I had not previously considered that drugs in use today may cause diabetes. I discovered that children who had suffered from tonsillitis or ear problems were very likely to have been exposed to antibiotics, and that eczema and exfoliative dermatitis are side effects of exposure to antibiotics. Asthmatics and sickly children were also probably more likely to visit the doctor, and be prescribed drugs. It was at this stage that I began to consider the possibility that drug exposure may be the common factor.

I then undertook further studies to look at the drug exposure of the diabetic child in utero, during birth and from birth to diagnosis of the disease.

Of 35 diabetic children attending a hospital in Kent, whose parents agreed to participate in the study which involved the examination of their child's medical records, I discovered that if the child had been exposed to drugs in utero or during birth, he was more likely to have earlier onset of diabetes. Of the 65 per cent of children exposed to drugs (excluding pethidine) the average age of onset was 4.9 years, compared with an average age of onset of 6.5 years among those without drug exposure. Furthermore, the percentage of children in the study who were induced was 32 per cent, compared with a rate of 18 per cent for the S.E. Thames region in 1989.

If anything, my results represent considerable underreporting of drug exposure since in many cases mothers were being interviewed years after pregnancy or labour.

I went on to study diabetic children diagnosed before three years and attending a hospital in the Wessex region, and before four years and attending a hospital in the Cambridgeshire Area Health Authority.

Within these two groups of patients, some 58 per cent had a history of in utero or drug exposure (excluding pethidine) during labour. In one case, a six year old had been given amoxycillin five times before his first birthday, twice during his twos, and three more times during his third and fourth year. Besides nine other courses of antibiotics, he received a cephalosporin, antihistamines, a powerful anti vomiting drug and one for gastrointestinal spasms, and Bactrim, a very potent antibiotic (see Q&A, p 10).

I got hold of another study group from local diabetic groups and The Daily Telegraph newspaper readers in the UK. Both groups had a high proportion (60 per cent) of children who had experienced drug exposure either in utero or during labour.

In another case, a child diagnosed as diabetic at 11 years had very little history of drug exposure after birth. However, the child's drug exposure from conception reveals a pattern similar to that of some other diabetic children in the study: his mother took antibiotics while she was pregnant, and was given oxytocin (to induce labour) and/or ergometrine (to control postpartum bleeding) during and after labour. If antibiotics are implicated in the onset of diabetes, the records I examined suggest that it may take many years after exposure before clinical diabetes becomes apparent (see box p 3).

This long latency period between exposure and disease is likely to occur if drugs were administered to the child at a very young age, causing the destruction of only a proportion of the insulin producing cells of the pancreas. The disease would only manifest itself only when the demand for insulin increased, possibly due to increasing consumption with age. And of course the younger the age at which drug exposure takes place, the earlier and greater the potential damage, since organs are small and immature and may therefore take longer to remove toxic chemicals from the body. The sheer profusion of drugs to which these children had been exposed prompted me to compare the chemical structures of drugs known or thought to be diabetogenic (capable of causing