Dexcom G7 regrets

Hi, all. Within the past few months I received my first CGM, ahead of getting my new T-slim pump, so I can get used to this new system in increments. Now I’m questioning my choice of the G7 over the G6. At the time I placed the order, I assumed that the newer model would be improved. I should have done research first. Obviously it is much better than fingersticks for convenience. Today I read an article by a seemingly knowledgeable person that stated the prior model, G6, works better overall for corrections, with a pump or manually. The algorithm was better, according to the author. Since I haven’t started training on the pump yet and I have a three month supply of the G7s, I’ll see how it all works out. Has anyone on Medicare been able to switch from G7 to G6 if you weren’t happy with G7?

The G6 filters the results, averaging a few readings before displaying a number or saving results. The G7 doesn’t. There are advantages to each. The historical values the G6 records are smoother and likely more accurate, but the history the G6 records isn’t a recording of the current value that it’s showing you in bigger text. Instead it’s a smoothed value. That means that sometimes you (or you pump) will take actions that you have a hard time explaining by looking at the G6 history. The G7 doesn’t do any of this.

Both the G6 and G7 are prone to “compression lows” - if you put pressure on the sensor (for example lay on it at night), then the sensor will very rapidly report a drop of 40 points or more. Then after it wakes you up and you roll over it immediately pops back to normal. On the G6 this process is slower, both going down and up, but just as deep. So with the G7 you get notified about this and fix it more quickly. I’ve seen my pump respond to the resulting incredibly fast rise in reported blood glucose by providing an unneeded bolus of insulin. Not large, but not needed is not needed.

I’ve switched between them 3 times now (G6 to G7 to G6 to G7) and I don’t think there’s much real difference. Dexcom certainly didn’t do anything to reduce those compression lows.


I’m sticking with G6 (and BIQ) for as long as I can. Several reasons: I prefer to set my own target bg. My preferred target is 80, the CIQ target is 110. I want to be at a fasting level prior to any eating. My T1D mgt is very tight - preferred range is 60-140 (and yeah, I get rogue highs, but not often. I will correct at 110 if I have no IOB and may have some “food on board”) I use temp basal function a lot - both for decreased and/or increased basal for a couple hours. That function is lost with G6. I don’t mind the 2-hour warmup - small price to pay for the flexibility of managing my own T1D. And the Oregonian in me prefers to dispose of a transmitter every i90 days rather than very 10 days.

1 Like

Thank you, Greg. It makes me feel better to know there isn’t a great deal of difference.

Thank you for sharing, artwoman. I, too, do a lot of interacting with my old Medtronic pump because I like tight control. Since I haven’t made the switch to my new Tandem pump yet, I have yet to experience what you are describing. A target of 110 sounds too high but I guess I will just have to deal with it when the time comes.

What I plan to do is to use “sleep mode” overnight - to avoid the possibility of overnight lows that I might sleep through. I will then turn off CIQ during the day and go on manual mode. I frequently titrate corrections, say if I am 100-120 I will dose 100% of correction. 120-140 it will be 120% of correction setting. I also use temp basal to help bring things back where I want them. I can do that now with BIQ and the G6, but I gather BIQ is being discontinued. It will be CIQ or manual. I’m up to the task, so I will do manual during the day.

1 Like

I was an early adopter of G6. It’s was my first CGM. At the time, it seems like most of the Dexcom users here were clinging to the previous G5 version for all they were worth and didn’t want to upgrade. There were things they liked about G5 and didn’t want to give up. I didn’t understand it all. Utterly baffled. G6 was all I knew and it was fabulous. It struck me as such odd behavior.

Now, 6 years later, I’m in the same boat they were. There are a LOT of users who love G7, and here I am… Clinging to my G6, nervous to make the switch being forced on me. I know G6’s quirks and I’ve adapted to them, I know how to extend sensor life so I have ample supplies. I know how to work with the data it gives. G6 is second nature, while G7 is still largely unknown to me. I’ve only tried one sample G7 sensor and couldn’t even pair it to my pump. It’s hard to give up something that’s working for me in the face of the unknown.

I highly doubt G6 is any better than G7, though, it’s just different. Change is hard. There are pros and cons to both systems, but at the end of the day they both provide reliable data that helps us make good choices in managing our diabetes.

G6 is a little more cumbersome with 2 different disposable components that expire at different times. Takes some juggling to manage shipments, especially if your insurance doesn’t allow you to order 3 months at a time. It’s another part to pay for. The upside of that 2-part system is that you can reuse a sensor to get more life out of it than Dexcom and the government want you to. That’s worth a lot to some, nothing to others, and others yet will actually consider it a high-risk detriment. It’s got a 2 hour warmup where you get no data. It also uses a smoothing algorithm to make really pretty graphs with smooth transitions between data points.

