Updated glucose management indicator (GMI) better aligns with HbA_1c than current GMI: implications for clinical...

The continuous glucose monitoring (CGM)-derived glucose management indicator (GMI) is valuable to people with diabetes, and healthcare professionals and organisations for assessing overall glucose levels, optimising management plans. However, the current GMI can both over- and underestimate HbA_1c, which can create clinical difficulties. Our aim was to improve the agreement between these two markers across clinically relevant ranges of glucose.

An updated GMI (uGMI) model based on physiological processes was evaluated using clinical trial and real-world data. The empirical relationship between average glucose (AG) and HbA_1c was evaluated over 100 equal-sized data bins ordered by the sum of their rank positions. Alignments were assessed using biases in various ranges.

In 18,860 individuals with 26,647 AG-HbA_1c pairs, the uGMI significantly improved alignment with HbA_1c and reduced the proportion of pairs with clinically significant discordance. Specifically, the regression slope of HbA_1c vs GMI decreased from 1.4 to 1.0 when using uGMI, effectively eliminating proportional bias by reducing the deviation from unity (p<0.0001). Furthermore, absolute bias at HbA_1c below 31 mmol/mol (5.5% in National Glycohemoglobin Standardization Program [NGSP]) and above 75 mmol/mol (9.0% NGSP) was reduced from >4.4 mmol/mol (0.4% NGSP) to ≤1.1 mmol/mol (0.1% NGSP). Consistent performance across both Abbott Freestyle Libre and Dexcom CGM devices confirmed that the uGMI is robust and device-independent, supporting its clinical utility and incorporation into standardised CGM clinical summary reports.

Compared with the CGM-derived GMI, the uGMI provides more accurate and consistent agreement with HbA_1c, particularly at lower (<42mmol/mol or 6% NGSP) and higher (>58 mmol/mol or 7.5% NGSP) HbA_1c levels.