TIR expectations

I had to laugh recently. Various advertising from diabetes companies comes through my feeds, and this time it was an ad from Medtronic Australia, touting the improvements in TIR (Time In Range) for their 670G pump.

Here it is reproduced in its entirety:


I’m not sure exactly where they sourced this data, but presumably from various clinical trials they’ve conducted. I’m going to assume they’re talking about the percentage of time spent within the classic “in-range” blood glucose definition of 3.9-10 mmol/L (70-180 mg/dL).

There are two “take-home” points they’re making:

  • The TIR of many people is…. sub-optimal.
  • The TIR using the 670G pump’s Auto Mode can be a lot better.

It’s hard to argue with those points. Any tools that can help people improve their glycaemic control are welcome. And Medtronic have certainly been able to deploy lots of 670G pumps.

My own reaction

I do have to laugh when Medtronic is proud of getting people to 70 or 75% TIR. This is certainly better than 50%, but not aspirational. I’m sure the average TIR of the T1D population is not up to the 70% mark yet, but our reactions are always coloured by our own experience.

With my own T1D I currently use an Accu-Chek Combo pump and a Dexcom G5 CGM, controlled by AndroidAPS. This is “DIY” tech. Open source, bespoke, customisable, whatever you want to call it. And as such it’s not yet approved by the regulators. But it is accessible to many people.
Incidentally, before the Combo pump I used to use an old Medtronic pump and the OpenAPS system, with similar results.

So what TIR do I get with minimal effort? I went to my Nightscout site and generated a Distribution report for the last 90 days using the 3.9-10 range. The answer: 95.4%.

This is actually a bit low: usually I average up around 97%. But there’s always variability in diabetes and life (even averaged over 90 days) so I’m not surprised by this figure. 5% of a day equates to 1.2 hours out of each day that I haven’t been in this range. Often it’s a lot less than that, but we’re talking about long-term averages.


Admittedly this is a case of n=1. Just me: not the average of a large clinical trial cohort. I presume there are some “outliers” on the Medtronic system, and I know there are people using AndroidAPS with larger (and smaller) TIR than me. Some of the data reported at openaps.org/outcomes mentions mean TIRs in the 80%s. There are outliers everywhere: even some MDI users manage a TIR of 90+%.

But even so I thought it might be interesting to see this graphically alongside Medtronic’s clockfaces:

Pictures can tell quite strong stories. And this one made me laugh!

The tagline on Medtronic’s video is:

Get Back Up to 45 hours every week with the MiniMed 670G System.

What about this?

Get back up to 82 hours every week with an open source APS.

Just a cheeky thought…

12 thoughts on “TIR expectations”

  1. The question we’ll likely have to wait for long term studies to answer is “How much TIR is good enough to avoid complication?”

    I’m guessing over the coming years, the Medtronics of this world will improve their tech ‘inspired’ by the open source movement and their TIR will come down (although it would be almost impossible for them to match an individually tailored looping system ).

  2. 95% is an outlier amongst even the DIY systems though. If you look at the reported results on (admittedly small) samples at OpenAPS.org/outcomes, the various reported studies show that the mean TIR is around 80%-85% rather than more than 95%.

    The key point though, is that TIR improvements lead to reduced Hba1C and better overall outcomes, so while you find the marketing disingenuous, it has to be better for someone with little engagement to go on to a supported system and see improvements than continue, badly, on a system that doesn’t work so well for them.

    I know the post is meant to be tongue in cheek, but there are far too many people that won’t take it that way.

    1. Oh indeed. As I noted, any TIR improvement is a good thing. If the 670G can improve lives then that’s great overall.

  3. I am around 94% TIR using the 670g. I have been as high as 97% and as low as 88% with very minimal effort, a full warranty and 24 x 7 call center support. My point is that it is easy to pluck any one user and compare it to an average. It really depends not on averages but rather how devices work for us individually.

    I love my choice, and best part about it; I dont need to minimize other choices to demonstrate how good I have it.

    Note: I am a Medtronic ambassador. My opinions are my own. They did not pay me to say nice things about Medtronic devices or the company. In fact, they do not pay me at all. OK, they sent me a shirt and a cup but even I am more expensive than that.

    1. Thanks Rick. Well done!

      You’re right, different systems will suit different people in different ways. This is why choice is good.
      I did of course point out that there are outliers everywhere.

      There are some natural biases that most people have. Sometimes it’s a superiority bias. Sometimes the outliers believe that “everyone else is like me”. And many more.

      But as soon as anyone (in advertising, clinical trial data, whatever) shows us a result, it’s instinctive for most of us to compare ourselves against it. Especially in diabetes where we have so many numbers to use in comparisons.
      Sometimes the results are amusing.

  4. I think it’s a very important time for people with diabetes and their care providers to be discerning. Not all loop systems will be equal. When choosing an off-the shelf system don’t just look at glycemic outcomes. Look at what percentage of trial participants stuck with the tech for entire trial period vs how many gave it up. You don’t want your next tech to increase your diabetes burden. I’d only sign up for something that guarantees automatic upgrades as new software becomes available. Read plenty of actual online user reviews. See if you can ‘try before you buy.’ Agree with Tim’s points.

  5. Andrew Bassett-Smith

    Just checked my last 3 months, which includes over a week of crappy numbers due to corticosteroid injection in the shoulder, thus almost 10% of that time, my TIR is still in the 85% mark

    That’s using the TSlim pump, so no automated adjustments happening.

