RÉSUMÉ
The DCCT syudy report (in 1993) showed that long-term maintenance of near normoglycemia markedly delays the onset and/or progression of microangiopathic complications of type 1 diabetes. Now, intensive insulin therapy has been considered to be a standard treatment of type 1 diabetes. However, it brings about serious hypoglycemia three times as frequently as conventional therapy with split-dose insulin mixtures. If IDDM patients could predict their blood glucose levels, it would be beneficial to prevention of hypoglycemia and to strict glycemic control. We intended to research whether insulin-treated diabetic children in a diabetic summer camp could predict their blood glucose levels. The number of subjects was 28. They were elementary and junior high school children 9 to 16 years of age. They predicted their blood glucose levels before every meal and bedtime, and then monitored their blood glucose levels. Their HbA<SUB>1</SUB>c was 7.6±0.3%, total daily insulin dosage 36.0±3.2U/day, and infection times 3.3±0.2/ day.<BR>Measured blood glucose levels and predicted blood glucose levels were not distributed normally, but their natural logarithms (Ln (mBG), Ln (pBG)) were. Ln (pBG) (5.01±0.02) was significantly higher than Ln (mBG) (4.92±0.03) (p<0.01)<BR>A positive correlation was found between Ln (mBG) (x) and Ln (pBG) (y) (y=0.359 x+3.239, r=0.495). Events of which Ln (pBG) was within±20% and±30% of Ln (mBG) were 124 (31.5%) and 175 (44.5%), respectively, of the total 393 events. As for the relationship between Ln (mBG) and Ln (pBG) at each time, a change of Ln (mBG) was significantly larger than that of Ln (pBG).<BR>These results showed that prediction of blood glucose levels was difficult for type 1 iabetic children, especially when blood glucose levels were extremely high or low and when they fluctuated sharply.