Severe hypoglycemia and ketoacidosis over one year in Italian pediatric population with type 1 diabetes mellitus: a multicenter retrospective observational study

Background and aims: Evaluation of incidence and correlates of severe hypoglycemia (SH) and diabetes ketoacidosis (DKA) in children and adolescents with T1DM. Methods and results: Retrospective study conducted in 29 diabetes centers from November 2011 to April 2012. The incidence of SH and DKA episodes and their correlates were assessed through a questionnaire administered to parents of patients aged 0e18 years. Incidence rates and incident rate ratios (IRRs) were estimated through multivariate Poisson regression analysis and multilevel analysis. Overall, 2025 patients were included (age 12.4 � 3.8 years; 53% males; diabetes dura- tion 5.6 � 3.5 years; HbA1c 7.9 � 1.1%). The incidence of SH and DKA were of 7.7 and 2.4 events/ 100 py, respectively. The risk of SH was higher in females (IRR Z 1.44; 95%CI 1.04e1.99), in pa- tients using rapid acting analogues as compared to regular insulin (IRR Z 1.48; 95%CI 0.97e2.26) and lower for patients using long acting analogues as compared to NPH insulin (IRR Z 0.40; 95% CI 0.19e0.85). No correlations were found between SH and HbA1c levels. The risk of DKA was higher in patients using rapid acting analogues (IRR Z 4.25; 95%CI 1.01e17.86) and increased with insulin units needed (IRR Z 7.66; 95%CI 2.83e20.74) and HbA1c levels (IRR Z 1.63; 95% CI 1.36e1.95). Mother' sa ge was inversely associated with the risk of both SH (IRRZ 0.95; 95%CI 0.92e0.98) and DKA (IRR Z 0.94; 95%CI 0.88e0.99). When accounting for center effect, the risk of SH associated with the use of rapid acting insulin analogues was attenuated (IRR Z 1.48; 95%CI 0.97e2.26); 33% and 16% of the residual variance in SH and DKA risk was ex- plained by center effect. Conclusion: The risk of SH and DKA is mainly associated with treatment modalities and strongly depends on the practice of specialist centers.