Why A1c can be different from eA1c and GMI


A couple things here (if you are talking about looking at your average BG on your meter). On your meter, a low BG will average out a high BG. That is reflected on the number it shows on your meter.

A 40, 40, and 200 average out to 93 on your meter. But you still had the 200, which elevated the amount of glycation you had, despite the 93 average.

Also, BG meters are not a random sample of values! You don’t test at random times everyday. You test when you eat, when you feel bad, when you are high, when you are low, when you are taking insulin. So the BG meter average is only an average of the times you have tested. I know that seems like a dumb simple comment to make, but it really is relevant. I know that I test much more when my BG is bad compared to when it is good. I think that is a very common thing to do…

The A1C, it is not a measurement of average. So lows don’t reduce it. A1C is a measurement of the amount of glucose that has bonded to the hemoglobin. So “average blood sugar” when used to describe A1C is somewhat over-simplified

You can get somewhat of a ballpark idea of your RBC lifespan by getting your reticulocyte count and doing the calculation. It is still a ballpark, but it can give you a little better guess of your RBC lifespan, and how it might compare to normal lifespan.

But with a 5.1, you can be happy. Even if you had a very short RBC lifespan, a 5.1 is great no matter what. So congratulations.


I’m actually talking about my Dexcom average. I recognize that that average is not perfect, but it’s actually a better reflection of my true average than my meter average. As you said, I test more when I’m out of range or think my bg levels aren’t aligned with my Dexcom.

The Dex seems to be more accurate for me than it is for others on here. It’s generally behind about 15 minutes and off by as much as +/- 10 mg/dl (when considering the delay), but it’s not usually off by more than that. Though it does happen sometimes. My Dex also doesn’t consistently run lower than my real numbers (unlike others report). It’s sometimes higher and sometimes lower.

While I agree that it is over-simplified, multiple studies have established that there is a correlation between the two. Based on my average glucose, the correlated A1c based on population data would be somewhere between 5.6% and 6.0% (which I fully recognize are good numbers :slight_smile:). Some studies have shown that there can be external factors that “falsely decrease” the A1c. This creates two concerns… first, why is mine lower? Do I have one of these external factors, but I haven’t been diagnosed with it? https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3912281/

My other concern is that the studies that show that a low A1c reduces the likelihood of complications are being done on a population where there is a correlation between A1c and average blood glucose. I think.

So, is my A1c truly representative of how much glycation is occurring on other proteins in my body? Or since my red blood cells aren’t living as long as they are for the general population is my average actually a better measure of how many AGEs I’m accumulating?

I think my average is fine, so I’m more worried about the first concern. Both are something I would like to know though, especially because while my average is fine now it hasn’t always been this good. I think my A1c has always been lower than my average glucose would predict it to be.

Thanks for the tip. I’ll ask for this next time I see my doctor.

First off, congrats on a maintaining a great A1c, everyone here knows how hard this is, and shows that you really understand how to manage yourself.

My comment below is really just meant to reality check what you are finding. i.e. what is your goal. I like to think of A1c as a rough measure of how high and long your elevated blood sugar excursions are. The higher and longer your excursions the higher your A1c.

In the paper you referenced, they also mention this:

So what happened recently can have more impact that what happened towards the end of the period. Can this explain any of your differences in measurement? i.e. have you checked your average for the last 3 weeks against the average for the whole period.

Finally, and I don’t mean to be flip. But what clinical difference do you think there is between 5.1 and 5.6? According to the DCCT anything below 7 is giving you approx. 95% of the risk avoidance, and anything below 6, is basically gluco- normal give or take. My approach would be to say A1c <6, awesome and move on.


My average over the last 45 days is 118, and the average of the month before that is 118. The standard deviation is 40 and 39 respectively. It changes a bit from week-to-week, but remains within the average range in my message. It was not better or substantially different (as far as I can tell) in the 2-6 weeks before the test by any measure.

I’m sorry if what I wrote was annoying. I can understand why my post might be annoying. My main concern was not about the higher likelihood of complications based on my A1c right now (which I know is negligible but maybe I could have made that clearer in my post). My main concern was about why it was so much lower. However, I would also like to understand which measure is more significant.

