#23: The HbA1c Test – Does It Just Reflect Average Blood Glucose?

Everyone with diabetes is familiar with the glycated hemoglobin laboratory test, abbreviated HbA1c, that is used to assess the adequacy of glycemic control and more recently to diagnose prediabetes and diabetes. It has also been correlated with and reported on laboratory reports as an average blood glucose (BG) over the previous 2 to 3 months in an attempt to make it more meaningful to patients. But what you may not know is exactly what it measures and how it may not reflect your average BG.

What HbA1c Measures

HbA1c measures glycated hemoglobin contained in red blood cells (RBC). Glycated hemoglobin is formed by a slow reaction between hemoglobin and glucose. Hemoglobin is a large protein in RBC that carries oxygen from the lungs to the organs and tissues throughout the body. HbA1c, first described in 1969, is formed when glucose is covalently bonded to the N-terminal valine residue of the beta chain of hemoglobin. Because RBC have an average life span of approximately 117 days in men and 106 days in women, the HbA1c reflects the average BG over that time period at least in theory. Of this period, the last 30 days contribute 50% to the HbA1c result. This is because of the distribution of young versus old RBC in circulation. See here.

Other Measures of Glycation, The Glycation Gap, and The High-Low Glycator Hypothesis

In addition to HbA1c, glycated serum proteins (fructosamine) and glycated albumin can also be measured in clinical laboratories. Albumin is one of many different proteins in the serum (blood). These proteins also become glycated (react with glucose) and can be measured. The time span that these proteins spend in the blood is shorter (about 2 to 3 weeks) than that of hemoglobin. Typically, measuring these glycated proteins is done when rapid control of BG is desired e.g. gestational diabetes. However, when simultaneously drawn fructosamine and HbA1c are compared, 23% of subjects had an HbA1c value 1 percentage point higher and 17% had an HbA1c value 1 percentage point lower than the value predicted from fructosamine. This is referred to as a glycation gap between the results of an intracellular (HbA1c) and an extracellular (fructosamine) protein target of glycation or an integrated measure of glycemic control.

The within-subject inconsistency of two precise measures of glycemic control supports the validity of the high glycator–low glycator hypothesis, i.e., physiologic as opposed to technical causes for differences in A1C.

See here.

Factors Other Than Glycemia That Affect HbA1c

Several factors affect the HbA1c results that have nothing to do with average BG. Any medical condition that alters the structure of hemoglobin, lifespan of RBC, or rate of hemoglobin glycation will necessarily alter the HbA1c. Examples include the hemoglobinopathies (sickle cell disease, thalassemia), B12, folate, and iron deficiency anemias and hemolytic anemias, kidney failure (uremia), hypertriglyceridemia as well as drugs that modify HbA1c by a variety of mechanisms. Interestingly, vitamin C supplementation can simultaneously falsely decrease HbA1c and falsely increase BG readings: the higher the dose, the greater the effect. In addition among hematologically normal people, RBC survival varies sufficiently to cause clinically important differences in HbA1c. See here. This study reviewed some of the factors that affect HbA1c including race and ethnicity. There findings suggested that rate of hemoglobin glycation or red cell survival time may differ among racial and ethnic groups.

Previous studies in nondiabetic individuals have shown that A1C levels in the same individual change little over time but that levels vary markedly between individuals. Additional variation in A1C levels between individuals has been shown to be related to factors independent of glycemia such as female sex, sex hormones, differences in visceral fat, and biologic variation in hemoglobin glycation or red cell survival. Recent studies have suggested that interindividual differences in intra-erythrocyte 2,3-diphosphoglycerate, which catalyzes the production of A1C, may in part account for the variability of A1C observed in nondiabetic subjects. Similarly, interindividual variation in intra-erythrocyte fructosamine 3-kinase, which deglycates intracellular fructosamines [i.e. HbA1c], might partially explain nonglucose-mediated interindividual variation in A1C. Evidence from diabetic twin studies have suggested that A1C levels are genetically determined. Interindividual variation in A1C may also be explained by differences in erythrocyte survival. Studies in both type 1 and type 2 diabetes have, for example, demonstrated that hyperglycemia is associated with decreased erythrocyte survival.

