Long-Term Performance of Two Systems for Automated Insulin Delivery in Adults With Type 1 Diabetes: An Observational Study

Aims To compare glycaemic outcomes for two automated insulin delivery (AID) systems, the Tandem Control IQ (CIQ) and the MiniMed 780G (MM780G). Material and Methods In this observational study, we evaluated 60 days of glycaemic data from 139 persons with type 1 diabetes (CIQ: 79 persons, MM780G: 60 persons), who had an active glucose sensor time >= 85%. Results The time with AID was median 620 (IQR, 439-755) days for CIQ users and 509 (429-744) days for MM780G users (p = 0.26). The last HbA1c before initiation of AID was 59.7 mmol/mol in CIQ and 60.1 mmol/mol in MM780G (p = 0.88). The time with an active glucose sensor was higher for CIQ than MM780G (median 98.5 (97.4-98.0)% vs. 96.5 (94.9-97.0)%, p < 0.001). Time in range (TIR, glucose 3.9-10.0 mmol/L) was lower in CIQ than MM780G (mean 68.9% +/- 11.4% vs. 73.7% +/- 12.0%, p = 0.02) as was time in tight range (TITR) (glucose 3.9-7.8 mmol/L) (43.0% +/- 12.2% vs. 48.4% +/- 12.7%, p = 0.01). The difference in TIR (4.2 (95% CI 1.0-7.5)%, p = 0.01) and TITR (5.0 (1.4-8.6)%, p < 0.01) remained statistically significant in a multiple regression model controlling for various baseline variables. Time with an absolute rate of glucose change > 1.5 mmol/L/15 min was higher in CIQ than MM780G (9.4 (IQR, 7.2-13.3)% vs. 7.4 (5.2-10.4)%, p < 0.001). Conclusions The CIQ system had a higher active glucose sensor time but a lower TIR, TITR, and a higher time with a rapid glucose rate of change than the MM780G system.