I have had vitamin deficiencies before and they ended up having a pretty big impact on my life. So I do think it’s important to think about why my A1c is consistently betweeen .5-.9% lower than my average bg would predict when I’ve read that there are reasons (such as anemia) for a falsely lowered A1c. Maybe it just wasn’t appropriate for me to discuss this on the forum. I haven’t had a lot of luck in talking about this with doctors, and I guess I just wanted to see if anyone here had an idea.

1 Like

No, by all means feel free to discuss in the forum, I was just asking what your goal was since it could be construed to be micro-managing.

Your goal seems to be reasonable, i.e. understand why your individual results are lower i.e. vitamin deficiency or anemia etc.

Although, I am assuming they did the anemia check in the blood work you just had pulled at the physician visit.

Also, if you are wearing a CGM most of the time, have you done anything to try and quantify if your high blood sugar exursions have changed in time or severity over the period.


Ah, makes sense.

So, I didn’t have this test done. I’ve had my iron checked the last few years, and it’s come back normal each time. I don’t really know what other forms of anemia there are that cause this. When I asked my doctor last time (4 mos ago), he didn’t really have any good answers. I didn’t know that my A1c had dropped until after my appointment, so I wasn’t focused on it at that time. I should’ve asked again though. We’re building rapport now I think, so maybe he’ll listen a bit better to me at the next appointment. He said he usually does a full panel once a year on his patients, so I suppose a good step for me would be to do some research on other forms of anemia between now and then. If I have more background then I can ask about different tests.

I think that Afrezza has helped me spend less time in the high range after eating or correcting, but that started around 5 months ago. I think that’s what resulted in the A1c drop from 5.5% to 5.1% because a similar drop occurred in my average bg levels. Time in range is similar in the last 6 weeks to the month before that (about 6% above 180 mg/dl). In the time of the 5.5% A1c, I was closer to 10%-13% above 180 mg/dl. So there is definitely a reason the A1c decreased, and that’s comforting.

I’ll do some research about other forms of anemia and talk to my doctor at the next appointment.


Right, but all highs are not the same. For instance, in both of these cases my son was high for 10 hours, but the first case would result in much more glycosylated hemoglobin.



In both cases the time above range would be the same, but the amount of glycosylated hemoglobin would be vastly different.


So you’re saying that perhaps the general population experiences the first graph more than I do?

Before Afrezza, I generally spent about 2% of the time above 250 (looking at past AGP reports).

Last 3-4 months it’s been less than 1% thanks to Afrezza. 1% at 90 days, .3% at 30 days. .7% in the last 2 weeks. So I suppose there’s a small decrease before the test, but I don’t know if that’s significant or not.

Not sure how this compares to everybody else. My A1c was lower than predicted by my cgm the whole time I’ve had a cgm (around 2.5 years). Though it does seem like a larger gap now.

@Katers87 - good job on the control. I have previously complained about being the opposite - A1Cs higher than CGM 90 day and 30 day averages, but I have almost gotten over it :slight_smile:

Just curious if anyone knows the answer…

The DCCT finished up in 1993, which is 25 years ago - Since then, we have all sorts of new diabetes technology like rapid acting analogs, long acting insulin (levemir and lantus), and CGM’s to name a few.

Since then have we seen anything big an new in terms of large scale complications studies like DCCT? Is there anything else in the works?


Unfortunately, big studies cost big money, and if there isn’t a clear way to make a bunch more money from the results i.e. increasing indication etc. then the money isn’t available from industry to do a continuing large study to verify the results.

So until Europe or China, or perhaps JDRF and the ADA decides to invest in a big way to answer that question, we will be stuck with 25 year old data and its conclusions.

No, I am saying that an average could easily be unchanging if following the first graph you had a period of lows (which isn’t uncommon after a period of highs) and although that should show up in the Standard Deviation I don’t think average and standard deviation are the whole picture.

I hope that someone comes up with comprehensive metrics that track more without all of the effort. For instance, when trying to understand some things, I will go through our daily tracks and try and roughly quantify the time above a benchmark and the severity of the high to understand the data and changes we have experienced in treatment or response.

1 Like

I put the last 6 weeks in the first report below, and the month prior in the next report. I kept the little baby graphs at the bottom so you could see the daily trends. It is true that there were more highs in the month prior. I certainly won’t pretend that my numbers are perfect. I’m not flatlining over here. :slight_smile: I suppose that’s the point though… my A1c is lower than I’d expect based on my numbers.

