P-130: The relationship between CGM metrics and HbA1c using different sensors in children with type 1 diabetes
R Hanas, P Adolfsson, A -L Brorsson, K Eeg-Olofsson, O Eriksson, M Fredriksson, A -L Fureman, E Jelleryd, A Pundziute Lyckå, A Rydin, A Sjöstrand, K ÅkessonAbstract
Introduction
HbA1c has been the gold standard for glycemic outcomes. Increasing emphasis is now placed on continuous glucose monitoring (CGM). CGM data over several weeks can be converted to the Glucose Management Indicator (GMI), which estimates the HbA1c expected if glucose control remains stable over 2–3 months.
Objectives
To investigate the relationship between HbA1c and sensor metrics with different CGM systems.
Methods
Data from the most recent visit in the Swedish National Diabetes Register (extracted up to 2023, < age 18) were analyzed. Mean CGM glucose over the preceding 14 days was used to calculate GMI. Analyses were performed separately for 3 sensor systems: Dexcom G6 (D), FreeStyle Libre 3 (F), and Guardian 4 (G).
Results
5,789 individuals (85.7% used pumps, 14.3% multiple injections) with type 1 diabetes were included. Mean HbA1c was 50.4 mmol/mol (6.8%), 50.1 for D, 51.6 F, 51.5 G. Mean glucose was 8.5 mmol/l (154 mg/dl), GMI 52.9 mmol/mol (7.0%). HbA1c 48 mmol/mol (6.5%) represented a mean glucose of 8.4 (D), 8.4 (F) and 7.9 mmol/l (G), and TITR (Time in Tight Range, 3.9-7.8 mmol/l, 70-140 mg/dl) of 50.3 (D), 51.3 (F) and 57.6% (G). Fig 1 shows Bland-Altman plots.
Conclusions
In automated insulin delivery (AID) systems, insulin dosing is guided by sensor glucose values. Therefore, systematic differences between sensors may lead to variations in HbA1c despite similar mean glucose levels. In this study, there was a negative overall bias for F and D sensors (higher GMI compared to HbA1c) leading to a relative increase in insulin delivery in an AID pump. The bias was rather similar over the span of HbA1c values for F, while D and to an even greater extent G sensors had a negative bias at lower HbA1c levels and a positive at higher. This would result in lower mean glucose at higher HbA1c levels, leading to a relative decrease in AID insulin delivery. These biases should be considered when interpreting CGM metrics and advising patients on AID insulin pump use.