Anyone else get wildly inaccurate Dexcom readings?

[Bill_St]

Try taking a number of finger pricks and compare them. Remember that CGM take readings every minute(Libre) or 5 minutes (Dexcom) and average them. So many reasons for blood strips to be inaccurate, never mind the meter itself. Look up the medical trials on all the systems and see the real differences. Then consider why you are taking readings; and consider the variability of insulin itself! ( Look up the medical trials of that in the US and then be really shocked) Nobody dares to test it in the U.K.

 

[hboyt]

I have been using the Dexcom for about 4 years and find it very inaccurate in the last 2. I have had to change sensors often, sometimes within 1 or 2 days of putting it in. Now I am finding that at any given point it just loses the signal or reports the sensor is not responding and for no discernible reason. And this is very troubling for me as I have been thinking of going on an insulin pump that uses readings from the Dexcom.

Sounds like a faulty transmitter. I've been on the Dexcom G6 for over 2 years with the tslim pump and only once had this problem with it not responding. I went over all the details with dexcom and they sent me a new one. Might be worth a ring to dexcom...

 

[hboyt]

I'm using a Dexcom G6, and with this one you calibrate it in the app, just go to 'settings' then 'calibrate' and type in the finger prick reading. I've never used a Dexcom One, but if it has an app then I would have a nosey in there first. If not, maybe you could you try calling your Diabetes Team and asking them? Best of luck.

Just ahve to be careful you don't calibrate it more than I think it's 10% otherwise it will tell you to replace the sensor. I had this problem when I was using Medtronic pump and sensors. With the Dexcom G6 I never calibrate

 

[jackois]

I tried the Dexcom One for a month and found it to be very poor, literally the only use it worked at was as a trend monitor. I was blood testing two or three times a day and then calculating the blood sugars myself.

I can confirm that there is no calibration facility. That these sensors are prescribed via the NHS is a total waste of public funds.

I'm currently using the G7 and this seems to be mostly accurate.

 

[Harebell1111111111]

To me the modern CGM sensor seems to be a techie miracle. The fact that the little bit of wire stuck in to your interstitial fluid reacts to produce a tiny electric current and then do this consistently for 10 or 14 days is amazing. Over the last 6 years I have used G5, Libre 1, Libre 2 and am now trying G6.

The one thing I can say with conviction is that every sensor is different. the odd one is almost perfect but the next one can be a bit rubbish. However I have had very few bad enough to be replaced. My consultant, who is also type 1 and has tried everything says he could not say any one was better between Libre 2 and 3 and G6 and G7.

I am currently wearing a Libre 2 and a G6. I am reading the Libre 3 ways. First with the German patch on to xDrip, second with the MiaoMiao transmitter on to xDrip on a second phone and then using NFC and LibreLink. All 3 readings are different even though I calibrate the xDrip ones. The differences are generally within 1.5 mmoll/ml at the lower end but sometimes 2 or 3 when 12 or above.

I would say that if you are getting a variation of more than 2.5 mmoll when below 6. Then this is potentially dangerous.

There are various reasons why one person may have a different experience to another. One is the persons physiology then the make of phone and operating system. Then which version of Android and IOS they are running.

I was unable to get my fitbit watch to show my blood sugar until I changed to a new phone. For the last week it has been running perfectly. Somehow my original phone did not seem to be connecting reliably even though they were both Android devices.

Dexcom of course has a list of compatible devices and my new phone is one of these. In theory this means they should be able to help you with any problems and are likely to have previous knowledge.

An interesting comment from Bill-ST
about the inaccuracy of blood glucose strips. The actual standards they have to meet are really quite low and when you add to that the sensor itself is out by an average of 10% you realize that what looks so compelling on your phone's screen is a bit of a rough guess. However much better than anything that has gone before. However everyone should get the most accurate kit they can to improve their outcomes over the coming decades.
I have an article on a blogsite I used to run called bgonmywatch.com. The article is titled "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)

 

[Bill_St]

To me the modern CGM sensor seems to be a techie miracle. The fact that the little bit of wire stuck in to your interstitial fluid reacts to produce a tiny electric current and then do this consistently for 10 or 14 days is amazing. Over the last 6 years I have used G5, Libre 1, Libre 2 and am now trying G6.

The one thing I can say with conviction is that every sensor is different. the odd one is almost perfect but the next one can be a bit rubbish. However I have had very few bad enough to be replaced. My consultant, who is also type 1 and has tried everything says he could not say any one was better between Libre 2 and 3 and G6 and G7.

