Wednesday, February 24, 2016

A plug for a couple of other blogs...

The search for a cure

If you are, like I am, trying to stay informed about the current state of T1D cure research you can either read a lot of scientific papers or visit this near encyclopedic blog

Current Research into Cures for Type-1 Diabetes

Even a nasty old nitpicker such as myself finds very little to complain about in the author's coverage and analysis. As a side note, the conspiracy theorists thinking JDRF actively obstruct the research for a cure should have a look at this page (but keep in mind, as the author explains, that JDRF is not usually the sole or even a constant funding party). 

One caveat though: the blog is slightly depressing because... reality is a bit depressing.

New insulins

Then, since a lot of "advanced" insulin patents have expired or are about to expire, the industry has suddenly become a bit hyper active. A lot of new formulations or slightly modified molecules should hit the market shortly (at what price?) : Olga, from Diabetes Lab has posted a nice summary of what we can expect (or has already been released in some markets).

Saturday, February 20, 2016

More on Dexcom G4 (505 algorithm) trauma induced drops.

A month ago, it gave an example of what I considered to be a micro-trauma induced CGM sensitivity drop. You can see that post here. While, to be honest, I consider the issue to be almost settled for us thanks to previous observations, additional information never hurts. Let me start by giving a bit of background on the Saturday tennis training: training starts at 16:00 with a small warm up and about 10 minutes of controlled exchanges. After a 10 to 15 minutes, Max coach will start hitting a bit harder, make Max move forward and backwards and send a few lobs so Max can smash as he sees fit. The consequence of that heavy smashing and active play is that Max often finds himself in hyper extension, stretching himself as much as he can to catch those high balls and smash them down the court. That sequence happens around 16:10 to 16:15, after which we have our first BG check/recarbing pause.

We rotate Max's sensor location, but here is what happens when the sensor is on the same side of the abdomen.

The first chart is the one I previously posted. Max woke up with a perfect range that day and we more or less kept the levels until tennis with a bit of recarbing before the training began. After those first ten minutes and a bit of delay, we saw a non physiological (in terms of speed) sudden drop that almost exactly matched the later recalibration offset.

The second chart is a new one. This time, the unpredictable dawn phenomenon struck and Max woke up a bit higher than a month ago. This led to a somewhat more chaotic preparation, but identical values for the start of tennis. Then, after the exact same training chronology, we got the same non physiological drop at the same time. Likewise, we observed the same compensating offset when we recalibrated. (The recalibration came later this time because Max had another training scheduled from 18:00 to 19:30) . And of course, a delayed hypo hit us at about the same time, followed by an additional one, possibly because of the additional training...

This is so strikingly timed that I did not have to adjust the legend position in the code used to generate the chart.

On the Libre, one alternate explanation could be the use of a predictive algorithm. That explanation can't be excluded on the G4 505 since it can support an eventual limited predictive capability. However, had some predictive algorithm over estimated the drop rate, it would have corrected itself as additional data came in.

The most likely explanation as far as I am concerned is

- the sensor is stable and well embedded (on Saturdays, our sensors are typically 5 days old)
- the extreme extension of the abdomen during exercise breaks whatever scar tissue has built around the wire
- sensitivity drop suddenly because of the presence of blood or fluids on site (fairly typical in micro traumas of all kind).

This remains, of course, a working theory. But I like observations upon which theories are built to be reproducible and that one definitely is.

Wednesday, February 10, 2016

Dexcom G4 non AP vs SMBG bias and some double checks.

In the Summer of 2014, a few Dexcom users of the "CGM in the Cloud" facebook group kindly shared their data with me so I could analyze it and see how the G4 performed in real life. Some of the results were posted in the facebook group, some of the results were posted here.

One of the most striking results I observed (other than the fact that real life accuracy wasn't as good as advertised) was the existence of a significant bias in the Dexcom measures when they were compared to SMBG values.

Here is that chart again

What it shows is that the Dexcom G4 non AP had a tendency to over-estimate low values (outside of incidents such as compressions) and under-estimate high values. 

At the time, I did my best to validate that interpretation, clean up my data of any eventual biases, double check. But of course, there's always room for mistakes. That's why I read with great interest that large scale study by Lori Laffel MD, MPH, of the Joslin Diabetes Center where the very same bias is clearly visible. The scale, presentation and units are different but one can clearly see that in the low range the average errors tend to be on the up side, from 70 to 200, the errors are balanced and that above 200-250, errors are on the low side)

These results are mostly irrelevant today as the standard G4 is living its last days and the G4 with 505 update is now available outside the USA.

However, this result comparison shows that, with the kind collaboration of normal users, it is possible to extract valuable - and valid - insight.

I'll get back to the paper in more details on this blog if I can find the time.