Evaluation of HbA1c and glucose management indicator discordance in a population of children and adolescents with type 1 diabetes

Abstract

Background Glucose management indicator (GMI) is a useful metric for the clinical management of diabetic patients using continuous glucose monitoring (CGM). In adults, a marked discordance between HbA1c and GMI has been reported. To date, no studies have evaluated this discordance in children/adolescents with type 1 diabetes (T1D). Methods HbA1c and real‐life CGM data of the 12 weeks preceding HbA1c measurement were collected from 805 children/adolescents. The absolute difference between HbA1c and GMI was calculated for both the 12‐week and 4‐week periods preceding HbA1c measurement and the proportion of discordant patients was defined according to specific thresholds in the entire study population and in subjects stratified by type of CGM, insulin therapy, gender, age and puberty. Regression analyses were performed with HbA1c‐GMI discordance as dependent variable and patients' characteristics as independent ones. A new GMI equation for children and adolescent was derived from the linear regression analysis between mean glucose and HbA1c. Results HbA1c‐GMI discordance calculated on the 12‐week period was <0.1, ≥0.5 and ≥1.0 in 24.8, 33.9 and 9.2% of the subjects, respectively. No significant differences in the proportion of discordant patients were found comparing patients stratified by type of CGM, insulin therapy, gender, age and puberty. GMI‐HbA1c discordance was not significantly explained by age, gender, BMI, type of CGM, insulin therapy, hemoglobin, anemia and autoimmune diseases (R2 = 0.012, p = 0.409). HbA1c‐GMI discordance calculated on the 4‐week period was comparable. GMI (%) equation derived for this cohort was: 3.74 + 0.022x (mean glucose in mg/dl). Conclusions GMI could be meaningfully discordant respect to HbA1c in more than a third of children/adolescents with T1D. This discrepancy should be taken into careful consideration when the two indices are directly compared in daily clinical practice.

Continuous glucose monitoring metrics (Mean Glucose, time above range and time in range) are superior to glycated haemoglobin for assessment of therapeutic efficacy

“…When a laboratory HbA1c value is known, it becomes possible to calculate an estimated value for average glucose, using a weighted least‐squares regression of Mean Glucose on HbA1c 18 . Many studies have observed that these two variables are linearly related 3,4,17‐20 . Likewise, one can obtain an estimate of HbA1c from a measured Mean Glucose value, using an unweighted least squares linear regression of observed HbA1c on Mean Glucose 19,20 .…”

“…The principal method used in the present study was to compare Spearman correlation coefficients among HbA1c , Mean Glucose , %TIR , and %TAR for T1D, 14,16 and Pearson correlation coefficients among Mean Glucose , %TIR , and %TAR for T2D 15 . We did not examine correlations with proportion of time below range, which is very weakly correlated with the other four CGM metrics 4 . Our goal was to identify the metrics most closely correlated with HbA1c and Mean Glucose , based on the premise that metrics with the highest correlation with Mean Glucose would be associated with the long‐term complications of diabetes 8 .…”

“…18 Many studies have observed that these two variables are linearly related. 3,4,[17][18][19][20] Likewise, one can obtain an estimate of HbA1c from a measured Mean Glucose value, using an unweighted least squares linear regression of observed HbA1c on Mean Glucose. 19,20 This estimate has been designated the "glucose management indicator" (GMI) to emphasize that it is not a laboratory measurement of HbA1c.…”

“…There is a linear relationship between %TIR and Mean Glucose. 14,21 The strength of this relationship has led many clinical diabetologists to regard %TIR as nearly equivalent to HbA1c for use in managing individual patients, 3,4 and as a basis for a primary endpoint in clinical research studies. 5 %TIR has a higher correlation with Mean Glucose than with HbA1c, implying that Mean Glucose may be more informative than HbA1c.…”

Monthly video consultation for children and adolescents with type 1 diabetes mellitus during the COVID-19 pandemic

“…Furthermore, the quarterly HbA1c is not always adequately reflected by software GMI, mean sensor glucose, TiR, and TbR, and so cannot be simply replaced by software measures in all patients [36] . For this reason, from our perspective, before starting primarily video consultation-based care, it is advisable to compare an HbA1c value from the outpatient clinic with the corresponding GMI from a 4-week evaluation and check whether the GMI deviates clinically significantly, i.e., more than 0.5%, in case of incorrectly low calibrated sensors, for example.…”