Using Hemoglobin A1C ( HgbA1C ) for diagnosing Diabetes

The use of HgbA1C for diagnosis of diabetes changed in 2010, when the American Diabetes Association (ADA) revised criteria for the diagnosis of diabetes mellitus. In addition to the fasting glucose (FPG), oral glucose tolerance test (OGTT), the Hemoglobin A1c > 6.5 (HgbA1C) was accepted as another way to diagnose diabetes

This cutpoint ( 6.5% ) is acknowledged to classify fewer individuals as diabetic than using the standard FPG >125 mg/dL. The FPG >125 better correlates with plasma glucose of 200 or greater. Several studies have suggested increase of diabetic retinopathy when plasma glucose is greater than 200, so a marker that correlates would be helpful.

Hemoglobin A1C is accepted as the best measure of average glucose. HgbA1C has been shown to correlate with likelihood of diabetic retinopathy in a linear fashion, and as expected specificity increases as the cut-point is raised.  A HgbA1C of 6.5% is estimated to have a specificity of 99% for identifying diabetics.

The cut-point of 6.5  was chosen because moderate to severe retinopathy was rarely found below that level, but there were no mathematical models able to identify a clear, objective cut-point. This is a concern, because above 6.0% there is increasing retinopathy in some studies.  In a Japanese study the optimal cut-point or inflection point would have been 5.3-5.5% for the development of retinopathy. There is growing data that retinopathy, microalbuminuria, chronic kidney disease and peripheral neuropathy may have a continuous linear relationship to the HgbA1C which makes choosing a cut-point difficult.

Comparison of the 3 ways of diagnosing diabetes shows that HgbA1c is the least sensitive and may miss 70% of the diabetics. The 2 hour post prandial plasma glucose (OGTT) better identifies diabetics by almost a 3:1 margin compared with HgbA1c and 2:1 margin compared with fasting glucose >125. This suggests that HgbA1C, FPG, and OGTT measure different facets of glycemia.

Lowering the cut point on the HgbA1C might be suggested, but the positive predictive value has been shown to drop from 93% to 24% if the HgbA1C is moved from 6.5% to 5.8%. Suspicion of diabetes may require more workup then just a HgbA1C.


  • A HgbA1C of greater than 6.5% is a specific diagnosis for diabetes, but many diabetics will not be identified using this definition. 
  • It does appear that HgbA1C in this range is a risk factor for development of coronary artery disease (CAD).

It would be nice to have studies of people with a HgbA1C between 5.5 and 6.5 to make a more rational treatment decision.  Evidence suggests that there are some people in this range (5.5-6.5) that are diabetics, and will have neuropathy, retinopathy, chronic kidney disease including macro and microalbuminuria. 

  • Therefore, while a HgbA1C of 6.5 will diagnosis a diabetic, a HgbA1c of 5.5-6.5 does not exclude diabetes. 

    Bryan E Fuhs MD                                              Reviewed 5/18/2011
  1. Implications of Using Hemoglobin A1C for Diagnosing Diabetes Mellitus; Samir Malkani, MD, John Pl Mordes MD The American Journal of Medicine (2011) 124, 395-401
This entry was posted in Diabetes Type II and tagged . Bookmark the permalink.