A tale of two sensors

[CherryAA]

As I have reported elsewhere, my first freestyle libre developed a fault, there is a faulty connection on the reader that means that it only reads 1 in 10 times. At the pint it developed the problem I was only a couple of days into a new sensor. Abbott agreed to supply me with a new reader and a new sensor.

When that arrived I put the new sensor on my left arm and then just kept them both going until the natural expiry of the sensor linked to the old reader at which point I will send the sensor back . It had not been my intention to do that, simply that the very first readings from the new sensor were 1.5 higher than the old and as I had not waited first I figured I might as well rely on the old one until the new one "settled down" . It never did settle down

I am currently therefore in the position of being able to see what is happening on both sensors for 5 days.

Five days in the readings are extraordinary.

The two sensors are reading almost exactly in sink for the entire five day period, creating remarkably equivalent graphs and shapes, however the new sensor is 1.5 mmol higher than the old one at every point in the proceedings. Blood pricks of my left hand show the lower reading on both sensors. What is very clear is that the new sensor has just not been correctly calibrated - i.e. every single reading is going in exactly the right direction pretty much exactly the same reading as the old one, just 1.5 higher. Thus in order to determine what it is really telling me I have altered the target glucose range to 5.5 - 8.2 whereas on my old reader the target is set at 4.0-6.7. having amended it this way, both sets of readings are giving the same proportion of high and low readings and the same shape of recordings in the white area above the blue band after eating . There has been zero change in the position the more days into the sensor I get.

I am not sure what to make of this in terms of the future - however going forward, I think my solution will be to do blood pricks in the first couple of days and then amend my target range to show the blue band at whatever level the blood pricks are showing me i.e. giving an allowance for the difference if another one develops between blood test and sensor.

Abbott already agreed to replace the "faulty " second sensor however I will be reporting this to Abbott tomorrow and I will send them a copy of the readings from both sensors for the period in question and both sensors along with the damaged reader. It does call into question a little bit why there is no facility to calibrate the readings based on the actual blood results.

What is clear though is that the new sensor is doing its job very well and consistently, just that every reading it is giving is 1.5 too high which has seems to be something to do with the quality control on calibration as opposed to my personal body chemistry - unless its entirely normal for the left arm to read 1.5 mmol higher than the right even if the blood prick tests going into both sensors were actually from the left arm!

velly interesting as the Chinaman said !

 

[tim2000s]

You might find this interesting @CherryAA: http://www.diabettech.com/diabetes/freestylelibre-the-only-way-is-up-or-down-or-maybe-up/ I ran two sensors side by side for 5 days in early 2015.

 

[CherryAA]

@tim2000s, Thank you for this, it shows pretty much the same data except that in my case I have found the finger prick test to be a lot closer to the old sensor than the new rather than midway between the two. The data has also been very consistent so that I can pretty much predict what the other one will say, once I have scanned the first one, in either direction. I find it hard to believe that its the difference in the interstitial fluid when all of the blood samples have come from the same arm that is the one with the too high new sensor on it.


I think the idea of the Abbott Freestyle is fabulous and certainly I much prefer it to constantly pricking my finger the overall shape and patterns are extremely useful and even with some of these issues it still enables you to understand exactly how your body reacts to food easily.

As a result of the freestyle I know understand that on a very low carb diet ( i.e. under 30g carbs and less than 100g protein) my body doesn't really react to anything I eat per se, instead I have a morning liver dump regardless of food and the only way I then bring down both that liver dump and my fasting levels is by coupling the carb diet with an actual calorie deficit overall Once both factors are in place and I become ketogenic, then I start to see real reductions everything including weight, fasting blood sugar and morning highs Low carb per se without conscious calorie restriction seems to keep weight stable, fasting levels in the 5.5-6.5 range and a 3 hour period when my levels are too high due to the liver dump. I'm guessing that if I can lost the remainder of my excess weight ( 25kg down 20kg to go) then at that point I might have got rid of the excess stuff causing the liver dumps.

As a tool for really understanding long term trends I think it falls short at present because one can only really rely on the two week trend within one sensor and that data then needs to be seen in the context of some real blood samples for each sensor so you can gauge where your current sensor is currently reading. My biggest gap was a 15mmol high after a fish and chip lunch compared to an 11 high on the old sensor. Unfortunately I didnt have the fingerprick to hand at the time, which is a pity because the 11 would have been a real cause for celebration being my biggest carb blow out since I started testing and at the wrong time of day for me as well!

I must admit I was expecting to find a much more uneven correlation between blood and sensor as opposed to what in both our cases appears to be clear - "under" or "over " reporting sensors which seems to be connected to the sensor calibration not the vagaries of interstitial fluid. That seems to be something Abbott should be able to fix eventually. Did you report your findings to Abbott? I am going to do that and send them the new and old sensor at the same time.

 

[tim2000s]

As a tool for really understanding long term trends I think it falls short at present because one can only really rely on the two week trend within one sensor and that data then needs to be seen in the context of some real blood samples for each sensor so you can gauge where your current sensor is currently reading.

I think that depends rather on what you mean by long term trends. That you get wider variation in the evening than the morning still shows up regardless of sensor variation, for example, and seeing the change in spikes after adjusting bolus timing is also something that you can see. It really depends on what you're looking for in the long term data.

 

[tim2000s]

As a tool for really understanding long term trends I think it falls short at present because one can only really rely on the two week trend within one sensor and that data then needs to be seen in the context of some real blood samples for each sensor so you can gauge where your current sensor is currently reading.

I think that depends rather on what you mean by long term trends. That you get wider variation in the evening than the morning still shows up regardless of sensor variation, for example, and seeing the change in spikes after adjusting bolus timing is also something that you can see. It really depends on what you're looking for in the long term data.

 

[CherryAA]

well I'm type 2 and my goal is to get Hba1C in the normal range. So my long term is the 90 days of data it gives me, which should show an improving trend if my hba1C is coming down

The Abbott freestyle is set up to report an hba1c figure which I had assumed was coming from the 90 days of data it purports to hold ? If there is a fluctuation of 1.5 between sensors just based on each sensor calibration performance, its not going to give any real idea of whether you can expect an improved Hba1C or not - for me its the difference between -non diabetic 5.6% (old sensor) and fully diabetic 6.5% (new sensor) so that aspect of its reporting is not much use at all.

Neither is the "time within target" much use over more than a 14 day period, if on has to adjust the target to reflect the calibration error. The readers contain 90 days of data which will be a mish mash of these conflicting sensors. Before I had to change the reader If I had taken that reader into the doctor he would have said - great no drugs required, if I take the new one he will say -oops time for the metformin - those seem quite material issues to do with calibration.

The other thing that is a bit annoying is that having got one bad sensor with a few days data on it, you are stuck with that data in the averages and statistics you are reviewing for the next 90 days even if Abbot have agreed to replace the sensor .

Like I said before I like the facility and it has told me an awful lot, not least even if I were to abandon it now, I now know just how consistent the shape of my readings is when following a consistent diet - which is a lot more reassuring that taking individual samples but not be able to see the trend lines.

Overall I will be keeping it, and putting up with these discrepancies, however it does feel like its something that Abbott should be able to fix in terms of quality control of its output product, rather than human blood vagaries. it boasts that this does not need calibration whereas other systems do. My interpretation would be that it probably also needs calibration!