Libre view report estimated A1c calculation appears to be wrong.

[CBeresford]

Hi I recently had a HBA1c blood test with a result of 32 mmol/mol. I had the test done privately as my practise only offers the test annually it was 44 when tested by the surgery in March. I spoke to my Dr yesterday and agreed to send a copy of the report to my Dr for my file along with my libre 2 report. My average glucose with libre 2 has been 5.7 over the last 3 months giving me an estimated HBA1C of 5.2% or 33 mmol/ mol. However my libreview report calculates my average glucose level of 5.7 to be an estimated A1c of 5.8% 40 mmol/ mol. My understanding is that A1c is the same as HBA1c. I am reluctant to send this report to my Dr as it is so different to my actual HBA1c.

I am being really stupid and missing something obvious, or is the calculation in libre view wrong. My libre app shows the correct calculation of average blood glucose of 5.7 being 5.2% or 33 mmol.

 

[harbottle]

The GMI is a different calculation and isn't the same as estimating a hba1c. I think it uses shorter time spans.

 

[CBeresford]

The GMI is a different calculation and isn't the same as estimating a hba1c. I think it uses shorter time spans.

Looking at studies regarding this the variation between GMI and A1c has been shown to be up 0.9. In my case it appears to be 8.

 

[Lucyr]

The clue to why they’re different is in the word “estimated”

 

[helli]

The Estimated result Libre provides is 10 lower than my blood test.
Therefore, I consider it a useful alternative thing to track between HBA1C blood tests but not the same thing.

 

[CBeresford]

The clue to why they’re different is in the word “estimated”

The variation between estimated and actual in this case is way out, the acceptable variation is up to 0.9 mmol/mol in this case it is 8 mmol/mol. I am happy that my HBA1c is 32, and I can accept an estimate of 33 which the app suggests, but why does libre app think it is 33 and libre view think it is 40. It really doesn’t make sense.

 

[Robin]

Looking at studies regarding this the variation between GMI and A1c has been shown to be up 0.9. In my case it appears to be 8.

What measurement system is the 0.9 difference using? If it’s using the old percentage system, where your HbA1c was something like 6.5%, this equates to 48mmol/mol under the new system. A discrepancy of 0.9% probably equates to 8 in the new system. I can’t believe that the discrepancy was only 0.9 if measured in mmol/mol, that’s tiny, and my estimate and actual HbA1c has never been that close.(it’s usually at least 2 or 3mmol/mol)

 

[harbottle]

Looking at studies regarding this the variation between GMI and A1c has been shown to be up 0.9. In my case it appears to be 8.

Not sure what the issue is.

The report is using GMI, which is not going to be the same as the estimated hba1c and the GMI can be a lot different - mine was 4 higher than hba1c when I last looked at the reports on the Contour website (Which is where my data is uploaded from my finger pricks) and estimated hba1c was 33 from the same data in some random app I'm using that also records the data.

Not sure why you are worrying when the hba1c from a lab was 32!

 

[CBeresford]

What measurement system is the 0.9 difference using? If it’s using the old percentage system, where your HbA1c was something like 6.5%, this equates to 48mmol/mol under the new system. A discrepancy of 0.9% probably equates to 8 in the new system. I can’t believe that the discrepancy was only 0.9 if measured in mmol/mol, that’s tiny, and my estimate and actual HbA1c has never been that close.(it’s usually at least 2 or 3mmol/mol)

I’ve just read that studies have shown that for 80% of people the difference between GMI calculated A1c and average glucose calculated A1c is 0.5%. Which is considered acceptable, mine is 20% different. Is anybody out there
using libre 2 who can tell me if their estimate is equally inaccurate.

 

[harbottle]

The variation between estimated and actual in this case is way out, the acceptable variation is up to 0.9 mmol/mol in this case it is 8 mmol/mol. I am happy that my HBA1c is 32, and I can accept an estimate of 33 which the app suggests, but why does libre app think it is 33 and libre view think it is 40. It really doesn’t make sense.

The GMI (the 5.7) estimate uses 14 days of data to estimate an a1c
The estimated hba1c (5.2) is using a lot more data.
They are different measurements and can differ by a lot more than .9

GMI is used to look at short term 'mean glucose' as a 3 month measurement may be masking a lot of highs or a lot of lows. Using a smaller window can capture that sort of info.

 

[harbottle]

From

What Is The Glucose Management Indicator (GMI)?

What is GMI in diabetes? Learn more about another important metric that can help with diabetes management.

healthmatch.io

• 50% of patients had differences of 0.5% or less. For example, if the GMI was 6%, the A1C ranged from 5.5% to 6.5%.
• 11% of patients had differences of 0.1% or less. For example, if the GMI was 6%, the A1C was between 5.9% and 6.1%.
• 22% of patients had a difference of 1% or greater. For example, if the GMI was 6%, the A1C was below 5% or greater than 7%.

