Welcome back to our weekly diabetes advice column, Ask D'Mine — the first edition of the New Year, so Happy 2013 to all!

Your host is veteran type 1, diabetes author and educator Wil Dubois. This week, he explores in detail a question about how reliable the A1C test is and how much we should really rely on it in gauging our D-health. Good question!

{Got your own questions? Email us at AskDMine@diabetesmine.com}


Toby, type 3 from Texas, writes: For the past year, we have been struggling with our son's A1C coming in higher than his average BG readings. Our software lets us know that we averaged 11 checks a day at an average BG of 130 mg/dL.  His intermittent CGM readings came back with a slightly higher average BG of 135. Yet 3 times in a row, over a 9 month period, his A1C was 7.5. Our doctor made the assumption the fault was with us, the parents, and only provided us with 'Diabetes 101' information on how to lower his A1C.  How accurate is an A1C? Are there things that can throw it off? While I'm at it, is there anything else about A1Cs I should know?

Wil@Ask D'Mine answers: Considering that A1C is viewed as the bedrock of diabetes control, I'm going to ruin everyone's new year. I hope to hell you all have some bubbly left over from ringing in 2013 because you'll need it to drown your sorrows by the end of this post.

Because the bedrock is quicksand.

For background, the A1C wasn't always the big deal it is today. The test only became a big deal in the aftermath of the seminal DCCT trial in the 80s and early 90s. I wouldn't say that the researchers ran A1Cs as an afterthought, exactly, but I don't think at the time anyone thought the test had the potential to become the gold standard of diabetes control, either.

But in sifting through the data at the end of the study, the researches noticed something: Lookie here! Type 1s with really high A1Cs suffer some really bad shit; and type 1s with lower A1Cs don't suffer so much bad crap. Huh? Do you think there could be a connection? Believe it or not, at that time, many docs didn't even believe there was a connection between high blood sugar and diabetes complications. It was the DCCT trial that proved there was.

It quickly became obvious that A1C was a great way to look at overall average control for the past three months, changes in control, and risk of complications. There was just one teensy, weensy little problem. No one, anywhere, ran A1Cs the same way. Different countries used different assays and different procedures. Hell, even within countries, different labs did them differently. As you might expect, the results were all over the map.

There were no standards, and reaching a consensus on how to create standards took a long time. I'll spare you the details of this scientific and political food fight, but rest assured that nowadays any A1C anywhere has, for all intents and purposes, pretty much the same result. That said, while there is now a global standard for lab procedures, there's still plenty of disagreement....

We'll talk about that in a bit.

How Accurate Is It?

According to the National Diabetes Information Clearinghouse (NDIC), the "official" accuracy of the A1C test result is 0.5 on either side of the "true" number. So if your honest-to-God A1C was 7%, then you could expect your lab test to read anywhere between 6.5% and 7.5%.


So like our meters, A1C machines aren't exactly 100% accurate. But they are getting better. The standards organization in the USA for A1Cs is the NGSP. NegSpa, as I guess you would have to pronounce it (but I'm only guessing, as I didn't call them), is like KFC—the restaurant formally known as Kentucky Fried Chicken—which is now only known by its initials. Before NegSpa became NGSP, it was called the National Glycohemoglobin Standardization Program. They went the KFC route because as the program grew, it became international in scope, so they could hardly keep calling themselves "national." They dropped their name and adopted their acronym.

(The same way JDRF is now known only by its initials because the word "juvenile" no longer represents their full constituency-!)

But frankly, KFC rolls off the tongue better than NGSP.

And while the A1C isn't perfect, it's gotten more precise during the past several years. According to NGSP, accuracy has steadily improved since 2007, when the acceptable error was plus or minus 15% of the score's value.

How Much Does Accuracy Matter?

Just to play Devil's Advocate (someone has to do it, and I'm sure everyone will agree that I'm fully qualified for the position), I don't think it would matter, even if the accuracy were much worse than it is today. That's not where its value lies. I view A1C a little bit like the early CGM (continuous glucose monitors): it may not be 100% accurate in its individual readings, but I think so long as the test source is stable—same machine or same lab—the trend can be trusted and is highly valuable. Is your diabetes stable? Getting better? Getting... um... like mine, worse?

We also know that very high results are very bad, as are very low results, at least for those of us on insulin. Beyond that, we can't compare the long-term outcomes of, say 6.7 vs. 7.2 yet, so maybe it doesn't matter how precise the test is, anyway.

Falsies (Not Like You're Thinking)

There are actually several different types of A1C tests. They all conform to the international standard, but some types of tests don't work right for D-folks who are of African, Mediterranean, or Southeast Asian heritage.

