Accuracy of Freestyle Libre

[mountaintom]

I have a free trial of the libre on my arm and have had it on for 1 week. The numbers are usually 2 points higher than a strip test making it quite dangerous in my mind - if it says 5.1 I could be having a hypo and not know. Does everyone experience these discrepancies and if so, why bother? I’ll admit my fingers have been given a bit of a rest since putting it on but...

 

[Perro]

Yes i also experienced the same.. I tried to talk to Abbott here in Norway about it, but I was totally ignored.

 

[Deleted Account]

It seems to be that the Libre (and CGMs) work well with some people and not with others. I have not worked out the difference between us; I am in the same boat as you.
However, I would not consider them to be totally useless. For me their value is in detecting trends: you may not get an accurate reading about how high pasta, for example, raises your BG but you can get a good idea about how soon the peak happens after eating to be able to adjust your dosage time next time you eat pasta.
With thus in mind, I found value in occasional use but not all the time.

 

[Scott-C]

The big selling point of libre was that it was "factory calibrated" so that, unlike dexcom, you wouldn't need to calibrate against blood twice a day.

But, erm, regular users know that it hasn't lived up to that claim. There are ways of working round it - I found that most of my sensors ran fairly consistently about 1 below actual, so it was easy enough to say when libre says 5 it's probably 6.

I'd normally test a fair bit on the first day or two of a new sensor just to get a sense of where it was in relation to blood and make adjustments from there.

Most folk find the difference fairly consistent but some get no consistency at all. It takes time and a few sensors to get an idea of how to read it and decide whether to trust it. I eventually, after a few months, decided I was ok bolusing from it.

One thing which has made a big difference is using a blucon transmitter with it. It's a small bluetooth transmitter from www.ambrosiasys.com, costs about £100 one off cost plus £25 shipping and import dues, reads libre every 5 mins and bluetooths to a free android app xDrip+. The app lets me input a bg test to calibrate the sensor reading against blood. Makes a huge difference - provided I calibrate a couple of times a day when relatively stable, it's generally only about 0.2 to 0.5 out. Plus I get hypo alerts. Not bad for 100 quid.

 

[mountaintom]

Yeah. Mine is usually two points higher. But I’m not willing to take that risk. Even if it’s 2 points higher 98% of the time I still don’t want to risk that 2% if it could potentially be hypo from hell.

 

[Boo1979]

Hope its ok to jump on as a T2
I purchsed 4 sensors to work out some stuff re trends and to see the impact of a couple of self experiments and they've been great for that - but its a different story when it comes to accuracy
First 2 sensors seemed pretty bang on with my meter readings ( about 0.5mmol difference)
Sensor 3 began to show me having hypos when the blood meters read at least 1mmol higher and out of hypo territory
Sensor 4 has been all over the place and Ive now abadoned it. It seemed accurate about 75% of the time, but wildly out by 3 mmol plus at others eg libre showing LO (under 2mmol) meters showing anything between 4 and 6.6, Libre showing 9.6, meter showing 5.8
I wouldnt trust the Libre to guide any immediate medication decisions unless I was taking a matched reading with my meter whenever a decision was necessary

 

[Scott-C]

Each sensor seems to have its own "personality" in terms of accuracy and trustworthiness.

Maybe something to do with manufacturing processes, maybe something to do with it's interaction with the body - is it close to a bunch of capillaries, has it been biofouled and clogged up with glucose eating cells? - dozens of reasons why they can differ.

The trick is to be able to figure out the good ones from the bad ones, and, unfortunately, that's a trick which is only picked up after lengthy use.

I know I keep on mentioning blucon a lot (I'm not on commission, honest!), but I'm just very enthusiastic about it. The ability to calibrate irons out a lot of libre inaccuracies.

Many say that dexcom is more accurate. That's probably because it requires calibration twice a day. Without calibration, I imagine dexcom would be pretty sketchy at times too.

William Lee Dubois wrote Beyond Fingersticks... back in around 2009, which is like pre-history in cgm terms. It's a bit dated now, but a lot of his points still hold true, about what to expect, and just as importantly, what not to expect.

I never regard a reading as "true": it's more of a clue or adminicle of evidence which I'll use alongside other indicators to decide whether, in each instance, I'm ok bolusing with it or need bg test confirmation.

 

[mountaintom]

Thanks everyone. I’m going to write a lovely email to Dexcom and see if I can try one of their sensors out.

