This small attached report is based on data that I've collected over the last few weeks.
Let me know if you have questions!
1960-cgmcomparison.pdf (162 KB)I am not sure it helps since you do not include meter readings. If you want accuracy, how do you know which one is the more accurate? There are places where one cgm is high when the other one is low. Which was correct?
I understand that the meter has a wide range in accuracy but at least it is an accepted check on the cgm.
Which cgm did you think was more accurate? I can see that the two were reasonably close a fair amount of time but I would be concerned about the wide divergence when it did occur.
Thanks.
Note that the plots do include the BG readings for the calibrations (the red dots). Over time there are quite a few of those, even without the additional BG checks that I haven't plotted. The calibration data is also used in the scatter plot at the end.
Great set of data, thanks for posting! It is apparent from both sets of charts that the mean error is smaller for the G4. This could be verified statistically with the raw data. The medtronic overall has better user interface features.
You're very welcome! And yes, I completely agree to both points. Now if only they could combine those strengths :).
This is really neat, thanks for sharing. I am surprised how much both CGMs correlate. How did you get both systems on at the same time? Did you have to pay out of pocket for one of them? Did you wear both systems for the federally mandated time?
Thank you for sharing!!!
Unfortunately Dexcom is not available where I live (in Hungary), but the Medtronic Enlite is. I was waiting for Dexcom (in vain), but I think that this tells me that Medtronic is worth a try, especially since I have a VEO pump. I hope it will work this well for me!
Very interesting data and something to take note of. Do you happen to know if both systems have the same guidelines for calibrations? Some of the days seem to have an awful lot of them.
Interesting how the calibration in the last plot on Day 4 (which was lower than the Dex reading, higher than the MedT reading) seemed to "overcorrect" the calibration factor on both systems. Also, the calibrations on the top of Page 5 seem to have been done at some pretty steep rise/fall times. (The flexibility to calibrate when "flat" is easier said than done!)
There's a lot to study here, but I'm not so sure I'd jump to any conclusions about one versus the other as it relates to accuracy. In the second plot on Page 3, for instance, the MedT shows an attention-worthy low, while the Dex does not. Did it really exist?
You're welcome! I've been using the Medtronic Enlite sensors together with the Guardian monitor for almost two years. Now that the Dexcom G4 sensors became available here in Australia / New Zealand, I trialed the Animas Vibe pump. Thus I was carrying around both the Guardian and the Vibe.
Actually I had to pay out of pocket for the sensors for both systems. CGM isn't covered by health insurances here, unfortunately. But for me it's really worth it, despite being very expensive. For those reasons usually I wear the sensors as long as possible, but for the comparison experiment I changed each sensor after its official lifetime (6 / 7 days).
Yes, I'd strongly recommend trying it out. I'm really happy with the Enlite sensors so far, and I've been using them for almost two years. Personally I wouldn't want to imagine living without CGM anymore, it helps so much. Hopefully you'll get it covered by your health insurance, which often requires a justification letter from your diabetes physician.
The Dexcom system has much more relaxed guidelines in comparison regarding the calibration (here, see chapter 4), basically they say that one can calibrate any time (unless you're outside 2.2-22.2 mmol/l), and at least once every 12 hours.
Medtronic recommends 3-4 calibrations a day, whenever the BG is stable.
There's some more interesting background information about how the Medtronic calibration works here.
And exactly as you're saying, it's often not that easy to judge whether the BG is really flat ;). Sometimes I try to guess where the peak of the curve is (which probably has happened on page 5), when I expect that there won't be a real plateau for any time soon.
And yes, this data certainly doesn't justify to say that one system in general is more accurate than the other, as it's a single-person study :). But the scatter plot in the end shows pretty clearly that at least in my case the Dexcom seems to work better.
The low on page 3 in the second plot didn't exist. This was at the end of the official Medtronic sensor life span, which is why I replaced the sensor that night (that's where the straight line starting at 21:00 comes from, where no CGM data was present).
