A1c and CGM don't agree, I've finally discovered why, w/ update


I reviewed the American Diabetes Association 2019 Standards of Care and found many references to anemia. This one is in the context of using the A1c to diagnose diabetes.

When using A1C to diagnose diabetes, it is important to recognize that A1C is an indirect measure of average blood glucose levels and to take other factors into consideration that may impact hemoglobin glycation independently of glycemia including HIV treatment (11,12), age, race/ ethnicity, pregnancy status, genetic back- ground, and anemia/hemoglobinopathies.

Anemia is noted as a complication of chronic kidney disease (CKD).


The current standards of care recognize that anemia is tied in with type 1 diabetes but apparently not all docs have integrated this knowledge into their everyday practice mindset.


Terry, I think you might expecting too tight of a correlation between A1C and average bg to begin with. I think it’s misleading that the “conversion tables” imply that the correlation is accurate to 3 digits and it has never been.

The scatter before A1C standardization (15 years ago?) was often more than +/-1 and today it is still +/-0.5. Even today after standardization, I can still see consistent 0.5% A1C difference in my A1C tests run at Labcorp vs Quest (my insurance company has bounced between the two in recent years) that has nothing to do with my average bg and everything to do with the different labs.


I didn’t expect the A1c and glucose average to tightly correlate. I’ve seen the graph you produced many times and I accept the truth it portrays. I’ve written many posts and comments here over the years about my A1c/glucose disagreement situation.

My expectation is that a robust root cause investigation that could reveal some of the reasons for individual discordance should be sought. Is this unreasonable?

Anemia is one of those things that can drive this divergence between the A1c and average glucose. Clinicians should be aware of this situation and respond appropriately.

In my case, three endos, over ten years and many office visits, failed to think about and follow-up with what amounts to a basic health screening test.


Anything that affects the life of a red blood cell can conceivably affect an A1C test.

What do you think the cause of the anemia is?

Mean corpuscular volume (MCV) as part of a standard complete blood count is used along with the above RBC indices of MCH and MCHC to help classify the cause of anemia based on red cell morphology. Often the first value to look at with an anemic patient, an MCV blood test measures the average size of red blood cells. If the MCV is greater than 100 fL/cell, that could be macrocytic anemia, and a common cause is B12 and folate deficiencies. If MCV is less than 80, a common cause is iron deficiency. This may or may not apply in your case.

After I gave blood last year, I was one micro point below normal for blood levels.

Ah: I see you said it’s iron deficiency anemia.


I agree. I’m also sure that there’s a great deal of variation as mentioned by @Tim12 , but I was also bothered by how different my A1c was from my predicted A1c.

I did some research on different reasons why these may not line up and discussed it in depth with my doctor a few months ago. He ran more tests than usual, but my bloodwork all came back normal- though I suppose my LDL cholesterol level creeped up from “stellar” to “average.” Hopefully it lowers by my next appointment.

I feel much better now that I know the discrepancy is not a cause for concern but likely just an attribute of the variation @Tim12 mentions.

I’m glad you know about the anemia now @Terry. I’m really surprised your doctors never tested you for iron-deficiency anemia. My endos have tested me many times for this. I had a couple of vitamin insufficiencies/deficiencies (B12/D) a couple of years ago that I had to convince my doctor at the time to test for. Raising my levels really improved my energy levels. I felt frustrated that my doctors had never tested for these things or been helpful in determining the problems. I can empathize a bit with your frustration.

Hopefully the symptoms you mentioned will leave. It’ll be interesting to see if your A1c lines up with your eAG. Hopefully your post will inspire others who notice similar discrepancies to discuss possible causes with their doctors.


Kate, how big of a difference did you observe?

And the variation is not purely random, I’m sure that randomness we as scatter on the points in the graph, represents thousands of individual patient variations, as well as scatter between different labs and even scatter between the same lab on different days.


My A1c is generally .5-.6 below my predicted A1c.

I’ve switched doctors 3 times and maintained the same approximate discrepancy, so I don’t expect that it’s due to a lab error


IMO this topic is taken far too literally. It is just an estimated correlation. It’s not a scientific mathematical correlation. That’s not how physiology works. It’d be alarming and worth investigating if someone had a tremendous spread— eg average bg 80, a1c 12…

But when we’re talking about half a point or a point of a1c we are way inside the margin of error of the equipment we’re using, individual physiology, and random chance



A1C simply measures the how much the hemoglobin is glycated. Trying to correlate it with an average BG or an average CGM number is not worth a whole lot.

Also, everyone has a different rate of glycation. So two people comparing their A1C’s is also useless.

The way to use A1C is simple. Get your A1C test every few months and look at the number, and compare it to what you had previously.

If you start at 7.0, and then get to 6.5, and then 5.7 - as long as there are not any other metabolic issues going on with your body - you know you are doing the right stuff to have better control. If your A1C is getting worse, then try to fix that.

That’s how to use it.

It’s really a very simple thing that does not need to be over-complicated.


I disagree. I think that a discrepancy can alert you to other medical conditions.

Sure, there’s plenty of variation. I definitely think it’s worth investigating a spread of .5 though. Clearly it would’ve been helpful for Terry if his doctors were aware that such a spread could be a result of iron-deficiency anemia.