    Rolled the 3 months back a bit to before the injection and it’s up around the 93% mark. I don’t eat low carb, but do enjoy a low carb high protein feast from the Weber bbq

    Have downloaded Dana Lewis’ book, to see how to use the Nightscout data etc and look at what recommendations it might have. Using tools to better our control at whatever level it might currently be should be embraced

  6. I’m on shots and currently hitting 85.5% TIR, even though I’ve been abusing Fiasp with all the carbs.

    Getting a basal profile really nailed down and then understanding how insulin works makes the biggest difference. I suspect most people don’t take the time to learn and just wing it. No surprises that doesn’t lead to optimal clinical outcomes.

  7. I have been using the 670G since it became available in Australia 6 months ago and my overall TIR hovers around 85%. However it isn’t the Medtronic algorithm doing the hard work but me setting the alert before high alarm at 9.0 mmol/L and either going out of auto to manually bolus or inputting “fake carbs” to get a bolus which will keep me in range. My pump will happily pump micro basal doses while my SG hovers at 12-15 mmol/L for hours.
    I am really happy when it does work especially with zero overnight hypo and I bolus correctly for meals. It’s just when I start to spike the pump is too conservative to manage my BG levels.
    After 42 years on injections my hybrid loop pump is a game changer for me.

  8. Rick Phillips,

    as Medtronic’s ambassador, you know that the 670G is not available everywhere, and it won’t be nearly as soon.
    Therefore, loop systems such as OpenAPS and AndroidAPS, which do what may get into these countries in a year, two, three and maybe not at all, are much more affordable for diabetics in these countries and work with insulin pumps that are in these countries’ health systems available to patients.

    The fact that OpenAPS and AndroidAPS are DIY projects and are not approved by state authorities does not say anything about the quality of these solutions. Of course, it is not for everyone, but whoever wants to, they can learn to control and use them.

    And when it comes to phone support … well, there is a problem with the corporate hotline that in every country there are only a few people who do not solve system weaknesses but only extinguish hot user problems.

    In DIY projects, a person has a large number of people at their disposal, who, like a corporate hotline, will help him solve his current problem, but moreover, the user himself, based on his knowledge with the use of the system can contribute to its improvement …


  9. It’s depressing that the bar is set so low, and that so many people don’t have access to the tools [including hardware, but mostly education, guidance and confidence to act] that might let them do better. Seeing all the posts on the T1 fb groups where people doing new pump starts are given basal and bolus rates wildly different to their experience on MDI and are then left running sky high for weeks on end while their DE tinkers with +/- 5% adjustments makes my blood boil, but makes it easy to see why 70% TIR with a helping hand from the machine sounds appealing to many.

    I come from a biomedical research background and do analysis & modelling of population health in my day job, plus my wife is a medical specialist, so getting a handle on diabetes management has been not easy but achievable for us. What hope has the average punter, having to take the advice of whatever DE is available to them as gospel and figure out the rest based on the reports of well meaning strangers on facebook groups (albeit including the author of this blog and some other commenters, whose opinions seem more educated than any DE I’ve dealt with!)? Even though CGM is now a reality for most kids under NDSS, so many parents seem to only use it as a reactive tool for managing hypos and hypers, and even their health team don’t appear to have used the wealth of data available to optimise dosing or treatment strategies and instead fall back to textbook age/weight/%TDD formulas.

    My son is currently sitting at 76% TIR (4-8mmol/L) for the last 30 days (T-slim + Dex G5), and that’s without being hypervigilant about it. He’s only 8 and we don’t want to go texting him during class all the time, so if he goes a little out of range he’s not going to chase it with a correction until his next meal break a couple of hours later. If I set the target range to 4.0-10mmol/L the stats jump to 90% TIR. Most of the time not in range is due to his honeymoon switching on and back off over a matter of days (carb ratio was 1:21, then down to 1:45, now back up at 1:18 all in the last 30 days). We’ll see how we go cranking TIR up from there assuming he stabilises with honeymoon in the OFF setting! I’ve finally got around to setting up Nightscout & NightscoutLoader and running AutotuneWeb in the last couple of days so have a little tinkering to do, but the TDD comes out almost identical, it suggests about 8% decrease in carb ratio, 15% increase in ISF and -10 to +20% increase in basal in a few time slots.

  10. Interesting to know TIR of others. The main issue is that trying to adjust TIR too strictly to 90% or higher means your day is a routine, even if you don’t have hypos and excursions to high. I mean, if you practice some sport, the normal issue is to be high/low for some time. Diabetes varies invariably over a variety of circumstances. To put it simply: I was driving in the highway under heavy rain and suddenly a crash car happened. I went from 6.7 to 10 in nearly 20 minutes (when normally after breakfast is may reach a maximum of 7.8).
    Yet I’m using Freestyle libre paired with a MM2 bluetooth and collecting data in my smartphone. I am not a user of a pump, and use tresiba to control the basal (an injection every 24hr) and the usual rapid acting insulin before meals. My HbA1c is around 5.9 thanks to that control and I practice road biking. So I think one can get TIR around 80-85% without a pump. It is more a question of having the right information to make decisions and thanks to these close-to-CGM applications one can review the treatments. I upload everything to NS and use autotune to adjust basal, which is a perfect tool.
    To summarize: I think the claiming of this or that pump insulin to help get TIR in any range is artificial. Furthermore, my target range is 4.4 – 8.3 and I get a 85% in the last 90 days. Education is what helped me more and open source apps has been a definite step to get that. Thanks to that open source movement we have this. Otherwise, we would have to pay or wait for multinationals to get their money without considering totally to users.

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.