We could pick these apart, but I just don’t really see how we’d arrive at a 5.1% A1c from the last 6 weeks…




My numbers were very similar…and my a1c came in at 6.7!! Go figure?!


@Chris. Actually, someone has. PGS, or Patient Glycemic Score, is a measure of the glucose variability index, time in range, and measures of hypoglycemia both below 70 and below 55, over a specific time period.

It is a difficult computation but there are some apps that do it for you, xDrip+ being one of them. When taken with the GVI (Glucose Variability Index) it is easy to gauge both the altitude and amplitude of a track.


Silly question… Can/will anyone please elaborate on the iron deficiency comments referred to above? I have pernicious anemia. For most of my life…but I don’t comprehend the diabetes part of the relationship. Maybe I shouldn’t have skipped biology in 8th grade or something, but I have never read or heard of a correlation between the two before.

1 Like

I’m not entirely sure if there’s a direct connection between the two, but studies have shown that your A1c may be higher or lower than you would expect (based on studies showing a positive linear correlation between A1c and mean glucose) because of anemia. I’ll paste a few studies below, but I haven’t read through all of them. Most conclude that iron-deficiency anemia results in a falsely elevated A1c and hemolytic anemia results in a falsely lowered A1c. I came across this possibility while wondering why my predicted A1c (based on average bg level) and actual A1c were pretty far off.

I don’t know a lot about pernicious anemia, but I think that means that you can’t digest B12 properly. I’d guess that you get B12 shots? Please correct me if I’m wrong.

A few years ago, I had insufficient B12 levels by US standards (deficient according to a few other countries) and deficient Vitamin D levels (by all standards) . It took my doctor time to figure it out, and my quality of life was significantly affected. It was like I was in a fog all the time. There were lots of issues going on. My eyesight was affected during this time frame too, but I can’t prove that it was a result of the vitamin problems and it did not get better once my levels normalized.

The fog and the other symptoms mostly left once my vitamin levels normalized by taking supplements. I’ve had my B12 levels tested since then, and they’ve been normal so I don’t think I had pernicious anemia. I wasn’t vegetarian, so there’s no reason I should’ve had a B12 problem though. The research article below indicates there’s a relationship between B12 deficiency and diabetes.

Here’s an article about the relationship between diabetes and pernicious anemia:


@Katers87 Thank you very much for the articles. Looks like I have some reading to do. My B levels have been sub threshold since I was in knickers. I take sub-lingual b 2x daily, and it’s still low. The vitamin D thing though I have a handle on thx to the weather in California.

I appreciate the articles very much!

1 Like

I couldn’t revisiting this thread full of interesting data and sources after the recent discussion we had on GMI vs A1c:

I’m late to the party here, but I think that the theory behind the relationship with anemia and A1c kinda goes as:

Anemia=low numbers of RBC
A1c=measurement of glycation of RBC (by proportion of RBC)

Glucose within serum will bind to hemoglobin on RBCs, and A1c is our way of measuring that. If you have two individuals with the same blood glucose levels, but one is anemic, then the anemic individual has less RBCs for glucose to bind to. The absolute number of RBCs could be the same within the two individuals, but there’s just less RBCs overall in the anemic individual. So the proportion of glycated RBCs/total RBCs (A1c) will be higher in the anemic individual because of a reduction in the total amount of RBCs.

This does not consider the lifespans of RBCs though

Also, if I’m off, then someone feel free to correct me!

If you are referring to iron-deficiency anemia, yes it can cause an increase in A1C.

Your theory is reasonable, but the reason that an iron-deficiency anemia can increase A1C is not fully understood. Some believe it has to do with oxidative stress. Others believe that with iron-deficiency anemia, since the production of RBC’s is reduced, the RBC’s will last longer, and will ultimately lead to higher A1C numbers.

Unfortunately, there is no cheap, easy, and accurate way of determining RBC lifespan. You can do a reticulocyte count, but that is just an estimate. You could do a reticulocyte staining procedure, (REBEL - Reticulocyte Based Estimation of Red Blood Cell Lifespan), but that’s not something insurance would likely cover.