Conversely, G7 is all one piece with a smaller/less wasteful inserter. You don’t have to juggle reusable transmitters or worry about how many more days use you can get out of it. It only takes 30 minutes to warmup. There’s even a grace period where you keep getting data for half a day after the 10 day session is up, which is a pretty big deal. That actually allows you to overlap new and old sensors. You can get the new one warming up up while the old one is still doing it’s job, and basically eliminate any gaps in sensor data. (It’s funny how nervous I get now during the warmup gap. I think I fingerstick more now in those 2 hours than I did all day before I had CGM!) They also did away with the graph smoothing, for better or worse. You get real time data now, not a weighted average of the last 15 minutes. Means faster information about changing blood sugar, but requires more brainwork to see the patterns.

Most of that sounds really good, right? It’s the inability to restart and having to adjust our behavior to the new way data displays that has some of us clinging to the old device. For instance, if the G6 told me my BG was +15 mg/dl in the last 5 minutes, I would have a tiny panic attack and rage bolus to head off the wicked spike that foretold. My instincts know I’ve likely been trending up for the last 15 minutes and I’m only just now finding out about it. BG isn’t smooth and pretty, though. It jumps all over the place as our bodies use it and move it. Without that smoothing algorithm, jumpy data is normal on G7. +15 still might mean something significant, but it also might mean nothing. I need to be patient and see what it’s doing. I’ve got to unlearn that rage bolus instinct and look at the bigger picture. Surely real time true data is better than pretty data, though, even if it means I have to change everything I know about using Dexcom. Right? You don’t have any experience with these things, though, so it won’t make a difference to you.

There was also some issue with weak Bluetooth signal in early G7 sensors, but to my knowledge that’s been remedied. That shouldn’t be a deciding factor anymore.

At this point, I wouldn’t try to fight it. G6 is being phased out and you’re already on the right side of the equation. There’s just no point in getting attached to G6 only to have to go through the stress of changing a few months later.


As it is now, with my old Medtronic pump, I frequently use temporary Basal Rates and keep a close eye on CGM and give correction boluses. So your information is so helpful. Knowing that I can put the T-slim in manual mode and in CIQ overnight sounds like a perfect plan for me to maintain the control that I want.

Excellent points you make, Robyn. Having lived with Type 1 for 33 years, I should be thankful to have any CGM these past few months. It has helped me a lot. In the beginning I was panicking every time I saw an upward arrow and bolused for it. Then I would see the downward spike arrow with impending lows and I would shut my pump off. Recently I’ve learned to just wait a bit and see what happens so I don’t get into the seesaw of up and down, which drives me insane. Thank you for sharing. I really appreciate all the helpful information. Now I don’t feel so bad about having the G7.

“For instance, if the G6 told me my BG was +15 mg/dl in the last 5 minutes… My instincts know I’ve likely been trending up for the last 15 minutes”

Do you find that the reduced data smoothing of the G7 causes Control-IQ to resume normal basal rates more quickly when BG starts to rise after basal has been reduced for a “low”?

I’m also attached to the G6, but when Control-IQ reduces/stops my basal it doesn’t seem to start increasing it again until it sees 3 consecutive data points trending upward. That ends up meaning that every time my BG goes under 100 Control-IQ sends it back up to 140+.

With Basal-IQ this never happened b/c it immediately resumed the programmed basal when BG started to rise even one point. I could stay 70-100 for hours despite occasional basal pauses to prevent lows.

G7 can’t fix the Control-IQ algorithm, obviously, but any sense of whether the reduced smoothing might help a little? (Or whether Tandem has made the algorithm even more conservative when used with G7?)

So far I’m still wearing my old Medtronic pump and my G7 independently. My T-Slim is here but I am waiting for my cartridges and infusion sets to arrive, which should happen around the end of the month. I’ve just been trying to think ahead. From what you’re describing, basal IQ sounds so much better. I’m assuming the algorithm for CIQ is not the greatest for those of us who want tight control. Maybe Tandem is more concerned with it’s customers getting too low. What I might do is put the pump in manual mode during the day and switch to CIQ at night to prevent the lows when sleeping, if that is possible. At this time my knowledge is limited since I am not connected to the pump yet. There’s a ton of learning I’ll have to go through.

I don’t want to discourage you about Control-IQ!

For me, Control-IQ is definitely better than what I can do manually while living a busy life. (To do better manually, I would need to consistently prioritize avoiding carbs an adjusting insulin/carbs based on BG every hour or two.)

