Dexcom vs. Freestyle accuracy

Dear user’s,
just a quick note to show you a graph which I got yesterday after simultaneously wearing a Dexcom on one arm and a Freestyle on the other. For those of you really excited with the recent US approval of Freestyle…
I decided to switch to Dexcom because of the transmitter and because Freestyle was always reading too high compared to blood samples, after wearing it for 6 months.
As you can see in the graph (sorry it is in mmol/l, 10 mmol/l is 180 mg/dl, and 80 mg/dl is 4.4 mmol/l), Freestyle trend is above and Dexcom trend is below, consistently, regardless of 6 calibrations (red dots). The shapes and trends do seem to resemble each other.
Sometimes the differences between the two are above 2 mmol/l , which is 36 mg/dl, enough to cause a false low. For example at around 6 pm, Dexcom was already signalling very low and Freestyle did only give a slight slow. Both Dexcom and Freestyle gave more than their quoted less than 10 % error, consistently… :frowning:
And the most accurate of the two is:
A free app developed to read Freestyle, was only away by 0.2 - 0.3 mmol/l throughout the whole day!

1 Like

Aside from Glimp (I don’t know what that is, I’ll have to look into it), based upon the red calibration dots, it appears that Dexcom was tracking more accurately. Or am I reading this data incorrectly?

Indeed, Dexcom was more accurate and the calibrations were closer and closer to the real value after the first one. With the Freestyle it also depends on which sensor you have, as you might get lucky and get a good one (rare) that is not too far off the real value.
Glimp is an independent app for the smartphone, nothing to do with Abbott, can be downloaded from the playstore, and allows calibrating the Freestyle and tweaking a few more things. In my case I found its results closest to the actual blood reading. Maybe someone got different experiences?
All the best!

@JorgeM - Does the Glimp app give different BG readings then you would get from delivered Abbott Freestyle Libre functionality?

Also what do you mean when you say the Glimp allows calibrating the Freestyle? I thought this was not only not required but also not possible? How does this work?

1 Like

Definitely, Glimp gives different (in my experience closer to real) results than LibreLink app, and it allows calibrating the results you see on the screen.
As I understand Glimp scans the raw data of the Freestyle and applies your own calibration to it, whereas Abbott have got their own secret hidden algorithm which is the same for everyone, and for me turns out to give generally higher readings than real.
Although Freestyle sells as a real advantage that one would say goodbye to fingersticks, given its inacuracy, it would come in handy to be able to calibrate it.

1 Like

@JorgeM - How close do you have to have your SmartPhone running the Glimp App in order to pick up the cgm data from the Freestyle Libre? Are you are to hold / wear the phone normal or do you actually have to wave it directly in front of the Freestyle Libre?

Can the Glimp App on the SmartPhone be used for realtime alerts as the Dexcom gives or would that require additional technology/hardware?

Unfortunately you need to scan your phone for Glimp as you would do with the LibreLink app, and there are no alerts unless you keep scanning or add some additional diy transmitter. I know there are already options out there like Limitter, but do not fancy fiddling with these by myself. Therefore I switched to Dexcom, just for the transmitter feature.

Jorge Mendoza

Libre + Xdrip + blucon provides a calibration possibility as well, for those who need a cheap CGM solution. Comparing Libre to Dex is not fully fair since Dex costs three times as much, if not more.

I would argue not only fair but something many people are probably doing (ie comparing the systems). But include the price as part of the comparison as well. Include everything.

Always a bit more difficult to get real prices instead of guesses or ballparks (whole nuther discussion about how there is a serious lack of pricing transparency in the entire medical industry) but certainly possible.

The Dexcom really does look like the readings matched your fingerstick readings pretty closely. But if the Freestyle tended to consistently run a little high, I’d just handle my records with it like I handle my readings from two of my ReliOn meter readings - adjust.

Every time I have a lab draw, I take my meters with me and do a test within minutes of the draw. One of my meters has averaged close to right on compared to lab results. The other two meters have averaged about 10% high. So on my spreadsheet of readings I have one column for “actual reading” and a second for “adjusted reading,” which for the two meters is 91% of actual. When using those meters I just dose insulin on the basis of the adjusted reading.

The problem with the Freestyle, though, is that one only has ten days to figure out how far off it is. Every new sensor starts a new accuracy situation all over again. But I’m using the ReliOn meter because of the difference in cost between that and one that might be more accurate. My thinking would be the same for the Freestyle.

