CGM Accuracy

Has anyone looked at the overall accuracy for their CGM?

I was curious if anyone has put anything together that looks at the CGM readings vs Blood Glucose meter readings that they use. And if so, what criteria do you use for accuracy? Something like 5% i.e. CGM 105 BG 100 or more like 7 - 10%

I was at the doctor's with my son last week and this came up. He is not on a cgm and I am (sometimes). The diabetes nurse educator said they were 70% accurate. We both have a medtronic paradigm system. I have found that if I calibrate when my blood sugar has been stable it is fairly close and very helpful. If I have the meter syncing to the pump and uploading all of my BG results it is not. So without the auto-uploading and calibrating 2x daily when the BG is stable I have had good luck.

The numbers aren't as valuable as the trends. However, I tested a lot in the beginning and discovered trends that make the numbers more useful. When I'm higher than ~140, the CGM shows about 20-30 mg/dl higher, but when I'm lower than ~100, the CGM shows about 20-30 mg/dl lower. So as long as my CGM is showing between 70 and 200 (the 200 is usually after meals), I don't worry too much about it. If it shows higher or lower than that, I test with my meter before making corrections.

I have a Dexcom Seven Plus and have been using it for about a year now. The question of accuracy comes up once in a while, and I can say that it's really not a concern (at least with Dexcom; haven't tried others). If you calibrate too often, or compare the numbers closely, they will not be spot on all the time. But like Natalie said, it's much more about the trends. The Dexcom is always a bit delayed (~15 minutes), so if you compare numbers while your glucose isn't steady, than they won't be spot on. But the trend is there and gives you an idea of how fast you're headed up or down, and that is the valuable information.

For the first month that I had the Dexcom, I was still testing with a meter many times a day. I tried to track the accuracy by comparing those numbers, but ultimately learned that I was over-calibrating. Now I only enter a calibration once every few days. In that time, I kept a chart of this data though, which I have uploaded here as a picture. I didn't correct for the time delay either, which is why the "line" is a bit fatter. The average accuracy was 21% off during that time, however the average glucose reading between the two methods was only 4mg/dl off, which is amazing.

I should probably do this test again sometime, since I now know how to calibrate better. I'm sure doing it right can get to 90% accuracy with even less variability. If you account for the time delay also, it would come even closer.

5525-dexcomerror.jpg (26.2 KB)

Thanks for the replies!

I know the trends are what I need to look at, but there are those days where I don't feel the trend is right because the readings have gotten off. I don't like the thought of my Endo using bad trends to make decisions. Maybe the new MM Carelink software they are rolling out to Professionals will help eliminate the bad data trends.

In my opinion, the endo is there for suggestions, but it doesn't have to be a one-sided conversation. If they are suggesting basal rates or bolus amounts based on your CGM reading, but you think they shouldn't change in that direction, than let them know. I only see an endo 1-2x a year, and have learned how to just make the adjustments myself. For example, lately my Dexcom was showing a slow rise after 3am, so I upped the basal during that time by about 25%, and for the next few days watched the cgm closely and the problem was resolved. I don't think every basal change needs the approval of an endo, and especially when we get 250+ readings in a day, our judgments are easily seen in real time. Before I had a cgm, that just wasn't the case.

You could check your blood glucose numbers against the cgm numbers and make a graph to see for yourself. It's a simple graph, just made with two columns. Take a bg reading a few times a day, and write down the cgm reading from the same time. Than you can use Excel or OpenOfficeCalc to have it whip up a quick graph like that. As long as there's more or less a diagonal blob, than it's pretty accurate. If you see a lot of points off in the distance, than there may be some calibrating issues.

I really hope my endo doesn't put too much faith in the CGM. When my sensor is failing, I start to get VERY low readings that are inaccurate. Last night, my sensor tried to tell me I was 59, but when I tested, I was 125. The biggest value of the sensor to me is the trends -- if it says I'm high, then I probably am, even if the number is not the same. It stimulates me to check my BG more often than I probably would have otherwise, and I'm sure I'm going to get a lower A1c because I'm correcting more often.

I hope your endo turns out to be sensible! :-)

Trends Come and Go… After all these Yrs? Nothing is Consistant, except the InConsistantcy of Both the Insulins, Test Meters and Our Own T1…

Gain or Loose a Few lbs?

Eat Differently

Different Seasons

More or Less Acitvity

Liver functions

and of course? STRESS

and the list goes on…

All that Contributes to Long Term Trends are not reliable

Remaining Fexible, doing the usual 8-12x testing and more if your are Going to Be Physically active are still the keys to getting that magic 6% A1c and avoiding as many hypo’s as you can… But , if you want to Get Tight Control? You’re still going to get those hypo’s, just test enough to catch them in their 60-70’s and not less , is key I found…

And w/o Insurance willing to pay for all those Test strips?
Or To Get the Gov’t to take the Mfg Co.'s to court and force them to Either make the Price alot less and not use those profits to fund their other Research of everything else they do ( Not for DIabetes) , which is just their way to get us to fund their Reserach for everything else… Source: Story by John Stossel-FOX News

In Mexico and India and other countries they pay 50% Lower Prices for the same Test strips we have… Forcing Us Americans to be funding everything and everyone else…
If you don’t have Insurance and have to Pay out of Pocket? Your screwed… and I beleive over 55% of T1’s fall into that category… See why Obama’s Plan was so Vital for them?
And the same for Seniors… That cannot afford to pay another $3-$500/mo for Drugs… and the donut hole is a joke.

Ask the Pump companies How many Pumps are out there? What’s the % of T1’s that have them? And how come , if they ARE the ultimate Solution currently? How come ALL of them don’t have one?

And ever wonder why? How come we have so many of Older t1’s living such long lives after having so many yrs of High BG’s?

But other than our Eating and Physical Activity , the only other thing we can Control is our BG’s… everything else ? Not so much… and It’s all those “everything else”, I think plays a much bigger role than they want to tell us… If they did tell us the Truth? Maybe we would Do more to Find a Cure, rather than use these Costly Band Aids that only the Few can afford to have…

And of course, alot less $ would be made from the Test strip , Pumps & CGM Mfg. co.'s…

I keep hard copy logs, that look like this: @ whatever time = Bg# and cgm is = SG#.

Looking at those each day, sometimes they do appear to be off. However, after uploading and looking at the trends, they are almost spot on, for the most part. Even checking the data table, they look to be fine. It is looking at the Guardian screen, that tends to make me wonder. No pump here.



You may have already started going back up. Don’t forget about the lag time.

Like you mentioned in your other post, I also will not correct from the cgm, nor do an injection based on the cgm.

Having said all of that, there are many times, that I believe the cgm to be more accurate than the meters. I also thought that, when I had the navi.