Hba1c Test results vs Libre, ACnow

[Sax]

I promised to share my Hba1C test results from lab versus personal tests using Libre 2, both the app estimate of Hba1C which is low, and the Libreview "GMI" correlation. Numbers given in bold are actual values, plain text are my conversions between units.

AC1now ______________ 31_____5.0% (19/8; suspected as only 90% of tube blood filled)
Libre 2 app ____________ 29 _____4.8% (est. from blood sugar readings, known to be lower than reality)
Libreview GMI_________ 37_____5.5% (based on 2 week period ending same day as test)
Actual NHS test_________34_____ 5.2%

Recent diet a little carb heavier than usual in the run up to the NHS test so results are in line with expectations. The Libre results are both equally wrong, which is amusing and shows that it probably works best as a monitoring tool for trends.

BTW if anyone is curious, previously Hba1C was 76, or 9.1%, when measured in Q4 2020, and this is with 120g or so daily carbs and 500 mg/ d metformin.

 

[ianpspurs]

I promised to share my Hba1C test results from lab versus personal tests using Libre 2, both the app estimate of Hba1C which is low, and the Libreview "GMI" correlation. Numbers given in bold are actual values, plain text are my conversions between units.

AC1now ______________ 31_____5.0% (19/8; suspected as only 90% of tube blood filled)
Libre 2 app ____________ 29 _____4.8% (est. from blood sugar readings, known to be lower than reality)
Libreview GMI_________ 37_____5.5% (based on 2 week period ending same day as test)
Actual NHS test_________34_____ 5.2%

Recent diet a little carb heavier than usual in the run up to the NHS test so results are in line with expectations. The Libre results are both equally wrong, which is amusing and shows that it probably works best as a monitoring tool for trends.

BTW if anyone is curious, previously Hba1C was 76, or 9.1%, when measured in Q4 2020, and this is with 120g or so daily carbs and 500 mg/ d metformin.

Stunning transformation from 76 to 34. I've seen a few over 6 years here but that is one of the most amazing.

 

[ianf0ster]

I promised to share my Hba1C test results from lab versus personal tests using Libre 2, both the app estimate of Hba1C which is low, and the Libreview "GMI" correlation. Numbers given in bold are actual values, plain text are my conversions between units.

AC1now ______________ 31_____5.0% (19/8; suspected as only 90% of tube blood filled)
Libre 2 app ____________ 29 _____4.8% (est. from blood sugar readings, known to be lower than reality)
Libreview GMI_________ 37_____5.5% (based on 2 week period ending same day as test)
Actual NHS test_________34_____ 5.2%

Recent diet a little carb heavier than usual in the run up to the NHS test so results are in line with expectations. The Libre results are both equally wrong, which is amusing and shows that it probably works best as a monitoring tool for trends.

BTW if anyone is curious, previously Hba1C was 76, or 9.1%, when measured in Q4 2020, and this is with 120g or so daily carbs and 500 mg/ d metformin.

Congratulations, but caution in the light of my thread on sudden BG drop and possibility of permanent eye damage.

 

[LittleGreyCat]

I promised to share my Hba1C test results from lab versus personal tests using Libre 2, both the app estimate of Hba1C which is low, and the Libreview "GMI" correlation. Numbers given in bold are actual values, plain text are my conversions between units.

AC1now ______________ 31_____5.0% (19/8; suspected as only 90% of tube blood filled)
Libre 2 app ____________ 29 _____4.8% (est. from blood sugar readings, known to be lower than reality)
Libreview GMI_________ 37_____5.5% (based on 2 week period ending same day as test)
Actual NHS test_________34_____ 5.2%

Recent diet a little carb heavier than usual in the run up to the NHS test so results are in line with expectations. The Libre results are both equally wrong, which is amusing and shows that it probably works best as a monitoring tool for trends.

BTW if anyone is curious, previously Hba1C was 76, or 9.1%, when measured in Q4 2020, and this is with 120g or so daily carbs and 500 mg/ d metformin.

Just checking - and by the way outstanding results!

How long was your testing baseline for Libre 2/Libreview?
For an accurate comparison it needs to be a full 90 days, as that is the period over which an HbA1c is supposed to measure.

 

[Dandelade]

I promised to share my Hba1C test results from lab versus personal tests using Libre 2, both the app estimate of Hba1C which is low, and the Libreview "GMI" correlation. Numbers given in bold are actual values, plain text are my conversions between units.

