As a long term user of the Libre and a recently started user of the Dexcom G4 with xBridge (note that I don’t own a receiver – I’ve built all of the gear myself), using an Algorithm that is similar to the 505 in the US Dexcom G4, it’s quite interesting to see the similarities or differences between the two. My main reason for building the xBridge and associated pieces was to start working with open loop artificial pancreas software, ahead, hopefully, of using a closed loop, but that doesn’t stop me being able to observe what I see with the two different platforms.
I think there are two use cases to look at in comparing the two. The first is day to day usage and the second is the stats packages and how to interpret the data in a look back. I won’t go in to accuracy in too much detail or compare the delays, timing, etc, other than to say that my user level views initially reflect what Pierre posted in his blog relating to Dex on Type1Tennis. That’s to say that what I see is:
Libre overstates highs when you check, then reverts back to a lower level in historical data when compared to the G4/xDrip combination, which understates them
The G4/xDrip seems to understate lows more than the Libre does, although I’ve seen occasional overstatement of lows from the Libre
When the arrow is level, then they are similar in the numbers they report back
The Libre’s predictive algorithm when the blood glucose level is moving and not flat is better at “keeping up” with what’s going on in a blood test. The G4/xDrip solution shows a noticeable delay
The alarms that G4/xDrip provides can be really helpful, especially when you are struggling to adjust your basal overnight
Having a real-time feed to a smartwatch is cool, but also really useful. And even less noticeable than scanning!
And finally a note on “Accuracy”. I don’t like the term “Accuracy” as blood testing meters aren’t “Accurate”. What’s more important is consistency and my limited experience of all three is that on a single sensor, or batch of blood testing strips, I haven’t seen wide variability (at least after the first day of a new sensor). Others experiences may vary, but as long as the result delivery is consistent across a testing device, I’m happy.
So let’s look at the two use cases first and foremost. First up is:
Day to Day Use
Fundamentally, the two systems operate in different ways. One is broadcasting data every five minutes. The other is sampling data every minute, then every fifteen minutes, aggregating that into a single data point.
Thanks to the broadcast nature of the CGM system, you don’t really have to interact with it to capture the data. it is done automatically on the collector. On the Libre you have to scan.
Likewise, you need to calibrate the Dexcom sensor twice daily to maintain reasonable accuracy and I found myself doing something similar with the Libre, although once a day, just to make sure I wasn’t missing anything. While there is no systematic reason to calibrate, it gives you a “brain reference” so you know how it tracks compared to your blood.
Another difference is in the applicators. The Libre one is a lot less fiddly to use and feels less like a medical contraption. The Dexcom one reminds me of the Palmer Injector.
But beyond getting set up and going, what’s the difference in day to day use? Set up as I have it set up, xDrip works with my watch so I can glance at my watch at any point and see what’s going on. This is easy and totally non-intrusive (not that I’m bothered by that). It is also feeding Nightscout so that should I want to I can share the data with anyone else (and I can use the reporting function in NightScout, more of which later). My main use is keeping an eye on what I’m doing though, so it works well for that.
Using the Libre is more like, well, Fingerpricking. As I discussed here, it’s fingerpricking with the added advantage of not sticking anything in your fingers and still having a form of continuous recording. If you are using the Libre regularly (like me ~40-60 times scanning a day) then it’s not that dissimilar to using a CGM, but you can only ever find out what’s happening at your discretion.
This is where the two diverge. Even on xDrip, there is a hardcoded low alarm. You fall below 3.1 and the alarm goes off loudly. This can be set to vibrate only, which on the phone still makes a racket, and if you are using an Android Wear watch, you can set it so that notifications only work on the watch and not the phone. It’s still an alarm that you can’t totally disable though. You have total control over any others, as you set them and can decide whether they work in silent mode, etc. I have an alert for when I am too high and a couple that pick up as I drop, given the delays and differences apparent in the first set of bullet points.
These alarms are incredibly important if you are not hypo aware. I had them set whilst I adjusted my basal insulin to avoid lows and highs. But, and here’s the big but, they can also be really, really annoying and if they are going off through the night as a result of fluctuations caused by the aforementioned adjustments, seriously tiring. As a result, I can see where the term “Alarm Fatigue” comes from.
If you are struggling to manage your glucose levels and you are up and down all the time, using the Dexcom receiver or even xDrip would be thoroughly exhausting. The alarms would be all consuming. For some it would encourage them to get “in the zone” (like it does me) and adjust yourself to get to a good place:
I don’t want to hear the alarms, so I’m avoiding them as best I can. That means keeping in range. It’s a challenge but one I was used to from trying to stay in the “Blue Zone” on the Libre.
For others, the risk of alarming, or even regular alarming has an opposite effect, especially if you aren’t paying for the benefits it gives you, and the sensors come off and stay off. Peaceful sleep and life then ensues, but it’s not necessarily helpful to you.
