OBJECTIVE To determine on a large scale the multiple medical and nonmedical factors that influence glycemic control in the general population of children with diabetes, we performed a nationwide French cross-sectional study.

RESEARCH DESIGN AND METHODS We enrolled 2,579 patients aged 1–19 years with type 1 diabetes of > 1 year's duration. The study was center based: 270 centers were identified, 206 agreed to participate, and 147 included at least 90% of their patients. Questionnaires were completed by physicians interviewing patients and family, and HbA_1c measurements were centralized. To identify explanatory variables for HbA_1c level and frequency of severe hypoglycemia, we performed multiple regression analysis using all the quantitative variables collected and stepwise logistic regression for the qualitative variables.

RESULTS Mean HbA_1c value for the whole population was 8.97 ± 1.98% (normal 4.7 ± 0.7% [SD]). Only 19 children (0.7%) had ketoacidosis during the 6 months before the study, whereas 593 severe hypoglycemia events occurred in 338 children (13.8%). Control was better in university-affiliated hospitals and centers following > 50 patients, reflecting the importance of access to experienced diabetologists. Children had a mean of 2.3 injections, allegedly performed 2.8 glucose measurements per day, and were seen an average of 4.6 times per year at the center. In the multiple regression analysis, 94% of the variance of HbA_1c was explained by our pool of selected variables, with the highest regression coefficient between HbA_1c and age (R_c = 0.43, P < 0.0001), then with daily insulin dosage per kilogram (R_c = 0.28, P < 0.0001), mother's age (R_c = 0.26, P < 0.0001), frequency of glucose measurements (R_c = 0.21, P < 0.0001), and diabetes duration (R_c = 0.14, P < 0.0001). Logistic regression identified quality of family support and dietary compliance, two related qualitative and possibly subjective variables, as additional explanatory determinants of HbA_1c. The frequency of severe hypoglycemia was 45 per 100 patient-years and correlated with diabetes duration, but not with HbA_1c levels or other variables.

CONCLUSIONS Although overall results remain unsatisfactory, 33% of studied French children with type 1 diabetes had HbA_1c <8%, the value obtained in Diabetes Control and Complications Trial adolescents treated intensively. Diabetes management in specialized centers should be encouraged.