New-onset type 1 diabetes mellitus (T1D) in pediatric patients represents a clinical challenge for initial total daily insulin dosing (TIDD) due to substantial heterogeneity in practice and lack of consensus on the optimal starting dose. Our INSENODIAB (INsulin SEnsitivity in New Onset type 1 DIABetes) study is aimed at (1) exploring the influence of patient-specific characteristics on insulin requirements in pediatric patients with new-onset T1D; (2) constructing a predictive model for the recommended TIDD tailored to individual patient profiles; and (3) assessing potential associations between TIDD and patient outcomes at follow-up intervals of 3 and 12 months. We conducted a comprehensive analysis of medical records for children aged 6 months to 18 years, hospitalized for new-onset T1D from 2013 to 2022. The study initially involved multivariable regression analysis on a retrospective cohort (rINSENODIAB), incorporating baseline variables. Subsequently, we validated the model robustness on a prospective cohort (pINSENODIAB) with a significance threshold of 5%. The model accuracy was assessed by Pearson's correlation. Our study encompassed 103 patients in the retrospective cohort and 80 in the prospective cohort, with median TIDD at diagnosis of 1.1 IU/kg BW/day (IQR 0.5). The predictive model for optimal TIDD was established using baseline characteristics, resulting in the following formula: TIDD (IU/d) = ([0.09 × Age2] + [0.68 × %Weight Loss] + [28.60 × Veinous pH] - [1.03 × Veinous bicarbonates] + [0.81 × Weight] - 194.63). Validation of the model using the pINSENODIAB cohort demonstrated a significant Pearson correlation coefficient of 0.74. Notably, no significant correlation was observed between TIDD at diagnosis and partial remission markers (IDAA1C, C-peptide) at 3- and 12-months postdiagnosis time points. In the context of new-onset T1D in pediatric patients, we identified key influencing factors for determining optimal TIDD, including age, percentage of weight loss, weight, veinous pH, and bicarbonates. These findings have paved the way for the development of a dosing algorithm to potentially expedite glycemic control stabilization and facilitate a more individualized approach to treatment regimens.
Read full abstract