The accuracy of your Dexcom G4/G5 CGM readings depends upon the quality of your calibration. If your calibration is off by 20 points, your next 12 hours of readings will be off by, likely, more than 20 points. So, if you want accurate Dexcom readings, you need technically perfect calibrations.
The basics
Much of what you need to do for a good calibration is already a part of your Dexcom training:
Start with perfectly clean hands. Before calibrating, wash your hands thoroughly with soap and water for 20-30 seconds. Shake the water off your hands. Do not dry your hands with a used towel – instead, use a paper towel. Let your hands dry thoroughly before doing a calibration. Please note that alcohol will NOT clean food residue (sugar) off your hands.
Do not milk your finger to get blood. It is OK to apply pressure to your finger pad to get more blood, but do not repeatedly milk the finger down towards the tip: you might dilute your glucose reading. If you have a hard time obtaining blood from your finger, first wash your hands in very warm water for up to a minute. Then swing your arms violently back and forth several times, to get blood into your extremities. Poke your finger, keeping your hand as low as possible, way below your heart. This should allow you to get plenty of nice, warm blood.
Do not calibrate from very cold hands if you can avoid it (as it may not have the most recent glucose index). Use very warm water to warm your hands first.
calibrate when within your in-goal range, if possible close to your BG target. Your CGM cannot be perfect everywhere between 40 and 400. You want it to be most accurate within range.
Do not over-calibrate: calibrate two or three times per day, no more – unless, for some reason, your CGM readings are more than 20% off. Your CGM algorithms may become unstable when you calibrate too often in a day.
Only calibrate when your CGM arrow is horizontal. This is critical. If it is not horizontal, your calibration will almost certainly be off. Yet a horizontal arrow is not a precise enough trigger: the next post will give you a more accurate way to proceed.
Hydrate well. Your CGM is known to function poorly when you are not well hydrated. A typical example is at the end of a hot day walking in the heat.
Once you have mastered the basics, there are a few more points to consider.
When starting a new sensor, your first two calibrations must be prompt. Dexcom has confirmed multiple times that separating the first two calibrations by 10 minutes is not better, but worse.
You should be ready to calibrate more than usual, if need be, during the first 24 hours, which are always more tentative, or when you are trying to make your sensor last longer: it always becomes noisier and less precise in these circumstances.
Don’t get obsessed with calibrating right when your Dexcom asks you to. Wait for just the right moment, the moment when calibration will be perfect.
Calibrate when your CGM curve is perfectly flat. The arrow on your CGM is actually a rough approximation: your eyes are better. It is best to look at the curve, and pick a moment when your curve is nice and flat, and has been for a good 30-40 minutes: sometimes, it may take that long for your CGM to catch up to your true readings after large BG variations. For instance, don’t calibrate right after a meal or a correction (wait at least 2 hours or more after taking insulin or carbs).
Calibrate when your TRUE BG is perfectly flat if at all possible. Since it measures interstitial fluid glucose concentration (rather than blood glucose), your CGM is 10-20 minutes behind your real BG, and is often not aligned with your real BG curve when your BG levels are varying quickly. So a flat CGM curve may not be a true flat BG. For instance, my son’s BG rises when he gets up from bed --but the CGM will show it after a delay of 15 minutes. So, when he calibrates in the morning, he does it straight from his bed, right as he wakes up, if his curve is flat enough.
Remember that, when you calibrate, you are calibrating the transmitter, not your hand device. So, if you calibrate from a receiver, don’t calibrate again from your phone. There is a lag between the two, so a calibration on one won’t show up on the other for 10-20 minutes, occasionally longer.
If you are more than 20% off (or 20 points off under 80 mg/dl), calibrate, then test again 15 minutes later. If you are still too far off, calibrate, then test again 15 minutes later. If you are still too far off, calibrate one last time and check again 15 minutes later. If you are still off, it is time to call Dexcom and tell them your sensor is inaccurate: more than likely they will request you change the sensor, and will send you another one.
Be ready to validate that your true BG is flat. When we are unsure of what the true BG is doing, we often fingerpoke twice within an interval of 5 minutes. If the second reading is near enough the first one, then we calibrate.
If you are desperate for a calibration but do not have the opportunity for one, look for a “saddle point,” a moment when you expect your BG curve to stall and invert, i.e. a maximum or a minimum. Because your CGM is 10-20-minutes behind, you may get a decent calibration when your CGM shows that you are almost at min/max. Hopefully, this means that your true BG is barely past the max, and, on the whole, neither your true BG nor the CGM measurement vary much at the time.
When your BG changes quickly and across a wide range, your CGM may be out of sync with your true BG for a long time. Don’t calibrate at those times! You can find an example of what happens here.
In rare situations, your CGM may still lag behind your true BG even when your BG has been stable for a while. I have seen multiple circumstances when a CGM needs 30 to 60 minutes to catch up to a large variation of BG, after the true BG has stabilized. When that happens, if you calibrate during the period of misalignment, your CGM will actually catch up to the true level after you have calibrated, and show a new discrepancy in the opposite direction:( Recently, for instance, we found our CGM to be 60 units off (!), triple-measured manually, calibrated, and found out one hour later that the CGM was off in the other direction by half as much as before.
Do not always assume that your glucose meter is right and your Dexcom is wrong. When in doubt, test two or three times. Your Dexcom has been averaging your glucose meter errors (only guaranteed to +/- 20%, remember?) for several days – it could well be closer to the true number than your glucose meter.
