1st A1C since starting on TandemX2 CIQ

Just a reminder that Control-IQ 's behavior is pretty much entirely controlled by your settings. There are absolutely valid criticisms of the system (“I don’t like the 110 target”, “I don’t like the locked 5 hour DIA”, “It only gives 60% of a correction”, “it doesn’t give corrections often enough”, etc…), but if you feel like it’s ineffective at doing what it’s programmed to do, then that comes down to settings.

Correction factor is the aggressiveness control, and it works on both ends of the spectrum. It controls how much power the pump has to alter both highs and lows. If your CF is set to too high of a number, i.e. less aggressive, then the system is less capable of withholding insulin when a low is predicted. It thinks that it only needs to withhold a smaller amount than is actually true. Too high of a number here incapacitates the system, so it can’t effectively prevent lows. Too low of a CF number of i.e. over-aggressive, can also cause lows by over-correcting highs. You would recognize that pattern by really fast changes, such as highs plummeting into lows, which then rebound into highs again. You’re really looking for that sweet spot in the middle.

There’s no guarantee that CF is the problem, though. A basal rate that is set too high can hide insulin on board from both you and Control-IQ. It can’t properly foresee how much power IOB has to drop your BG in this case, so doesn’t know to suspend. If there is no insulin on board and the system has reduced your basal, then it expects your BG to turn around at any moment, but it doesn’t have correct information to know better. That’s why basal testing is so important. Good basal rates are the foundation of Control-IQ (and the DIY algorithms, too), and then Correction Factor just sorta amps the algorithm up and down.

If you find you have lots of IOB on board when the lows happen, then I:carb setting may be the culprit, or it may just come down to bolus timing.

If you’re having so much trouble with lows and it’s a recent development, if could be that you’ve made changes that make you more sensitive to insulin in general. More active during the summer, dietary changes, weight loss, medication changes, etc… In that case, all 3 settings might need changed!

Control-IQ is the most powerful commercial system, but it’s only as good as it’s settings. I really think it’s best suited for patients who understand how the settings work and/or those who have a medical support team who understands how the settings work, and can make appropriate adjustments. Tandem is working on software that can recommend adjustments, but until that’s released, this nature of the system may be a downside for some. Unfortunately, not all doctors are good with the evolving tech, so there’s definitely an argument for Medtronic and O5 in these cases.


I made the switch two years ago and I’m in love! I’ve fluctuated between 5.7-6.5 depending on life, but my favorite part has been that flatline over night. I rarely have a night where I have to correct! I keep kicking myself for putting up with Medtronic for so long!


In order to get CiQ working correctly you need to get your basal rates correct. I suggest thst you fast for a day.
Record the basal rates thst keep you flat lined and use those rates in your settings.

It took me a few months of fooling around before I realized this. My old Medtronic pump settings just weren’t right for my Tandem. Once I got it set up right my sugars are nearly always stable. I still go high once in a while and I still go low but never low enough to cause any issue.
Hopefully you can meet with a dietitian or a pump educator who can get it set up right

Your coefficient of variation number, at 16.4%, is stellar! This is a little appreciated diabetes statistic. It is simply the Standard Deviation divided by the Mean. In your case, it is 18 mg/dL / 109 mg/dL.

Why is variability such an important measure? Most people (doctors included!) look at the downstream measure of % hypoglycemia and never consider the role of glucose variability.

Reduced variability permits safely lowering the blood glucose average or mean without raising the risk of hypos.

This risk reduction is usually built on a judicious eating style that avoids food and meal timing that undermines good control. Your status as T2D is likely a benefit to you in this regard. For comparison, as a T1D, I rarely measure a CofV < 18%. I speculate that you likely enjoy some limited beta cell natural insulin release that I don’t.

I encourage you to see this as a “strong card” in the hand you’ve been dealt.


Enjoy your success! Even though you’re well aware of the shortcomings of the A1c measurement, knowing most clinicians fixation on that number allows you to demonstrate competence in their world. I hope you didn’t have to listen to the almost obligatory warnings about hypo risk based on A1c alone.

