First A1c on Libre (instead of Dexcom)

Are you sure about all that? The “MARD” figure of merit for CGM is the difference between what a CGM says the BG is right now and what a laboratory meter says the BG is right now. The CGM algorithm uses the history of past readings plus its estimate of lag to predict (extrapolate to) the current BG. They’re not trying to show a number that your BG will reach in 15 minutes. Their predictions are generally pretty good; that’s why the FDA says it’s generally safe to dose insulin based on the CGM reading alone. There are known exceptions, such as when a sensor is new, or when there’s a pressure low from sleeping on the sensor, or when the BG trend is changing fast (like shortly after eating glucose to correct a falling BG.) In those circumstances the glucometer is our friend.

And I wonder if your “15 minute lag” figure is up to date. The studies on the G6 showed an average lag of about 5 or 6 minutes, and it would be surprising if the Libre was 3 times slower.

I think the 15 minute lag was true for the G5 in our experience. The G6 is better with lag, but still the sensor predictions are wrong for us when blood sugar is changing rapidly. With that being said, I think MichaelS’s comment that lag should be taken into account is valid, however as you point out the specifics of the lag may not be correct. I don’t think I have seen a sensor lag study with the Libre.

The Libre does have a lag, similar to Dexcom. But I used Dexcom for years and didn’t have problems with it running consistently low. Even when my blood sugar is steady and unchanging, the Libre reads lower. This morning Libre read 3.6 mmol/L, xDrip read 5.6 mmol/L, and meter read 4.9 mmol/L. The difference in Libre versus meter reading is the difference between me treating a low or not. Since I’ve been testing more often throughout the day and basing my actions off the meter results, the Libre thinks 20% of my readings are low, which is definitely not the case.

Well it is possible that the Libre isn’t going to work well with your particular set of conditions. There are certainly a number of people that couldn’t use a Dexcom due to this. When we were using the G5, we had a long period of time where we didn’t believe the number and just used it for the trend (i.e. up, down, or flat). Which was enough information to make us continue using, even though we didn’t believe the number. The G6 for us has been accurate enough to dose as long as the trend is flat.

I’m using it for trends now, and it is useful for that and for prompting me to test. Since it’s covered by my insurance, I’ll definitely continue to use it. I think maybe eventually I’ll just round everything u0 by one mmol/L instead of testing so often.

First regarding any reading as completely accurate leads to madness!

As we have discussed before blood glucose meters can be 20% at least out so a reading of 180 could be 216 or 144 if compared to the reading at your clinic visit. Best use a Contour Next where a reading of 180 is likely to be between 171 and 189.

Libre claims to be accurate to about 9% so should be between 162 and 198.

My experience with 18 months of the 14 day UK one judged against the Contour Next is that over 180 it reads up to 20% high and at 72 it reads at least 10% low. But not entirely consistent as my current one is low at all levels.

Memory suggests that somewhere I have read that accuracy figures are based on readings between 72 and 180.

Of course you should try to take readings when you have a straight line for 20 or 30 minutes.
The lag figure I had heard was 7/8 minutes ( a bit tricky to be precise if readings are every 5) I allow for 10. The lag is because the sensor reads the interstitial fluid which gets its glucose from the blood but not instantly.

If you put the Libre on 24 hours before you activate it you should get it up and running without the early wobbles. In my experience it still goes for the full 14 days. If using Xdrip+ with an extra 12 hours.

If you are interested in bg meter accuracy I have an article on my blogsite bgonmywatch.com called “Calibration needs accurate blood glucose meters” that goes in to this in more detail with links to various meter surveys.

The most accurate meter is generally accepted to be the Contour Next with a Mard of around 5. (With Mard the lower number the better) I attach a table showing 17 popular meters in order of accuracy. My practice had prescribed the Jazz before I realized that it was one of the least accurate.

If you are using or are threatened with the US 14 day Libre then I do have an article on the blogsite detailing the competition to produce a transmitter that will will read it satisfactorily.


Thanks for the detailed post. I’ve beeb using CGM for almost five years, so am familiar with most of what you’ve mentioned. I’ve been using Contour meters since they first came out. I’m currently using a Contour Next One as my main meter.

I don’t expect a CGM to be totally accurate. I agree it can make people crazy if they’re expecting total accuracy all the time. But being within 20% or so would be nice, especially for a system that advertises it can replace fingersticks. This morning’s Libre reading (straight arrow) was 4.1 mmol/L and meter reading was 6.8 mmol/L. That falls way outside 20%…more like 65%. I think all the other examples I’ve posted fall outside 20%, too. I wouldn’t be complaining at all if my sensors were only 20% off.

Occasionally the Libre is right. Yesterday it read 2.9 mmol/L while my meter read 3.1 mmol/L. But since it’s so often so off, I don’t know how I’d know it’s accurate.