Reduced erythrocyte (i.e. RBC) survival translates to a falsely low HbA1c. Also see here and here.

Correlation Between HbA1c And Average Blood Glucose or Lack Thereof

I discussed the Diabetes Control and Complications Trial (DCCT) in blog posts #10 and #12. This paper used the data from DCCT to derive regression lines by plotting HbA1c vs mean BG for both the conventionally treated, intensively treated, and both groups of T1D patients in the trial. The results from their paper are shown in the figure below.

Post 23 DCCT regression lines graph

The authors made note of the fact that the intensively treated T1D patients had lower mean plasma glucose (MPG) concentrations for any given HbA1c value compared to conventionally treated patients. The authors state that

A plausible explanation can be found for this difference if the concept of high and low glycators is assumed. High glycators have consistently higher HbA1c than expected for their MPG, whereas low glycators have lower HbA1c than their MPG would suggest. The proposed reasons for this between-individual variability in hemoglobin glycation rate include differences in erythrocyte survival and other genetic elements.

In the A1c-Derived Average Glucose study, a total of 507 subjects including 80 without diabetes, 268 with T1D, and 159 with T2D were included in the analysis. For each subject in the study 2,500 continuous glucose monitoring (CGM) and 230 finger stick BG measurements were obtained during the 3-month period. The relationship between the HbA1c level and the calculated average glucose (AG) during the preceding 3 months and the linear regression line are shown in the figure below.

Post 23 Translating the A1C Assay Into Estimated Average Glucose Values

You can draw a vertical line at any A1c value to see the range of AG amongst the study subjects. For most values of A1c you might choose, the AG has a range of 75 to 100 mg/dl. Likewise, you can draw a horizontal line at any AG value to see the range of A1c amongst the study subjects. For most values of AG you might choose, the A1c has a range of 2 to 5%. Both of these ranges are quite wide and emphasizes that HbA1c is not simply a measure of AG as many of us probably surmised. This linear regression line shown in the figure above is currently used by some laboratories to report an estimated AG along with the measured HbA1c. This is intended to give meaning to the HbA1c result, but as we now understand other factors influence the HbA1c result in addition to average BG.

This small study measured HbA1c and fasting BG over 20 weeks in twelve normal subjects. They found

that nondiabetic HbA1c values vary markedly between subjects, while values in the same individual change little over time (i.e., the interindividual variation for HbA1c is well in excess of the intraindividual variation). This implies that the nondiabetic reference range for HbA1c measurements is composed of subjects who each have their own narrow “reference range” and that each of these personal reference ranges differ substantially between individuals. If similar interindividual differences also exist in diabetic subjects, then patients with the same glycemic control may vary by at least 1-2% [in HbA1c], which has implications in setting glycated hemoglobin targets. The large interindividual differences between subjects may also help to explain why diabetic patients with apparently similar glycemia do not necessarily have matching HbA1c values.

A comparison of the regression lines from the A1c-Derived Average Glucose study as well as the DCCT study to my personal data from July 2007 to May 2015 is shown in the graph below.

Post 23 Mean BG vs HbA1c Graph

The fact that my HbA1c does correlate very well with my average BG (red circles and line above) with an R squared of 0.9076 means that there is very little intraindividual variation in agreement with the conclusion of the study above. In the graph above for example, an HbA1c of 5.5% would translate to an average glucose of 151, 130, 111, and 95 mg/dl in the DCCT conventionally treated patients, DCCT intensively treated patients, A1c-Derived Average Glucose study patients, and me, respectively. Thus according to the high vs low glycator hypothesis, I am a high glycator. Individuals with diabetes could identify themselves as high, medium, or low glycators were they to plot their average BG values over the previous 60 to 90 days against their HbA1c results.