This shows quite a difference between GMI and A1C, far more than .9!

I'm not sure why you are worried about it, as the GMI is more aimed at people who have to inject and it's a means of determining what day to day control is like.

 

[rebrascora]

For many of us the predicted HbA1c on LibreView is significantly lower than actual blood result. Quite surprised yours is higher and no idea why there is a discrepancy between LibreLink and Libreview which is I think what you are saying. How much data is your LibreLink result based on.... bearing in mind we had an update recently which wiped a lot of data.

 

[CBeresford]

From

What Is The Glucose Management Indicator (GMI)?

What is GMI in diabetes? Learn more about another important metric that can help with diabetes management.

healthmatch.io

• 50% of patients had differences of 0.5% or less. For example, if the GMI was 6%, the A1C ranged from 5.5% to 6.5%.
• 11% of patients had differences of 0.1% or less. For example, if the GMI was 6%, the A1C was between 5.9% and 6.1%.
• 22% of patients had a difference of 1% or greater. For example, if the GMI was 6%, the A1C was below 5% or greater than 7%.

This shows quite a difference between GMI and A1C, far more than .9!

I'm not sure why you are worried about it, as the GMI is more aimed at people who have to inject and it's a means of determining what day to day control is like.

I suppose I am worried because my difference is 20%.

 

[harbottle]

I suppose I am worried because my difference is 20%.

Just ignore these estimates.
My GMI was 4 higher than actual hba1c
MyShugr was estimating 3 lower than actual hba1c.
They're both estimates based on finger prick data.
The lab result was slap bang in the middle of them both.

 

[CBeresford]

For many of us the predicted HbA1c on LibreView is significantly lower than actual blood result. Quite surprised yours is higher and no idea why there is a discrepancy between LibreLink and Libreview which is I think what you are saying. How much data is your LibreLink result based on.... bearing in mind we had an update recently which wiped a lot of data.

I took a snap shot of my average glucose and estimated HBA1c on my phone back in June, before deleting the app. There is a huge discrepancy between libre app and libre view, 20% difference between my estimated HBA1C on the
app compared with the libre view report.

 

[rebrascora]

Oh and congrats on a great HbA1c result!

Out of curiosity, how was the private test done? Was it a spot of blood or did you attend somewhere to get blood drawn? I do wonder about the validity of the spot blood tests.

 

[helli]

Is anybody out there
using libre 2 who can tell me if their estimate is equally inaccurate.

Yep! Mine is equally inaccurate.
The Estimated value reported on LibreLink is consistently over 20% different from my blood test HBa1C

And I don't care!

I know they are measuring different things. I care that my HBA1c is as low as possible without experiencing too many hypos and it is not rising.

 

[CBeresford]

Oh and congrats on a great HbA1c result!

Out of curiosity, how was the private test done? Was it a spot of blood or did you attend somewhere to get blood drawn? I do wonder about the validity of the spot blood tests.

It is not a spot blood test. The lancet is a blade rather than a needle and you drip blood into a pink topped blood collection bottle about half the size of the routine vacutainer bottles used in the surgery, you have to fill the bottle to a pre marked line, the bottle is laced with anticoagulant (the same anticoagulant as in the pink and purple top bottles). It doesn’t need a huge blood sample to do the test. The sample you give in the surgery is often divided up into smaller amounts for several tests such as a full blood count, as well as HBA1c. But thanks for casting doubt on my HBA1C test and the validity of Boots online laboratory. As included in the attachments my average glucose over the past 3 months has been 5.7 mmol/mol which equates to a HBA1c of 32/33 which is what my HBA1c says it is, it is what the libre app says it is, but not the libre view report for some reason.

 

[Robin]

I took a snap shot of my average glucose and estimated HBA1c on my phone back in June, before deleting the app. There is a huge discrepancy between libre app and libre view, 20% difference between my estimated HBA1C on the
app compared with the libre view report.

If you’re asking why the discrepancy between what Librelink and Libreview show, are you sure they are using the same data? The screen shot you’ve posted shows it’s using '33 of 90 days, so one months worth, is the Libreview the same?

 

[CBeresford]

Yep! Mine is equally inaccurate.
The Estimated value reported on LibreLink is consistently over 20% different from my blood test HBa1C

And I don't care!

I know they are measuring different things. I care that my HBA1c is as low as possible without experiencing too many hypos and it is not rising.

I am just worried about sending the report to my Dr. I reckoned everyone must be in the same boat if an average blood glucose of 5.7, is calculated as 5.8 % A1c 40 mmol/mol on their libre view report.