Table 2

Reasons not to prefer A1C compared with plasma glucose determination for diagnosing diabetes

Diabetes is clinically defined by high blood glucose and not by glycation of proteins.
A1C is a poor marker of important pathophysiological abnormalities featuring diabetes.
A1C has a poor sensitivity in diabetes diagnosis and would change the epidemiology of diabetes.
2-h glucose level and IGT are stronger predictors of CVD than A1C.
Fasting is not essential to identify perturbation in glucose metabolism.
Standardization of A1C assay is poor, even in Western countries, and standardization of glucose assay would be easier to implement.
In many subjects, A1C assay is unreliable and cannot be used.
A1C has significant differences in various ethnic groups, which are poorly understood and characterized.
Within-days biological variability of plasma glucose might unveil disturbance of glucose metabolism.
Individual susceptibility to glycation of hemoglobin is not relevant to diabetes diagnosis.
Using the same biomarker for diagnosing and monitoring diabetes might have negative effects.
Cost of the assay: glucose is unquestionably cheaper than A1C, and A1C assay is not available on a large scale in most of the countries.
A1C levels vary not only according to glycemia, but also to erythrocyte turnover rates (e.g., hemoglobinopathies, malaria, anemia, blood loss) as well as other factors.
Correlation between A1C and FPG is ~0.85%, which means that as many as 30% of the variation in FPG is not explained by A1C and vice versa.
Nothing is known about changes in A1C during the development of diabetes.
A1C levels of 6.0–6.5% do not predict diabetes as effectively as FPG and 2-h PG (OGTT).
Sensitivity of A1C to detect diabetes defined by the OGTT is <50%; thus, the majority of diabetic individuals will remain undiagnosed if A1C is used.
The levels of A1C predicting future retinopathy, nephropathy, etc., in the population is not well established (<6.5%?).
No diabetes prevention trials have selected their populations based on A1C.
Using A1C will delay the diagnosis of diabetes in ~60% of incident cases.