There are times when, for some reason, your Dexcom is doubting your calibration, or when you are so far off that several calibrations 15 mns apart (as per the previous post) still leaves you far from reality. In those rare cases, a last chance technique allows you to force the Dexcom to go exactly to one specific value. It should be used rarely if ever, since (a) each BG meter has significant measurement noise, and (b) the Dexcom, averaging these errors over several days, efficiently reaches, in general, a better value. If you need to use this technique, here is how:
average at least two BG meter readings (we often do three to be sure) to get a precise number
enter the value as a calibration
immediately re-enter it as a calibration again
this will force your Dexcom to the value entered. If it does not, your Dexcom system is defective, and you need to call it in to Dexcom customer support.
A Dexcom is an addicting tool. We would never be able to manage BG with the same accuracy without one. But it is also easy to forget how to do it without one! So consider withdrawing from your Dexcom 2-3 days every 6-8 weeks to remember what it takes
I am in debt to @rgcainmd for many of these thoughts!
Thank you so much for posting this! One question right off the bat…
I’m curious - is this just for calibrations, or is this a general rule for finger sticks? I remember hearing conflicting opinions when I was checking my BG during my first pregnancy (borderline GD), but hadn’t really thought of it since. I bruise very easily - anything more than the most shallow setting on my lancing device - which means most of the time, I don’t get enough blood without milking my finger for it.
It is a general rule for fingersticks! My son was like you to start with, but now he gets blood very reliably. Have you tried the tricks listed up in the wiki?
When you swing your arms, you want to do that truly violently, projecting your hands down strongly.
My guess, btw, is that, very quickly, you will be able to increase the penetration of your lancets (your skin and fingers are going to become rougher).
[EDIT] Of course, if you try everything and don’t get blood flowing, then milking it is . I was surprised at how much blood you can get when you apply pressure to the right part of the pad though.
I will be trying them now that I know about them - oops! I did know about the warm water trick, but I’m not always in a position to be able to wash my hands before a stick.
I have already noticed this to be the case - it’s easier to get more blood than it was when I first started checking 10 months ago (ha, just the other day I had my first spray when I applied a little pressure), so I guess it’s just a work in progress!
Btw, to make you feel good about this: this other night I had to test my son in the middle of the night. Here I am, holding a flashlight between my teeth, holding his hand in one hand and the lancet in the other - I could not get good blood – so I ended up milking too… It happens But I think that was the first time in 3-4 months?
Should we add something about the possible need for more calibration when using alternate locations? We end up calibrating more on lean body parts such as the thigh.
Just for kicks, I am trying the 15 minute wait between the two initial calibrations. I’ve never done that (at least on purpose). So I’ll see if it makes a difference.
Although, based on the back-and-forth above, I do think additional calibrations within the first 24 hours might be helpful to get it started on better accuracy faster. Next sensor start, I am going to try for an additional (obviously have to do initial double calibration to start) 3 calibrations within first 24 hr and make an effort to spread them around at various BG levels. Not going to change the BG intentionally for the Dex but maybe catch a level at the low end of the range and catch a level at the high end and such.
After that back to the 1 or 2 calibrations per day. If we calibrate in the morning and the difference is single digit then often times we don’t bother with the evening calibration and just ignore the “request” from the Dex for the calibrate.
I do not believe that it will have a positive effect for measurements when you are in range. But I think it would make out-of-range measurements more precise.
I will be interested in what you see. It may take many experiments to see a pattern emerge, though.
We just did a sensor restart. Back-to-back calibrations. Actually used the same drop of blood. This new Contour Next One meter (thanks @Eric !!!) uses so little blood it is insane. After two strips on the same drop there is still this huge drop left. Obviously we are not used to the new meter yet. So anyway, the readings are 97 and 95. Entered into the Dexcom and it comes back with 96. I think the Dex simply takes the two readings and averages them and uses the result as a single calibration point. Obviously a total guess.
Yes, that’s always been the case for me on the back-to-back calibs, it’s just the average of the two. Every single time I have done them back-to-back it is an average. And my readings are always close because I use a good meter.
[deleted my rambling story]
[but then it got long again]
[sorry]
Anybody use a “double tap” calibration technique when the Dex is really far off?
For example.
cgm reads 42. Meter reads 86. Sensor is only 2 days old.
regular calibration causes cgm to pull up about halfway to 65 or so (more or less - forget exactly). I am thinking the calibration is toast based on how far off it is. So, IMMEDIATELY, I enter the SAME meter reading into the dex calibration again. The cgm immediately jumps up to exactly the meter reading.
I have done this a couple times in the past without thinking much.
But with all our back and forth about how the internals of the calibration work, this got me wondering.
Is it possible the Dex has a built-in calibration reset sequence which is triggered by a “double tap” calibration? Could this erase prior calibrations and allow a “do over” without the need to actually stop and start the sensor and put up with a 2-hr blackout?
Our calibrations are normally within single digits and typically only with 1 calibration per day. So being off by over 40 points is most unusual for us.
I do exactly the same thing when things are out of whack. I am not 100% sure of exactly what happens inside the Dexcom though. Let me review my notes on that.
. I was surprised at how much blood you can get when you apply pressure to the right part of the pad though.
I did know about the warm water trick, but I’m not always in a position to be able to wash my hands before a stick.