December 20, 2006

The Formulas equating HbA1c to Average Glucose Level Don't Work with Near Normal Blood sugars?

NOTE: Important New Information Added to this post in OCT 2007. Please scroll to bottom to read!

I got my new A1c result yesterday, the first in six months.

It was 5.5%.

During this period, my blood sugars have been significantly better than they had been for years. My fasting blood sugars have dropped about 20 mg/dl and my post-meal values have dropped by about 30 mg/dl. These changes have been measured many times using reference meals with known blood sugar outcomes.

In the past, my A1cs were almost always 5.7%.

The drop in A1c doesn't seem to capture the significant lowering of my blood sugars over this period.

The usual formula to estimate the relationship between A1c and mean plasma glucose was derived from the DCCT study.

That formula is:

Mean Blood Glucose = (A1c * 35.6) - 77.3

Applying this formula, an A1c of 5.5% is supposed to correspond to a mean plasma glucose of 118.5 mg/dl and an A1c of 5.7% is supposed to correspond to a mean plasma glucose of 125.6 mg/dl.

Neither of these values correspond to anything I have ever seen in my testing, and I test a lot and at many different times of day. My 30 day meter average, based on 150 measurements, has ranged between 98 and 103 during this period.

There is another, less cited formula that works better, at least for me. It is called "The Nathan Formula" it is:

Mean Plasma Glucose = (A1c * 33.3) -86

This formula yields a mean plasma glucose value of 97 mg/dl, which comes much closer to what my meter's 30 day average has looked like throughout this period.

Here's a calculator that will give you the Nathan Formula A1c/Mean Plasma Glucose equivalents:
A1c Calculator

However, I think the most important point is this: despite doctors' reliance on A1c it is not a particularly accurate measurement of what your blood sugars have been over the past three months. And not only that, the DCCT formula seem to work a lot better in people with very high blood sugars which is the group from which it was derived than they do those of us with near normal ones.

Here's a study published in the Journal of Family Practice that looked at the evidence connecting observed glucose measurements and A1c.

Does daily monitoring of blood glucose predict hemoglobin A1c levels?

The conclusions of this review include the following statements:

The relationship between HbA1c and blood glucose levels is such that blood glucose levels from the preceding 30 days determine about 50% of the total HbA1c. (10) This relationship may be altered by uremia, intake of vitamins C or E, and conditions that affect erythrocyte turnover. (11)

It remains unclear whether management strategies that focus on minimizing HbA1c levels are optimal for prevention of diabetic complications.

Although HbA1c levels correlate with the risk of some complications, aspects of glycemia not reflected in the HbA1c level, such as the heights of glycemic "excursions" from the mean, may independently affect the risk of complications of diabetes. (12) If so, quantitative analysis of day-to-day blood glucose levels might yield a better estimation of the risk of diabetic complications than HbA1c levels.

So what this means is this. The improvement I've made in my blood sugar by keeping my "excursions" (i.e. post meal values) around 110 instead of 140, is probably a lot more significant, healthwise, than the measily .2% improvement in A1c.

Keep this in mind if your A1c results don't match your observed daily testing results, and trust what you see on your meter not a cheering (or baffling) A1c.

UPDATE OCT 31, 2007

The American Association of Diabetes Glucose Trials has come up with a new and improved equation to relate A1c to mean glucose developed using "hundreds of thousands of readings" and CGMS data.

It is the A1C-Derived Average Glucose (ADAG) formula:

Measured in MMOL/L: Average Glucose = 1.583 * A1c - 2.52

Measured in mg/dl: Average Glucose = (1.583 * A1c - 2.52)*18.05

Using this formula:
4% = 69 mg/dl
5% = 97 mg/dl
6% = 126 mg/dl
7% = 155 mg/dl
8% = 182 mg/dl
9% = 211 mg/dl
10% = 239 mg/dl

In short, 1% of A1c equates to a difference in mean glucose of 29 mg/dl.

UPDATE: 4/23/2008.

I've put together a calculator that will convert A1c and average blood glucose using the new ADAG formula.

You'll find it at


Anonymous said...

very informative post. thank you. i often have wondered why my a1c is higher than the readings i get after every meal. nathan's calculation is much closer to what i usually have.

your blog is really one of the best, most informative on the web. thanks.

