Response to Comment on Nathan et al. Relationship Between Average Glucose Levels and HbA_1c Differs Across Racial Groups: A Substudy of the GRADE Randomized Trial. Diabetes Care 2024;47:2155–2163

We appreciate Dr. Selvin’s thoughtful response (1) to our article recently published in Diabetes Care, in which we conclusively established racial differences in the translation of HbA_1c values into average glucose levels (2). We recognized the complexity of race as a construct in our discussion and deliberated whether we should write anything regarding the practical consequences of our findings with regard to treatment and diagnostic guidelines. Ultimately, we decided that it was necessary to point out the potential for overtreatment, resulting in increased risk for hypoglycemia, or overdiagnosis, if race was not considered.

While the widespread concern that followed the decision to incorporate race into the CKD-EPI equations and other guidelines (3) was not lost on us, we decided to include the statement that race “should be considered” in setting treatment goals. While the complexity of race is clear, it was the self-identification of race in The Glycemia Reduction Approaches in Diabetes: A Comparative Effectiveness Study (GRADE) that was used in our analyses and was associated with differences in the relationship between HbA_1c and average glucose levels. As Dr. Selvin suggests, there are likely genetic or other biological differences between racial groups, perhaps owing to differences in red cell turnover or glycation itself, that underlie these racial differences. She states that race, as a social construct, is a flawed surrogate for these other biological variables (e.g., genetics) and, therefore, should not be accounted for in therapeutic decisions. However, we concluded that these racial classifications, while imperfect, still provided information that would improve the quality of therapeutic decisions. Furthermore, while there are other patient characteristics, including interindividual variability, that contribute to the differences in the relationship between average glucose and HbA_1c, self-identified race provides one readily available means of improving the fidelity of the relationship. Finally, the use of precision medicine based on genetics for most people with type 2 diabetes remains largely aspirational (4), and establishing an individual correction of HbA_1c levels by performing continuous glucose monitoring in all patients would be onerous and impractical.

Waiting until individual genetic or other biological differences are understood and are widely available puts Black patients at heightened risk for hypoglycemia, as was seen in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study (5). We thought that safety demanded consideration of race-specific adjustments in HbA_1c-based treatment targets and did not see how implementing such race-based targets that might prevent overtreatment and hypoglycemia marginalizes self-identified black populations.

Primum non nocere was, and remains, our guiding principle, but in retrospect and considering the absence of more complete data that usually guide the formulation of policy, we should have substituted “could be considered” for “should be considered” in setting treatment goals. A more cautious look by organizations, such as the American Diabetes Association, that set guidelines should consider whether, on balance, using race-guided HbA_1c treatment goals is advisable.