Abstract

Dr John’s talk will be chaired by Professor Sir David Spiegelhalter, Winton Professor of the Public Understanding of Risk, University of Cambridge.

Any claim about an individual’s risk of harm is always made relative to some reference class (or body-of-information): there is no such thing as an individual’s “actual” risk of lung cancer, but, rather, his risk “as a smoker” or “as a 40 year old”. This familiar claim in the philosophy of statistics gives rise to a simple question: which risk estimate should physicians communicate to individuals in the clinic? In this paper, I explain why this question is so difficult; I\argue that some standard ways of addressing it in the medical literature are deeply unethical; and I investigate the relationship between my question and the more familiar question of how we should present risk information.

Unfortunately, the term “best” is ambiguous between “most effective way to change behaviour” and “most effective way to ensure uptake”. Furthermore, there is a surprising silence on an even more fundamental question: what risks should we communicate in the first place?