Libre (in)accuracy on HBA1C

[searley]

Hi!

Now I'm not talking about trying to get my finger prick to match my scan, as they are measuring different things that respond at different times etc etc

BUT the difference between my Libreview HBA1C and the hospital blood test has been consistently very different, latest being 32 vs 44.

The higher number is the blood test obviously! This equates to approximately 1mmol/l difference in all my readings on average.

Has anyone else made this comparison?

Obviously there are several factors to consider...

1, blood glucose meters are not 100% accurate
2, Libre is not accurate
3, most software like libreview only estimate hba1c over a short period, my last discussion with the consultant told me not to trust the libre view or decom clarity for this reason, more use it as an indication of whether the results are going up or down
4, hba1c results also have a margin of error, i know someone had 2 hba1c in different labs in 1 day and had differing results

these for factors above alone could make quite a difference in the hba1c results

i use the dexcom which I find much more accurate than the libre, and i run it through night scout which the reporter is quite good over a 3 month period.. my estimated hba1c was 51, actual was 47.. so i can imagine the libre being slightly further out than that.

either way the hba1c is still good, and im sure the libre has helped you achieve that... and as time goes by you'll manage better in what to expect from the libre.. and hopefully when the libre 3 comes it will be slightly more accurate as they are aiming for iCGM approval

 

[LittleGreyCat]

most software like libreview only estimate hba1c over a short period

Are you sure?
LibreView and LibreLink claim to use data from the last 90 days which is pretty close to 3 months.

 

[searley]

Are you sure?
LibreView and LibreLink claim to use data from the last 90 days which is pretty close to 3 months.

they claim that.. but the algorithms aren't great.. so the Futher away from 4 weeks the less accurate (apparently) just what my consultant says.. in reality the hba1c is slightly biased towards the current month less so by the following 60 days which is why 1 good month before an hba1c can have a significant effect

most software calculate the hba1c by averaging the last 3 months... where as a some software like nightscout and xdrip bias there calculations to closely resemble a true hba1c... this is why some software will give better results than others. I'm typically running 2 or 3 apps which calculate hba1c from my dexcom all giving different results because the algorithms they use are different... the closest for me has always been nightscout and xdrip

either way they are still good enough to give you an impression of how you are doing and whether getting better or worse.. just don't expect total accuracy

 

[searley]

they claim that.. but the algorithms aren't great.. so the Futher away from 4 weeks the less accurate (apparently) just what my consultant says.. in reality the hba1c is slightly biased towards the current month less so by the following 60 days which is why 1 good month before an hba1c can have a significant effect

most software calculate the hba1c by averaging the last 3 months... where as a some software like nightscout and xdrip bias there calculations to closely resemble a true hba1c... this is why some software will give better results than others. I'm typically running 2 or 3 apps which calculate hba1c from my dexcom all giving different results because the algorithms they use are different... the closest for me has always been nightscout and xdrip

either way they are still good enough to give you an impression of how you are doing and whether getting better or worse.. just don't expect total accuracy

infact attached is an extract from my consultant review where I stated my hba1c reports it should be 48 to 51... he thinks I'm using standards software.... he blatantly stated the software is only good for 2 week accuracy so in the letter stated 2week results

I use nightscout and hba1c was 47

 

[Mr_Pot]

Are you sure?
LibreView and LibreLink claim to use data from the last 90 days which is pretty close to 3 months.

I got an estimate of my HbA1c and I only had the one sensor, so 14 days of results. Surely Libre calculates an average BG as soon as it has enough data (7days?) and then refines it as more data comes in. The estimated HbA1c will just be a "conversion" from the average BG, I can't see the relevance of 90 days since it is BG that is being measured not HbA1c.

 

[searley]

I got an estimate of my HbA1c and I only had the one sensor, so 14 days of results. Surely Libre calculates an average BG as soon as it has enough data (7days?) and then refines it as more data comes in. The estimated HbA1c will just be a "conversion" from the average BG, I can't see the relevance of 90 days since it is BG that is being measured not HbA1c.

the minimum for hba1c is 14 days in libre.. it's an estimate from the average bg as they are obviously closely tied together... some software estimate better than others... but will be in the right 'ball park'

 

[Mr_Pot]

the minimum for hba1c is 14 days in libre.. it's an estimate from the average bg as they are obviously closely tied together... some software estimate better than others... but will be in the right 'ball park'

Have you compared the HbA1c estimate with the average BG to see if the conversion is the same as standard tables or this calculator?
https://www.diabetes.co.uk/hba1c-to-blood-sugar-level-converter.html
The clever part with algorithms will be the calculation of average BG from the data.

 

[jape]

The key word here is "ESTIMATED". My estimated hba1c at the moment is 5.7% (39), but I expect it at my next bloodtest to be somewhere between 6.2 to 6.4.

Maybe the formula to calculate hba1c from average blood glucose readings is not exactly correct, or the rate at which red blood cells are dying is not correctly reflected in the formula, or varies between individuals???

 

[Fenn]

Libre 51
Hba1c 61

booooo!

 

[StewM]

Hi!

Now I'm not talking about trying to get my finger prick to match my scan, as they are measuring different things that respond at different times etc etc

BUT the difference between my Libreview HBA1C and the hospital blood test has been consistently very different, latest being 32 vs 44.

The higher number is the blood test obviously! This equates to approximately 1mmol/l difference in all my readings on average.

Has anyone else made this comparison?

Whilst the points other major have made are very pertinent, I’ve found every HBA1C I’ve had has been within 1mmol/mol of my Libres estimate. I’ll be interested to see if my next one is, as the last three Libres I’ve had have been much more inaccurate than most of my other Sensors, e.g. claiming I was below 3.5 for four hours in which my Blood Sugar never dropped below 5 on a Blood Tests.