Still, I can see why people stop using it. The alarming is annoying and when it starts to affect your sleep and intrude on your life, it reminds you that you are struggling to manage your condition and can have a profound psychological effect. Even with the additional alarm controls that xDrip provides, without the alarms there’s not an enormous amount of benefit to using CGM over Libre.
So in a nutshell, essentially you use the Libre at your discretion, but the CGM approach has the capacity to be much more in your face and a constant reminder if you struggle.
There is a different discussion to be had as to what the availability of a continuous line of data does to a person psychologically, and Andy has covered that extremely well in his blog. I think there is a steep learning curve to be had on anything that provides continuous data, and whether that’s CGM or CGR (continuous glucose recording AKA Libre) you have to be able to handle both the data and the way it makes you feel. That is not to be underestimated.
Reports and Look Back
Before we start, let’s get one thing clear. As I’m taking advantage of the way Dexcom technology works and not subscribing to their platform (saving a number of 100s of pounds on a receiver) I have no access to their suite of software tools. As a result, I’m comparing what NightScout can offer me with what I can get from Abbott.
Oboard the collection device
The Abbott reader has a fairly large range of data on board, including daily charts for 90 days, a logbook of all entries, average glucose breakouts for up to 90 days, daily patterns, time in target and low glucose events. broken out by time.
Thankfully, the xDrip developers have included similar in the phone app. There are 14 days of glucose graphs, then stats from today through to 90 days showing time in range, median and mean glucose levels, estimated Hba1C and Standard Deviation for each period, plus a pie chart for time in range and a proper Ambulatory Glucose Profile.
I think the presentation from xDrip is better than that from the Freestyle Libre Reader, but then it has an entire phone screen to play with, so it should be!
Take away reports
Taking the official data set, the reports from Abbott are beautifully presented and include useful measures such as the Ambulatory Glucose Profile and hotspot analysis to help you identify where you are high or low and therefore what therapeutic changes you might wish to make. I find this one can be very helpful:
There is plenty more data in the report, and most of it is useful and revealing. It also looks very polished.
NightScout is the only way to get equivalent data for my xDrip set-up. It doesn’t have quite such pretty graphs and the plethora of reports isn’t quite the same, but you still get an Ambulatory Glucose Profile of sorts and some fairly useful other data, plus the breakout of all the daily information:
Okay, it’s not as pretty, but it provides what you need to review your historic data and see what the trends are and where. And there is enough information there to be able to take similar things away from the reports that you get from the pretty Abbott reports.
From a use case perspective then, aside from how pretty is it to look at, some of the data combinations in the Abbott report are perhaps better combined that from NightScout but the two provide the information you need to be able to review your management techniques, history, etc, and NightScout better presents basal insulin when this is available from a pump.
In terms of the look back reports then, you can get similar data from both, and read from it what you need to.
What do we take from this in terms of using the two platforms? Well they clearly do different things, and both of them do them well. The similarities are obvious in terms of the captured data and the variety of reports available, both using the collecting devices and in the reports (dependent on whether you set up NightScout, which I’d strongly recommend).
The real difference is in whether you want a continuous stream with alarms or something you manage in a similar vein to fingerpricking. Both will provide information pertaining to the continuous data but they are simply different.
Having used both of them and bought Libre as a cheap way into CGR, I personally prefer the xDrip/Dexcom approach. I don’t like the prices but I like the way the information is available to me and using the #wearenotwaiting approach has reduced the costs substantially. I also like the idea of having an alarm there in the background to warn about highs or lows overnight (even though they don’t happen very often). It’s certainly reassuring. I also am using it with an Open Source Open Loop APS, following which I want to close the loop, which sways my decision.
There are some interesting innovations due in the 12 months or so, and this arena is going to open up dramatically. In the meantime, I’ll stick with my current solutions and use them to optimize my levels to the best of my abilities, and to warn me when I’m not quite on track…
I love my xDrip. One feature I only recently started using is the "Speak Readings" option. Though I have a Pebble Watch I find that when I'm busy on the PC or driving, hearing the reading is useful. Bit embarrassing sometimes when I forget to switch it off.. If the alarms really bother you then the software is open source so you could theoretically make changes in your personal copy. The xDrip alarms are more "intelligent" than the Dexcom receiver. As an example: My personal "Low" alarm goes off during the night. I treat the low & settle down to sleep but after the chosen snooze interval the alarm will go off again if I'm still below my "Low" setting. The receiver doesn't make allowances for the fact that my bg is rising & will soon be OK. Interesting reading, especially compared to the Libre. Thank you.
Yes, the flexibility on the alarms (and that you can set them to only go off overnight) is fantastic. I also like the flexible snooze option. I think the main reason they got a mention is because I spent a night with them keeping me awake as my glucose levels wouldn't behave. That's really in reference to "Alarm Fatigue from CGM" than a specific shortfall of xDrip. Having it linked to my watch and making all the alarms function through that has worked really well (and is much more discrete).
I realised last night, when the battery ran out, that I'd managed to bust my resistors on the xBridge and need to get the soldering iron out… Oh well!