Because it is often difficult to conduct randomised trials in children, paediatrics can sometimes remain a bastion of untested dogma, as with the vexed question of recurrent urinary tract infections in children. The logic used to run: recurrent UTIs in childhood can cause renal scarring and lead to renal failure in early adulthood; many recurrent UTIs are associated with vesico-ureteric reflux, so all children with them should have micturating cysto-urethrography and if necessary reflux surgery; and those who are spared the knife need prophylactic antibiotics. Almost all of this is wrong, and thousands of children have been put through intense misery through misapplied clinical logic. Even the prophylactic antibiotics are useless: this study shows that they merely encourage microbial resistance.

Very occasionally, people taking chloroquine as an antimalarial experience hypoglycaemia, and pharmacological clues of this sort led a group of rheumatologists to analyse their clinic records for the past twenty or more years to find out if patients taking hydroxochloroquine for rheumatoid arthritis showed any difference in their incidence of diabetes. The difference is quite convincing: a drop of over 70% for those on HCQ for more than four years, and an overall 40% reduction among ever-takers.

Given that we’re not likely to be putting half the population on hydroxychloroquine in the near future, we badly need more and better drugs for type 2 diabetes. As we’ve seen with rosiglitazone, the problem is that HbA1c levels tell us little about vascular risk, so each new agent is on probation while we await the results of long-term studies with hard end-points. The latest drugs are those which either mimic incretin or slow down its breakdown by inhibiting dipeptidyl peptidase 4. The latter are oral drugs, but they have already shown a worrying trend towards increasing the rate of significant infection; the former (incretin analogues) need to be given by injection. At least they don’t cause weight gain, but I can’t see them becoming popular, as most diabetics of my acquaintance dread the day when they will have to inject themselves. Instinctively I’d infer that an inexorable impulse towards increasing initiation of incremental incretin-imitator injections is intrinsically improbable.