Freestyle Libre – really a public BETA?

Freestyle Libre – really a public BETA?
Freestyle Libre – really a public BETA?

Given some of the issues observed with the Freestyle Libre, the costs involved and the precious lack of information available from Abbott in relation to the in depth testing undertaken, I decided to add to the knowledge base.

Being in the unusual position of having an additional sensor and reader, I thought I’d give Abbott a little help and run a side by side test with two separate sensors and readers. I noticed on 9th January that I was starting to see some variance between blood glucose readings and sensor 20 minute delays. I noted that at the gym in the evening, prior to the oddities starting to happen, I’d had a sweatier than usual workout, and that the sensor looked like it was starting to lift a little. Since then I’ve also noticed some itching at the sensor site. On Saturday morning, with six days remaining on the “old” sensor, I plugged the new one in and off we went.
I can’t be sure what I was expecting to see, but being two days in so far, the output has created a couple of surprises. 
The start point was here, with a blood test and two scans twenty minutes later, one on the new and one on the old. When photos like this come up, I take the old reader on the left and the new on the right. For clarity’s sake, the old reader also has a sticker over the glucose strip slot.
At the start, there was a notable difference between the two sensors, with the new one tracking blood glucose levels closely:

At all point where I’ve made the comparison, this pattern continues, with the older sensor remaining below the blood glucose reading, sometimes as much as 2mmol/l and at other times, as little as 0.4 mmol/l.

There is clearly something not quite right with that older sensor as there is a noticeable difference between it and the new one. Did it lift? Did something get underneath it? I’m not sure…
Where it gets more interesting is this morning. Despite the odd goings on of my blood glucose level whilst going through a change of basal insulin, you’d expect the graphs on the two sensors to be similar, if not almost the same, wouldn’t you? I would… But the sensors now disagree…
Compare these two. The old sensor has a very smooth graph, the new one is full of bumps and lumps. Which is the real deal? Now that’s a more difficult question to answer. I don’t know. I would tend towards the newer sensor telling the “truth”, but you just can’t tell.
The bigger issue is that these two graphs could have come from completely different people!

There is then the further anomaly relating to hypos. The Libre seems to operate slightly differently dependent on the way a hypo is entered. If the velocity of entry is slow, in effect you drop in at a leisurely rate, the sensor shows you as in hypo or nearly in hypo when you get the awareness symptoms. If you drop in relatively quickly, you’ll know you’re there long before the sensor tells you anything useful, and the sensors have been found on occasion to not pick up hypos for some time. This is presumably a limitation of the interstitial fluid reading technique rather than the sensor’s fault necessarily, but it can be highly frustrating. While Abbott say that you should revert to blood testing during times of rapid change, it does beg the question of the value of the device at these times. You certainly can’t rely on it. 
Have I changed my mind on the Libre yet? Not fully. I’ve always felt that its greatest use was in being able to see trends and trajectory and that the point readings were less important (and have always been less important in managing blood glucose levels), as they are simply a calibration. I don’t think this has changed.
What has become apparent is that the initial public roll out of this product is more like a public beta than a true General Availability. Abbott themselves have admitted to a number of people calling up with issues that there were many more than they expected. 
All those of us who have participated have essentially done Abbott a huge service or favour in paying to help them identify how the product works in the real world. While I don’t have a problem with this – I have after all found it to be invaluable, it might have been better to couch the terms of the agreement as a BETA, much like those that Google have with Glass. There do seem to be high failure rates on the sensors (admitted by Abbott reps on the phone) and a recent Facebook poll, which is clearly not scientific or properly representative, suggested this may be as high as 20%.
I’ll continue to update on this sensor experiment as I think it’s a useful data set to capture but I’m not convinced that rolling this out on prescription is a good idea. There are still a few too many bugs that your average user will struggle to deal with and until these are dealt with, it remains a target for early adopters only.


  1. Good work Tim. I've used 5 sensors and none have them have been very accurate compared to BG readings also taken with the Libre. However, like you I found the trend data to be of more use than the actual readings – what is scary is how your two readers differ in their trend graphs – as you say they could come from different users. This for me is the straw that broke the camels back – I could never rely on the actual sensor readings but I cannot now rely on the trend data so the whole Libre system has become useless to me.
    I agree with you that Abbott have been using us as BETA testers and personally I think they should refund us at least the cost of the starter pack.

  2. I spoke with Abbott support this morning. I offered to provide them a download of all the data and a comparison with the current good sensor.

    The guy on the end of the phone seemed to be genuinely disturbed by the performance of the sensor/reader and pleased at the offer of more data. His biggest concern was that the blood test strip issue on the old meter may, in some way, be linked to the algorithms that the sensor uses to convert the signals to a useful data set. Once the sensor is expired, they'll be getting both of the components back (I've already had the replacements so I'm not bothered about that) to check out and test.

    Somehow, I don't think they saw this as good news!

Leave a Reply

Your email address will not be published.