Related Experiment Video
Updated: Jun 20, 2026
Improving IV Insulin Administration in a Community Hospital
Published on: June 11, 2012
Comparative Analysis of Capillary HbA1c and CGM-Derived GMI Estimates: A Real-World Data from a Single Center
Adi Dagan1, Michal Yackobovitch-Gavan1,2, Moshe Phillip1,2
1Gray Faculty of Medical and Health Sciences, Tel Aviv University, Tel Aviv, Israel.
Introduction And Objective:
Understanding the relationship between hemoglobin A1c (HbA1c) and continuous glucose monitoring (CGM)-derived Glucose Management Indicator (GMI) is essential for accurate glycemic assessment. This study evaluates the correlation between paired capillary HbA1c and GMI across different assessment periods and identifies factors associated with significant discrepancies.
Methods:
This retrospective observational study included 270 youth with type 1 diabetes (mean age 15.85 ± 5.99 years, 56% male, median HbA1c 7.4% (interquartile range: 6.7-8.1). Capillary HbA1c and CGM data obtained during the same clinical visit were analyzed to assess discrepancies between measured HbA1c and GMI values, calculated using 14, 30, and 90 days of CGM data. For each individual, two paired measurements were collected over time.
Results:
Across paired measurements, 37.3% of individuals demonstrated a HbA1c-GMI discrepancy greater than 0.5%, while 11.8% exceeding a discrepancy of 1%. Discrepancies showed no significant correlation with age, sex, body mass index, and disease duration. Glucose-6-phosphate dehydrogenase deficiency, present in 1.9% of the cohort, was consistently correlated with discrepancies >1% in all measurements. A significant proportional bias was identified (β = 0.260, P < 0.001), with HbA1c underestimating GMI at lower values and overestimating at higher values. Among CGM metrics, mean glucose, GMI, and time in normal glucose (TING) (70-140 mg/dL) showed similar associations with HbA1c, whereas TIR showed a slightly lower magnitude of association.
Conclusion:
A significant number of participants exhibited discrepancies between HbA1c and GMI. This discrepancy was unrelated to CGM sampling duration and stable within individuals, suggesting intrinsic factors may contribute. HbA1c and GMI each reflect different aspects of glycemic control, and neither is sufficient alone. Combining HbA1c with CGM-derived metrics, such as TING, may provide a more accurate and comprehensive assessment of glycemic exposure.
