjusting for intermediate steps);
• Adjust for time-independent confounders (i.e., confounder affects both risk factor and disease), where standard stratification, regression analysis, propensity score methods and matching/balancing approaches yield valid causal effects if all confounders are measured
• Adjust for time-dependent confounding (i.e., the confounder simultaneously acts as intermediate step in the causal chain between risk factor and disease), where standard regression analysis fails and "causal methods" such as marginal structural models or g-estimation must be used. 

Software Practicals: 
We will work through three practical analyses using the software R to demonstrate adjustment for baseline and time-dependent confounding in a randomized trial of cholesterol-lowering medications and mortality.

• Example 1: Drawing a causal diagram for a pragmatic trial in patients with elevated cholesterol levels with potential non-adherence; discussing the potential for bias.
• Example 2: Adjusting for baseline confounders to improve efficiency and interpretability of the intention-to-treat effect using standardization.
• Example 3: Focusing on the placebo arm only, demonstrating how to control for bias introduced by non-adherence using baseline and time-dependent confounding with marginal structural models.