I’m encouraged to read about your positive outcome with the Tandem X2 CIQ.


No, I seem to have a good record of lucking into endos (in this case an APRN) who are up on current standards and practices. Doesn’t hurt to live in areas where practices are associated with major research hospitals (last one was with Brigham & Womens in Boston; new one is with Yale-New Haven in CT) She was thrilled, not least b/c she had also helped advise me on the choice when I had to get a new pump this year…

1 Like

Since I now have more meaningful statistics I no longer pay any attention at all to my A1c. I’ll let my endo who once had an uncle in Mumbai die from hypo obsess over how low, it is while they ignore that I’m consistently 0% below range.

I think it’s sad that any doctor regard a broken clock metric as Golden, not know how to interpret statistics for dummies reports rationally and not have an hour per month to spend on continuing education. For the last 20 years of my career I had to spend 4 hours each night continuing my education just to remain employable.

Absolutely right. Folks in IT have an expression that perfectly describes it GIGO - garbage in garbage out. A programmed system can only work as well as its programmed data allows it to.

I spent my first month with a tandem pump intensively studying Its behavior, 3 months tweaking its settings rationally as I changed my behavior.

Those readings are now far from what the endocrinologist programmed in from their cookbook table. The only number that’s close is TDD - for a day where I’m sitting around doing nothing.

Since then I let Control IQ run automatically, only intervening when it starts beeping at me. If the alerts aren’t from a badly timed meal bolus, they’re nearly always indications of a failing infusion or sensor site.

1 Like

A1C is only useful for diagnosis, useless for modern management with a pump. TIR is better than A1C. Average and median glucose levels combined with SD are a much better metric set. You get your average glucose below 130mg and your SD below 20% and your A1C will drop. You won’t do that with a pump alone. It requires awareness and lifestyle refinement - intelligent small tweaks to your preferred lifstyle.

“Going high” means going above target, above 180mg/dL. TBR (low) is more important than TAR (high). How long you are relatively high is more important than
than the CGM number when it comes to A1C; a higj or low CGM number can be off by 15% and often is- unpredictably.

Three roles of thumb be used to work with Control IQ to reduce rebounds after ows.

  1. One unit of basal insulin drops BG about 30 mg/dL.

  2. For BG above 80 mg/dL, one glucose tab (4g of quick carbs) raises BG about 10 mg/dL.

  3. CGMs are slow and untrustworthy for making correction decisions. When a CGM is rising or falling rapidly, is above 180 mg/dL or below 100 mg/dL it can NOT be trusted for accuracy only trends.

The first rule means that in the absence of food, if your basal rate was 1 unit per hour and you stopped basal delivery completely, your BG would rise 30 mg/dL per hour.

If your glucose was dropping low enough for Control IQ to suspend delivery, it was dropping below 110mg/dL. Dropping low doesn’t happen without a cause. That could be sickness, bolusing too early, bolusing too much or exercising without prep.

After a Control IQ suspension, your glucose has to rise more than 80mg/dL to “go high”. Otherwise that rise requires food and/or a >2 hour suspension of insulin and/or an over-correction with too much carbs.

The second rule means that if you have an actual low, not a CGM compression, if you don’t over-react by taking too much quick carbs, your BG won’t overshoot into being high, and not in an hour.

Control IQ will detect a rise above 160 during the day, and if not within an hour of a bolus, will alert and will increase basal rate partially dampen the rise. If the rise continues, it will auto bolus to dampen it further.

The third rule means that when you are “out of range” to be accurate, your corrections whether with insulin or quick carbs must be made using a finger stick readings, NOT the CGM numbers.

Use a CGM readings to “stop a drop”. Wait 15 minutes between carb corrections,. Once glucose slows dropping, be very cautious about further carb “corrections”. If glucose is rising and you are still competent, wait and watch. Every 15g of excess carbs will raise your glucose 40 mg/dL Your correction “target” is 110-120mg/dL

If you still have a Tandem pump, check some settings. I’m assuming that your basal profile ICR and CF were set by careful observation and experimentation, not using a cookbook formula.

