The epistemological function of Hill's criteria.

OBJECTIVE This article outlines an epistemological framework for understanding how Hill's criteria may aid us in establishing a causal hypothesis (A causes B) in an observational study. METHOD We consider Hill's criteria in turn with respect to their ability or otherwise to exclude alternative hypotheses (B causes A; there is a common cause of A and B; there is no causal connection between A and B). RESULTS We may classify Hill's criteria according to which of the alternative hypotheses they are able to exclude, and also on the basis of whether they relate to (a) evidence from within observational study or (b) evidence independent of that study. It is noted that no criterion is able to exclude the common cause hypothesis in a systematic way. CONCLUSION Observational studies are typically weaker than experimental studies, since the latter can systematically exclude competing hypotheses, whereas observational studies lack a systematic way of ruling out the common cause hypothesis.