How consistent are Continuous Glucose Monitors - New Study

Many people here (including myself) are using CGMs to allow us to better understand how foods and medicines impact our blood glucose. There is research to suggest that if you can keep your blood glucose levels lower (e.g. 70 to 100) with fewer high spikes above 140, you may age more slowly and maintain health longer (all other things being equal). The problem is that the Current CGMs are not as accurate as we would like. See this new study:

Full Paper Here:


This sort of inaccuracy is one way that diabetics end up in the hospital: they overdose their insulin in response to an inaccurate measurement. I once had a bad batch of blood glucose test strips that all said my test was 250+mg/dL, when it really was about 100; I didn’t believe it, so I measured with my other meter–otherwise I would have sent my blood glucose plummeting with an insulin dose.

My friend who has diabetes warned me to test with the blood test strips in addition to using the CGM so you have an idea of what the “real” curve is. My CGM is pretty consistently 15 to 20 points higher than my finger prick measures - so I just adjust accordingly in my mind.


at steady state my CGM is right on with the test strips
during state changes, the CGM is about 20 minutes behind my blood test strips

1 Like

Which CGM brand do you use, Paul? My experiences with the Freestyle Libre 2 have been pretty awful (exaggerated hyper and hypoglycemic responses compared to glucometer). The last one I used even stopped working after 6 days. The company sent me a new one, but I don’t even know if I want to bother with it any more.

Freestyle Libre 2
this is the third one I’ve used and all were right on the money.

c) Misclassified meals were more likely meals with low carbohydrate content.

this makes the least sense - i’d have expected the opposite

I find the libre to be directionally accurate. Once i finger prick a few times, i can figure out how far off it is, and it generally stays around that far off.

The dexcom I believe lets you manually calibrate it using finger stick data

1 Like

I found the Libre 1 and 2 to be regularly around 0.5-1 mmol/l (10-18 mg/dl) higher than the finger sticks that I also have (I use two different ones, that always are quite close, within 0.1-0.3 mmol/l). But the Libre 3 (I used only one of those yet) was terrible, with at least 1 and up to 2.5 mmol/l (45 mg/dl) difference. The difference was minimal at lower value and maximal at higher value. With the Libre 1 and 2, I had a HBA1C of 33-34 mmol/mol (5.2-5.3%). With the Libre 3, I was almost diabetic at 6.1%.


Is there no way to calibrate them to your manual finger stick?

Factory-Calibrated Continuous Glucose Monitoring: How and Why It Works, and the Dangers of Reuse Beyond Approved Duration of Wear

CGM Tips and Tricks for Better Accuracy and Less Frustration

By Adam Brown

What I’ve learned from 50,000 hours using CGM

Published: 02/27/2017


Unfortunately the Freestyle Libre 2 can’t be manually calibrated, not sure about the new Libre 3. Only the Dexcom can, as far as I know.

Yes, but it’s complicated and not doable simply with the Libre app. You have to use some sort of hardware, a special transmitter, and a third party app to override the factory calibration. It’s a no-brainer for diabetics, but as a healthy user, I skipped it.


its still directionally accurate
i find that however it self calibrates, is how it is the entire 2 weeks
sometimes low, sometimes high, and sometimes spot on

keep in mind, that even if accurate, it wont fully line up with a finger prick
i usually finger prick a few times in the beginning, and then factor in the difference after

1 Like

When I used to do finger pricks I had to keep track of which device I was using because one device was always about 20 mg higher. I haven’t tested my CGM with finger pricks but I did take a reading when I was waiting to have my blood drawn. Quest came back with 79 mg and freestyle libre said 79 mg when my blood was drawn.