ABSTRACT
Background and Aims: In type 1 diabetes (T1DM), despite technological advances, results on glycaemic control or quality of life remain poor. Improvements could be achieved with system that automatically calculates doses and controls the pump such as DBLG1-system which is evaluated in our study. Methods: 140 T1D patients treated with insulin pumps and poorly balanced (HbA1c >8%) were recruited in early 2020 in 9 French diabetes centres. They were randomised to receive three months of automated insulin delivery by the DBLG1 system and a Dana pump (CL) or to continue their usual treatment (OL), with a randomisation ratio of 4 DBLG1/1 conventional treatment. The COVID crisis prematurely terminated the study. We present the results of the 99 patients treated for at least 14 days with CGM data for at least 75% of the time (88 CL/11 OL). Results: After adjustment for initial HBA1c and centres, patients treated with CL spent on average 12 min less per 24h in hypoglycaemia <70 mg/dl than patients treated with OL (p < 0.01). They spent 13.6% more time in the 70-180 target range (p < 0.001) per 24 hours, i.e. almost 3.5 hours, and 3 hours less in hyperglycaemia >180 mg/dl (p < 0.0007). The benefit was mainly nocturnal. There was a linear relationship between the time in hypo- or normoglycaemic targets at baseline and the amount of improvement achieved by the system (p < 0.001). Conclusions: In T1DM patients poorly controlled by insulin pumps, the DBLG1 system significantly improves the time spent in hypo- and normoglycaemia. The greater the initial imbalance, the greater the improvement.