What Makes the Difference? Predicting Glycemic Trajectories Following Initiation of Automated Insulin Delivery...

To identify glycemic trajectories following automated insulin delivery (AID) initiation in youth with type 1 diabetes and associated factors.

We used group-based trajectory modeling to characterize groups for time in range (TIR) and hemoglobin A1c (HbA1c), measured at baseline and at 3, 6, 12, and 18 months of AID use, in 713 youth (ages 6-18). Multinomial logistic regression estimated associations between clinical and sociodemographic predictors and glycemic trajectories.

Four TIR and three HbA1c trajectories emerged. All trajectories showed rapid improvement from baseline to 3 months, continued but gradual improvement between 3 and 6 months, followed by waning or sustained improvements after 6 months. Only a small subset (group 4, 12.7%) achieved TIR ≥70%, although more (43.3%) reached an HbA1c <7%. In the TIR model after adjusting for age, sex, type 1 diabetes duration, and prior treatment, participants with fewer user-initiated boluses (odds ratio [OR] 2.46; 95% CI 1.78, 3.39), higher insulin doses/kg (OR 1.54; 95% CI 1.25, 1.91), and fewer appointments (OR 2.69; 95% CI 1.21, 6.00) had higher odds of membership in the lowest TIR group (group 1) than in the highest TIR group (group 4). In the HbA1c model, lower baseline insulin doses/kg and more boluses were associated with membership in the most optimal group.

Distinct glycemic trajectories exist following AID initiation, with all groups experiencing rapid early improvements in TIR and HbA1c that plateau or decline over time. Baseline bolus frequency and insulin doses/kg remain key predictors, highlighting the persistent influence of preexisting behaviors and insulin needs on long-term outcomes.