Just started Dexcom G6 - all readings incorrect

When I calibrate it always reverts back to the same level.
So I stopped calibrating. Generally if it’s out of range it’s a bad sensor. Generally my dexcom is very close to my finger sticks. So I don’t test much.

1 Like

Dexcom specifically describes when to calibrate.

Over-calibration can “muck up” the algorithm.


thanks, I may give that a go. Do you know whether there is a cap on the range difference. For example, if the meter says 100mg but your sensor says 200mg, thats a massive difference, do you calibrate? When I was on medtronic cgms, this was a big don’t calibrate until the gap is much closer.

I wonder why they don’t have a mmol version of this rule published anywhere. I tried to look for it but cannot find it. I’ve converted it into mmol below. I think the same rule still applies in mmol. But the 20/20 name doesn’t really make sense when it’s in mmols :sweat_smile:

The 20 rule says - Dexcom G6 reading must be within:
    • 20% of the meter value when the meter value is 4.4mmol (80 mg/dL) or higher
    • 1.1 mmol (20 mg/dL) of the meter value when the meter value is under 4.4mmol (80 mg/dL)

I appreciate Dexcom’s advice on calibration, but some of that is never going to happen. I use an semi-closed system with my G6. I’m literally trusting my life to it. I DEMAND accuracy, and have never seen this “mucked up” algorithm by doing it. You just have to do it at the right time, then no problem. 20 points variation if my fingerstick is under 80 is not going to cut it. If my Dexcom reads out of range, when I’m really not, you better believe I’m going to calibrate.

The only time I’ve ever had trouble calibrating is after a restart, when there’s a massive difference. I just did a second restart the other day (currently in day 22), fingerstick was a nice, round 100, but Dex read 240. I I always have to divide those giant restart leaps in several different calibrations. I tried to calibrate to 190, but the Dex rejected the calibration. My pump told me to try again in 15 minutes. It also rejected the 2nd calibration (sugar had changed by then, but I kept the 50 mg/dl difference). By the time I entered the 3rd requested calibration, it finally accepted it and has been running great ever since. It took me a little longer than usual to get it fully aligned with my real numbers, but no problem at all ever since. I check at least once a day on days 20-30, more often if I see the raw data going erratic, and until the very end of the sensor life the calibrations work.

I will say that I use one of the highest ranking meters, though. The Freestyle lite. It’s either 2nd or 3rd place after the contour next in terms of accuracy, but does exceptionally well in the target range. I definitely wouldn’t recommend using a random number generator, like the One Touch meters I detest so much, for calibrating, though.

1 Like

Yes this can be quite a massive interpretation if you are close to the hypo mark.

At 80mg (4.4mmol) I’m still functional and normal. At 60mg (3.3 mmol), i will feel symptoms of hypo and probably be freaking out about getting some sugar into my system. If the sensor tells me i’m at 60mg when I’m at 80mg, I say damn you sensor and shut up with your alarms already :grin:. …which is pretty much what happened yesterday at 1am when the alerts woke me up.

I used to do this on the medtronic - slowly tricking the sensor closer to the number. But the medtronic sensor dies (like completely stops working and a message tells you to replace it) if you have 3 calibration errors in a row. Is there a similar thing on the dexcom?

Nope. After 3 or 4 calibrations on the Tandem, it sort of forces it to accept the calibration. Except, sometimes you really will get a bad sensor, or a legitimate sensor error. You can’t calibrate those away. The tech support at Tandem is really great about expecting your sensors to work for you. If you feel the need to calibrate more than 3 times in a day, they will send you a replacement. They’ve never even asked me to verify how big the discrepancy was. Also, if you go more than three combined hours without data in a 24 hour period, the screen will read “- - -” or “???”, they will also replace the sensor. The pump may also give you a " failed sensor, replace now" alert, which Tandem will replace.

1 Like

I actually got one of these in the first 24 hours during the middle of the night, I wasn’t sure what it was because it wasn’t a signal failure so I googled and the only thing I could find was wait three hours and then about 1.5 hours the readings came back. So didn’t really investigate why.

Thanks I will calibrate the next time I have that discrepancy outside the 20 rule. My sensor isn’t way off, but it’s off enough for me to continue to use my meter.

I calibrate whenever the Dexcom is out of an accurate range that I am comfortable with as long as my last three reading are stable and a finger stick is in range. For big differences between meter and dexcom I will calibrate it down in sections of no more than 40mg/dl, 2.22mmol/l with some time between each calibration (sometimes the first calibration actually fixes it).

