Retrospective analysis of impact of automated insulin delivery technology on acute care utilisation from 2019 to 2021

Abstract

Aims: The aim was to evaluate the impact of using an automated insulin delivery (AID) system versus multiple daily insulin injections (MDI) on the odds and costs of acute care utilisation among people with type 1 diabetes (PwT1D). Methods: This was retrospective, observational cohort study that leveraged electronic health record (EHR) data and administrative claims data from 15 January 2019, through 31 December 2021. PwT1D treated with MDI or newly starting an AID system were identified between 15 January 2020, and 31 December 2020 (identification period), with index date defined as the earliest evidence of an AID device or MDI therapy during the identification period. Participants were required to be continuously enrolled in their health plan for ≥12 months pre- and post-index date. Utilisation and costs of inpatient (IP) and emergency department (ED) services data from medical and pharmacy claims data were analysed. Multivariable modelling was used to assess odds and costs of IP and ED service utilisation among AID system users compared to MDI users while adjusting for pre-index characteristics. Results: This study included 7264 AID system users and 27,973 MDI users. AID system users were younger and had a higher proportion of females than MDI users. After adjusting for pre-index characteristics, AID users had lower odds of IP admission (point estimate [95% confidence interval]: 0.70 (0.63-0.77) or ED visit 0.76 (0.70-0.82)) at 12 months and incurred lower average 12-month costs for all-cause IP admission ($3054 vs. $4701; p-value < 0.001) and ED visit ($1230 vs. $1793; p-value <0.001) compared to MDI users. Conclusions: Use of AID systems among PwT1D was associated with lower odds of IP and ED utilisation and lower costs in the follow-up period compared to MDI.