Started my CGM - Should I eat my words?

As some of you may recall I ordered a Dexcom G5 based on my endo repeatedly saying that the only way I could have an A1c in the 5s is if I was going hypo and that was probably occurring overnight. Well the G5 came this week and I started it up last night and did my initial calibration about 9pm. I went to bed soon afterwards but was awoken to an alarm. I apparently missed the low alert because the first thing I heard was the severe hypo alert (< 55 mg/dl). The dexcom said I was in the 40s.

I was tired, so I didn’t even bother to go downstairs and test, I just treated it and within a short time I was back up in the normal range and remain fine the rest of the night. I know I should have confirmed the low with my meter. My bad. But I also am not all that clear that I should trust the CGM in a run up period.

So the question is whether others have found the Dexcom to give “wonky” readings during that run up period in the first day or so. If not them I might end up hanging my and sheepishly walking into my endo’s office admitting that I have probably been having overnight hypos despite my failure to find any with my meter. What is your experience?

Did you have low symptoms? That 55 could have been a 75 or even < 55. I regularly experience first day readings that range from highly accurate to wildly inaccurate. If I were you, I wouldn’t draw any conclusions regarding your overnight patterns yet. You will likely know after a few weeks.

Good luck with your continued CGM use. I’m currently at an ideal 82 mg/dl and drifting sideways enjoying my first cup of coffee for the day.


I didn’t have low symptoms, typically when I have had detected hypos before I would feel symptomatic and I would feel the symptoms when in the 60s and below. But that doesn’t mean I had not experienced undetected hypos with no symptoms.

My CGM is reading 93 mg/dl and steady but my meter says 113 mg/dl. I suspect that it may just take some time to refine the calibration.

I tend to calibrate more frequently during the first 24 hours but still limit it to the > +/- 20% of the fingerstick in the high range and +/- 20 mg/dl in the lower range. I also do two fingersticks and average when calibrating. If the two fingersticks are far apart, I do a third one and use the middle reading. I don’t know if this is better or not but it makes sense to me to moderate the inherent inaccuracy of meter technology.

For me it depends. Sometimes it is relatively close in the first 12 hours and sometimes it is way off. No clue what the difference is other than potentially something with the sensor placement.

Side question on the G5 receiver. Do you have issues with the buttons on the receiver?? I just got my g5 this week. My g4 was very quick to move from screen to screen when clicking the buttons. The g5 seems like I have to hold the buttons down for 3 or 4 seconds for it to register that I’m clicking the buttons. Very frustrating.

1 Like

I hope you sort it out, the CGM should help a lot.

as far as I know low sugar doesn’t take off sugar from the blood cells that the A1c measure, it only counts sugar stuck during highs. I happy to be told I’m wrong.

I think he was guessing you had lows to have A1c in the 5’s

this group follow a lot of Bernstein’s stuff and do ok.

Calibration is key when it comes to effective use of the CGM. I do many things to get my CGM up and running accurately as quickly as possible. One thing is to attempt to change the sensor when my blood sugars are steady and only calibrate when I’m steady. If you calibrate while your blood sugars are moving quickly it makes for a far less accurate sensor. I dose and treat based on CGM readings alone, especially after the first day when I’ve got the sensor well aligned with my meter. During the first day, I will still treat based on the CGM, but I will check with my meter for readings that are very high or very low.


I had heard that you should calibrate if you test with your glucose meter and find a discrepancy of more than 20%. I had calibrated like an hour ago, so I didn’t bother. From what I understand the Dexcom Seven only used the last 6 or so calibrations so overcalibrating could drop accuracy, but that is not the case with the G4 or G5. Perhaps someone else knows.

My G5 seems to respond almost immediately.

1 Like

The A1c measure the “damage” (also called glycation) of your hemoglobin from elevated blood sugars. Obviously low blood sugars cause less damage. And endo who is concerned about hypos for an insulin dependent patient who is tightly controlled is being prudent and is doing a good job. My meter averages are higher than would be anticipated by my A1c suggesting I am low at times not being tested.

