• Guest - w'd love to know what you think about the forum! Take the 2025 Survey »

Explaining Libre discrepancies???

barrym

barrym

Well-Known Member
Messages
829
Location
North Wiltshire
Type of diabetes
LADA
Treatment type
Insulin
I know everyone has there own story about the 'inaccuracies' of Libre compared to finger prick tests, and I've become a defender of the product as in my view, the differences are immaterial in so much as they don't really change the action you would take.

In my experience the Libre is generally close to a meter in-range and lower than a low meter or higher than a high meter.

I was sure this was down to Abbott's design intentions. This opinion partly because some of the third party sensor apps do seem to be closer to a meter which just slavishly analyses blood.

I'm about to use some ugly figures now, but don't get distracted.

I had a bad evening and when I tested in the early hours of this morning I was 12.2 on the LibreLink app. I took the appropriate action and all was well this morning.

I need to explain what I use.

I scan the sensor with Libre Link. This uploads automatically to Libre View. I have a Diasend account which logs in to the Libre View account and hoovers up the 15 minute CGM readings for my HCPs to view through their Diasend suite.

On my Phone I have a Diasend App too which will look at the Diasend data. I took a look and noticed that there was a spike at 2am as I mentioned but it wasn't as high as 12.2! Hmm. I haven't done so, but I bet if I exported from Libre View the data would show the same.

I'm convinced that what the sensor sees and calculates for the scanner software is much closer to a meter reading, but that Abbott want to help us <?> by emphasising direction of change and encourage the appropriate treatment.

Whether that is true, or whether you want it or not, I believe it does show that the whole interstitial method of testing IS reliable, and for those that insist on wanting matching numbers with a meter, why I don't know, then switching scanning app, or looking at the raw data on LibreView would be a way forward.

I look forward to comments on my theory as I could well be totally wrong, but from experience and what I noted this morning, and have seen before but not realised the significance, I reckon that Libre IS accurate.
 
Well, when you consider that a Libre sits in the same place, and is relatively undisturbed for two weeks, compared to a blood meter that actually has many variables:

Location of finger stab
Temperature of meter
Temperature of strips
Dust and bits in the strip port that could affect conductivity and therefore reading
Cleanliness of hands
Age of strips
Battery life remaining on meter
User error
Reusing lancet - could introduce dirt/sugar into the sample

Plus a load more that I imagine I can’t think of...

Why do we assume it’s the blood meter that’s the accurate one? I only fingerprick to drive now, I’m more than happy to bolus from my Libre.
 
Am also new to the forum so hi there!. (think am yet to introduce myself on the site, yet already posted and replied to threads).

I have never quite ever trusted electronic gadgets apart from digital radios, so not at all surprised that you are getting conflicting readings from the Libre meter, even though never used one myself, have extensive knowledge of digitalised technology itself. If the readings are off the Richter-Scale, then absolutely essential that you contact the manufacturer is essential as well as reporting it to your GP will allow them to review and improve the performance of the Libre or invent a new one - if there are already there isn't a new model on the market right now due to complaints about the product.

Have read online that 'FGM and CGM are reading interstitial fluid, not blood so there are likely to be discrepancies here and there' can and does occur may also significantly affect blood-sugar level results, just something to consider in advance, yet have no idea what are the ideal ratio interstitial fluid are supposed to be. Blood sugars themselves can also fluctuate rapidly throughout the day, regardless of what lifestyle choices and diets you might follow if you are stressed or suffer with anxiety conditions. My GP is aware that I suffer with high levels of stress and it always affects the results of my blood-pressure readings, so goes to show you that the human condition is very complex indeed and deceive programmed tech rather astutely.
 
barrym said:
I know everyone has there own story about the 'inaccuracies' of Libre compared to finger prick tests, and I've become a defender of the product as in my view, the differences are immaterial in so much as they don't really change the action you would take.

In my experience the Libre is generally close to a meter in-range and lower than a low meter or higher than a high meter.

I was sure this was down to Abbott's design intentions. This opinion partly because some of the third party sensor apps do seem to be closer to a meter which just slavishly analyses blood.