On top of that, a number of things can give you a false low A1C—meaning your real sugar is higher than your results indicate. One example is a bleeding ulcer or other "chronic" bleeding. If you lose blood at a higher than normal level, your body will produce new blood at a faster than normal rate to replace it. This gives you a higher percentage of "newer" blood that hasn't been hanging around long enough to hyglociate at the expected rate, giving you the false low. Folks with anemia are also prone to false lows, as are those on dialysis.

A1C tests can also give false highs—meaning your real sugar is lower than your results indicate. The most common cause is from low iron in the blood, a.k.a. iron deficiency anemia, not to be confused with garden-variety anemia above.

And then there is the crap shoot called hemoglobinopathies. Most of us have only hemoglobin A. But... there's always a 'but' in diabetes, hence my recent suggestion to change its name to Buttabetes... some folks have one of several other types of hemoglobin that are lumped together as "hemoglobin variants." These can cause either false highs or false lows, depending on the type.

Oh, and kidney failure and liver disease can also make A1Cs wacky.

And although this remains highly controversial, some researchers believe in variable glycation, the theory that some people just naturally "glycate" at higher or lower rates than others, regardless of diabetes control.

Got a headache yet?

Don't forget the post we wrote last year, where a leading voice in the diabetes world, Dr. Irl Hirsch, expressed his concerns about the A1C and how it's not always accurate. Especially in that it doesn't actually measure a full 90 days, but can be weighted for the past 30 days so that someone with a higher or lower blood sugar readings can skew the results.

Still, it's considered the "gold standard" and is the best we've got at this point.

The A1C Tower of Babble

By the way, most of the rest of the world doesn't report A1C as a percentage like we do here in the States, but instead use something called International Federation of Clinical Chemistry units, or IFCC units. For an A1C, the IFCC units are mmol/mol.

Check out this poster from Accu-Chek UK, where we are told our A1C should be under fifty-three! No shit. Of course they mean 53 mmol/mol, which would be 7 % to our way of thinking here in the States. A 53 percent A1C would be a blood sugar of 1,474 mg/dL, at which point I'm pretty sure you are dead.

Stateside, the supporters of the percent method of reporting have their heels dug in pretty deeply, so I'm not worried about having to learn mmol/mols anytime soon. Going forward, most scientific literature in the diabetes space will probably dual-report data in both sets of values.

And Then There's eAG...

To the "rescue" is yet another measure, this one a statistic, well, more of a formula. Once upon a time, the various powers that be decided that to make this less confusing (!), we PWDs would benefit from being able to convert our A1C scores into meter numbers, called estimated average glucose, or eAG (estimated average glucose). Most lab reports worldwide now include an eAG, but like the rest of this mess, there's not 100% agreement on how to go about doing this. The leading formula for converting a percentage reported A1C into a mg/dL number is:

28.7 x A1C in percentage points — 46.7 = eAG in mg/dL

And there are formulas for all the mmol/L and mmol/mol stuff, and every other perverted back and forth you might need. Here's a calculator to help you convert to your heart's content.

Hey. Don't shoot the messenger. Check out the 'Mine's report on eAG from back in 2008; it actually appears the eAG idea was more of a "power grab" by dueling medical factions, and probably won't ever be widely accepted.

Last Not Least: You, Your Son, & Your Doctor

As to the discrepancy between your in-house data and your son's A1C, any number of things could have happened.

By the newest formula, an A1C of 7.5 means your son's average blood sugar is 169, about 40 points higher than your meter readings. It could be you had a bad meter that was running low. As to the CGM, of course, it doesn't know what the blood sugar really is; it only knows what the meter tells it that the sugar is (by calibration), so if the meter runs low, so too, does the CGM. But that's a pretty big difference you're seeing. It could also be that your son has one of those oddball things that throw the tests off.

But bottom line, I think an endo should consider that a ton of data might tell more truth than three little drips of blood. It says a lot about how we blindly trust data we really shouldn't. It says even more about medical arrogance and the medical decision-making process.

Frankly, I don't think much of a provider who doesn't at least wonder why an 11-times a day tester is showing an average that far off from the A1C. Not just once. Not just twice. But three times.

That should raise a red flag. And not a red flag that says "bad parent."


This is not a medical advice column. We are PWDs freely and openly sharing the wisdom of our collected experiences — our been-there-done-that knowledge from the trenches. But we are not MDs, RNs, NPs, PAs, CDEs, or partridges in pear trees. Bottom line: we are only a small part of your total prescription. You still need the professional advice, treatment, and care of a licensed medical professional.

Disclaimer: Content created by the Diabetes Mine team. For more details click here.


This content is created for Diabetes Mine, a consumer health blog focused on the diabetes community. The content is not medically reviewed and doesn't adhere to Healthline's editorial guidelines. For more information about Healthline's partnership with Diabetes Mine, please click here.