Anna in San Diego, CA said...

Your website and blog has been truly eye-opening for me since I found it almost two months ago. After reviewing my lab results for glucose for ten years, I am pretty sure I have a glucose tolerance problem, despite it never catching a doctor's attention outside of pregnancy (high "normal" fasting glucose around 100-108; 5.5 A1c; I do better on a low carb way of eating; gestational diabetes with my one pregnancy 8 years ago). After reading about "truly normal" BG levels here and then in Richard Bernstein's book, I bought a glucose meter and started testing. Even with my mostly low carb diet, I was alarmed at how high my post-meal levels were and how long they took to go down. My fasting BG was rarely under 100 and post-meal BG (that had only small portions of starchy foods) were sometime 130+. A rare (for me) small serving of pound cake or cookies will push it to 150-170+. Lowering carbs further made a huge difference after a few weeks. I have eaten hi-carbs a few times to see how glucose intolerant I really am, but after learning how damaging all that sugar is and seeing pp tests of 150-170, I hesitate to continue testing high carb effects, even just to find out how bad the problem really is. I'd rather just stick to very low carb food because I could see that my BG went down over time.

Why is it that I always have to figure this stuff out myself? If I hadn't fired my doctor of ten years, and tried out another doctor, eventually settling on a third "enlightened" doctor out-of-network, out-of-state, and out-of pocket, I would never haver had my low thyroid condition diagnosed and treated last year.

I suspect my glucose issue is the same sort of situation. Minimal screening is done (despite my GDM history) and my FBG will need to go out of the "normal" end of 110 to concern them. I have an appt in March for an annual exam with a new doctor in my network who has a more holistic reputation than than the doctor I "fired", so that gives me time to do lots of testing and data collection before I discuss this with him.

Btw, before I started testing with a meter again, my lab test from the last annual physical noted an average BG value of 97 with my A1c of 5.5%, so that corresponds to your "nathan's calculation".

And thanks for all the information you post online. It is a wealth of easily-understood and useful information for folks like me, who find that the conventional medical system is just not doing much to promote good health and avoid future medical problems. Cutting-edge, unbiased, and well-researched information is hard to find, depsite the plethora of medical sites out there (including drivel such as WebMD). No one is more responsible for health than the patient and that means being informed and constantly vigilant, like a health detective :-).

Anonymous said...

thanks for your information.
i just want to remind you that MPG is different to MBG. Usually use:

The DCCT formula is:
MPG (total)= 1.98 *HbA1c-4.29 mmol/l
MPG (Fasting)=1.51 *HbA1c-1.35 mmol/l

using simple statistics, we may have
HbA1c= 0.34 *MPG (total) +4.118 mmol/l
HbA1c= 0.315 *MPG(Fasting)+4.687

1 mmol/l=18 mg/dl

Jenny said...

In the U.S. MPG (mean plasma glucose) is the same as the readings on our blood sugar meters (mean blood glucose or MBG) because these meters already convert the blood value to that plasma value which is 12% higher.

In the UK, this is not the case, some meters do convert, others do not. I've written about this in an earlier blog posting.

Pegasus said...

Hi tried using your formula in Excel and although your calculater site is a very nice, would prefer to obtain the calc w/o having to go on the website each time. The formula your blog provided produces an unintelligable result: This is what I used in Excel:
= (1.583 * 0.07 - 2.52)*18.05
Produces: -43.4859. What gives?

Specifically, I am interested in the formula you use to take Average Glucose and convert it to an A1c measurement. Great site.. learned some things today. Thanks~

Jenny said...


Your problem is that you are converting the 7% a1c to .07. Use 7.

Bob Lloyd said...

The anon post of July 13 2007 giving conversion formulae for the DCCT calculations were incorrect.

Given the DCCT formulae:
MPG (total)= 1.98 *HbA1c-4.29 mmol/l
MPG (Fasting)=1.51 *HbA1c-1.35 mmol/l

the correct rearrangements are:
HbA1c = 0.505*MPG(total) + 2.167
HbA1c = 0.662*MPG(Fasting) + 0.894

It wasn't my post but I was trying to reconcile the inconsistent conversion figures and the got the algebra wrong. The above conversion is correct.