Control-IQ in sleep mode keeps my “steady state” fasting BG higher and less stable than I’d like, but it also prevents enough higher highs that my A1C has not changed (~5.6-5.9 range). Fasting, it mostly keeps me oscillating between ~95-130 every 3-3.5hrs (with the high end BG increasing a bit if I eat a lot of carbs or try to push my BG lower than 100).

Basal-IQ tended to produce similar results if I had a lot of stress and not much control of my diet and schedule, but could do much much better with only a tiny bit of extra attention. If avoiding carbs I could easily stay between 70-100 any time I hadn’t eaten in 2-3 hrs. (Unfortunately, Tandem has stopped selling Basal-IQ.)

It’s crazy that Tandem discontinued the Basal IQ in favor of Control IQ. Someone must have thought CIQ was a better idea. I’m used to fiddling with my pump a lot since it isn’t connected to a CGM so I’m basically in manual mode now. I’ll see what works best for me once I’m up and running with my Tandem pump. Thank you for sharing. It gives me a picture of what I will be experiencing with my new Tandem pump. My A1C has always been under 6.0 and I’d like to keep it that way.

So, I haven’t actually tried Control-IQ with G7 yet. The only G7 sensor I wore was a sample from my doctor before there was any pump compatibility. I can’t speak from personal experience.

BUT… Control-IQ isn’t dosing insulin off of any single CGM data point. It’s not working in the moment at all. It does not see rapid changes/jumpy data at all. It’s not waiting for 3 data points of rise, even though it might LOOK like it to you. Control-IQ is completely based on 30 minute predictions provided by the CGM. Control-IQ is doing the math, using your correction factor and basal rates settings, to determine how much insulin it needs to rob from your basal insulin to bring you back to BG target. If it’s truly suspended, it stays that way until the 30 minute CGM predictions shows you’ll be out of the danger hole, and then basal resumes at a reduced rate until the prediction shows you back at target in 30 minutes. And it’s redoing that calculation every 5 minutes based on the latest CGM predictions.

G7 is basically still doing the smoothing algorithm in the background, just not showing it to us on the graph. It’s looking at past data, looking at how much that current data has changed from that, how consistent that change is with the other charges in recent history… And from all that, extrapolating a prediction of where you’ll be 30 minutes in the future. THAT’S what Control-IQ is adjusting our insulin to, and it’s the same as it did with G6, so I don’t see how a switch between G6 and G7 would affect Control-IQ at all.

What you described doesn’t strike me as a sensor issue at all. That sounds like your pump settings still need tweaked. And I do mean TWEAKED, because 140 is still an awesome BG. It’s only slightly over the target. I’m going to make a big assumption here that your basal rates are pretty good if you’re annoyed about 140, which primarily leaves correction factor to question. You might have a a CF that is SLIGHTLY too aggressive, making the pump withhold too much insulin to correct any lows. The lower the correction factor, the more extreme the roller coaster of ups and downs gets. The higher the CF, the smoother the ride gets and the more likely you are to stop on target, but the slower the system can effect change.

It could also mean your basal rate is a touch too low there, but as mentioned earlier, I’m putting this secondary to CF intentionally because I think CF is more relevant to you. Basically, when Control-IQ is doing correction math (to treat/prevent both highs and lows), it’s only counting how much more or less insulin you have than your basal rate, what’s known as “net IOB.” You want a negative value (less insulin than your programmed basal rate) to correct a low, and a positive value (more insulin than your programmed basal rate) to correct a high. The correction math gets all screwed up, though, if your basal rate doesn’t accurately reflect your body’s need. A basal rate that’s too high will mean there’s more potential to lower BG than the pump knows about, makes you feel like you’re fighting lows, and conversely, a basal rate that’s too low means you’ll be struggling to get BG back down to target because you don’t really have as much IOB as the pump thinks you do.

Your own behavior also factors in. Did you treat the hypo with carbs? If so, you double-treated the low. Both you and the pump tried to fix the hypo, so it’s no wonder you over-shot the target. The advice is usually to treat with 5 g of carbs with pump automatic, instead of 15 g… but in reality it’s pretty subjective. I’ve got a good feeling for what Control-IQ can handle and what it can’t. I don’t treat in the 60s or above anymore, unless I’m still dropping fast and trying to head off those critical 50s. Thankfully, the days I feel like I need to treat with carbs are exceedingly rare anymore. If you ate carbs, you can pretty much ignore anything I said about pump settings.

1 Like

Control-IQ IS better than Basal-IQ. At least for the average diabetic. The average diabetic who still struggles to reach the 70% TIR and HbA1c < 7 international treatment goals. The average diabetic who’s life is threatened by glycemic variability and the severe hypos it brings. It is better at preventing lows than Basal-IQ, it’s smarter at turning around and resuming insulin after a suspension, AND it’s treating the high end of the spectrum, too.

