Experience with Precision in CGMS values

As I was reading this article that appeared today in the news:
(it appears to me that the description of “flocking away from finger pricking to CGMS” may be an exaggeration), I was wondering about the precision of the values you guys see in your CGMS monitoring and how they compare to the values you obtain when you test with your good ole glucose meters…

On the first day, it’s sometimes pretty shaky for both Minimed (I’m in a 30 day Guardian study and I used the 722 for 6 months in a different study) and Dexcom (owned STS since June 2006, upgraded to Seven this past summer). After the first day, my Dexcom is usually very close, meaning within 10% of the meter number, or about as close to the meter number as a second finger stick usually is. The 722 was right on with BG over 120, but tended to lose it in the low range, consistently saying I was 95 when I was 60, 130 when 28, etc, etc. With the Guardian, it’s usually closer in mid-range, say 120-200. It has the same problem with lows that the 722 did. And I’ve also noticed that it isn’t great with highs, usually if I’m over 200 it may say I’m 220 when I’m actually 280, things like that.

The bottom line is, I almost never use the monitor number to dose insulin. The exception is when I have a sensor that has been right on track with the meter for some time, and I’m doing a small correction say 90 minutes before a meal, when I’ll be doing a fingerstick anyway…

This article, which ran in my daily newspaper looks like a Press Release from Dexcom.

No one can use a Dexcom without also using finger sticks because you have to keep recalibrating.

And you can’t dose insulin based on the monitor either because they can be very off.

That said, the more insurers are motivated to pay for them, the better.

What I’ve found (consistently) is that my MM CGMS is very accurate (within 5 - 10% of glucometer) on days 2 - 6 of a sensors life, as long as I keep my BG in decent control (80 - 150 range). Day 1 tends to be less accurate. Days 7 and on are hit and miss (some sensors good until day 12, others fail at day 7). If I mess up and shoot over 200, the CGMS won’t go as high as the glucometer (maybe 20% in error, on the low side).

While the manufacturers can’t recommend (it hasn’t been approved by the FDA for that) a patient bolus their insulin based on the CGMS without a confirmation finger test, some users (myself included) actually do that some of the time. With experience you can know when the CGMS can be trusted and when it can’t. I don’t do this all the time, but on those occasions when I’m confident in the CGMS accuracy and it’s not convenient (or possible) to do a finger test, I’ll bolus just based on the CGMS. I also frequently do additional small correction boluses based just on CGMS readings.

When I have a CGMS sensor in (on the known accurate days) I generally only do 4 - 6 finger tests. On days when I don’t have an accurate CGMS sensor in, I generally finger test 8 - 12 times. So it does reduce the number of finger pricks, while simultaneously helping me achieve tighter control. To get the equivalent level of control, I’d probably have to finger test 15 - 20 times.

I’ve been using the Dexcom (original version) since February, and I have to say that it really has been very accurate to my glucometer readings. Day 1 and then the last couple of days of sensor use (can range from 10 days to almost 3 weeks depending on when the sensor starts to fail) are the most inaccurate days. I do bolus correction doses based on Dexcom’s readings. If it’s a large dose or if something just doesn’t seem right, I do a fingerstick confirmation. I do fingerstick calibrate throughout the day, also, not just the 2 daily recommended times. I have found that it definitely helps keep better control. My A1C went from 6.9 (January) to 5.9 (August). More importantly, it really helps with the trend data to prevent lows before they happen.

Does anyone have experience with the Freestyle Navigator trials? I’d be very interested in knowing if the accuracy was improved enough that confirmation finger sticks could be done away with (other than when calibrating, or when sugars are changing rapidly). I want to do away with fingersticks as much as possible.


I just found out that Dexcom has a serious problem with readings when the individual takes Tylenol. I was told it affects the old one and the dexcom 7. Anyone else heard of this?

When I had a one on one training with the Dexcom rep, she told me not to take Tylenol. She also gave me some reading material about it. I never take Tylenol so I knew it wouldn’t be a problem for me, but I do think there is some isuue with it when using the Dexcom STS or Seven.

Dang. Any device that won’t work if I take Tylenol is not going to work well for me. That is the ONLY pain pill I can take and I have serious orthopedic problems. I do my best to keep doses as low as possible, since I know that at a certain dose over a lifetime it takes out the kidneys and I’m probably heading towards that dose. But still.

I was thinking of investing in a Dexcom, too.

I have found the accuracy with the FreeStyle Navigator to be very good. I used to do 10+ finger sticks per day and I am comfortable making bolus decisions based solely on the CGMS data. However, the key is to have very good blood glucose control and understand the disease where one is able to make informed decisions. Since I typically range between 70-140, I cannot attest to the accuracy for extremely high levels (although it has been accurate when I’m in the 50s). The accuracy hasn’t changed from day 1, 5, or even 10 (assuming you use the 5 day sensor 2X). On a side note, I’m still in my honeymoon phase (c-peptides are still measurable) and this has helped me with my control.

With that said, there are definitely problems with the Navigator (durability, etc) that won’t be fixed until after FDA approval. Thus, I would recommend obtaining either a Minimed or Dexcom until Navigator’s next generation model becomes available. I do believe the Navigator can be accurate enough to eliminate finger sticks.