We’ve said before that EBM is about using the ‘best available’ evidence, and that while we’d love all our treatments to be supported by large RCTs and systematic reviews, Archimedes is living proof that this often isn’t the case.

Sometimes we’ll be in the situation where the treatment could be quite reasonably subjected to a randomised trial. But there isn’t a trial that’s up and open to be accessed. So if we offer that treatment … is it experimental? What makes a treatment ‘experimental’? What does ‘experimental’ even mean?

The concept I have is that an ‘experimental’ therapy is one where we haven’t used it for this indication. The drug may well have been used in humans, in children even, but the basis for the choice is entirely pathological. Each time we use a new drug though, it’s surely an ‘experiment’ for each child? This is reasonable to suggest, but it uses a different flavour of meaning: it suggests that anything with an uncertain outcome is an ‘experiment’, and this is not a common reading of the word. Experimental therapies don’t have any clinical experience, though they may have pathological reasoning, and they are necessary to advance medical care. These advances are quite reasonably undertaken within the strict guidelines and guardianship of research ethics frameworks, with legal and moral codes and checks and balances. To undertake this outside of this framework could risk the patient, family and clinician. The frameworks in law and governance are to safely support the advancement of medical science… experimenting outside of them can be seen as rogue.

Now I’m not saying all rogues are not attractive, not useful nor quite a lot of fun, but do we want them looking after the children?

• Archi