 

[Sibyl]

My latest batch of 4 - first 3 were always a tad under but as has been said I use it to spot trends. However the 4th is reading higher but it is consistent so I’m aware I’m actually a point or so lower. Got to remember that even our finger tests have a margin of error.

 

[Gork]

I removed a sensor after 4 days of obviously low readings. I matched the last 10 readings with test strip results on my Breeze2 meter and found the Libre system to be 60% to 90% lower. Although Libre is replacing that one there were rather belligerent towards me because I wanted them to replace the test strips I had to use from my original supply. They refused.

My next sensor is shows a few accurate reading to many at 20% low and libre says that this is acceptable.
Really?
20% error is acceptable?
Anyone else see this?

 

[becca59]

I removed a sensor after 4 days of obviously low readings. I matched the last 10 readings with test strip results on my Breeze2 meter and found the Libre system to be 60% to 90% lower. Although Libre is replacing that one there were rather belligerent towards me because I wanted them to replace the test strips I had to use from my original supply. They refused.

My next sensor is shows a few accurate reading to many at 20% low and libre says that this is acceptable.
Really?
20% error is acceptable?
Anyone else see this?

It’s in the blurb. Yes 20% is allowed and that is after a 15 minute gap. Annoying I know. What interests me is the variation with them all. Have had useless....replaced, 20%....hard to rely on and perfect....brilliant! Oh that they were all brilliant. I feel there is a long way to go yet with this equipment.
Also they made me test the accuracy with the Libre reader before replacing and sent me test strips to do so. I have found there is a variation between the Libre reader and my Freestyle Verio. The Libre reader always being nearer the sensor reading. Deliberately calibrated that way?????

 

[Gork]

I have heard of said "15 minute gap". What is this?
Did they issue test strips or did you have to purchase them?
How many?

 

[becca59]

I have heard of said "15 minute gap". What is this?
Did they issue test strips or did you have to purchase them?
How many?

You test blood sugar on a metre and 15 minutes later scan your Libre. It is not expected that the results will be the same if you do them at the same time as the very way the Libre works there is a 15 minute delay. Even with that 15 minute gap they allow a 20% differential. People expect perfect accuracy, that is not going to be the case overall. In my opinion it is the trends and speed of movement that is key.
Re the strips, the first time they sent 10. I then bought 50 off Amazon. I only ever use them if I feel the sensor is not accurate.

 

[Knikki]

I'm on a Libre trial as well and so far have found the sensor to be fairly accurate when compaired to finger pricks.

Guess should do some more checking.

 

[porl69]

I have been using the Libre since January this year (thank you NHS) and have found all bar 1 to be very accurate. If there is a big discrepancy between BGs and the Libre then Abbotts are very good at replacing them. Unfortunately not every person is so lucky and will get very random results. Don't forget it is 15mins behind a finger prick due it testing interstitial fluids.

 

[smc4761]

I started on Libre about 7 weeks ago and find my reading compared to the finger prick test are pretty much the same, even in low, normal or high BG readings

 

[becca59]

I started on Libre about 7 weeks ago and find my reading compared to the finger prick test are pretty much the same, even in low, normal or high BG readings

Jealous!

 

[EllieM]

Mine were accurate for about 8 months and then I developed a skin allergy to the sensor. The readings then became consistently inaccurate, generally under reading by drastic amounts - frequent LO readings which could be anything between hypo and 8. Half a dozen sensors later (Abbott replaced sensors and even the reader) I had to admit defeat. Recently I used my last sensor and though the allergy was greatly reduced the readings were still all over the place. My body just doesn't agree with the sensor.

 

[becca59]

Yes @EllieM I find the more they start to itch the more inaccurate they become. It’s a sort of rejection. They do vary in how I react. The last one was amazing for 10 days then the itching started and it kept telling me I was about to keel over. Which I wasn’t.

 

[Bluey1]

This is a comparison between the Libre (top) and the Enlite (bottom)
The Enlite was calibrated with Abbott Neo. I used 3 meters Abbott Neo, Beyer Contour next and the Accu-Chek Performa. The Neo and Performa were always very close and the Contour next was frequently over 1.5 mmol/l over (and occasionally under.
all blood was from a single finger stab.
I know this is not scientific, I got some control solution that only works with the Accu-chek but has a range of 1.7 - 3.3 or 14.1 - 19.1, so absolutely no use. The grey bands are 4-8 mmol/l. I know this was on one of my more challenging days.