It’s not just Tandem that thinks that. There’s research behind it. It’s helping diabetics reach what used to be unobtainable goals. Endocrinologists stand behind it. Patients stand behind it. The results are so undeniable that the American Diabetes Association added these automated insulin delivery systems (or at least access to them) to the official Standards of Care for insurance and medical providers in 2023, then doubled down on their importance again in 2024.

Control-IQ just works very different than Basal-IQ. Basal-IQ didn’t use your settings to make adjustments at all, so you didn’t have to worry about tweaking them for performance. Basal-IQ did exactly 1 thing and 1 thing only. It turned off your basal delivery when you got down to 80. It turned it back on when your BG started going up again. Great sometimes. Dreadful other times if CGM data is jumpy, like when your BG goes from 59 to 60, insulin resumes, and the next reading is back in the 50s and Basal-IQ needs a few more cycles to think about it d decide to suspend again. Control-IQ on the other hand is using your personal settings to correct future BG predictions. Good settings are VITAL to it’s performance, and they will need tweaked until you get the results you like. It’s reducing insulin to head off the lows before they happen, and it’s rarely suspending insulin altogether, making the smoother transitions. The caveat is that it’s behavior is entirely dependent on your personal profile settings, and you have to provide it good settings to do it’s job. It might take some time. It took me nearly a month of subtle tweaking to fall in love with Control-IQ after I made the switch.

Now, there is a very small subset of diabetics who probably wouldn’t benefit from Control-IQ, at least not in it’s default mode. (Sleep Mode 24/7 it’s incredible tight control and HbA1cs in the 5s of you’re good at bolusing.) Those who are in truly excellent control and can frequently stay in the 70-110 range without any help. Those who’s goal is to be below 110 the majority of the time. Those who don’t need any help whatsoever to avoid the highs and get mad the pump doesn’t like us below 110.

Thankfully, those people should be appeased by the big update coming out later this year. It’s going to have lower BG targets amongst a bunch of other things.


Ah, still helpful to hear others’ G7 experience!

Based on just my own data with G6, it appears that Control-IQ is fitting the projected BG curve after an inflection point using at least 3 smoothed historical data points. While my issue isn’t caused by the sensor, I think it’s possible that less smoothing of those 3 data points could mean a slightly higher slope as BG starts to rise and earlier and/or more aggressive basal adjustment… but, yeah, probably a very small improvement.

RE tweaking of pump settings:
I’ve actually been sort of impressed by how resistant the system is to tweaking (and therefore robust to small errors in settings). I’ve tried tweaking both basal and correction factor both up and down by up to ~20% (in small increments) and it seems that I can change the frequency of the oscillation, but the magnitude is pretty stubborn.

Admittedly, I’m contributing a bit to the instability by intentionally running my basal just slightly high so that I won’t stabilize at the top end of the algorithm’s target range. :slight_smile:

That was a concern for me in deciding to make the switch from Medt, but I’ve found it actually matters a lot less than I thought it would. It’s using that number for various operations, but the design is very open to user input so you can modify a lot of that stuff. For instance, when you’re bolusing it will ask if you want it to adjust the amount up or down depending on your BG relative to the target, but it asks, and “no” is an option. I almost always reject it if I’m in the 80s-90s. You can also tweak the recommended bolus yourself—it’s “recommended,” not hard-wired. The whole system is designed more as an assistant than a robotic controller that takes over and doesn’t tolerate user interference, which was my experience with Medt. I’m getting A1Cs in the high-5 to low-6 range, TIR in the mid-90%, which meets my target goals. (An average BG of 110 corresponds to an AIC of ~5.4, so it’s fairly aggressive if you think of it that way.)

1 Like

Thank you, DrBB. I hope all of this makes more sense to me once I get hooked up to my new Tandem. What you said is all very encouraging. I’m happy that I will be able to have some control over the system.

I think if you’re a hands-on pumper, as you suggest, you’ll find it pretty easy to get along with. I had given up on the whole AID thing after getting burned with the 670G and went back to my old Paradigm dumb-pump for 5 yrs, so kinda like your experience. Stand-alone Dexcom CGM, which I vastly preferred to the Guardian thing. Did fine until my faithful old dumb Paradigm crapped out. Had a lot of trepidation about going back into the AID space again but Tandem sounded like it might be a better fit for my own idiosyncrasies and that has been borne out in practice. I have some gripes with the insets and the reservoir filling stuff (I vastly prefer the Mio Advance inserter and the syringe business is a little kludgy), but in terms of the software, the UI and the results I’m very happy with it.

No. I switched from G6 to G7 last year. I have not had any problems at all. I like it as much or even better than G6.

1 Like