Blood or Urine Glucose-Based Insulin Therapy and Control of Glycemia: Computer-Simulation Study

Adjustment algorithms for conventional insulin therapy must be tested for safety and efficacy before clinical implementation. We did this by computer simulation. Accordingly, a computer simulator of human intermediary metabolism created 10 randomly chosen diabetic subjects for study. All were well defined with respect to compliance (i.e., medication and diet) and life-style (i.e., physical and emotional stress). Insulinadjustment algorithms that were tested calculated daily insulin dosages for these computer-simulated patients based on either blood or urine glucose concentrations self-measured 4 times/day before breakfast, lunch, dinner, and bedtime snack. The twofold purpose of the simulation study was to determine the ability of the adjustment algorithms to improve initially poor metabolic control and to compare the outcomes when either blood or urine glucose measurements were the basis on which glycemic control was implemented. A significant improvement in metabolic control could be achieved with either blood or urine glucose measurements as input to the algorithms. Detailed comparisons between blood and urine glucose-based treatments showed no significant advantage of blood glucose-based algorithms at breakfast (122 ± 21 vs. 131 ± 16 mg/dl) and dinner (117 ± 27 vs. 130 ± 23 mg/dl), whereas mean glycemia at lunch (122 ± 24 vs. 164 ± 21 mg/dl) and bedtime (117 ± 25 vs. 150 ± 21 mg/dl) after 120 days of simulation did differ significantly (P < 0.01). Hypoglycemia was not provoked by either treatment. Total daily insulin doses evolved by blood glucose-based algorithms were significantly (P < 0.05) higher than the doses used by urine glucose-based algorithms (53 vs. 47 U). We conclude that either blood or urine glucose-based insulin-adjustment algorithms that we devised promise to achieve improved glycemic control with only minor differences between the two. Clinical studies can now safely be conducted.