Afrezza: Extreme Storage Challenge

During the early stages of Afrezza’s development, it was claimed that it could go without refrigeration for more than two months. However, the packaging instead says to only store at room temperature for 10 days. This extreme test goes 16x longer than the labelling to see how it still performs.

Since starting on Afrezza I have been wondering if Afrezza really does require refrigeration. As you will see in the video, where I stored the insulin was certainly much warmer than average room temperature.

The blister pack spent 161 days without refrigeration before this test. It was first put in the heated hothouse on August 22nd, and then taken out on December 5th (105 days). After that, it spent 56 days in the hottest room in my house during a typically hot Australian summer.  Melbourne had 20 days above 90° F (32.2° C), and 5 days above 100° F (37.8° C) during the test period.

Unlike the US, summer officially starts in Australia on December 1st. For an official temperature record of Melbourne over this period, visit this Bureau of Meterology page.

To test if Afrezza still worked, I used a can of Australian full strength Coke, (which has sugar as 100% sucrose in Australia). Sucrose and caffeine together make this drink raise glucose levels quickly. A control test using newly-ordered Afrezza stored in the fridge was also conducted (video will be uploaded soon). Basal insulin was the amazing Tresiba, and no food or other insulin had been taken prior to the tests.

Long story short, there was NO decrease in Afrezza’s efficacy after the five months. It worked flawlessly and glucose never left the non-diabetic levels. Glucose started at 4.2 mmol/L (75 mg/dL) and ranged from 3.8 mmol/L (68 mg/dL) to 4.6 mmol/L (83 mg/dL) during the test. I certainly won’t be requesting refrigerated shipping from the US anymore.

The significance of this really can’t be overstated. The Frio insulin cases we use when travelling barely work in tropical humidity. For diabetics who travel frequently, those who live in hot climates, or anyone who doesn’t want a fridge packed full of insulin, this is a game-changer.

Put simply, Afrezza does not need refrigeration and there is no other insulin on the market that can make this claim.

The manufacturer, Mannkind should capitalise on this feature, and submit revised storage conditions for approval. Extended room temperature storage is yet another major point of difference between Afrezza and all other insulins.


PS:  Full length footage of experiment will be uploaded soon.

Update: First HbA1c Result

An update with my first HbA1c result since starting Afrezza and the FreeStyle Libre.

My previous HbA1c results have been around 6.5%, which is inline with the recommendations from the Australian Diabetes Society, who recommend “a general target HbA1c of ≤7.0%” for Type 1 diabetic patients.

My latest test result, which reflects the average glucose level over the previous three months, was likely somewhat affected by the basal mix-up I described previously. The result of 6.0%, although a lifetime best, was therefore a little disappointing – especially with some incredible A1c results being posted online by other users.

If I continue with the same level of control as I have now, my next A1c test is likely to be better – as it will not be impacted by the basal mixup. It’s also worth noting that unlike most users posting great A1c results, I do not have a continuous glucose monitor, and therefore do not benefit from alarms when my glucose is going out of range. The FreeStyle Libre samples blood every minute and provides CGM-like historical data, but requires a manual scan to view this data.

I suspect that using a CGM might change the way I use Afrezza, and would provide a further reduction in my HbA1c due to alarms notifying me earlier than my manual scanning with the Libre.

Although I am a little disappointed, the author of the Crick, Tech, Munch blog gave me some encouraging words that put the result in context: Only 7% of Type 1 diabetics in the UK reach an HbA1c result of 6.0% or better.

The HbA1c test is also limited in that it can potentially reward poor glucose control, by being lowered through periods of hypoglycaemia. A supplementary test —1,5 Anhydroglucitol — is a historical indicator of hyperglycaemia only.

Coupling the HbA1c result with this GlycoMark 1,5AG test gives a far better picture of glycemic control than HbA1c alone. Together, they can show whether an HbA1c result is being artificially lowered by hypoglycaemia and give an accurate indicator of glycemic variability.

