I have to agree with most others here. The times that I DO calibrate, the sensor eventually ends up failing. I think it is best not to calibrate it and let it, sometimes very slowly, trend back to what it should read. This may take a full day after sensor insertion, but hopefully not that long.
In the mean time, you’ll have to do more finger sticks to let Loop know the actual BGs.
My G6 experiences have differed from some of the reports in this thread. I’ve had decent accuracy with no calibrations (using the code) except for (1) restarted sensors seem to need a few calibrations to bring them into range and (2) the last death throes of a dying sensor. Lately I’ve given up on the whole concept of restarting sensors or a predictable sensor 10 day life. I’ve seen sensors die like this:
This is on Day 8 for this sensor. The odd thing is I can’t seem to get more than 8 or 9 days per sensor lately. The way they give out (False low alarms and 1-2 hour sensor errors) is a PITA.
Three image above, that’s without calibrating? At that late stage then i would agree it’s just a dying sensor and a good excuse to get a full replacement.
No calibrating, I was half asleep for all that! After I posted the image the sensor came back to life and was about 25% lower than actual BG, so I calibrated. Replacing the sensor later today regardless.
I never calibrate. It never makes it better. If the number is wrong, I just wait it out, to be honest.
It never makes it better but it can sure make it worse
Some pics of the old sensor and transmitter. No blood noticed.
The current session was set with no code. Will see what happens when I single and double calibrate when necessary.
Hey! When you do start a new sensor and notice the readings are off (say, much higher or lower) what is your general calibration procedure? Do you enter the exact number or increments of 20%? I started a new sensor yesterday, but it’s been reading 40 pts lower this morning. I am too afraid too calibrate given the calibration loop my last sensor threw me on so have been waiting it out. Was curious what others did 
I calibrate it with what exactly I got from my meter. I don’t know why you would do otherwise? However, I will say the times that it is really off, it is because there has been blood in the sensor. So it doesn’t matter how much I calibrate it, it’s never going to be right.
Agreed, the assumption is that the number you’re correcting is being corrected with a correct number. 
So if my CGM reads 180 and Liam is only 110, I would enter 110 only 1 time and it would end up settling somewhere in the middle on the next tick. If the CGM reads 300 double up and Liam is actually only 110, I would enter 110 TWO TIMES back to back to “reset” the CGM…with the G6 that may result in calibration loops / sensor errors, but that’s OK with me…better than showing 300 for god knows how long, waiting on the CGM to fix itself.
Try using one for a full 10 days without any calibrating. See what you think. In my observation the overall accuracy improved tremendously… all it seemed to accomplish by calibrating when it was way off was to throw it further and further off when it other wise wouldn’t have been… quite a bit different from how the g5 worked
My advice is don’t calibrate at all, especially if within first 24 hours. If I do, if anything it takes longer to adjust back to on track. In my experience so far, 95% of the time, it will either sort itself out or it’s a bad sensor (rare, but has happened, usually bc of blood). But 40 points off within first 24 hours makes me think more likely the former.
This isn’t new with the G6 for Liam. The G5 system was never accurate once his BGs were very high (> 250) or very low (< 60). The CGM has always been used by us mostly for seeing trending data. But from my experience (extensive testing of my own over the years) if I wait for the CGM to “fix itself” that means I’m low/very low as well as high, hours longer each day than he actually is. His low/very low, for the past year, has been around 1%… Sometimes near 0%. His very high is around 1% and high of between 6-10%. But when I just wait… That results in his low/very low being between 5-8%… Sometimes higher…and I find that unacceptable. As a parent who has to have discussions around this topic every few months with the Endo, I want the data to be as close to accurate as possible. My system works great most of the time, so I’ll be sticking with my own process but it definitely isn’t for everyone… For one thing, it’s a lot more work on me… But I’m ok with that if it means having correct data being reported and acted on.
For highs, if the number remains high until it corrects itself, that also means Loop is giving him more insulin than he needs which means corrections later. Much easier for me to just FORCE the CGM to display correct data… I have observed no long term inaccuracies resulting from this process. His CGM has just never been accurate when outside specific BG ranges.
It’s not new to anyone, but how the system responds to calibration is different with the g6…
Yeah I would agree with Sam. Though frankly, I found both the G5 and G6 quite accurate except for exaggerating extreme blood sugars and lagging on steep change like recovery from lows.
I do find that G6 self-correcting leads to more accurate blood sugars faster than my fiddling with it, like @Sam. That said, my endo only looks at my Dexcom data in whatever ways I ask him to, so I don’t feel like I have justify anything, so maybe I’d feel differently if working with a pediatric one. It’s kind of frustrating that they can’t just factor that in though!
I’m on my second session. No sensor code entered and it’s been night and day performance since the 1st session. Multiple calibrations done so far this session and zero calibration loops or sensor errors and outside of the extremes where I calibrate, the accuracy has been spot on. So, this is probably what we’ll stick with moving forward. Although I may try a session WITH code and no calibrations until the lagging system forces my hand again.
So far I’m loving the no sensor code session.
I may have missed it, but how was accuracy the first 8-12 hrs with no code? That is my major gripe with G6.
May have to try the no code option.
Accuracy and performance have been great so far. The BGs only went off the rails once so far (indicated 250 double arrows up when he was actually only 113), but the double calibration fixed it immediately w/o any calibration loops / sensor errors. Two ticks later, the CGM read 121 and the finger stick read 118. That one calibration saved me probably 1 to 1.5 hours of “false high” readings since, as we all know, it takes the CGM MUCH LONGER than it should to come up from a low and to come down from a high.
Just to throw my $0.02 into this highly charged topic…My habit has been to judiciously use calibrations in the first day or two of a G6 (with code) and I have been satisfied with the accuracy going forward from that. By judiciously, I mean calibrating when BG is in the range I expect to find most useful. I am not using Loop, but I want the low alerts to be accurate and timely. So I try to calibrate only near the low end of my range.
Example: About 26 hours after starting my current sensor, my meter read 64 and the G6 was alarming at 59. I calibrated to 64 and moved on. That was the only calibration this sensor received. Today the accuracy is fine. I am not convinced that judicious calibrations will harm the accuracy going forward. It seems logical that over calibrating would cause accuracy issues but occasional calibrations don’t seem to harm anything in my experience and I think they help.
Glad you have hit upon a strategy that is working for you. Hope it continues working.
