Association, cause, and causal association, revised: reasoning and methods

Association and causation are often confused; the principle difference is that causation is directional. In this chapter we discuss the initial description of causation as it applies in biomedical studies, occurring first is a rather linear manner directly in a laboratory experiment, then expanding outward toward observational studies where many effects occur simultaneously. In doing so we introduce causal “mediation analysis” as methodologic technique to address multiple and complex confounding effects and intermediate or mediator factors by creating instrumental variables. We discuss how Mendelian randomization can also be used to address confounding through a similar instrumental variable approach, and how the explosion of genomic knowledge might also be incorporated into the direct causal framework. Finally, we muse about all these effects in the context of the public health.