AC1now ______________ 31_____5.0% (19/8; suspected as only 90% of tube blood filled)
Libre 2 app ____________ 29 _____4.8% (est. from blood sugar readings, known to be lower than reality)
Libreview GMI_________ 37_____5.5% (based on 2 week period ending same day as test)
Actual NHS test_________34_____ 5.2%

Recent diet a little carb heavier than usual in the run up to the NHS test so results are in line with expectations. The Libre results are both equally wrong, which is amusing and shows that it probably works best as a monitoring tool for trends.

BTW if anyone is curious, previously Hba1C was 76, or 9.1%, when measured in Q4 2020, and this is with 120g or so daily carbs and 500 mg/ d metformin.

Wow, what a result! Interesting fluctuations on the figures too, thanks for sharing.

 

[Antje77]

How long was your testing baseline for Libre 2/Libreview?
For an accurate comparison it needs to be a full 90 days, as that is the period over which an HbA1c is supposed to measure.

Yes, it needs to be a full 90 days, but the results are remarkably similar to what I saw the last 3 times I had my hba1c tested (last time was almost 2 years ago so that was still Libre1), and I did have my full 90 days.

Libre 2 app ____________ 29 _____4.8% (est. from blood sugar readings, known to be lower than reality)
Libreview GMI_________ 37_____5.5% (based on 2 week period ending same day as test)
Actual NHS test_________34_____ 5.2%

(I've left out the a1cnow test because I've never done a home test.)

My Libre app (both for Libre 1 and 2) always estimates my hba1c 8 mmol/l lower than the LibreView GMI, just like yours.
What's different is that the 3 times I have now compared the app's estimated hba1c to a lab test is that lab was also 8 mmol/l higher than than the app.
I'm therefore assuming LibreView is correct in my case, although I hadn't found the GMI on there when I last had a lab test, and algorithms may have changed in between.

edit: And very well done, congratulations on your excellent results!

 

[Sax]

How long was your testing baseline for Libre 2/Libreview?
For an accurate comparison it needs to be a full 90 days, as that is the period over which an HbA1c is supposed to measure.

Sorry should have made that clear it was only 28 days of data - 2 sensors worth, so not completely and absolutely comparable. There was a bit of a gap between sensor 1 and 2 also.
However my diet is very regular at present (not surprising given global circumstances) and I doubt there would be significant difference from doing 6 sensors worth of data - the predictions on both sensors were virtually identical the whole period. The main thing was actually to show how they compared to ACNow test as someone had asked about that the other week right after I'd just trialed it.

On the other topic of actual results, I suspect much of the results is probably due to a short duration between onset and diagnosis and my starting to take action, rather than particular skill in management on my part. I'd been tested previously and had no concerns on blood sugar just 2 years earlier, not even pre-diabetic.

(edit: and of course performance enhancing drugs, in the form of the metformin).

I note the comment about vision. I did take several months to work out where all the carbs in my diet were when I started, so I probably had a more gradual decline than the headline figures might suggest. Other than that I've had optician tests this year, and my 2nd retinopathy screening is coming up in a few months, so unless I go to a private clinic for more checks early I don't see what else I can do - the horse has bolted if I did do go down too fast.

 

[LionChild]

I promised to share my Hba1C test results from lab versus personal tests using Libre 2, both the app estimate of Hba1C which is low, and the Libreview "GMI" correlation. Numbers given in bold are actual values, plain text are my conversions between units.

AC1now ______________ 31_____5.0% (19/8; suspected as only 90% of tube blood filled)
Libre 2 app ____________ 29 _____4.8% (est. from blood sugar readings, known to be lower than reality)
Libreview GMI_________ 37_____5.5% (based on 2 week period ending same day as test)
Actual NHS test_________34_____ 5.2%

Recent diet a little carb heavier than usual in the run up to the NHS test so results are in line with expectations. The Libre results are both equally wrong, which is amusing and shows that it probably works best as a monitoring tool for trends.

BTW if anyone is curious, previously Hba1C was 76, or 9.1%, when measured in Q4 2020, and this is with 120g or so daily carbs and 500 mg/ d metformin.

Very thorough experimentation Sax! Well Done! Well it looks like you are doing very well anyway?
I have found with Libre 2 so far that the readings tend to be higher than the finger pricks, whereas with Libre 1, the readings were lower. I do agree with you though, that Libre is best thought of as indicating trends rather than having a high degree of accuracy in terms of actual scanned readings. Sadly. But it is still very useful I think?