1 Like

This is really interesting and for you worrying.

For the last 18 months I have tested twice a day and written down the results. While at the higher levels I do not expect great accuracy below 108 I am within 20% both with Xdrip and Libre. Just twice in the last 6 months Xdrip has been noticeably higher than 20%. I’ve just gone back over the figures for the last six weeks and the remarkable thing is how close Libre and Next are to each other at the lower levels.

Libre has always been level with BG or lower apart from on one occasion which as my fear is of hypos is what I want.

From what you are saying Libre stuck to your arm works differently to Libre stuck to my arm so does this mean that the same sensor will react differently to diferent peoples physiology or is the sensor you use (I am assuming you are in the States, I am in UK where we do have different sensors) made in a different way?

If different people react differently to the same sensor it does open up a whole new can of worms.

You have my fullest sympathy I spend my time both with my blogsite and personally telling people how CGM has made my life so much easier and how I just love the tech.

1 Like

I think CGMs work differently for different physiological. The Dexcom was very accurate for me, but for some it’s just not accurate. I’d guess the same is true of the Libre or other CGMs.

I’m in Canada, so I may well be using sensors that are identical to yours.

CGM for sure has made my life easier. Even this inaccurate Libre is helpful for trends, just not dosing and treatment decisions. Unfortunately, it’s the only CGM that’s covered for me. If I went back to Dexcom, I’d be paying out of pocket. I’m going to see what my endocrinologist thinks and what my next A1c is before making any decisions.

1 Like

I used the Dexcom G5 before moving to Libre + Blucon as a cheaper alternative.

I did not really notice much difference in accuracy when moving over but I think the Libre accuracy has improved a bit over the nearly 2 years now.

I have been surprised by how good my control as measured by my HbA1c has been compared to the average so maybe it is just I am lucky with my physiology that sensors like me!

Good luck with the next doctor visit.

This always happens to me, but happened with Dexcom too. Do you notice once you get up and start moving around you get an arrow and it catches up?

Dexcom sometimes uses to read low for me in the morning, too. But this discrepancy is more persistent than just the morning. Currently (it’s noon) I don’t have my meter, but the Libre is reading 3.8 mmol/L and xDrip is reading 6.1 mmol/L.

I think this sort of thing has been happening consistently since I started using the Libre and resulted in many falsely treated lows.

According to the Manufacturer of the Libre system, there is a fifteen minute delay between the number you get from scanning the Libre sensor (because the sensor is measuring the interstitial fluid. The delay is the time it takes the glucose level in your blood to change the glucose level in the interstitial fluid in your body. (Please see this link for the post on the Freestyle Libre website: https://www.freestylelibre.us/cgm-difference/benefits-of-cgm.html , scroll down to the section titled: Are glucose readings different between CGM and BGM?)

1 Like

The Libre itself continues to be very inaccurate, even with a new sensor. At times, it’s been up to 5.0 mmol/L lower than my actual reading. Most often it’s 1.5 to 2.0 mmol/L lower, and regularly 3.0 mmol/L lower.

I’ve been using xDrip and it’s been much more accurate. Frequently within 0.5 mmol/L of my readings. When it does drift off, I just calibrate and it comes back. So this will be my setup for the foreseeable future. I only have the app on my iPhone, so do need to figure out how to get the readings into some format that my endocrinologist can review.

Last time I used xDrip I just handed my phone to my endo and let him browse through the statistics, but I know not everyone would be comfortable with that.

I’m using xDrip for iPhone, which is very different from xDrip+ for Android. As far as I can tell, there are no statistics or other historical information available in the app. All you get is the current reading, trend, and eight-hour graph.

Too bad - Xdrip+ for android can export to Tidepool for endo viewing which is really nice.

1 Like

I think maybe xDrip for iPhone can connect to Nightscout, but I don’t have that set up and need to look into it all…

1 Like

That sucks. How do you find the accuracy? When I used xDrip with Dexcom I barely calibrated and found it to be very accurate whether or not I did. I just got a MiaoMiao and was hoping I could get away without calibrating like with the Libre reader, but it was reading way too high for the first couple days and doesn’t seem nearly as responsive to rising and dropping blood sugars as the Libre reader/app.

Interesting how different our experiences are!

I’m finding the MiaoMioa with xDrip (on iPhone) generally more accurate than my Libre reader was. I calibrate it twice a day. It does drift off by up to 2 mmol/L sometimes. But overall, it’s a much better reflection of my blood sugar. It also responds more quickly to changes in blood sugar (although I notice that the LIbre, in general, responds much more quickly to changes than the Dexcom did).

I actually haven’t scanned with my Libre reader in a week. In part because I can’t find it. :slight_smile: