Systematic Discordance Between HbA1c and Continuous Glucose Monitoring-Derived Glycaemic Metrics in Paediatric Diabetes: Implications for Treatment Decisions

Background: HbA1c guides paediatric diabetes treatment, but biological factors like erythrocyte lifespan cause differences between HbA1c and CGM measures, risking overtreatment, undertreatment, and misclassification. This study quantifies HbA1c–GMI discordance in children, exploring clinical and biological predictors, including residual beta cell function via C-peptide. Methods: In this prospective cross-sectional study at a paediatric diabetes centre in Dhaka, Bangladesh, 120 youths aged 5–25 years with type 1 or type 2 diabetes (duration ≥ 6 months) were enrolled between March 2024 and November 2025. Participants underwent 7–14 days of blinded CGM with the Abbott FreeStyle Libre Pro iQ system, and concurrent laboratory HbA1c; fasting C-peptide was measured in a subset, and demographic, clinical, and treatment data were abstracted from medical records. The primary outcome was HbA1c–GMI discordance, defined “a priori” as an absolute difference greater than 0.5 percentage points; agreement was assessed with Bland–Altman analysis, and multivariable linear regression examined associations between discordance magnitude and diabetes duration, C-peptide, total daily insulin dose, and diabetes type. Results: Out of 120 enrolled patients, 97 with concurrent HbA1c and sufficient CGM data were included in the analysis (mean age 14.8 years; 57.7% female; 70.1% with type 1 diabetes). HbA1c and GMI showed only a modest correlation, and a significant proportion of participants met the predefined discordance threshold. Bland–Altman plots revealed a systematic positive bias, with HbA1c generally exceeding GMI. The mean cohort HbA1c was notably higher than the mean GMI, suggesting that many youths would be classified as poorly controlled and considered for intensification based on HbA1c alone, despite CGM indicating closer-to-target glycaemia; lower C-peptide levels were independently associated with greater discordance magnitude after adjusting for clinical covariates. Conclusions: In this paediatric diabetes cohort, HbA1c often overestimated glycaemic burden compared to CGM-derived GMI, especially among young people with reduced residual beta cell function. This challenges the assumption that these metrics are interchangeable. Using CGM measures alongside HbA1c and considering discordance patterns—possibly guided by C-peptide—can prevent unnecessary treatment increases and promote more personalised, physiology-informed management.