LAUSR

BACKGROUND We aimed to observe, up to 2 years, the effectiveness of initiating first-generation automated insulin delivery for type 1 diabetes management. METHOD Retrospective, real-world, observational study using medical records, conducted across five sites in Australia. Adults with type 1 diabetes, who had automated insulin delivery initiated between February 2019 and December 2021, were observed for 6-24 months post-initiation (until June 2022). Outcomes examined included glucose metrics assessed by HbA1c and continuous glucose monitoring (CGM), safety, and therapy continuation. RESULTS Ninety-four adults were studied (median age 39 years [IQR 31-51]; pre-initiation HbA1c 7.8% [7.2-8.6]). After automated insulin delivery initiation, HbA1c decreased by mean 0.5 percentage points (95% CI -0.7, -0.2) at 3 months (P <0.001); CGM time in range 3.9-10.0 mmol/L increased by 11 percentage points (9, 14) at 1 month (P <0.001); these improvements were maintained up to 24 months (all P <0.02). Median CGM time below 3.9 mmol/L was <1.5% pre- and post-automated insulin delivery initiation. The sub-group with pre-initiation HbA1c above 8.5% had the greatest HbA1c improvement (-1.4 percentage points [-1.8, -1.1] at 3 months). Twelve individuals (13%) discontinued automated insulin delivery, predominantly citing difficulties with CGM. During the 150 person-years observed, four diabetes-related emergencies were documented: three severe hypoglycaemic events, one hyperglycaemic event without ketoacidosis. CONCLUSIONS Early glucose improvements were observed after real-world automated insulin delivery initiation, sustained up to 2 years, without excess adverse events. Greatest benefits were observed among individuals with highest glycaemia pre-initiation. Future-generation systems with increased user-friendliness may enhance therapy continuation. This article is protected by copyright. All rights reserved.