And what does Bernstein (or that facebook group) have to do with any of this? Please stay on topic.


I definitely would not make any major decisions based on an overnight result in the first 24 hours. I also find that it can sometimes give false or exaggerated lows due to sleeping on the side with the sensor (I can’t control what my body does when I’m sleeping), so if it gives me an outlandish reading I double check with a finger stick, especially if I don’t feel low (I can usually start feeling it when it’s in the 70s). I don’t usually get severe, unexpected overnight lows except in certain circumstances, evening parties with finger food and uncertain bolusing being a common culprit–if I’ve gone to bed with IOB and fighting a high, then I definitely take the alert seriously. But last night, e.g., I had a zero-carb dinner and when my CGM woke me because I’d hit the 80 alert level, but no down arrow, I backed my basal down to 1/3 and did something I’ve done a few times: I switched the low alert off. Probably would have been smarter to change the “low” setting to something lower than the default of 80, but hey, it was 2:30 a.m.

I don’t necessarily recommend this–I know it’s risky, but I also know my own patterns pretty well and there was really no reason to expect it was going to get much lower than it already was. Pretty severe dawn phenom is MUCH more of a problem for me, so at that hour I know my BG is going to start heading up without eating anything.


I make sure to bring my meter upstairs with me at bedtime so if I am genuinely as low as that, I am not stumbling around looking for the darn thing (and expending more energy).

ETA: Plus in the 40s my vision and balance is off, ugh.

Amen–you definitely need to keep this stuff handy, not least because the less you have to get out of bed and go wandering around, the easier it’s going to be to get back to sleep!

1 Like

I always try to predict my BG meter reading before I test. For example, dex is on an upward trend showing 90. The last time I tested, the dex reading was low by 15 points. I predict 90 + 10 (for the lag) + 15 (for the previous diff) = 115. If my BG meter reads between 105 and 125 then I consider this a good prediction. My dex is on its way to earn my trust. After a few good predictions I trust my dex.

The same goes for the dex. I always try to predict my dex reading before I look at it. If my prediction is off by a lot, I test my BG.

1 Like

When I was on MDI my endo did not give me her “Good Diabetic!” smile when my A1c got too close to 6 or god forbid below, for this specific reason. 6.3 was about the minimum she wanted to see. But this is largely due to the inflexibility of using a basal insulin, plus she knows I eat low carb and my basal-bolus ratio is much lower on the bolus side than the textbook recommendation, so I was relying on that inflexible Lantus to do more than it was really optimal for. With a CGM you can at least have a factual basis for controlling your basal dose–maybe splitting it–if you’re on MDI, and obviously with a pump you have much more flexibility in controlling its rate overnight, so A1c’s in the low 6’s and even 5’s don’t necessarily signal a problem.

As someone who would like to have a CGM and doesn’t, I have been following this from the sidelines with keen interest.

Stipulating for discussion purposes that cost isn’t a factor, my one major hesitation has to do with calibration. Since the CGM is dependent on a meter that is allowed to be off by 20% (actually more than that when far enough outside the target range), the CGM must inherently be at least that inaccurate. (You can’t be more accurate than your reference yardstick, by definition.) My inclination would be not to use it for determining my reading, but just for seeing trends.


I hear what you’re saying and agree that, in theory, a CGM cannot be more accurate than the [inaccurate] meter readings used to calibrate it. Unfortunately, those inaccurate meter readings are the best we have at this point in time. The reasons why I have my daughter use her Dexcom readings for treatment decisions (but not within the first 24 hours, of course) are: 1. They are almost without exception within 10% of her meter readings. I suspect a lot of this has to do with her particular physiology and the fact that we **calibrate wisely. ** 2. Her meter, a Freestyle Lite, is one of the statistically more accurate meters. 3. The way I view it is that her Dexcom uses her meter readings as a starting point only from which it then does its magical algorithmic thing in which (after the first 24 hours) I hold a great deal of trust. Two years (I know, not much in the grand scale of an entire life with D) of no adverse events plus kick-a$$ A1c’s from the get-go have earned this trust. As added benefits, it saves a lot of blood-letting hassle (thus reducing her D-burnout a bit), and she goes through test strips at a significantly slower pace (lowered D-expenses).