Possible contributors should be ruled out. I think it does their patients a great disservice if drs simply say that’s not how an A1c should be used.


According to the chart above, a spread of plus or minus 0.5 A1C is scatter on the chart such that I would consider plus or minus 0.5 A1c to be completely within bounds of what is expected.

I also agree with @Tim12 that comparing results also needs to be done within the same lab and if not then allowances need to be made. In addition to the lab test, the reference range as well can be different lab to lab. The result from a lab is intended to be compared to the reference range from that lab. This is also highly dependent on the particularly test. For some tests it does not make a big difference and the results and reference range are identical (or almost so) from lab to lab while a completely different test may have significant difference from lab to lab.

IMHO the specific case which this topic/thread was started on, rather than going down the road to discover an anemia problem based on a fairly elaborate comparison of A1c vs eAG, the better approach would be to run annual labwork.


Exactly. Thank you @katers87 for making this point. While I understand that normal variation may not be explained by anything more than the natural scatter of the data, sometimes there is an explanation of that variation. The lack of curiosity by a series of docs over many years missed the opportunity to intercede in a timely way.


Which is why I said:

A CBC (complete blood count) should be done annually anyway, and that is a better way to pick up an iron deficiency or anemia than looking at an A1C discrepancy.


I think that’s what this whole thread is about though. I guess I’m confused about posts saying that this thread is over-complicating how to use an A1c. Sure a CBC is clearly helpful and should be run.

There’s more than one thing that can result in an A1c discrepancy though.


Perhaps the doctors made the same conclusion about my observed discrepancy between my A1c number and the glucose experience reported from my CGM. They could have thought it was just a simple variation in the distribution of the data, but I still think it could have been used to provoke some thinking about other rational explanations. I suspect that none of my clinicians were even aware of the study that I cited in my post.

I agree that my anemia could have been easily detected with a blood test and, as I’ve stated earlier, I share responsibility in failing to get that done. I don’t need any doctor help managing my insulin regimen but I do depend on them to catch things like anemia.


So changing the subject away from A1C’s, and wanting to learn more about anemia:

Are you on any medication/diet/something else to help with the anemia?

Do you think your diet may have contributed to the anemia? A lot of us Mericans grew up with a lot of enriched cereal/pasta/bread but going low carb or even going “organic” can eliminate these from the diet.


This is really cool (although sorry it took that long for doctors to help you figure this out). I have to admit I’m surprised. Every bit of bloodwork I’ve had ordered for me by my doctors has included: A1c; fasting insulin and/or c-pep; metabolic panel; and CBC. I’ve had two rounds of antibody testing, and occasionally other things as well (micronutrient levels, etc.). But the CBC is something my doctors always look at precisely because it can help with early detection of things like autoimmune anemia, kidney problems, and some cancers (like leukemia).


I’m now taking a daily iron supplement. I’m also trying to up my red meat consumption since I suspect that real food is always better than taking a supplement.

It could have played a role. You make a good point that I’ve eliminated the grain-based foods enriched with iron and that could have contributed to anemia. At this point, what I don’t know about iron-deficiency anemia dwarfs what I do know. Since I’ve now won a front-row seat watching anemia play out, I’m sure to learn more in the months ahead.

Yours and others comments about how their doctors routinely check a CBC encourages me that at least some doctors follow this good practice. Unfortunately for me, my experience included 3-4 doctors who failed to perform. In the end, it’s the patient, with skin-in-the-game vested interest, who must own their health.



While at my regular 90-day endo appointment yesterday, I had a point of care A1c check performed. This is the first A1c after I started supplementing iron last month. I was curious to see if, as described in the cited study above, my A1c number would match up better with my CGM average glucose values.

My BG averages:
14-day - 93 mg/dL (5.2 mmol/L)
30-day - 94 (5.2)
90-day - 95 (5.3)

This American Diabetes Association web page has a calculator that converts an A1c into an average blood glucose and will also take an average blood glucose and translate it to an A1c.

I put in my 90-day CGM average of 95 mg/dL (5.3) and found that it is equivalent to an A1c of 4.9%. In the last 10 years, my A1c always floated about 0.5% above what the background average blood sugar predicted. What was my A1c yesterday? It came in at 5.0%! Only a 0.1% divergence from my average glucose.

I had a complete blood count (CBC) panel done yesterday as well. My hematocrit and hemoglobin tests both measured just above the lower lab limit, while the red blood cell count measured just below the lab limit. So, in comparison to the CBC drawn six weeks ago, my hematocrit and hemoglobin both improved and moved into range while the red blood cell count did not change and remained below range.

While I will continue to watch this over time, I think this circumstance suggests that low blood iron levels affected accuracy of the A1c to reflect my overall glucose exposure.

I’ve never received an A1c as low as 5.0% before and am blown away by what feels like a breakthrough for me. I know it’s just a number but it’s one that affirms the every day effort I put into controlling my BGs.

Some people have suggested that the A1c serves as an indicator of glycosylation on other tissues in the body. I don’t think this is true to the extent iron deficiency skews the A1c number higher than actual.