It is said of screening programmes that “all do harm; some do good as well, and, of these, some do more good than harm at reasonable cost”.1 Arguably the same can be said of testing done in the context of routine clinical practice to inform diagnosis or prognosis.2 Editorials and correspondence in the medical literature have discussed the problems associated with over-reliance on diagnostic testing, and cite examples of clinical harm and waste of scarce resource caused by inappropriate testing.2–4 …