But believe you me, any wonky-looking Dexcom readings warrant a finger-stick or three…

1 Like

I too use the Freestyle Lite, and for the same reason.

And I understand, from the shared experience of many people, that CGMs have saved plenty of otherwise necessary ER visits and the like. Being the empirical fanatic I am, I can’t gainsay the facts.

I guess some of it is visceral. I’ve just never been a fan of indirect measurement. But sometimes, it’s all we have and therefore better than nothing.

Yeah, I get it.

1 Like

I haven’t used a Dexcom but do have experience with the Medtronic Enlite sensors so here’s my input.
I first tried to use them starting in early 2014 and the first issue I ran into was being awoken by the low alarm at 3:00AM one night with the CGM telling me I was at 42 (of course I’d slept through several cycles of beeping as the readings dropped to that level). Once I finally woke up I tested with my meter (twice to verify) only to find I was in the mid 140s.
As a result I turned off the low alarm and threshold suspend and never turned them back on. This incident happened several days into a sensor so it wasn’t due to the startup calibration. I contacted Medtronics about it and they told me this behavior is not uncommon if I was sleeping on my side and putting pressure on or near the sensor which redistributes the interstitial fluid. I used the CGM for about a year and then stopped due to continual problems with inaccurate readings and the discomfort of wearing the things.

I recently (starting early June of this year) was given the opportunity to test Medtronics improved sensors. They provided me with 2 free sensors and loaned me a transmitter as mine is out of warranty. This time the experience was completely different, once the startup calibration period was complete (about 9-12 hours) the sensors were very accurate and I had no false lows or highs at all. They also improved the adhesive on the tape they provide which helped with the comfort of wearing them.
All that to say I wouldn’t place much faith in the readings during the first 9-12 hours after insertion of the sensor but after that they seem to be fairly accurate. I personally wouldn’t ever bolus based a CGM but if you’re comfortable with it then it’s up to you.

The last thing I’ll say is, even though Medtronics seems to have substantially improved their sensors I won’t be using them. The Medtronic rep I worked with reviewed my readings and said a CGM probably wouldn’t help me lower my A1Cs (I have really good control right now) so it’s only benefit would be to alert me of hypos. Hypos never sneak up on me, I feel them coming before I get below 80 so the added expense isn’t justified.


It really all depends. 95% of the time, the CGM is VERY accurate. Where you may run into problems is when there is a lag between treating a low and it showing up on the CGM. I usually recover before the CGM knows it, so after a low, I will test 15 or 20 minutes after treating.

The other side is, you may have gone low and then come back up and the CGM hasn’t registered the fact that you’re not low anymore. So, it pays to test so you don’t over treat. Also, why do you keep your BG meter downstairs? Mine is on the side of my bed.

1 Like

Within the first 24-48 hours, I generally always test when I get an -unexpected- low. If it’s daytime, and I’ve been slowly trending downwards and its been 5 hours since I last ate, I trust it to be low because that’s to be expected.

At night, I’ve learned to look at the trend. If it was a sudden, sharp downwards trend that set off an alarm it was probably a pressure-induced low from rolling over onto the sensor. I’ll wait for the next reading and see if it trends upwards. If it does, I’ll usually ignore it and go back to sleep. I’ve only had this a couple of times as I rarely sleep in a position to roll onto my sensor. If I’ve been trending slowly downwards and am being woken by a Low alarm, I’ll test to confirm if it’s within the first 24-48 hours of the sensor, or if I’ve restarted the sensor and it’s gotten to where when I go to calibrate it’s no longer close, which indicates that it’s reaching the end of its life. The last couple of sensors, after the first 24-36 hours or so, my Accu-Check and my Dexcom tend to be within 5-10 points from each other as I have a very strict calibration protocol and only start sensors and calibrate at the most optimum times.

Basically, between days 3-12 or so, I trust my Dexcom enough to use it to treat lows and highs, and to dose my insulin off of.