Evaluation of a portable ambulatory prototype for automated overnight closed-loop insulin delivery in young people with type 1 diabetes

Objective To evaluate an ambulatory, portable prototype, overnight automated closed-loop (CL) system and to explore optimal time of CL initiation. Methods We performed a randomized crossover study and compared automated overnight glucose control started at the time of an evening-meal or at bedtime. Eight young people with type 1 diabetes (T1D) on insulin pump therapy [M?=?4; age?=?14.3 (1.7)?yr; HbA1c = 8.2 (1.3)%; mean (SD)] were studied on two occasions at clinical research facility. A standardized self-selected evening meal [70 (11)g CHO] and snack [22 (4)g CHO] accompanied by prandial insulin boluses were given at 18:00 and 21:00?hours, respectively. In random order, automated CL was started at 18:00 or 21:00?hours and ran until 8:00?hours the next day. Basal insulin delivery was automatically adjusted by a model predictive control algorithm based on real-time continuous glucose monitor readings. Results Overnight plasma glucose levels (between 21:00 and 08:00?hours) were within the target range (71145?mg/dL) for 82 (59, 98)% of time when CL started at 18:00?hours and 64 (48, 70)% when CL started at 21:00?hours [median (IQR), p?=?0.036]. Time spent above 180?mg/dL [8 (0, 17) vs. 13 (3, 26)%, p?=?0.310] or below 70?mg/dL [0 (0,7) vs. 0 (0, 8)%, p?=?1.000] did not differ between the two occasions. Mean overnight glucose [121 (14) vs. 137 (13)?mg/dL, p?=?0.731) was also similar. Overnight insulin infusion rates were comparable [0.8 (0.5, 1.3) vs. 0.8 (0.6, 1.4)?U/h, p?=?0.263]. No interruptions to CL delivery were observed. Conclusion Automated CL delivery can be applied reliably and safely to control glucose levels overnight in young people with T1D. Tighter glucose levels may be achieved with an earlier time of CL initiation.