CGM Accuracy

Personal Case Study:
FreeStyle Libre 2 Study.pdf (178.1 KB)

From Abbotts website:
IMPORTANT: The device may inaccurately indicate hypoglycemia. The results of the clinical study conducted for this device showed that 40% of the time when the device indicated that user sensor glucose values were at or below 60 mg/dL, user glucose values were actually in the range of 81-160 mg/dL. Therefore, interpretation of the FreeStyle Libre Pro Flash Glucose Monitoring System readings should only be based on the trends and patterns analyzed through time using the reports available per the intended use.

I tried the Dexcom G5 in the past and got readings that were off by anywhere from 1 to 50 mg/dl. So, I stopped using it. I tried the Freestyle 1 (10 day) Libre and got similar results. I’m currently about 5 days into testing the Freestyle 2 (14 day) Libre. So far the readings are off by anywhere from 1-30 mg/dl. More accurate than the 10 day and G5. But, not accurate enough in my opinion for insulin dosing. I tested 5 times a day on an empty stomach before meals with the CGM and the FreeStyle Lite test strips.

I reviewed the details of several CGM studies. The methodology is to administer a bolus of glucose either orally or I.V. Then, measure glucose from the CGM and lab chemistry readings. This is all well and good. However, we don’t live in a laboratory under controlled conditions. We live in the real world where activity varies along with many other factors. It would be interesting to see others’ real-life experiences with CGM.

Update: 11/21/19 - FreeStyle Libre 2 vs finger Stick Study attached in pdf format.


It’s all what you’re comfortable with. Personally I wouldn’t consider around 30mg/dl off to be a big deal. I can’t dose accurately enough with the tools that are available and the million variables that affect blood sugar throughout the day for it to make much of a difference if I think my BG is 100 and I’m really 130 for example. I go by experience and feeling as much as what the CGM says. If it says I’m high and I feel high, it doesn’t matter much to me if it says I’m 200 and I’m really 250. The important thing is that I know I’m high, can take insulin and then monitor using the CGM to see if I need to take more. I can tell when I’m low very easily, so if it said I was 50-60 and I felt fine, I wouldn’t eat anything. However, if I felt low and it said I was 100 I’d probably take a few glucose tablets. The convenience of a CGM or Libre greatly outweighs having to fingerprick throughout the day in my opinion, and the trends are invaluable. If you need to know exactly what your BG is at a given moment then I guess you can’t rely on them completely.

1 Like

Since you asked, I’ve used the Medtronic system for over a year. I HATED it! It was a random number generator, would fall off from sweat despite me appling a variety of things to prevent that from happening.

then I got the G5. It is so accurate that yesterday that when I tested before breakfast, it matched my meter. More than 12 hours later when I tested it was…wait for it…a perfect match yet again. SOMETIMES it’s off more than I would like, but given that I’ve worn it for 2 years and have had just 1, maybe 2 sensors replaced by Dexcom for a perceived issue, I’d say that’s pretty freaking great. The sensors have saved me from harm more times that I could possibly count.

Because of my good luck with the G5 I’m quite concerned about the upcoming switch (I have Medicare) to the G6.

I have used the 10 day libre in the past and am now using 14 day one on my dog.

I actually dosed on the 10 day Libre. What I found is when it was off 20 points, it usually was always off 20 points. So I would just add the number I needed to. I used to call in any that were off by more than that as I was uncomfortable adding 40 points on the fly to a number to dose.

But the G6 with it’s calibrations has made mine stay within 5 points usually of accuracy. That’s important to me because I do dose by it and like to go days plus without finger sticking. I usually only finger stick a couple of times the first day to calibrate it and on a new sensor, the next day to to check it. And then around day 5. On the restarted ones after the first day I usually don’t check it unless I think somethings wrong.

A CGM gives you 24/7 numbers that you can respond to as needed. There’s no way you can collect as much information by finger sticking. Did my dose work for my meal, how high are my numbers going, how soon am I responding. What do my numbers do while I’m exercising and sleeping etc etc.

And the alerts have helped me refine my control, so much so that my last A1C was 5.1%. I would not have been able to do that without the Dexcom and it’s alerts!

I do calibrate mine at the range of 95-105, since that is the area I want to be the most and am deciding doses the most. If it’s off reading lower I know I have to snack on something anyways so who cares and if I’m high and it’s off, same thing I need a correction shot anyway that might fluctuate some anyway. So as long as it stays within 5 points around 95-105 I’m happy.


The Dexcom G5 is totally awesome. I trust it 95% of the time as I dose, eat and exercise totally based on CGM. It takes about 3 months to conquer all the ins and out of this CGM but once the time is invested, there is no way I would ever want to go back to the 10-12 finger sticks a day for far worse results. The Dexcom really shines when you learn to properly pre-bolus at exactly the right time with the exact amount of required insulin to keep your BG close to flatline based on the food you are going to eat. This can not even start to be accomplished in any meaningful way with finger sticks. Learning the BG curves of all your favorite foods again allows you to fine tune and flatline your bgs over time, which can’t be done properly with finger sticks.

