Dexcom G7

What do users think about G7 going to a non-calibration format? Have they increased the accuracy?

The MARD (percentage the data varies from laboratory values) is slightly improved in the G7, at 8.2% vs 9% previously. In the US, the standard expectation is that the MARD be less than 20%, so even 9% was already considered really accurate.

But you can still calibrate the G7, just like the G6. You don’t HAVE to, but you can.


Very glad that I will still be able to calibrate!

1 Like

The whole issue with the Dexcom being better is to be able to calibrate it. Without it, it becomes the same as the Libre and it’s inaccuracies. 20-40 points off. It makes a difference to me if I am at 130 and taking a correction dose or being at 105 and being fine with it. But at the supposedly 20% they want us to accept, I’m supposed to not need to know that difference. I beg to differ.

And at a 105 average, you would have an A1c of 5.3, but you are really at 130 and have an A1c of 6.2. But let’s take that further. You think you are at 130 with an A1c being at 6.2 but you are really at 160, within the 20% “allowable” error. Your A1c would actually be 7.2.

I realize all sorts of people don’t care. We have a tendency to be people that care more on this forum. But I think allowing 20% as acceptable is absurd. Especially when you are using insulin. And even if it tested out with an average of being 10% off. I bet that 10% was under the best conditions. I wouldn’t mind so much if it really was 10% off all the time. But mine is more than that until I calibrate it.

Plus on top of it, we are told to expect wild numbers those first 24 hours. Yet all of our numbers on our graph are used for the TIR. Mine has a tendency to come in low when a new sensor starts. I immediately calibrate it to get rid of that number. I tried to let it go once and it stayed low 30 plus points for over 6 hours and then I realized how much that was influencing my low readings. Too low for over 6 hours makes a difference. Without any other under 70 readings, that would end up being 2.8% for the week. They are generally not happy with that.

The ability to calibrate the Dexcom is what makes it stand out as better. Without that, it’s really not. The Libre 3 has a Mard score of under 8%.


Very well said Marie! I am the only person who ever looks at my Clarity reports, but I want them to be accurate. I calibrate so that my CGM will consistently match my finger stick readings.


The context that the G7s claims are made is a combination of reported issues and A1c results. The basic technology of the sensors hasn’t changed. imo Overall, the improvements mainly benefit the maufacturer, users as a group will see a slight improvement in convenience.

I expect that the G7 sensors will be easier to make, package, stock, order and set up than the G6. They will be smaller. I hope that they wil need less calibrations than the G6. I do not expect that every sensor will need no fingersticks for verification and no calibration.

Because without using a BGM, how can you tell whether systen built on a piece of coated wire in a vaiable enviroment is reporting accurately or not? If there is no way to calbrate the G7, I woiuld anticioate a higher replacement rate for those iof us with wonky sites.

I’m skeptical about it as I would be about anything that directly effects my health and safety. I’ll be a late adopter, as I am with all “new” tech. It’s just another tool. I know how to use the one I have now to get very good results.

I’m very skeptical about anything that the FDA approves actually being better performing than a prior approved tech.

The G7 approval was based on relative safety and performance compered to the G6, not absolute performance.

The true performace of a technology is in its results -outcomes- when used by the population. That something works faster in a benchmark test doesn’t matter unless that speed results in more actuak output or benefits to the users.

With diabetes technolgy, the gap between what the tech is capable of doing and what actual end users get from it is growing.

I figure that everyone here knows what an A1c is, but all may not realize that it is not an absolute value . The A1c itself is a statistical norm and lab testing for A1c can be +/-10% and meet accuracy requirements. A +/- 10% difference in A1c values is roughly the difference between A1cs of 6 and 7.

Secondary equipment like POS A1cNow, BGMs and CGMs are compared statistically to lab A1cs.

The A1c is the “gold standard” for diabetes management - done by endocrinologists to manage PWD, not diabetes. It can indicate that you are “out of whack” and whether a group of PWD are doing better or worse over time.

But if you use a personal CGM it doesn’t take long to realize that the A1c is not a good metric for managing your own diabetes. Its like an end-of-term pass/fail decision from a pariciple who doesn’t pay attentIon to what subjects you have difficulty with, or your teahers comments, just your avarage GPA. It’s not a good management metric for the students just the school.

A management metric is used to detect meaningful different results in response to changes. Changes in a management metric indicate what changes are beneficial and which are not. A good metric responds to a change quickly enough to let the manager make a decision.

Compared to financial metrics, the A1c is like how much cash you have in your checking account 4 times a year and hiw muych you have iu your wallet… It doesn’t measure your income, how you spent it, the maximum or minimum amount of cash you had in the account at any time or how you spent it or out some in a savings account.

Knowing those quarterly balances doesn’t tell you how to change your financial behavior.

The A1c is like that checking account balance and your fasting BG that day is like how much mioney you had in your wallet. Changes in them in a way you perceive as negative, don’t tell you what to do to improve them.

I believe that the G7 claims are carefully-worded advertising hype and do not in any way imply that the accuracy is good enough without calibration fo get excellent results. It is statistically more accurate than the G6. So if you were a cohort, not an individual, that is good, but it says nothing about the minute by miunute accuarcy.

The Dexcom G6 “no more fingersticks” and the FDA decision that it’s accurate enough “for making therapy decisions” has a lot of wiggle room.

The first can be interpreted two different ways, and the combination has a context.

The manual says to use a CGM when CGM reading don’t match expected values. So “no more” does not mean “never again”. I interpret “no more” as strips used compared to using a BGM withiut a CGM.

Practically, since I’m more accutely aware of my BG, whenever it is warming up I use a BGM to verify that it is tracking within the accuracy windows that Dexcom defines in their chart. Usually it takes 4-6 hours for me to verify that it is within 20% of my BGM at the low and high target levels. Diring this time I will use 4 BGM strips- as many as I use to use on a good day when using a BGM alone.

Then there’s the questions of context.

Accurate for therapy decisions is compared to using BGM. The average of all PWD who use the G6 is better than the average of all PWD who use only BGM- by 0.5%.
But both those are above the ADA and AACE targets. More than 70% of all CGM users have A1cs above 7.

It looks like you can still calibrate the G7.

"To log a BG meter value or calibrate:

  1. On the receiver, go to Menu > Events > Blood Glucose
  2. Select Log Blood Glucose, or to use the BG value as a calibration, select Use as Calibration.
  3. Follow onscreen instructions.


And this is in small print at the bottom of the Dexcom G7 page. Their liability escape.
“†Fingersticks required for diabetes treatment decisions if symptoms or expectations do not match readings.”

But also another note. If you don’t qualify for a receiver yet as @Terry4 said, you might have a problem getting Medicare to cover it? I would think new device, new receiver but on the other hand, Medicare probably is fine if you keep using the G6 until you’ve met the new receiver requirements.

On the Medicare page
“Receiver required for Medicare beneficiaries”

I want to keep using the G6 as long as possible anyways as I really like restarted sensors.

The G7 page