Authors’ Reply to: VanderWeele et al., Chiolero, and Schooling et al.

inference is of course appealing. However, pragmatism cannot be really considered as a scientific method; it is essentially a practice. Further, this approach remains relatively vague, which is problematic as there is no point in estimating a causal effect that is not well defined, especially for ‘policy makers who will be unable to translate [this vague causal effect] into effective interventions’. The counterfactual and interventionist approach is better defined, and fundamentally pragmatic, by making explicit the link with interventions rather than statistical association, will help us move from risk factor toward ‘consequentialist’ epidemiology.