Every intervention in medicine is built on a benefit–risk analysis. Clinical trials can formally assess this balance, estimating the number needed to treat vs the number needed to harm. But deciding the acceptability of this balance is not straightforward, because the nature, magnitude and frequency of the benefits and risks differ. Regulatory bodies in different countries make such decisions when licensing drugs. For example, they might decide that the trade-off is acceptable for chemotherapy agents, which reduce mortality significantly but frequently cause toxic neuropathies, but unacceptable when an antiseizure medication causes rare but potentially fatal liver failure or aplastic anaemia. The regulators must make complicated decisions, balancing the size of the benefit against the frequency and consequence of the adverse effects in the context of the alternative agents available. Once they are licensed, clinicians subsequently will discuss with their patients how best to use these agents and their individual benefit–risk ratio.

The benefit–risk trade off becomes more complicated when the benefit comes to one patient but the potential risk …