I am currently wearing a Libre 2 and a G6. I am reading the Libre 3 ways. First with the German patch on to xDrip, second with the MiaoMiao transmitter on to xDrip on a second phone and then using NFC and LibreLink. All 3 readings are different even though I calibrate the xDrip ones. The differences are generally within 1.5 mmoll/ml at the lower end but sometimes 2 or 3 when 12 or above.

I would say that if you are getting a variation of more than 2.5 mmoll when below 6. Then this is potentially dangerous.

There are various reasons why one person may have a different experience to another. One is the persons physiology then the make of phone and operating system. Then which version of Android and IOS they are running.

I was unable to get my fitbit watch to show my blood sugar until I changed to a new phone. For the last week it has been running perfectly. Somehow my original phone did not seem to be connecting reliably even though they were both Android devices.

Dexcom of course has a list of compatible devices and my new phone is one of these. In theory this means they should be able to help you with any problems and are likely to have previous knowledge.

An interesting comment from Bill-ST
about the inaccuracy of blood glucose strips. The actual standards they have to meet are really quite low and when you add to that the sensor itself is out by an average of 10% you realize that what looks so compelling on your phone's screen is a bit of a rough guess. However much better than anything that has gone before. However everyone should get the most accurate kit they can to improve their outcomes over the coming decades.
I have an article on a blogsite I used to run called bgonmywatch.com. The article is titled "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)

An interesting aspect of meter AND strip is the use of their test (control) solutions. Even with Contour. Look at the acceptable Range.

 

[Martindt63]

Have used a G6 for 4 years and only have an issue with low level blood glucose levels where its possible to identify why i am low i.e. user error - over estimating carbs or exercising too strenuously. This seems to keep me low for an extended period with no positive reaction to the 15 : 15 rule until I'm suddenly 9 and rising. However following consultants advice to finger test rather than automatically have a second round of 15 fast acting I found that just added to the confusion as readings from left and right hands varied from each other and the CGM..........

 

[wes.w]

My first post, so no links allowed I need 3 posts before I can link

I'm on my second working sensor (one failed during warm-up, Dexcom are sending me a replacement). When I googled "Dexcom One Accuracy" it took me to an Estonian Dexcom page in English that seems to give more info than you see on UK/USA web pages.

Quote: A difference between blood glucose readings and Dexcom CGM readings of up 1.1 mmol/L for readings lower than 3.9 mmol/L and up to 20% for readings higher than 3.9 mmol/l is within the medically acceptable range that is the industry standard.

and, of course, my test strips are only around 5% -10% accurate.

This is a table of expected accuracies on the Estonian page.

Interesting, isn't it?

 

[Garek]

Yes, interesting. I have definitely seen results outside of those margins on a regular basis, but this may be due to my blood glucose meter being inaccurate by more than the normal range. I use Accu Chek, not sure if they are considered fairly accurate or not though.

 

[wes.w]

To update my previous message, I filled in the Dexcom report on my failed sensor on Saturday, was contacted by phone message by Dexcom on Sunday (they use an 0800 number so no charge to phone them) , and the replacement sensor arrived on Tuesday. So full marks to Dexcom - if you are willing to take the time to fill in the report.

@Garek - Accu Check have a range of sensors. Mine is a Performa (first version) and the accuracy for the strips are (roughly converted from their USA measurements):
Over 99% of test results will be within ±1 mmol/l for readings less than 5.5, and within ±15% of actual result for readings 5.5 or higher.

 

[In Response]

As an experienced CGM user (I have used 5 different CGMs over the years), I find this thread disappointing.
Disappointing because people are being prescribed CGMs without their limitations being explained. CGMs are a fantastic piece of kit but only if you understand their limitations.

I don't use a CGM, but my supervisor's husband does. One of their complaints has been that it alarms in the middle of the night due to him being dangerously low, yet he feels fine and the finger sticks concur with him. I don't know if they're generally inaccurate, or if some people are more prone to inaccuracy, but you're not alone.

"Dangerous lows' in the middle of the night which are found not to be lows are likely to be compression lows. A known issue if you apply pressure on a sensor.

I use Freestyle Libre 3 and to be honest it’s very accurate - although being c15 min behind blood sugar readings because of measuring interstitial glucose …. So stable glucose level is most accurate and going up it’s behind the curve likewise going down - this you have to compensate for yourself

Many many years ago, Abbott changed their algorithm to convert the Interstitial fluid reading to BG to take into consideration this lag. Therefore, whilst ISR is behind BG, the readings you see from your CGM are NOT behind. Other CGMs do the same thing. So the "being c5min behind blood sugars" has been a fallacy for years.

