Unexpected HbA1c result

I had a blood test at my end appointment last week and got the results back today.

My numbers have been really good recently, TIR 97% (70-180), the Libre app showed an average blood glucose of 110 mg/dL and I’ve feel like I’ve been in a really good groove managing my levels.

My HbA1c at the end of June was 6.2%.

The FreeStyle Libre app predicted 5.6% on the day of my test last week. I know from experience that this isn’t accurate, but can give a sense of the direction it’s going. In June, the Libre prediction was 0.3% lower than my actual blood test result.

I was confident it would be lower than June.

Well, it was a bit of a gut punch to find it’s now 6.7%.

So what gives?

  • I’ve been eating more carbs recently (about 120g carbs a day rather than 100g carbs), but I adjust my insulin for the extra carbs
  • In the few weeks leading up to my test, I’d been ‘experimenting’ a bit bolusing for veggie burgers, fries and ice cream (there have been several autumn festivals in my town so we’ve had food truck meals and I’m still trying to figure out how to bolus for them)
  • I’ve been busy with work, so I haven’t been walking as much (I usually walk about 5 hours a week, but recently it’s been more like 2-3 hours a week)
  • I haven’t been sick in the last few months and besides a few veggie burger, fries and ice cream highs (all short-lived, one up to 280), I haven’t had unusually long or unexplainable highs
  • I was on my period on the day of the test (I’m not sure if this is relevant at all, but, ya’ know… it’s blood loss so maybe it affects HbA1c)
  • I usually do a few finger pricks at the start of a new sensor, then only if I’m low or feel different than the sensor is saying

Is there anything I should consider changing or paying attention to?

I read that HbA1c is more affected by recent highs and lows. Could the food truck highs be the thing that soiled my test result? If so, at least it was delicious. :laughing:

I’m not too worried about the number, but it went the opposite direction from what I was expecting, so it really threw me off.

GMI and HbA1c are not even close to being the same thing. One is a measure of the percentage of glucsated hemoglobin type A subtype 1c, The other is based on Blood glucose readings from the CGM. There is an attempt to make GMI correlate to HbA1c, but factors can skew the results.

This snip is from a paper:

“during short periods of acute hyperglycemia (illness, steroid administration, diabetic ketoacidosis), the average glucose and thus the GMI will be higher than a laboratory A1C measured at the same time, as the laboratory value reflects glucose levels primarily over the last 2–3 months. The opposite also can occur, where the GMI is lower than the laboratory A1C if there are short periods of much lower-than-usual glucose readings (starting a new carbohydrate-reduced diet, an intensive exercise regimen, or during the first few weeks after starting a new effective glucose-lowering medication).”

The whole thing is pretty interesting.


The last month right before the A1c test influences the test the most and I bet the Libre uses a basic 3 month average. I know I go with the Dexcom blood sugar average of 30 days to get a better idea of what I am at A1c wise versus their GMI number. I calibrate my Dexcom though, you would need to guess at a basic number to add to the Libre blood sugar average.

I used to add numbers to my Libre for dosing etc, but each sensor could be different, one might be off 25 points a lot and another only 10. I’m not sure if you check for variations, but if you weren’t adding maybe enough to the numbers, especially the last couple of sensors, that would also be influencing your actual numbers.

I eat whatever carbs I want and I’m a vegan. My last A1c was 5.1%. The carbs themselves aren’t the problem so much as if you are not used to it and staying higher longer, that would influence the numbers. Exercise helps you absorb insulin better, but that changes how much insulin you need or how long you stay high.

But just a few higher numbers for longer would influence your A1c, especially if it’s been in the last 30 days before the test, 280 for 1 hour means you have to be 100 for 5 hours to basically have an average of 130. And while you might have not been at 280 for a full hour, you don’t instantly come down to 100, it takes a little bit of time. So a little more of odds and ends of that could bring up your average.

So a recent history of experimenting with new foods could have made a difference, I’m not saying that’s bad. That’s the only way I figure out guessing at certain foods like taro donuts. We have a lot of food trucks where I’m at and a lot carry 1 or 2 vegan items, so I am always guessing. This is a constant experiment with various changes we have happen all the time. I’m just saying it could explain the differences.


Thanks Carlos.

The Libre 2 app calls it HbA1c, which I realise now is quite misleading, but I noticed the Libre 3 app calls it GMI. I guess I need to start thinking of the number in Libre 2 as being completely different from the HbA1c blood test.

1 Like

Thanks for this, Marie. This is what I thought might be the case, but it’s good to have someone confirm it.

I’ll spend more time working out individual sensor calibrations. From the reading I take at the start of each sensor and when I’m low, they seem to be within 10 mg/dL, but maybe I need to check more mid-lifespan to the end of the sensor to see if it stays consistent.

My veggie burger and fries highs were all short lived (going up quite rapidly, spending about an hour at their peaks, then slowly coming down), but as you said, a little time spent that high takes a lot of time spent lower to even out the average.

Yup. Every time I was high I knew I was getting information, so it wasn’t all bad. And it means I need to run another test of food truck fries to gather even more data. :wink: I just didn’t realise it would affect my HbA1c this much, if this is what it is.

It’s hard to know what’s in those veggie burgers - is it bean? just grilled veggies? what’s in the sauce? how many carbs is that bun?

And eyeballing the fries is an adventure - we live just a few houses from the truck where we get fries and once I brought my kitchen scale to see how much it weighed (I know it came to about 60g of carbs) but do they always give us the same amount?

For ice cream, I think I’ve found what works, but who know if that will work every time. Luckily, ice cream season is over so I don’t need to worry about figuring that out until next year. :laughing:


It took me about 4-6 tries to figure out taro donuts…

Luckily that truck is 30 miles away lol…


Also…the LIbre will be closer at the lower numbers, higher numbers it will be off potentially a lot more.


I know it’s only been one day and one sensor, but I did several finger pricks today at different levels (everything from 90 to 200 thanks to a delicious pretzel) and at different times of day when I had been steady for a while. Libre was always within 5-10 of my finger prick and Libre tended to be a bit higher than finger prick.


CGMs are simply not accurate; 6.2% is a fantastically good number for a T1, don’t diss it.

My Dexcom G6 quite consistently shows readings that are 20mg/dL or more higher than a fingerstick; that would give me an HbA1C which is also much higher. There is no way in a million years that any sane individual world rely on the guesstimates for HbA1c/GMI of a CGM. Because the CGMs are consistently wrong so, because of the consistency, they give estimates that are bogus.

Buy an HbA1c test kit. Trust that more than anything from a CGM.


Some additional info on why A1C and GMI can be different.

1 Like

I just had ANOTHER conversation with Liam’s endo about his “lows” during the last virtual visit. He only has 1% low and 2% severe lows even with the incorrect CGM data being used for those percentages. I tell her what I’ve always told her…there is no way you can recommend any tweaks based on CGM data. There is just no way. I told her what she should do, if she wants a true low percentage, is to go through Nightscout and gather all the TRUE BGs from the finger sticks that we do any time his CGM is trending downward and/or reading “low”.

Anytime we check his BGs we enter it into Nightscout so that era have a complete record of hour crazy far off his CGM is all the time. The CGM is really only good to see how his BGs are trending but not much else tbh.

I also tell them there is no way you are going to look at my sons numbers for 5 minutes and try to recommend any changes to me. I eat, sleep and breathe Liam’s Diabetes management. When I see PATTERNS, I can and do make adjustments. I refuse to tweak based on their two minute review of a snapshot of data that most regularly wrong.


Thanks. 6.2% was my HbA1c in June. The unexpected one was last week when it went up to 6.7%. I’m still happy with it, it was just a bit of a shock since I went in to the test with different expectations.

Everything in the Libre app suggested I was getting better control and I haven’t noticed sensors being off by more than 5-20 (my current sensor reads 5-10 higher than a finger prick), so I was certain it would be low, but it went in the opposite direction, which is confusing.

I’m aware CGMs are not accurate, but I’d hope I could trust them at least for the direction my reading are heading. Since it predicted a lower HbA1c/GMI than at my last test, that’s what I was expecting.

At my test in June, Libre predicted 5.9% at my test in June, which came back at 6.2%. Just 0.3% higher.

On the day of my test last week, Libre predicted 5.6% (I knew this wasn’t going to be correct, but led me to think my HbA1c would be lower than June’s) and the blood test came back at 6.7%, a whole 1.1% higher.

I now realised that the predicted HbA1c/GMI can’t be used as a forecast for my actual HbA1c. It’s just frustrating that’s it’s labeled as HbA1c in the Libre 2 app. I see they’ve relabeled it as GMI in the Libre 3 app, but I still won’t use it to base any assumption on.



I guess I’m not the only one who found Libre’s predicted HbA1c/GMI wildly off.

22% of people had a difference of 1 percentage point or greater, meaning if GMI was 7%, actual A1C was under 6% or over 8%.


I haven’t noticed Libre sensors to be that off for me, maybe 5-10 on most checks. But I only really check on Day 1, when I’m low or when I feel different to what the app is telling me.

When my endo reviewed my data last week, I got high praise for my 97% TIR (70-180) and average blood glucose of 110 mg/dL. So to get a higher HbA1c than I had in June was pretty unexpected. It’s a bit frustrating (not the result, but the expectation I went in with) and has got me doubting my Libre more since getting the result.

I think you’ve all heard me comment previously, “A man with one watch knows what time it is, a man with two watches is never sure.” All testing, including lab performed tests, have a margin of error, be they lab benchmarks, finger pricks (vary greatly by the meter), CGM, or whatever. The errors can be magnified by the tester’s actions and test materials: how was the blood draw done, what was the time period between draw and test, how was the blood preserved, what reagents were used, what machines were involved, what temperatures were maintained, etc. Not to mention environmental factors of what meds a person was on, what illness/medical conditions the person has or had, as pointed out by the writer in @MM2’s reference article. But even the article has a fallacy, the writer keeps referring to the “actual” A1c; “actual” based on what, a lab test perceived as accurate? I once had a lab test A1c of 5.2, that’s .7 below anything previously recorded; both the Doc and I figured it was an “lab test error” or anomaly because absolutely nothing else indicated it could be accurate, not diet, not exercise, not finger sticks, not the CGM numbers, no other data source supported the conclusion. But, we took it as a “data point”, albeit probably a skewed one, among all the others. Each of the tests has its uses, based on the phenomenology used, method of employment, etc., but all are still only an “approximation” or “indicator” of what is “actual.” Each of us has our preferred information source. I tend to like GMI and TIR, but still put A1c into the mix, all as indicators of how I’m doing. One number isn’t the focus except over the short run, and we’re playing a long game. Is everything pointing higher, lower, or about the same. All that said, I still get “excitable” when I get above 180 and can’t seem to bring it down…like night before last when I hit 218…but in the long game, its a datapoint…not a big deal unless its a consistent issue. Like I told my grandson yesterday, “Use what works for you and do your best!”


We were also higher based on labs we had done for Liam. He has been 5.7 for the past year or two and he is now 6.2. I am ok with that as well because Liam is getting more challenging to manage as he’s beginning to be so much more sporadically active (as any 8 year old is, chasing his brother around and such.). 6.2 (anything under 7 tbh) is very good, so I’m not sweating the small stuff.


I had this same thought. I’ve got two identical glucometers and when I got the second one, I tested them against each others. They were pretty close, but how do I know they’re not both wrong?

My GP’s glucometer seems to read about 20 points lower than what I expect (I compare her reading with my Libre reading in the appointment and her reading is always a lot lower than Libre).

I’ve been paying attention to the TIR, and when the predicted HbA1c/GMI ticks up or down a notch, but this unexpected HbA1c has got me doubting my app’s claim of 97% TIR, even though the occasional finger pricks seem to confirm my sensors are pretty accurate.



Half a percentage point certainly isn’t huge and I’m still under 7, but it is annoying when I was expecting it to go down a tad rather than up half a point.

I’m looking forward to seeing what my next test comes back as. No ‘experimenting’ with food trucks in the lead up to my appointment this time. :laughing:


@Finn, do they also do a BG check from the blood draw when you get your A1C? And I don’t mean just a BG check from their meter, but one where they send it to the lab.

If you want to check your meter, do a BG check at the same time they do the blood draw. Write it down and wait for the lab results.

It won’t be exactly the same as the lab value because our BG is not always exactly the same at all places, like capillaries, veins, arteries, etc. There can be some small variance there.

But it should be very close.

That’s a good way to check your meter.


Unfortunately not. I checked my results and the blood they draw for the test gives HbA1c and several other things, but not blood glucose.

I’ll ask at my next appointment how I can be sure my glucometers are correct. Maybe she can request the lab test the sample for blood glucose.

I know CGMs are less accurate when you’re high or low, but do glucometers have issues like that or are they pretty consistent?

1 Like