1 Like

I think you may have misunderstood me The “- - -” means you’re still generating data, but the software is confused by it. Basically there’s too much variation between data points and the algorithm can’t connect the dots. Sometimes it happens legitimately with a good sensor, like if your BG is changing too fast. If it happens a lot, though, it means it’s the end of that sensors life, though, no matter how long you’ve been wearing it.

1 Like

You’re in the perfect position to do a calibration. You’ve seen a consistent offset between fingerstick and CGM and it’s more than 20% off.

I don’t calibrate all my sensors because many (most?) never read more than 10% off. But when they are consistently off in the same direction it’s time to calibrate!

1 Like

I called customer care earlier about this one because it popped up again. They told me I was dehydrated and needed to drink more water :sweat_smile:. I did recall it flashed up as a sensor error in the end but everything just resumed back to normal eventually.

Do you know what causes the variation between data points? When it happened it was in the middle of the night and I was just lying in bed looking at my cgms reading come through :grin:. So it wasn’t something stressful or physical I was doing.

A lot of things can cause the erratic data, and yes, dehydration is a common one. The sensor is reading the glucose in the fluid between your cells. If you’re dehydrated, then there’s not enough fluid circulating to move the glucose being monitored. In my experience, that leads to stagnant, “off” data, not necessarily “- - -”, but I suppose it’s possible.

I use Xdrip on my phone, and it lets me see the raw data points right along with the CGM reading the Dexcom transmitter computes from that raw data. The Dexcom kinda smooths everything out. With a good sensor, I rarely see the raw data points because they’re under the Dexcom-generated number on the graph. But with a lousy sensor, I’ll see this scatter plot of extra dots all over the screen. The more scattered the raw data, the more likely I’ll see the -"- - -"

I’m human. I eat carbs and sometimes forget to bolus for them. If I see “- - -” on a good sensor, it’s usually because I forgot a big bolus and my blood sugar skyrocketed unexpectedly. If I get the “- - -”, it won’t display any of the really high numbers and the data won’t display again until I come down into the low 200s

I’ve also seen the “- - -” when I’m feverish, which pretty much ANNIHILATES any glucose I’ve got. I’m forever trying to feed the fever carbs to get it out of the hole. That wrestling match up and down can confuse the sensor, too.

1 Like

I actually calibrated my sensor yesterday night, the sensor reading was consistently lower than the meter reading.

Today, I’m finding the sensor reading (for the first time) higher than the meter reading. The calibration yesterday was done immediately after the finger prick and I had to take some adjustment bolus to correct at the same time. Could the adjustment have thrown the calibration off now? I’m thinking maybe I should calibrate again. How much is too much calibration ?

Well if you needed to take a correction or you were going to eat you should not have tried to calibrate as in my experience that throws the accuracy off but yes you should calibrate again. Just remember when calibrating your BG should be completely stable and in range and you should not have recently eaten/intend to eat soon or taken a correction.

1 Like

Does it have to be in range as well? The dexcom range doesn’t mention anything about in range so I didn’t consider that at the time.

So when should I have corrected? Say i was at 9mmol but the sensor says I was at 6 mmol. I go to calibrate. Then I go to the bolus menu and do a correction. Are you saying I need to hold off the correction and if so for how long?

@margiek sorry I don’t mean to hijack your thread, my issues are also to do with starting out on the G6 and getting incorrect readings hence the reason I’ve tacked onto the thread. I’m conscious that my g6 issues have overtaken your initial post. :sweat_smile: I’m at the point where I’m not sure whether I should start a new thread or continue on this one. Please do call out if you want me to move to a new post.

Yes, being in range matters. There is a margin of error with all BG monitoring tools. It’s actually less on the Dexcom G6 (MARD=9‰ variation) then most fingerstick glucometers (MARD can be up to 20% variation). But the margin of error is the absolute least when you’re in range.

If your true BG is 14 mmol, your G6 should be reading somewhere between 12.74-15.26 mmol, and your fingerstick can read 11.2-16.8 mmol. Meaning an “accurate” G6 can read up to 4.05 mmol different than your fingerstick.

However, if your true BG is 4.O mmol, then an accurate G6 can read anywhere between 3.64 - 4.36 and your fingerstick can read 3.2 -4.8, bringing the potential max difference down to only 1.16 mmol. So this would be the better option for calibrating.

It all honesty, the accuracy is probably even closer within range, but I didn’t look up the specific numbers.

You also have to remember that a CGM is going to lag by 10-15 minutes, which is why we emphasize the importance of your trend arrow reading flat when you do a calibration. If your BG is rising or falling, the numbers just can’t line up. Even if your data reads flat, but you just ate or bolused, you KNOW your BG is changing and the CGM just hasn’t caught up yet. That’s not the time to calibrate either.

Just remember “in range” and “stable” for your calibration rules.


Great explanation! Thank you.

1 Like