He or she that looks for the perfect CGM that is 100% accurate 24/7 regardless of diet, exercise, hormones, and does not want to learn all of the little quibbles associated with any CGM, there is nothing currently on the market. What there is, however, once properly self trained to use it will vastly improve results obtained from finger sticks.

1 Like

Yes, directional arrows are as important as the CGM values. e.g. If I was over 200, bolused heavily to get it down, then had unplanned activity, I might see the readings begin to fall faster and faster at some point. A reading of 120 with a straight down arrow likely means I’m under 75 at that point. Time to either test or just have some carbs. Otherwise, I might be under 55 in just a few more minutes.

1 Like

All part of the learning process which is why I really scratch my head when patients say they tried a CGM for a few days and it was not perfect. A lot is based on trends and it takes time and experience to optimize insulin dose or food intake based on those trends. That is much more important than having an exact match between finger stick and CGM.

I could keep a 4.6 a1c with just using finger sticks. After 60 yrs of type 1, I just got my first CGM last month. When my G5 is working well and is off by 10 or 20 I am happy with it, but sometimes I am off by a huge amount. I trust finger sticking always, I trust the GMC at times. I kept a 4.6 to 5.2 with finger sticks for over a decade. When I switched to a CGM, I didn’t really get any surprises. I eat close to the same amount of carbs everyday and I tested enough so that I usually knew about where my glucose levels would be.

I do find the CGM handy at times, but I could easily live without it. I always wake up at night if I am low, if I didn’t I wouldn’t be without a CGM.

I hope that the G6 will be more accurate for me. The G5 became more accurate when I really learned more about it and stopped getting compression lows etc, but I still can’t trust it’s accuracy.

I must be weird or extra sensitive, but when sleeping I like to be able to move all around without worrying about a compression low. I love the freedom of movement. I doubt that I will ever become a pump user.

Wow. 4.6. I know at least 2 endos that would complain about a T1 having such a low A1c. :slight_smile: One of them, years ago, would complain about my wife’s low A1c’s but she didn’t have lows, so he should have taken that into account and kept his mouth shut.

My GP at the time made me go to her favorite endo for another opinion about my low A1c’s. He looked at my readings and told her to leave me alone! He thought I was doing fine.

I was low carbing at the time and could keep my A1c’s very low.


A man with one clock knows what time it is. A man with two clocks is never sure.

The tudiabetes translation:

A tudiabetes poster with one meter knows what his bg is. A tudiabetes poster with two meters is never sure.

I know we’ve had folks here who demand that until bg meters all match each other to the last decimal point the FDA has to punish the bg meter makers. But in real life what I mostly need is to figure out whether I’m low, normal, or high. And beyond that it’s incredibly valuable to see on the CGM that I’m heading low, flatlining, or heading high.

I have found a CGM incredibly valuable. Just using for past 3 months and already many improvements in my control. After using it to adjust my basal (which was obviously too high as seen in the CGM graph drooping overnight) I’ve actually begun to lose weight.


Just be sure you aren’t one who’s CGM readings drop while in bed. If I were you, if I woke up during the nite, I’d do a finger stick to verify bg levels. My wife has issues with false lows with her sensors, while I have very few similar issues with mine.

1 Like

Yes, rolling over onto the sensor can falsely lower the CGM reading. Rarely is it enough to set off the hypo alarms but sometimes it does.

It’s the longer-term overnight trend that I’m interested in and it’s pretty easy to look at the graph and ignore the ups and down (real or not) and see the overall trend. That’s a huge benefit I get from the CGM in setting my basal.

I don’t mean low readings from pressure on the sensor. I mean lower-than-correct readings simply from lying down.

1 Like

Yes, I also notice sometimes if I take a correction and am lying down or very still (such as if I take a correction before a long drive), the CGM won’t show me drop until I get up and start moving around. However, I don’t know if it’s because the insulin isn’t absorbing until I move around, and then absorbs rapidly, or if it is the CGM.

I have the same issue, but I’m near certain it’s insulin absorption. I tried using Afrezza, which hits the blood stream very quickly, and the response was fast. With a pump bolus I could sit at the same BG value for hours if I don’t get up and move around. On the other hand, once my BG starts to fall with Afrezza, it’s not nearly as fast as when I’m up and moving around. So slower metabolism must be part of the equation. My endo concurs.

Mine is usually within 10 points, sometimes off by as much as 20. But still accurate enough for me to know when I’m low I’m low, and when I’m high I’m high, and show me the trends I need to know for how different meals affect me.

I’m also grateful to it for waking me a few times when I dropped low at night. The numbers on my meter and the Dexcom didn’t exactly match but both told me I was in an emergency.

So, off by 50 is a lot, but if you can calibrate and get it to within 20, I’d recommend it.

Editing to add I’m talking about Dexcom 6

1 Like

I’m gathering data over the next couple of weeks for my appointment. I’ll post it. It will include G6 vs manual check. You might have to remind me to post it. I just started today, so remind me in about a week.

How many times per day?


Well, I can post what I’ve got so far. I am looking at the accuracy, but the sensor just fell out…of course. If you look at the V(x) column - thats showing the % difference between the manual sugar and the G6. I made the reading bold whenever its less than 20% difference.