Inaccurate A1c

Looking for anyone with a similar situation - my A1c results are inaccurately low. It wasn’t until a recent change in endocrinologists that I found out this has been the case for the past 30 years. For example, my A1c is typically 4.8-5. In reality my blood sugar average is closer to 6.5 per CGM readings and fructosamine blood test. The doctor says it has something to do with how quickly I turn over my red blood cells. I’m curious if anyone else has this issue.

Funny you should ask: we had an endo appointment today, and the dr. and I were both surprised when my daughter’s A1c came back at a 6.5. I mean, that’s great (for us — others may have different goals), but based on the CGM data, we were expecting something closer to 7.0. We just figured we were falling victim to recency bias: she’s been running stubbornly high lately, but when we looked back at the 90-day graph, her pattern looked pretty good overall. Still, the GMI on Clarity said 7.0, so I’d love to learn more, too!

@Rina @srozelle check tha lab paperwork and they should give you the “lab normal range” for their test. My lab, LABCORP, shows a normal range 4.something (four point something) to 6.5 percent. Meaning, the lab results based on the lab assay, will read about 0.5% lower than reality, so my 6.2% reported by LABCORP is “really” a 6.7%. A “normalized” %. For a1c would be 4-6.0%. Just a thought. These are all really good numbers!

1 Like

I am grateful that overall my numbers are good but the falsely low numbers over the years meant that I had to wait much longer to get a pump. For years my endo said my numbers were too good to give me a pump.

I am very sorry to hear that. I consider my Endo one of my employees. If they ever behaved that way I would give them a warning, and they thay would be fired. Sorry that was your experience.

1 Like

Here’s what it says:

Your Value
6.5 %
Standard Range
4.0 - 5.6 %

But I don’t get it. What does that mean for accuracy? I thought it was just the reference range (which I thought was an automated way to flag items that may be of interest to anyone reviewing the lab work).

1 Like

What was the rationale, do you know?

I’ve often heard it said that someone’s numbers weren’t good enough to get a pump. I don’t understand that one, either, really. I mean, I get that if you aren’t checking your sugar and something goes wrong with your pump, you’ll end up high faster (including high enough for DKA faster) than if you were taking the longer-acting basal injections. But then I think, couldn’t someone do just as much harm with injections? I mean, if you don’t inject, period, you’ll end up in DKA just as fast as someone who doesn’t notice that the pump’s malfunctioning. And if you take insulin without checking your BG first, aren’t the risks the same with a pump as with injections?

As for someone who’s getting great results with MDI, I understand “if it ain’t broke, don’t fix it,” but if the patient wants a pump, presumably that’s because the pump seems like it will be an improvement to them in some way. And it’s not hard to imagine what the biggest, most obvious benefit might be, either. I mean, gee: I can take insulin by sticking myself repeatedly throughout the day, or I can take it by sticking myself once every three days. Which sounds better? :thinking:

So what’s up with being denied on either ground? Is it just some power trip, that people like the rush of denying someone something just because they can?

This was about 15 years ago and insurance wouldn’t authorize a pump because of what appeared to be very low A1c. It wasn’t until pumps became widely available that I was able to get one. Even then, my A1c was below 5.

I should clarify that I have Kaiser and my Kaiser endo wouldn’t authorize a pump for me. Fast forward and I have a new Kaiser endo and she is great.

I’ve been pumping since the mid '90s. My memory may be wrong🤪 but at the time I believe my doctor had to show medical necessity for me to get one. I don’t know that that’s still the case (unless a particular insurer requires it) - it seems like lots of new patients are started on pumps very early on.

if the patient wants a pump, presumably that’s because the pump seems like it will be an improvement to them in some way

I agree. However when I got my pump I do remember my doctor spending time telling me what it would and would not do, so patients would understand that a pump is not simply a matter of convenience and would appreciate the work involved. And for those who want to use a pump for tight/er control - this is my opinion but it might be possible to overdo it. I relent hearing about patients who were hospitalized trying to get their numbers lower, so there might be some concern that patients with very good A1Cs might take things too far. My theory.

Yes medical necessity was required. Had my A1c been accurate, it would have established that I did have a medical necessity, thus the frustration. My main question though does anyone else know of or have whatever condition causes a false negative A1c. Seems to be very rare.

I’m not sure it helps but I found this on a Google search:

The A1C ranges for normal, prediabetes, and diabetes are as follows: 1

  • Normal: Less than 5.7%
  • Prediabetes: 5.7% to 6.4%
  • Diabetes: 6.5% or higher

Don’t be confused by the 5.7-6.4 - if you get there (without frequent and significant hypos) that’s great, but according to these ranges your 6.5 is at the very start of the diabetes range so you’re at a good point.

Thanks, Dorie. But it’s not a question of understanding the reference range (we’re absolutely proud of the decisions she makes to keep her BG in range. It’s a tough job, as everyone on this forum knows!). I only shared it in response to Joe’s explanation, which I didn’t follow. Like Kathy, I’m trying to understand why an A1c might be artificially low compared to the GMI.

