[samhw]

Indeed, you're referring to the doctrine of double effect (which the linked Wikipedia article on R v Adams mentions).

That said, I don't think "[only] doses that are virtually certain to cause death lead to liability for the death" is correct. You had it right the first time: the relevant fact is not whether it's likely or even certain to cause death, but rather whether causing death is the primary intention. (I'm not a lawyer, though, I'm just speaking on the basis of my knowledge as a lay person.)