Unfortunately there is no coverage here, but I am considering paying out of pocket during pregnancy.
Thanks for such a helpful comparison.
How do you compare the convenience and ease of insertion of the sensors?
enlite straight 90 degree sensor vs. g4 sensor
Do you use the auto inserter of the enlite?
You're welcome! Note that I'm using the sensors on the upper arm instead of the abdomen. In the Enlite case I'm using the auto inserter, and it's a bit easier to use with only one hand available compared to the angled manual insertion of the G4. Pain-wise I find both very similar. Almost always I can definitely feel it, but it's really not that bad and stops very quickly. Luckily I never tried the infamous "harpoon" insertion of earlier models, which apparently was very painful.
It's convenient not having to recharge the transmitter for the G4. But of course that means that at least potentially you might need to replace it earlier.
Thanks so much for sharing this data set, Leo! Very useful.
Glad you find it useful. You're very welcome!
SO very interesting!
Hi, I have a few questions. Where the Enlite sensor went low, but the Dexcom G4 did not is that where the Enlite quit working? I know on the Sof set sensor we use now that the sensor data goes low when it stops working correctly
. Did either sensor seem more accurate than the other one?
I really liked how you combined the data. How did you do that? What did you export the data into?
What did you think of the Animas Vibe and the Veo?
Okay I looked again at your report and it seems you said by the scatter graph that the G4 is more accurate. Also, the graph dives with the Enlite is from a dying sensor. Still, other than that I thought the Enlite was okay.
Hi,
I just joined, have been a type I diabetic for about 40 years, and what attracted me to this forum was this topic. I have been running with an A1C of around 3 for many years, and I had to stop the severe hypos before becoming a victim. You may be surprised, a value of 3. Reason is that I am an active Marathon runner, not only marathon, but also supermarathon, 24 hr runs. A glycometer reading of 18 mg/dl is not strange to me, and I would feel low, but nothing special. The really severe situations were all with a glycemic reason above 45, and I can tell you that I am covered in scars from these hypoglycemia. They had to end and I got a Medtronic Enlite and Paradigm insulin pump. I'm an engineer from profession, and with no one knowing anything about the Enlite and refusing to give me detailed technical info, I started experimenting myself, testing my glucose about 20 times a day, defining a new practical measure for accuracy for a sensor, testing and comparing IM use with IS use, etc. I did this for over a year, and I am trying to get a final report finished, which will be submitted to diabetics center, and Medtronic, hoping that they will motivate improvement of the system.
Since February, the sensors don't work with me anymore. I get, what I call a current lock down, an ISIG value which stays around 10 nA, and a glucose insensitivity above 10 mg/dl/nA. To make it simple, the sensor is simply not sensitive anymore to changes in glycemic value. Medtronic, and I am in contact with an MD and the guy in charge for the next generation sensors, no help at all.
So, I have had it with them, and will be switching over to the Dexcom G4, combined with Vibe pump.
The report that is posted here, very informative. My sincere compliments for this report. If you look at fig 1, perfect correlation between BG and CGM reading for the Dexcom, hardly any correlation for the Enlite. And that is what I found too for many sensors. For me, the weak point of the Enlite is that every sensors behaves completely different. About 15% of the sensors don't keep up the 6 days, 50% of the hypo alerts are false alerts, etc.
I should be getting the Dexcom system next week, and I will then start a similar experimental period as I did with the Enlite.
Thanks for reading, and I look forward at exchanging more info with all of you.
George
I found the dataset very interesting. Did you happen to calculate an R squared value (goodness of fit of the data points to the diagonal line) for the Dexcom and the Enlite? I think that value would summarize best which CGM is more accurate. Just from eyeballing the data, it looks like there's less divergence for the Dexcom data points from the BG values than for the Enlite data points.
I'm at a point with almost 3 years on the Minimed SofSensors that the inaccuracy (especially for measuring "lows") may make me decide to continue on the Minimed pump but switch over to also carrying the Dexcom G4. I'd be interested in hearing about experiences of others that have done that. Thanks in advance.