I thought GMI was based on the average of the CGM readings from the last 90 days, and therefore should be a pretty good indicator of A1c, which I thought was also supposed to indicate average BG from the last 90 days.

Is it the difference between BG and interstitial fluid glucose? In other words, are CGM readings inherently higher than actual BG? (That seems backwards from a safety perspective: assuming there will always be error in any human endeavor, we should do what we can to ensure the error happens in the direction that causes less harm. And wouldn’t that be CGM readings that are lower than BG? To clarify, I’m not talking about the lag time, which I know is inherent. I figure over any given 90-day period, the lag time difference washes out.)

Regardless, that doesn’t explain what the A1c reference range has to do with it. @joe, what am I missing?

@Rina and @srozelle I am offering that the test is not absolute. That the test itself can be biased in some cases, specifically in my case LABCORP. So I know their reported a1c will be lower. It’s only 1 possibility.

The other possibility is that the CGM uses mathematics to “estimate” an a1c, and the CGM can be off a total of 20% from glucose, and there is a difference between serum and plasma blood sugar. You can read more here.

So one source of difference is the reported lab a1c, depending on the range reported. Both your results seem to be aligned with 5.5 which eliminates the lab reported value as a significant contributor.

Mathematically the CGM calculates the time at each CGM reported value and can integrate that data set to estimate an A1c so there is big potential for error in CGM accuracy and in integration versus blood cell Glycosylation. I am now thinking that’s where the potential error is.

1 Like

This is very cool, Joe, thank you. So, let’s see if I’m following:

  1. If the sample is sent out for analysis, then it’s going to be artificially low because of the time delay. But then shouldn’t LabCorp’s reference range be lower, rather than higher? Say a given sample, if tested right away, would be a 6.5. If sent out and tested later instead, it would be a 6.0. So if the normal range for samples tested right away is 4.0-5.6, then shouldn’t LabCorp’s late-tested normal range be 3.5-5.1? (My sample was tested right away, so this isn’t likely to be a factor for me, but it’s interesting regardless.)

  2. The CGM is only 90% likely to be accurate within +/- 20, so that’s definitely a possibility. Any data on which direction the error is more likely to run? I didn’t see anything in my quick search.

  3. And then lastly, there’s the serum v. plasma difference. I didn’t know those were two different things, so I went and educated myself. (Thanks, Google!) For anyone else who didn’t know, either, plasma is what’s left if you remove just the blood cells; serum is what’s left if the blood is allowed to clot first, so you’re removing not only the blood cells, but also the platelets, fibrinogen, and several (but not all) of the clotting factors.

Again, the delay inherent in allowing the sample to clot first means you get an artificially lower A1c from serum v. plasma.

Also of note in the article Joe linked, plasma glucose is about 11% higher than whole blood, and capillary glucose is 20% higher than venous glucose.

So, do they centrifuge the fingerstick sample they took and test the plasma, in which case it’s 20% higher than a venous sample would be? Or do they test the whole blood, in which case it’s 20% higher than a venous sample, but 11% lower than a plasma sample, and therefore that much closer to the immediately-tested venous plasma sample that seems to be the standard?

Did I miss anything?

Haha no I don’t think you missed anything. And those are all great questions and I am taking a week off so it might be si e time to really read your questions to be able to respond.

I’ll do one. The results from your lab are against known quantities in “reference ranges” so they, in this case LABCORP, tests a surrogate with a known value and their statistical results are significant between 4 point something to 5.5.which is skewed slightly higher than expected normal of 5.0. So my lab reports back a 6.2 from an unknown (aka me) I arbitrarily add 0.5 to their result to (I guess, unskew) correct it. Sorry I gotta go make dinne now. :grinning::four_leaf_clover:

Take your time, and thank you. Learning about all this was way more fun for me than the work I should have been doing. :nerd_face:

So I searched for “typical” procedures for blood analysis and the CDC has a protocol from 2008 online (link here)

If you look at the procedure, potential errors are in calibration, potential mishandling of the control reagents (-70C), potential mishandling of whole blood during shipment, storage, holding, as the temperatures are very specific and in some cases (4-8 C and -70C), very very difficult to maintain. .

It looks like this analyzer is using whole blood, so plasma versus serum is not going to be a major difference.

Lab conditions are, usually, vastly more documented and controlled than results obtained by finger stick and CGM… so if it is worth anything, I am still betting the lab is the “better” number and anything computed at home is suspect. just an opinion. your mileage may vary.

NDR: I crushed dinner, linguini with littleneck clams (in the shell) and my famous garlic/butter sauce.

My last A1C was 4.8. Based on my CGM data I was expecting 5.6. My assumption is that it is the A1C test that is more accurate, since, as Joe pointed out, CGM data can be off by up to 20%. So like you I have a discrepancy where my A1C was significantly lower than CGM data, but I’m wondering if there’s a reason you’re thinking the CGM data is the more accurate number (just for my own understanding)?..

Dinner sounds amazing!