Using HbA1c As A Glycemic Target For Persons With Diabetes

The interindividual variation in HbA1c makes having a specific HbA1c target for glycemic control in persons with diabetes potentially hazardous. For example, a low glycator might be deemed to be in good glycemic control when in fact they are not thus losing the opportunity to take corrective action. Conversely, a high glycator might be deemed to have inadequate glycemic control and corrective actions to lower BG could result in more frequent or dangerous hypoglycemia. In other words, using HbA1c alone to make adjustments to diabetes management has potential hazards. Using actual BG readings to adjust diabetes therapy is more appropriate in my opinion.

Using HbA1c To Diagnose Prediabetes or Diabetes

Finally, using HbA1c alone to diagnose diabetes has the potential to either diagnose diabetes in high glycators who do not have diabetes or to miss the diagnosis in low glycators. Since fasting BG can be normal on occasions in persons with prediabetes or T2D, it is not a perfect test either. The third test used to diagnose prediabetes or diabetes is the glucose tolerance test (GTT). However, it takes more time and expense to complete compared to either HbA1c or fasting BG. The HbA1c does not need to be drawn in a fasting state making it the most convenient test for patients. In other words, each test has its advantages and disadvantages, one is not necessarily better than another. These tests were reviewed in the previous blog post #22.

Using the current HbA1c cutoff of 5.7 – 6.4% to diagnose prediabetes or >= 6.5% to diagnose diabetes represents in my opinion a reasonable compromise between making a correct diagnosis in the majority versus missing the diagnosis in low glycators (false negative test result) and making the diagnosis in high glycators who do not have prediabetes or diabetes (false positive test result). In persons without signs or symptoms of diabetes, an abnormal HbA1c, fasting BG, or GTT result should be repeated and confirmed in a few weeks before making a diagnosis. In my opinion at the current HbA1c cutoff of >= 6.5% to diagnose diabetes, there would be only a small percentage of false positives. Using the 5.7 – 6.4% HbA1c cutoff for prediabetes would result in more false positives. But as pointed out in my previous blog post #22 since the initial treatment for prediabetes and type 2 diabetes should be lifestyle changes including a low carbohydrate diet, exercise, excess body fat loss, alcohol and tobacco cessation, adequate sleep and sunshine, a false positive test would yield health benefits, not harm.


For persons with diabetes, the HbA1c test is useful in comparing one’s own results over time i.e. if your HbA1c was 8% three months ago, but is 7% now, then your glycemic control has likely improved. However, an HbA1c of 8% in one person verses 7% in another person, does not necessarily mean that the average BG is higher in the former than in the latter. For those with diabetes (especially T1D) who are measuring BG at least 4 times per day, monitoring HbA1c would not add much information to the average BG results in my opinion. For those with diabetes who do not measure BG consistently, HbA1c measurements would be useful to determine whether glycemic control is improving or deteriorating over time. This later scenario is not recommended since the value of measuring BG is to be able to take an action (adjust insulin, medication, food, or exercise) based on the results. If BG is not measured regularly, no corrective action can be taken and will likely result in less than optimal glycemic control. And as I have pointed out in a blog post #10 and on my monthly T1D management updates, monitoring and striving to minimize BG variability measures including standard deviation, coefficient of variation, interquartile range, mean BG Δ per hour, and mean daily BG range are also useful. HbA1c will not reveal any information about BG variability.

I hope this post will give those with diabetes a better understanding of what HbA1c measures and how it may or may not be an accurate reflection of average BG due to the numerous other factors that influence its results.

Till next time …



  1. Svet Pavlovsky

    That is very interesting to know that the last 30 days of BG has major effect on the HbA1c results. As you mentioned before, I also agree that the daily BG control is more important. Since you can have lots of low BG and lots of high BG and get HbA1c as someone who’s BG pretty much stable. I did not understand though what high glycator means. Thanks, Svet


  2. Pingback: Ask the Doc: How Fasting Affects Females Differently, Causes and Treatments for PMS, Why HbA1C is an ‘Unreliable’ Marker, and More | Endurance Planet