Freestyle libre is hopeless, alerted me to being low and a scan showed 3.3, finger prick showed 12.4.

Whilst I would not expect a discrepancy this large (unless it was a compression low), CGMs are designed to be most accurate at "normal" levels. I would not expect an accurate reading over 10mmol/l.
It could also be an issue of "bedding in" if this was a new sensor, Many of us find it can take our body up to 48 hours to get used to an alien object in our arms. This is why we apply our sensor a day or two before activating.

Sadly, CGM manufacturers do not share their limitations but it makes me angry that HCPs are not educating their patients better.

 

[wes.w]

I'm now on my third Dexcom One sensor. The results from my second sensor were not as good as the first and third, with lots of differences between the general graph characteristics, which showed up as both isolated lower and higher points and sometimes as a group of unrelated points, so maybe this partially explains some of what Garek has seen. This was on the direct Dexcom monitor on my phone, and not in the 'Clarity' software daily reports. The second sensor also gave quite a few disconnects between the sensor and my phone - where the trend arrow was not visible, and a 'wait up to 30 minutes' message was displayed.

But overall the device has been very useful. I'm on 22 hour insulin and I suffer from chronic fatigue, but I've managed to reduce my fatigue by discovering that I had very low levels overnight and reducing my insulin to make my overnight levels reach only slightly lows. Next task is to try reduce my highs. I've accidentally found that gluten free rolls (mainly rice flour) are a big help - my 6 year old grandaughter is allergic to gluten and we both ate these rolls with results that surprised me when I checked my level changes, which were a lot less than I've seen previously!

 

[Poppygodmother]

I trialled a G6 in December and could calibrate that. However am now on NHS funded Dexcom one and you simply cannot calibrate. My meter readings are considerably lower than the Dexcom so results in me over dosing my insulin. Totally agree that these are false economics for the NHS

 

[wes.w]

Now on sensor no.7. A couple of observations:
1. Applied in the morning. Overnight readings stayed fairly level, rather than dropping down as they have on all previous nights. If this was 'bedding in' it is the first time anything unusual like this has ever happened.
2. Took my 3 year old grandson to a rural pub, and had a sandwich and a half of beer. At 7.0 and my levels usually start shooting up within a half hour. Went outside to take him into the playground area in a large field, and my low alarm (at 4.5) immediately went off. This was followed by lots of low 'random' readings for the next 45 minutes until we left the playground and drove away, then things went back to normal. I noticed that we had been very near to a phone mast. Perhaps .... ?

 

[plantae]

When I first started with Dexcom G6 I was calibrating every sensor after about 24 hours because it was way off, but after calibration the reading were fine. In the last few months (probably the last 8 or 9 sensors) I haven't had to calibrate once -- every time I see a weird reading I do a finger prick and it's very close to what the Dexcom is saying*. I don't know if I got a bad run of sensors at the start, or if the sensors have improved, or if my body has somehow come to accept the foreign matter under my skin now but there's certainly a big difference

* Except for the first 8-12 hours where it looks like a seismograph, but even that seems more reliable these days compared to when I first started

 

[Garek]

Dexcom One on iOS has just been updated and now looks like you can calibrate these sensors! Hopefully the Apple Health integration will now work properly too

 

[Garek]

So after a couple of weeks of testing, I can say that with calibration being added to the Dexcom Ones they are massively better. Integration with Apple Health now seems to be working well too.

 

[wes.w]

I got the android update 11 days later than the iOS one, so I haven't been able to test it properly yet.

But I'm still getting a few random lows with sensor No.9. It seems some sensors are better behaved than others

 

[Garek]

I got the android update 11 days later than the iOS one, so I haven't been able to test it properly yet.

But I'm still getting a few random lows with sensor No.9. It seems some sensors are better behaved than others

I haven’t seen that myself, but found I have to be very particular about sensor placement, otherwise I get very inaccurate results. For me, I have to place them on the back of the arms just above the elbow. If I position them near a tattoo or a stretch mark then the results are all over the place.

 

[wes.w]

I'm using the sensor on midriff which gives me six positions to try - left or right and above/in-line/below belly button. Not something I've recorded or noticed so far but I'll start doing it, so thanks for your feedback. I'm naturally skinny so don't have stretch marks or tattoos .