No abstract

SUMMARY In prescribing a diabetic diet more attention has traditionally been paid to the amount of dietary carbohydrate than to its type or structure. We have compared the effect on blood glucose of substituting unrefined, whole foods for refined, processed foods in liberal carbohydrate diets (50-55%/ of dietary energy) eaten by 10 diabetic children in a randomised crossover study. All measurements were made at home. The unrefined diet used whole foods (including dried beans) supplying 60 g/day of dietary fibre. The refined diet used processed foods supplying 20 g/day of dietary fibre. Diets were isocaloric for carbohydrate, fat, and protein. Glycaemic control was assessed by daily urine analysis for glucose, home blood glucose measurements, glycosylated haemoglobin, and by a 24-hour profile of blood and urinary glucose carried out at home after 6 weeks on each diet. Glycaemic control was significantly better on the unrefined diet. On profile days mean blood glucose levels on the unrefined and refined diets respectively were: preprandial: 5 * 5 and 8 4 mmol/l; postprandial 8 5 and 12-2 mmol/l. The mean 24-hour urinary glucose excretion on the unrefined diet was 9 3 g and on the refined diet was 38-0 g. Six months after the study the children were eating appreciably more dietary fibre than before (mean increase 13 * 6 g/day). Attention to food type and structure can improve blood glucose control in diabetic children and should provide an acceptable and more rational basis for dietary prescription than one based on carbohydrate quantity alone.Traditional diabetic diets have paid more attention to the amount of dietary carbohydrate than to its structure or type. There is however a substantial amount of evidence to suggest that changing the physical form of the food in the diet,'2 and increasing

The mean blood glucose concentration at breakfast (O minutes) was 11 mmol/l after the early injection and 10 mmol/l after the late injection. Subsequent concentrations were consistently lower with the early injection regimen than the late regimen. The greatest difference between values in the two groups was 3 7 mmol/l at 150 minutes. Mean plasma insulin concentrations were lower in the children on the early regimen than in those on the late regimen at 30 minutes before breakfast but higher at 0 minutes and thereafter. There were no significant differences in mean concentration of interUniversity Department of Paediatrics, John Radcliffe Hospital, Oxford OX3 9DU A L KINMONTH, MA, MRCP, research fellow in diabetes J D BAUM, MD, FRCP, honorary consultant paediatrician mediary metabolites between the two injection regimens. These were mainly within the normal range for healthy young adults except for the ketone concentrations, which were raised with both injection regimens until 180 minutes after breakfast.These results suggest that the timing of the morning injection of insulin is important in the control of postprandial hyperglycaemia in diabetic children. IntroductionIt is established that the management of diabetes mellitus should include a serious attempt to keep blood glucose concentrations as close to normal as possible.' The diurnal variation in blood glucose concentration with various insulin regimens has been well documented in diabetic adults2 3 and children (G Werther, MSc thesis, Oxford University, 1978) by using 24-hour sampling methods in hospital. Despite differences in regimen, a hyperglycaemic peak usually occurs after breakfast.2 3 Factors that may contribute to this hyperglycaemia include insulin deficiency, relatively high carbohydrate breakfasts, and rebound from nocturnal hypoglycaemia.Insulin deficiency should be improved by increasing the interval between the morning insulin injection and breakfast. A preliminary survey of 30 diabetic children attending our clinic showed that for most of them this interval was 15 minutes or less, either for convenience or for fear of hypoglycaemia. We performed this study to see whether the minimal disruption of giving the morning insulin injection 30 minutes rather than five minutes before breakfast could significantly reduce postprandial hyperglycaemia without producing preprandial hypoglycaemia in diabetic children. Our study also aimed at evaluating the acceptability of metabolic studies on such children at home.

subgroup within the Jewish community) have a particularly low incidence of carcinoma of the cervix. It is theoretically possible that all the apparent differences are the result of the proportions of Israeli women in the two groups.The authors state that the overall relative risk is of borderline significance. This must be some new use of the word significance. The 95',, confidence interval of the risk factor includes 1, and it would be more usual to state that the study had failed to show any significant difference between the two groups.It is clearly important that studies like this are published, even when, as in this case, the results are essentially negative and even when factors such as smoking and the sexual history of the partners are unaccountably ignored. However, to present associations which do not reach statistical significance as if they proved a causal relation is irrespon- SIR,-Dr Parkin has raised a number of issues. He states that no details of the method of age adjustment are given. This is clearly described in reference 17 of the paper, which is referred to in the last paragraph of the methods section; we see no reason to provide details of such standard statistical procedures when they are readily available elsewhere. They are certainly not needed, as Dr Parkin asserts, to assess the validity of the results presented.We recognise that the cases seen in the participating centres may not be representative of all cases in the populations of the countries in which the study was conducted. It was for this reason that the controls were selected not from the general populations of those countries but from among women admitted to the participating hospitals for conditions other than those under study. This procedure was instituted specifically to control for factors that are related to hospitalisation in the participating institutions.The fact that only 3",, of the cases but 20,, of the controls came from Israel does not explain our results. Risks were increased in oral contraceptive users in most countries, and the overall relative risk was adjusted for both age and centre.The 95",, confidence interval of the relative risk of 1-19 was 0 99 to 1 44. Rounding 0-99 to one significant figure gives us a lower 95),,

No abstract