Heterogeneity in the clinical course of type 2 diabetes in youth presenting with and without diabetic ketoacidosis (DKA) at diagnosis is not well described. We aimed to characterize if presentation of type 2 diabetes with DKA affects rates of insulin discontinuation compared to type 2 diabetes without ketosis (non-DKA).

This single-center retrospective cohort study included patients (BMI Z-score 2.4, IQR 2.07,2.65) with mean age 15 years (IQR 13,16), hospitalized for new-onset diabetes with DKA (n= 79) or non-DKA (n=356) at an academic pediatric institution from Jan 2019-Dec 2021 with follow-up until May 2023. All patients were initiated on insulin and titrated per their care team. Type 1 diabetes was excluded by presence of autoimmune antibodies. Primary outcomes were time to insulin discontinuation of insulin therapy and maintenance of discontinuation.

Time to insulin discontinuation and proportion who discontinued did not differ between DKA and non-DKA groups (48.1 vs 44.7% respectively, p=0.58). Trajectory analyses combining DKA and non-DKA groups identified 3 insulin discontinuation groups: early (20.1% within 2 months), late (12.5% within 6 months), and Never (67.4%). Multinomial regression shows that DKA at presentation is not associated with early (p=0.48) or late insulin discontinuation (p=0.70) compared to never discontinuation. Insulin was restarted in 37 participants with median of 20 months (IQR 16,24) after insulin discontinuation.

Despite not showing differences in insulin discontinuation in youth with new-onset type 2 diabetes with and without DKA at presentation, we identified heterogeneity in duration of insulin treatment in the combined group.