[Accuracy of blood glucose self-monitoring in children with type-I diabetes].

Children with type I diabetes often believe that they are able do gauge their blood glucose value without measurement. This study investigated how accurate these estimates are and which of the patient characteristics might affect accuracy. Nineteen children with type I diabetes (age 7.3-17.3 years, duration of diabetes 0.1-12.0 years) estimated their blood glucose immediately before a blood glucose measurement by reflectance meter. Ten to 54 estimations and measurements were done by each individual during a summer camp. According to the method of Cox et al. accuracy of estimates was evaluated by error grid analysis. Five zones were defined: accurate estimates (zone A); clinically benign errors (zone B); clinically dangerous errors (zone C, D, E). An overall accuracy index was calculated by subtracting the summed percentage of estimates in zones C, D, E from the percentage in zone A. The mean percentage of estimates in zone A was 41.5 (10.0-75.0)%, the mean overall accuracy index was 9.8 (-47.4-62.5)%. Except for mean blood glucose which was inversely correlated with accuracy index (r = -0.53, P = 0.02) none of the other patient characteristics showed an effect on accuracy index.Accuracy of blood glucose estimates is lower in children with type I diabetes than in adults and varies substantially between individuals. High mean blood glucose values are associated with a lower accuracy of estimates.