First, if you have the CGM Alerts/Low Alert target level set at 70mg/dL, that can make it harder to avoid going low. If you become more proactive you can reduce the severity of the lows that causes Control IQ to “misbehave”. It’s first priority isn’t yours - It’s not a good or a low A1C- it’s to keep you from hypo.

Raising the low CGM alert to 75, 80 or higher mg/dL will give you alerts sooner, more time to respond. It won’t change Control IQ’s settings, but the higher your level is when CIQ starts cuts back delivery, the more effective your intervention will be, the quicker the BG response will be, and the shorter the periods when Control IQ is reducing or suspending insulin.

If you have an Android phone, you can use the XDrip+ app to set as many level alert as you want to keep you proactively informed as to your glucose level,

Second, check the setting under My Pump/Personal Profiles/Pump Settings/Basal Limit. If it is too low or high that can make Control IQ overly insensitive or aggressive responding to rebounds with basal increases (I’ve set mine to twice my highest basal level.)

Third, the CFs in your profile directly change how aggressive Control IQ is at auto-bolusing. If it is too high, the auto-bolus delivered will be reduced from 60% of what’s needed to something less, and the rise will go higher. Lowering the number increases the insulin delivered for all boluses during that interval.

When I see an auto-bolus has happened, I expect to need to do a manual bolus later, and I try to figure out why my BG rose faster than my programmed response.

There are pumps that have algorithms that aitomatically calculate the numbers that determine how well Control IQ works. The clinical trials of these pumps have overall results that match the Tandem pump - 80% of the users have A1Cs over 8%. The FDA accepts these because that’s a 0.5% improvement over not using a pump.

But that level of hands-off “control” isn’t good enough to meet the AACE’s or ADA’s minimum targets. An 80% failure rate isn’t good. No matter what pump system is used there will be work required to get a “passing” A1C and more to get an “A”.

Thanks, Pstud123, for this advice. Honestly, my SD is 29-30. I think that reflects my inability to “refine my lifestyle” as you say! :blush: Nonetheless my TIR is 93-95% so I don’t feel like a complete failure, and A1Cs run 6.3-6.5. I will only be using the t-slim for another couple of weeks at this point, but I am going to try a couple of your suggestions to see how that goes! I have been using sleep mode 24/7 for several months now and that seems to work better for me that daytime regular CTRL-IQ. Thanks again.

Hopefully you are in the trial or rental period with the t:slim so you won’t need to fight with your insurer for a different one to be covered.

If sleep mode works better than CIQ for you, you might do better turning off Control IQ completely and use manual mode to figure out what basic settings are wrong. With any algorithm, garbage in, garbage out.

It takes time to adjust a pump to you and for you to adjust to it. The most efficient way is for you to become more consistent and predictable long enough that you can figure out how a CGM and pump react. That’s the best way to figure out how to tweak any pumps settings. Loop systems that haven’t user refinement controls, put you at the mercy of a black box dictator, not the beneficiary of an obedient, predictable and responsive tool working as your servant.

I find it’s easier to live with something mediocre that can be trained or adjusted to more match my needs and make small adjustments to my behavior than to live with something potentially great that was optimized for someone else and can’t be adjusted. That is as true with machines as it is with mates. You can get great results by making compromises, horrible ones when something or someone won’t.

eg I got my biggest improvement in SD by setting my carbs per meal to 40 grams and scrupulously using a digital scale and labels to just make sure that was exactly what I was eating. Any number would have work, 40 was a 10% reduction from what I had been eating on average previously and I wanted to lose about 10% of my weight.

With that variable fixed, I used dietary fiber to even out my digestion, and dietary fat and protein to slow post-prandial rises and durations of peaks. After 3 months I had standardized menus with just my favorite things to eat. Then everything stabilized. Last and I figured out what vitamins and minerals I needed that I wasn’t getting enough of. (It would have been smarter to do this continuously, but I’m a linear thinker).

After 18 months living with G6 and t:slim CIQ, I pretty much run on “auto-pilot”. Since I’m retired , my biggest remaining problems are sensor “failures” from degrading sites and remembering to eat on schedule.