A1C wildly different from Dexcom Clarity?

I use the Dexcom G6, and my Clarity app says my estimated A1C over the past 90 days is 6.8. I check the app frequently and it’s always right around 6.8. I got my blood drawn in the doctors office today, they ran my A1C and it came back at 6.1!! That seems like a huge difference to me. My number last year was 6.8 and the year before that it was 7.2. I had been using Tandem tSlim for years and just switched to Omnipod 5 in December, I’ve noticed somewhat better control over the last several months but I’m just confused as to how there can be such a big discrepancy between my Dexcom numbers and my actual A1C. Has this happened to anyone else before? Did the phlebotomist check the wrong person’s blood? lol

My A1c is typically .5 less than my Dexcom. Maybe I should calibrate more. Also, not considered a big discrepancy, just norma,l ± .5.

Dexcom has been making quite a few software tweaks recently and that may be part of the difference. You must also have a different version of Clarity than the rest of us as for the past few years Dexcom has been showing % GMI, not % A1C, so perhaps you need to update your version of Clarity.

Mine is up to date and does say GMI but my doctor has always referred to it as “estimated a1C” which I suppose is incorrect!

My A1c is always lower than my GMI, though usually by .3 or .4. At any rate, your trend is moving clearly in the right direction by any measure which is the main thing. Maybe your next A1c will be closer to your GMI or maybe you’ll keep seeing a largish gap

1 Like

OK, the next question is, how close is a finger stick to your Dexcom reading? I suspect that your occasional finger sticks are reading lower than what your Dexcom BG readings are showing; if so, you would need to consider that. Some patients are anal or have too much time on their hands and go crazy with calibrations until their finger stick BGs exactly match their Dexcom BG. Many of us don’t have the time to spend on multiple calibrations, accept the difference, and factor the difference in our insulin doses. Do what works best for you.

1 Like

My lab A1C runs .5 to .7% lower than GMI. GMI is a correlation between blood glucose and A1C based one of the studies conducted after DCCT. The HbA1C to eAG ADA is an older correlation from the DCCT. Also, inter species (individual) variation is to be expected.

Time in range is, IMO, a better measure of blood sugar control.

Clarity says my A1c is 6.0. Lab draw is 5.6. So I just ignore until my next labs.

This is all really interesting, because my A1c is always around .5 higher than my Freestyle Libre 3 GMI. I guess I’ll get a Dexcom Stelo when available, and average the two out. :grin:

It sounds like the man with a watch can tell you what time it is, yet a man with two watches is never quite sure. Diabetes is not about a number, regardless of the device or standard used. It is all about the range and the trend. If you always use the same device and the same standard, trending down shows improving control while trending up is a call for action.


The Dexcom clarity A1C is based off of the Dexcom sensor reading of interstitial tissue and is not 100% accurate (20/20 rule). All of the times that the Dexcom is inaccurate, those inaccuracies are being calculated into Dexcom clarity’s A1C. When the doctor checks your A1C it’s a blood draw which should always be a more accurate 3 month average of your blood glucose levels.

It has very little to do with the accuracy of sensors. The notion that you can extrapolate information about your diabetes management/success from a HbA1c result itself is inaccurate! It’s a statistical model, not a fact. There’s a pretty good correlation that as average BG of a population goes up, so does their HbA1C, but any individual person/data point can be all over the plot. They don’t all sit perfectly on a straight line like the medical people would have you believe. It’s very useful as a cheap diagnosis aid, but that’s it. It’s misused for continued evaluation.

There are lots of studies you can pick and choose from, from this is the data from one of them plotting average BG against HbA1c. It’s an absolute scatter plot!

Every one of those dots is a person. A woman, specifically, in that example. The researchers looked at her average glucose, tested her HbA1c, and marked it on the chart. How many of those dots sat in an office and got scolded by their doctor for being non-compliant and out of control because their doctor has no clue that they’re averaging really good numbers, but their HbA1cs keep coming back high no matter what they do? What about the dots in the upper left quadrant? How many of those dots got the help they desperately needed but HbA1c didn’t convey? Or did their doctor(s) brush them off as doing well, just because the HbA1c said they were good?

Not only is there a lot of natural variation in HbA1c vs Average BG, but there are an unfathomable number of things that can further influence the relationship. Iron levels, genetic differences, sleep patterns, stage of life, hemolytic disease, activity levels, elevation…

The CGM manufacturers and programmers can’t possibly know how closely your body aligns with the fabricated statistical HbA1c model. They can only analyze your stats and venture a guess at what your HbA1c would be in a perfect world, if you actually sit on that fictitious line someone drew in the sand.


In addition, A1C can have a 1% variance between testing units used, as A1C equipment only needs to be within +/—0.5% to be in compliance. Most doctor’s offices use tabletop non-certified equipment. Patients then throw GMI with its variances into the mix and claim equivalency between GMI and A1C. We are pretty lucky that all these variances overall still give us a trend and range metric we can use to manage our diabetes.

1 Like

They are 2 different methods, one measuring the estimated glycogen va hemoglobin but can be affected by other physical processes and can be off by 0.7-1%. The other is a more direct measure albeit through interstitial fluid, and that itself has issues that affect its accuracy. One shouldn’t expect them to match.

Some of the factors that influence HbA1c and its measurement - Use of Glycated Haemoglobin (HbA1c) in the Diagnosis of Diabetes Mellitus - NCBI Bookshelf.

I had my A1C done last week, and as you can see it’s unusually low (should have been 5-ish according to 90 day average blood glucose)

For individuals like me (Stage 4 CKD and Anemia), A1C’s do not reflect a 90 day average blood glucose. My red cells don’t survive very long and the new ones don’t have time for the hemoglobin to become glycated.

You could always have your fructosamine tested.

The fructosamine test measures glycated proteins (not glycated hemoglobin) that circulate in the blood for only 14 to 21 days. Measuring these proteins provides a picture of the amount of sugar in your blood for just that time period.

1 Like

Clarity A1C is based on Dexcom CGM readings which read glucose in the interstitial fluid not the blood. These readings are then “translated” into an equivalent blood glucose reading. So Clarity is giving an estimated A1C while the lab is drawing blood and reading the A1C from the blood, the gold standard.

You’re comparing interstitial glucose with blood glucose. They are not always the same and often different.