While it’s disappointing that the Freestyle doesn’t track more closely, I’m going to order one as soon as they’re available! I realize that some will need all the alarms provided by the Dexcom. But for me, knowing the trend is what’s important. If I were to see a double-up or double-down arrow, that would be my cue to stop what I’m doing and investigate. On the other hand, if I’m seeing a flat line, I can probably forego a meter test for a while. I think the Freestyle would allow me to avoid a meter check at some very inconvenient times. My objective is not to eliminate fingersticks, but to be smart about when I need to do them.

And while the Freestyle won’t alert me in the middle of the night, I can see the overnight trends, which could lead me to make future adjustments.

I think maybe I misunderstand how the Libre works. I thought it replaces, thereby eliminates finger sticks. If users verify with a finger stick, isn’t that the equivalent of using two meters before making a dosing decision?

The Freestyle IS supposed to replace fingersticks. But the graph presented suggests that this may not always be the case. Still, the FDA and government agencies around the world have approved the Freestyle as being accurate?

But as I said, even if that’s not the case, I want one. Yes, you would be double-testing, but ONLY at times when you were getting the double-arrow up or down. For me, that would not be frequent. YDMV!

In Europe, some 50,000 users were found to have read the Libre more frequently than they had with their meters AND their A1c’s were improved.

I do understand that some users will need the added features of the Dexcom, but I think there is a place for the Freestyle as well. And I think we will see improvements in both devices, as they go mainstream.

1 Like

Yes, this graph indicates that it may not always be the case. This is only one user’s experience, and doesn’t necessarily indicate the results everyone should expect, but I do understand how it would incline a user to verify with another source, like their reliable meter. :slight_smile:

EXCEPT when it doesn’t. Below is part of the warning in the Abbott announcement of the Libre approval for USA. I believe the “check blood glucose” symbol will appear when BG shows under 70 or over 180, although I haven’t seen those exact figures in any Abbott literature.

WARNINGS/LIMITATIONS: Do not ignore symptoms that may be due to low or high blood glucose, hypoglycemic unawareness, or dehydration. Check sensor glucose readings with a blood glucose meter when Check Blood Glucose symbol appears, when symptoms do not match system readings, or when readings are suspected to be inaccurate.

1 Like

And to be fair, the Dexcom also has situations where they recommend not doing a Bolus from the cgm reading. The obvious being a rapidly changing (double arrow up or down) or a situation where you do not feel (for whatever reason) that the cgm number is not proper.

1 Like

True indeed, but it’s not marketed as not requiring finger stick calibrations. This footnote is interesting to me. I thought the no finger sticks was a large factor in why Bigfoot changed its direction. This information puts that in an interesting light.

I think Caleb finger sticks as much to confirm Dexcom as to do the required calibrations (the confirmations, when done, negate the calibration need). If Libre requires the same, there’s no finger stick upside in practicality. At least for our experience.

Agreed and the same. Although I try not to go more than 24 hrs (more or less) without a calibration. So if the meter and the Dexcom CGM are in complete agreement but it has been a full day since the last calibration then we enter this as a calibration.


I think with Loop, he needs a calibration at least every 24 hours - might actually be 12 - or it will time out and stop Looping. Geez, I’m not even sure if that’s right. I used to know everything about his care and now I do so little and know so little…

1 Like

For me this logic is flawed.

The problem is that this comparison takes the results from the test stripes as the correct blood glucose values. Here the Dexcom is calibrated with the test stripes and therefore it follows the numbers of the test stripes.

Here the Libre takes a very different approach. It does not calibrate at all - and this for the simple reason that test stripes show deviations of +/- 15% (sometimes even more). The team of Abbott provided long term data to the FDA to proof that the system works more correctly without calibration. Obviously the FDA followed this argument and allowed the Abbott Libre to enter the market.

In other words: people are satisfied with their cgm if the numbers are comparable to their test stripes. This however is an illusion because the correct glucose values can only be derived from calibrated lab equipment.

The hard reality is that test stripes often deviate - and this more often than they should - and despite the improvements in accuracy from +/- 20% to +/-15% in the recent years. Statistically they can still produce massive outliers. With glimp you just calibrate with these outlier prone numbers. At the end this will produce less reliable numbers than the Abbott approach without calibrations.

The only situation where the glimp might be usefull is when the sensor has not been applied correctly. In these situations the Libre might systematically deviate into one direction. Out of 48 sensors this only happened once to me. For something this rare it makes no sense to introduce systematic deviations from calibrations to the sensor numbers.

1 Like