I'm about to use some ugly figures now, but don't get distracted.

I had a bad evening and when I tested in the early hours of this morning I was 12.2 on the LibreLink app. I took the appropriate action and all was well this morning.

I need to explain what I use.

I scan the sensor with Libre Link. This uploads automatically to Libre View. I have a Diasend account which logs in to the Libre View account and hoovers up the 15 minute CGM readings for my HCPs to view through their Diasend suite.

On my Phone I have a Diasend App too which will look at the Diasend data. I took a look and noticed that there was a spike at 2am as I mentioned but it wasn't as high as 12.2! Hmm. I haven't done so, but I bet if I exported from Libre View the data would show the same.

I'm convinced that what the sensor sees and calculates for the scanner software is much closer to a meter reading, but that Abbott want to help us <?> by emphasising direction of change and encourage the appropriate treatment.

Whether that is true, or whether you want it or not, I believe it does show that the whole interstitial method of testing IS reliable, and for those that insist on wanting matching numbers with a meter, why I don't know, then switching scanning app, or looking at the raw data on LibreView would be a way forward.

I look forward to comments on my theory as I could well be totally wrong, but from experience and what I noted this morning, and have seen before but not realised the significance, I reckon that Libre IS accurate.
Click to expand...

Tend to agree, Barry. There's a few of us using libre with a blucon transmitter on top to send 5 min readings to the android app xdrip+, primarily so we can calibrate to improve accuracy, and get hypo alerts, and there seems to be broad agreement that the official libre apps tend to overexaggerate/amplify lows and highs, whereas xdrip+ is less of a drama queen about it all.

I've not approached it systematically by logging results, but for a while after getting blucon/xdrip+, I spent a bit of time both scanning with the reader, and looking at xdrip+, and bg testing. xdrip+ was 9 times out 10 way closer to the bg test. I've read about glimp users reporting similar things - glimp allows for calibration too.

I'm not a tech guy, but, just from my impressions as a long term user, I reckon two things are going on. First, Abbott's factory calibration, compared to Dexcom's user calibration from bg tests, isn't as good as they hoped it would be but can be tightened up a lot by calibrating through glimp or blucon/xdrip+. Second, their algo exaggerates stuff when out of range.

So, basically, a sound sensor, but can be improved with some third party tweaks.
 
Well this is weird. I thought I'd actually look at the data stored at LibreView and that at Diasend. There is a difference. LibreView only seems to store (well export) scanned readings not the 15 minute interval ones, whereas the Diasend 'hoover' pics them all up. So the difference is what we see on the screen, Libre View doesn't seem to show actual 15 minute values, just the 'manipulated ones'.

So definitely looks like there's a place for the third-party scanners. I just wish I didn't have to populate Diasend for my HCPs. A small price to pay though for keeping them happy:-)
 
Hi @barrym - there is a difference in what is recorded with a scan and what is captured on the longer term data.

If you look at a scan, it can be significantly different from a glucose meter and historic data for a number of reasons.
  1. The scanned data includes a predictive algorithm which estimates what your blood glucose would be now based on the previous fifteen minutes of individual readings. It's not exactly what the raw data is. Pierre here (http://type1tennis.blogspot.co.uk/search/label/Freestyle Libre) has written a lot about the difference.
  2. The historic data (apart from when you scan) is "averaged" every fifteen minutes to give a glucose data value. As that is an interpolation of the previous fifteen minutes data, it is not exactly what the scanner has predicted when you've scanned (you see this happens quite frequently with spikes).
So yes, there are differences, and they are fully explainable.
 
I'm just using the scanner with the sensor and find it surprisingly accurate, I only finger prick to drive as I can bolus from the readings, and the trends are a definite help.

Currently I'm costing the NHS less with the Libre than I do finger pricking, Hba1c from 51 to 45, reduced insulin usage by 9u/day :arghh:overall, no longer chasing hypos and haven't been under 3mmol or over 16 since acquiring it....
 
Back
Top