The interdependence of targets for continuous glucose monitoring outcomes in type 1 diabetes with automated insulin delivery
Javier Castañeda,Bastiaan E. de Galan,Sander M. J. van Kuijk,Arcelia Arrieta,Tim van den Heuvel,Ohad Cohen
DOI: https://doi.org/10.1111/dom.15955
2024-09-28
Diabetes Obesity and Metabolism
Abstract:Aim The aim was to determine the interdependence of targets for glucose management indicator (GMI), time within the ranges of 70–180 mg/dL (TIR) and 70–140 mg/dL (time in tight glucose range [TITR]), time above 180 mg/dL (TA180) and 250 mg/dL (TA250) and time below 70 mg/dL (TB70) and 54 mg/dL (TB54) and its implications for setting targets in automated insulin delivery (AID). Materials and Methods Real‐world data from individuals with type 1 diabetes using the 780G system were used to calculate the receiver operating characteristic curves and establish interdependent targets for time in ranges based on several GMI benchmarks. Correlation, regression and principal component analysis were used to determine their association and dimensionality. Results In individuals aged >15 years (n = 41 692), a GMI 81%, >58%, <15% and 90%), whereas these values were poor for time in hypoglycaemia and GMI, which had a modest correlation (−0.21 to −0.43). Two dimensions emerged: (1) GMI, TIR, TITR, TA180 and TA250, and (2) TB70 and TB54, explaining 95% of total variability. GMI (or TIR) and TB70 explained >81% of the variability in the remaining continuous glucose monitoring (CGM) metrics, providing accurate predictions. Individuals aged ≤15 years (n = 14 459) showed similar results. Conclusion We developed a methodology to establish interdependent CGM targets for therapies with CGM data outputs. In AID with the 780G system, a GMI <7% requires time in ranges close to consensus targets. Targets for GMI, TIR, TITR, TA180 and TA250 could be reduced to targets for GMI or TIR, whereas targets for time in hypoglycaemia are not inherently tied to GMI/TIR targets.
endocrinology & metabolism