GPH 40: Causation and Causal Inference

In this episode, Medlock Holmes steps into one of the most fundamental questions in public health: What does it mean to say that something causes disease?From early infectious disease models and Koch’s postulates to Bradford Hill’s criteria and modern counterfactual reasoning, we unpack how causal thinking has evolved. We explore the difference between association and causation, and why that distinction matters profoundly for public health decision-making.Holmes guides us through the architecture of causal inference: confounding, bias, mediation, and effect modification. We examine how randomized trials approximate the counterfactual ideal - and why observational studies must rely on rigorous design and analytic discipline.We also introduce Directed Acyclic Graphs (DAGs) as visual tools for clarifying assumptions and avoiding analytic traps such as collider bias. Finally, we consider multicausality and the sufficient-component cause model, recognising that most public health problems emerge not from single agents, but from interacting systems.This episode provides the intellectual scaffolding for everything that follows in epidemiology. Without causal inference, there is no credible intervention.Key Takeaways* Association alone does not establish causation.* Bradford Hill’s considerations guide - but do not dictate - causal judgement.* Counterfactual thinking underpins modern epidemiology.* Confounding and bias threaten valid causal inference.* Randomization strengthens internal validity but is not always feasible.* Most diseases arise from interacting component causes rather than single agents.* Causal inference is foundational to ethical and effective public health policy. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit drmanaankarray.substack.com/subscribe