Afrezza: Extreme Coke Challenge

This impressive video demonstrates how Afrezza works very differently to previous treatments for Type 1 diabetes.

Although I’m only 3 weeks in and still working it all out, I’m beginning to see how Afrezza can really widen the range of foods I could eat while still improving my glucose control. This is the best demonstration I could think of to show off its capabilities. For non-diabetics, read this for real experiences of would happen without Afrezza.

I did this test on an empty stomach. I started at 11:57am, the only thing I had consumed was a coffee at 9:27am. For this I had the small Afrezza dose (blue), so there was no residual glucose lowering going on.

I couldn’t have done this even a week ago, and feel like I am getting better control of my diabetes and a better understanding of Afrezza every day.
Carbs: Ice-cream 15.2g + 375mL Coke 40g = TOTAL 55.2g sugar
Lantus: 7u AM, 9u PM – (before Afrezza, Lantus was 10u AM, 9u PM)

Also see a follow up experiment using emergency glucose here.

The unedited files can be downloaded below.


Update: The results for this experiment may be even more impressive now that I have further optimised my basal insulin type and dose. Expect a re-run in the future.


iPhone recording seems to have an 8GB size limit and stopped recording briefly, at an uninteresting point. This is why the timer can be seen jumping at around the 1hr mark.
Unedited Pt 1:
Unedited Pt 2:


  1. I have had diabetes for seven years and although I will turn 64 early next year, I have been diagnosed with Brittle Type 1. Professional fishing and cane farming rule out invasive CGMs and I have been advised by my endocrinologist that a pump would not work for me because of my varied and unpredictable responses to insulin. I use NovoRapid and Levemir and most days both doses have to be changed, sometimes drastically, depending on my day’s activity. Levemir is usually a 40% morning – 60% afternoon and NovoRapid with meals.
    Although my HbA1c is always about 7, I would like to see that number lower but because of the large swings I see with BGL, this would be hard to achieve. Aferzza could be useful for me but Tresiba would not work in my case.
    It is good to see you are trialing Afezza, Matt and I will be interested in how the trials progress.
    (Although you do exactly the same thing day in day out, diabetes can scramble your day by giving you completely different results.)


  2. Matt,

    It’s great that you’re trialing new ways to improve your type 1 diabetes. I hope that Afrezza continues to be available after Sanofi ended it’s partnership with it’s producers (MannKind). As a type 1 for 30+ years I can see the benefit of such a quick acting insulin.

    I do wonder why you decided to have a low glycemic index (G.I.) food (ice cream) before having the high G.I. coke? I have been told that if you eat a low G.I. food with a high G.I. food, the low G.I. food reduces the G.I of the other. Is that the reason you combined them or was that not a consideration? It’s still very impressive to have a minimal increase in you BGL after drinking coke. I wonder if your results would be any different if you weren’t on the Tresiba, just Lantus or Levemir…

    The other thing is that you said the Freestyle Libre tests your glucose every 90 seconds or something similar. I thought it only records the BGL every 15 minutes? I have one. I like it but it did do my head in a bit at first – not used to finding out if my BGL is going up or down quickly. At $95 per 14 day sensor I’ll only use it 2 – 3 times a year though. I don’t think they’ll get it subsidised anytime soon (or maybe only for 21 year old or younger – thanks government).


    1. Hi,
      The above test you are commenting on is using Lantus, not Tresiba. The results you see in that video are not as good as my later tests with Levemir and Tresiba. Watch the videos below for comparison.

      To see a similar test without the ice cream, (using emergency glucose instead), with Levemir instead:

      Or best of all, see this test without the ice cream, with Tresiba:

      The Libre samples interstitial glucose every 60 seconds, and uses those samples to generate a predicted blood glucose value on demand. The sensor generates a prediction every 15 minutes automatically too, and stores 8 hours of these in the sensor (These are also copied to the reader too).


Leave a Reply

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