Dear Dani Grace and Jude,
Note: This page is advice for Grace, Jude and John. For any other reader, it’s information only. No therapeutic relationship is formed – read this.
The amount of information on AID systems is overwhelming.
I know, it’s taken me six years to get my head around it. Trying to fathom which one to choose is a challenge.
I have condensed six years of learning to provide you with the most important questions to guide your decision:
- What AID systems are available to you?
- Which CGM device do you prefer?
- How much control do you want of the algorithm?
- How aggressive do you want the algorithm to be in preventing after-meal spikes?
There is no best system, only the best system for you. Don’t worry, you cannot make a bad choice, its just like choosing a super car!
So, work through these questions and see which one jumps out at you!
- What AID system are available to you?
The table below tells you what devices are needed for the different systems. Also, if they have European CE Marking. The table includes, as of November 2023, the license they have regarding age, weight, insulin brand, and total daily insulin dose (TDD).
Just because they are commercially available in your part of the world and you fit the criteria, it does not mean they will be available in your country and diabetes centre.
You will need to ask your diabetes team.
The DIY options are not on the table here as they are not approved or regulated.
My current opinion is:
The commercially available AID systems is good enough for 99% of people with type 1 diabetes.
However, if you are in the 1%, a nerdy-obsessed person (like me) who wants 100% control and is happy to do the work required, go and check out what devices are compatible.
· Open APS
Let us dig a bit deeper.
2. Which CGM device do you prefer?
If you are already using CGM you will have a bias to what you like. Use your own experience and go with what you like.
“Are you going to let a CGM device determine your insulin doses before you have tested it, and you are confident it’s accurate enough?”
I also wrote a scientific paper assessing the accuracies of all sensors in 2023. There were four clear winners. The Dexcom G6 and G7 and the Freestyle Libre 2 and 3.
The Dexcom G6 & G7 and the Freestyle Libre 2 & 3 are the only sensors with FDA iCGM approval (Nov 2023). To attain iCGM status, the accuracy of the sensor must reach a very high bar, tested on a wide age range of people with type 1 diabetes!
The Enlite 4, used with 780G, does not have published accuracy data to allow a thorough accuracy assessment. However, the MiniMed 780G has excellent clinical trial data and is therefore an excellent and safe choice.
Full disclosure: I have tried ALL the sensors and AID systems and done vigorous accuracy checks, testing them head-to-head. Every time the Dexcom G6 wins, for me, so I use the G6! The peace of mind provided by being able to calibrate when the accuracy is off proves invaluable for my mindset.
Being able to calibrate may not be useful for you though, and the G6 is bulky compared to the others. You best bet is to “try before you buy”.
Remember, no CGM device is risk free. They all have inaccuracies. But, some are lower risk than others. At the moment, the iCGM approval (G6, G7, Libre 2, Libre 3) is the only marker for low risk. However, the safety record of the 780G in clinical trials also gives confidence the Enlite 4 is accurate enough when used as part of an AID system.
3. How much control do you want over the algorithm?
This is a fundamental question, so take some time to think.
Do you want the system to do 90% of the heavy lifting by constantly updating the settings?
If yes, the best choices are the Medtronic 780G, CamAPS FX, or Omnipod 5.
All you must do is enter minimal settings and the target glucose level you want the algorithm to aim for!
You only need to keep your carb ratios up to date, and the algorithm takes care of the rest. This means you are not reliant on yourself or your diabetes team to update settings such as basal rates and correction factors correctly.
These systems reduce user error and minimise diabetes hassle.
However, if you are a person who has different sensitivities to insulin on different days, this can cause some problems.
For example, if you are not very active during the week but extremely active on the weekend, it’s likely the algorithm will be too strong on the weekend.
Another example is if you are on medications that spike the glucose levels that you take sporadically, such as steroids, the algorithm will struggle to be able to keep up. There are some limited settings to help with this on these systems.
Finally, if you are a person who likes to tinker, you will not be able to do this.
Alternatively, do you want control of how the algorithm works and the flexibility to adapt as necessary?
If yes, the best choices are T-Slimx2 or a DIY option.
All these systems require the user to set and regularly update the basal rates, correction factors, and carb ratios.
For adults, this usually means working hard for the first month but only infrequent adjustments to fine-tune after that.
For children, this means grafting the first month and regular updates every three months or so.
The success of these systems is quite reliant on the user and the diabetes team who support them.
This does not sound very attractive; however, these systems offer flexibility.
For example, you can have a different profile of settings for weekdays and weekends, and even holidays.
Also, you can make settings much more aggressive for times of illness or medications.
Reading this I am sure you will know which camp you’re in.
4. How aggressive do you want the algorithm to be in preventing after-meal spikes?
The major limitation of all AID systems is that they use “Fast Acting” insulin.
This is misleading because the “Fast-Acting” insulins are not as fast as the carbohydrate absorption from meals.
This means glucose spikes after eating high carbohydrate meals still happen when using AID.
Of course, you want to do the basics of eating balanced meals, carb counting accurately, and giving insulin 10-20minutes before eating.
However, we all know this is not possible 100% of the time.
How effective AID systems are at tackling after-meal glucose spikes depends on two things:
- Target level
- The aggressiveness of the algorithm
This table shows the different target levels and adjustable settings that impact the AID algorithms. Very simply, the lower the target can be set, the quicker the algorithm will start increasing the insulin.
From most to least aggressive:
- CamsAPS FX
- Omnipod 5
- T-Slimx2 with Control-IQ
The second and more important consideration is how you can adjust the algorithm settings to make it more aggressive.
It all comes down to active insulin time.
Active insulin time is the duration the algorithm thinks the insulin lasts. When insulin is given for food or corrections, the algorithm uses active insulin time to calculate how much insulin on board there is that might drop the glucose level. This is then used in the calculations of how much extra insulin to give to tackle highs. The shorter you set it, the more aggressive the algorithm will be.
From most to least aggressive:
- DIY (if active insulin set 3 hrs and correct carb absorption time set)
- 780G (if active insulin set 2 or 2 ½ hrs)
- Omnipod (2hrs aggressive, 2.5 hrs normal, 3 hrs safe)
- CamsAPS FX (calculated by artificial intelligence)
- T-Slimx2 with Control-IQ (fixed at 5 hrs)
There is a trade-off to consider.
The more aggressive the algorithm is the fewer highs but the potential increased risk of lows.
There we have it.
First, find out what’s available.
Second, think about what CGM sensor you want to use.
Third, consider how much hassle and flexibility you want from the algorithm.
Finally, ponder how aggressive you would like the system to be after-meal spikes.
Great, you have made your choice.
It’s time to access the tips and tricks for your system