Letter: Response to: “Lack of Association Between Hemoglobin A1c and Continuous Glucose Monitor Metrics Among Individuals with Prediabetes and Normoglycemia”

To the Editor,

Rodriguez and colleagues ¹ address an important question: whether standard continuous glucose monitoring (CGM) metrics correspond to hemoglobin A1c (HbA1c) across the full glycemic spectrum, including populations without diabetes. Their observation that CGM metrics correlate strongly with HbA1c among individuals with type 2 diabetes but far less so among those with prediabetes or normoglycemia is valuable for both clinicians and researchers interpreting CGM-derived data. However, the authors’ conclusion related to limited interpretability of CGM outside of population with diabetes warrants reconsideration.

HbA1c reflects cumulative glycemic exposure but is also influenced by multiple non-glycemic physiological factors, including interindividual variation in erythrocyte lifespan, differences in glycation rates, and red blood cell turnover. ² These sources of biological variability can lead to substantial discordance between observed glucose profiles and measured HbA1c. As Beck et al. ³ demonstrated, individuals with similar mean glucose levels may exhibit markedly different HbA1c values, raising legitimate concerns about the precision of HbA1c as a marker of true glycemic exposure, especially in individuals with near-normal glucose levels.

A second methodological concern relates to the short CGM sampling window used in the study. Estimating long-term glycemic exposure from only 7 days of CGM data—combined with a minimum acceptable CGM coverage of merely 70%—is highly restrictive. Recent evidence demonstrates that short-term CGM sampling introduces considerable bias when used to approximate longer-term glycemic control. ⁴ Furthermore, our analysis of more than 97,000 individuals showed that days with lower sensor coverage are not missing at random and are systematically associated with lower time-in-range. ⁵ Such coverage-dependent bias may disproportionately distort CGM-derived estimates when monitoring periods are very short.

Taken together, these considerations suggest that the observed discordance between CGM metrics and HbA1c in individuals without diabetes may reflect limitations of HbA1c and short-term sampling error rather than limited interpretability of CGM itself. Rather than concluding that CGM has limited relevance in populations without diabetes, it may be more appropriate to ask whether HbA1c is sufficiently precise to detect early, subtle dysglycemia.