 

[Sax]

My Libre app (both for Libre 1 and 2) always estimates my hba1c 8 mmol/l lower than the LibreView GMI, just like yours.
What's different is that the 3 times I have now compared the app's estimated hba1c to a lab test is that lab was also 8 mmol/l higher than than the app.
I'm therefore assuming LibreView is correct in my case, although I hadn't found the GMI on there when I last had a lab test, and algorithms may have changed in between.

I believe the Libre will tend to under-estimate more than over, due to its origin as a type 1 safety device that would be the safer way to interpret any ambiguity. And this means it will tend to give a reading lower than true for estimated Hba1c values. Abbott suggests the online Libreview GMI is calculated using normalised correlations between the Libre value and "actual" values. In my case the differential in sensor to blood sugar values might have tended to be lower in the sensors I used than the average and so the algorithm overcorrected.

I'm going to have to look into buying more Libre sensors to repeat this test now over 90 days aren't I....

 

[LittleGreyCat]

My Libre consistently under estimates compared to HbA1c.
Also LibreView, although giving a higher figure, are also low.

Example:
Libre 6.1%
LibreView 6.4%
HbA1c 7%

Edit: but Brexit alone knows when I will next get a blood test!

 

[Chris181]

I’m not sure these measurements are comparable. According to the calculator on this site, an Hbaic of 6.0 (42 mmol/mol) is equivalent to an average blood sugar reading of 7 mmol/L. They are measuring different things (glycated haemoglobin vs actual glucose). The Libre is measuring something different yet again ( the glucose in interstitial fluid…not the blood). In my experience the Libre system measures 20% lower than finger pricking…leading to false hypo alerts. Your results may vary!

 

[Ronancastled]

I'm watching this thread like a hawk as I'd a similar issue with the Libre Link App quoting 4.6% & the Libre View GMI quoting 5.2%. which gave me the hump although the algorithm must be based on solid mathematics.
My thread on it is here https://www.diabetes.co.uk/forum/th...ed-a1c-on-libreview-vs-libre-link-app.182614/
What I find weird is that in the Dexcom non-diabetic trial their average CGM glucose 99mg/dl exactly matched their average A1c 5.1%.
https://www.diabetes.co.uk/forum/th...ine-normal-glucose-the-results-are-in.182221/
Does that mean the Dexcom is way more accurate than the Libre or is there something more at play.

I did a bit of research & found that the GMI was introduced due to the large variations seen between average glucose of T1s & their actual A1c.
Perhaps the larger deviations in diabetics play a part, no matter how well controlled their BG is.

 

[MrsA2]

I don't know any of the answers or reasons, but for 10 days now, after my finger prick test on waking I then scan my libre on my phone 15 minutes later.
Most days the libre is between .6 and 1 lower than the finger prick and I just factor this in when using the Libre for the rest of the day (ie its low aren't that low and it under reports the highs)
BUT this morning both were exactly 6.4. They have never ever agreed before and I find this spooky

 

[Ronancastled]

BUT this morning both were exactly 6.4. They have never ever agreed before and I find this spooky

I found my lows were lower on the Libre but my highs were also higher even allowing for the 15 min lag.
Also, if I didn't scan post meal until say 3 hours postprandial my spikes were less than when I constantly scanned.
I found scanning after eating carbs things were dynamic & the Libre was trying to estimate based on it's last reading plus some interpolation based on trajectory.
Often a high scan would be an outlier on the Libre View record.

 

[Sax]

I’m not sure these measurements are comparable. According to the calculator on this site, an Hbaic of 6.0 (42 mmol/mol) is equivalent to an average blood sugar reading of 7 mmol/L. They are measuring different things (glycated haemoglobin vs actual glucose). The Libre is measuring something different yet again ( the glucose in interstitial fluid…not the blood). In my experience the Libre system measures 20% lower than finger pricking…leading to false hypo alerts. Your results may vary!

The Libre app and Libreview online tool can give estimates of Hba1c/GMI values in mmol/mol and % which is what I was quoting - not the blood sugar readings which I appreciate are different units.

 

[Sax]

Going to buy 3 months of meters and repeat this test, although I might have some unavoidable gaps in the period as I might not be able to wear the libre for visiting some work sites.