I hope to have another HbA1c test in about two months. I have also found Australian laboratories that can perform the GlycoMark 1,5AG test and hope to have that conducted around then too. I am looking forward to seeing these results.


HbA1c 6.0%

First HbA1c since Afrezza. Historical results are also shown on document.

Simple Sugar Surfing

A new technique of using insulin is causing quite a buzz within the T1 community. Some think it will become the new standard of care, replacing DAFNE. Read why Afrezza greatly simplifies this method.

Also called Dynamic Diabetes Management, Sugar Surfing is an intensive glucose management technique for tight glucose control that requires a continuous or flash glucose monitor. It de-emphasises the traditional insulin-to-carb and correction ratios, and introduces a whole new lexicon derived from surfing such as shove, nudge, drift and flux.

Timing is the most important part of Sugar Surfing, more important than dose size. The technique involves learning the lags of your insulin and each type of food and taking into account the delay of the glucose monitor. The goal is to optimally time more frequent interventions. High glucose is treated immediately, rather than waiting until the next mealtime. In practice, the approach tends to result in about 9 pump boluses or injections of mealtime insulin in a typical day, which the author compares to the 11-12 waves of insulin each day in a non-diabetic.

Afrezza users who are posting the amazing A1C results have most likely mastered a simplified variant of Sugar Surfing without realising it. Although I’ve only recently discovered Sugar Surfing, I’m amazed how it resembles how I have ended up managing my diabetes, using Afrezza and the Freestyle Libre…. except the Afrezza way is, again, far simpler.

At first, the technique looks quite daunting. It involves a lot of trial and error – and “low expectations” are advised for the first 12 months. However, Afrezza greatly simplifies Sugar Surfing, and precisely addresses some of its weaknesses. Surprisingly, this means that less doses of Afrezza per day are needed to achieve the same glucose control.

When using injected insulin, the technique requires “pre-empting” dosing as long as 45 minutes before every meal, then waiting to start eating at the optimal time for each meal – once a curved pattern appears on the glucose meter display.  With Afrezza, pre-dosing isn’t required with any meal, which greatly reduces the scope for timing errors.

The frequent dosing of injected insulin brings the challenges of stacking doses, that increase the likelihood of hypos. Sugar Surfing introduces the “i-chain” concept to deal with this problem, but the lack of a long tail of action makes Afrezza so much easier.

With Sugar Surfing, in addition to interventions triggered by CGM alarms, the next intervention decision is to be made 2-3 hours after dosing insulin at the latest. At this point, a decision whether additional carbohydrate “nudges” or additional insulin “shoves” is to be made. Additional injected insulin at this stage is problematic, as additional carbs will be required later to avoid hypoglycaemia. Afrezza eliminates this problem, as it is finishes its action so much sooner. Post-meal lows requiring “nudges” seem non-existent with Afrezza, presumably for the same reason.

A single basal rate is recommended for Sugar Surfing, making Tresiba the perfect basal for it. Tresiba is equivalent to a pump set to a fixed basal rate all day, and takes out a whole host of variables, further simplifying management.

As the technique requires taking into account the lag in the glucose monitor, the FreeStyle Libre is brilliant – with changes showing six minutes before the Dexcom G5. Although the Libre lacks alarms, Sugar Surfing highlights how its factory calibration is not just for convenience. Each calibration required every 12 hours with the G5 introduces variability and often reduces accuracy.

If you’re considering using Afrezza, it’s worth checking out the book. It will introduce you to the mental leap that you will have to make with Afrezza – that timing is more important than dose size. You can begin applying the principles with your existing regime – and see how labor-intensive the technique is. I would think that after a few months Sugar Surfing with injected insulin, it would be obvious why you would want to upgrade to Afrezza.

Users with a background in Sugar Surfing would hit the ground running with Afrezza, without the learning curve and “unlearning” that I encountered. A simplified Dynamic Diabetes Management protocol would serve as the best protocol for Afrezza use in Type 1 diabetics. Most users would eventually figure this out themselves within 6-12 months, as it becomes quite obvious when using Afrezza with a CGM.

However there’s no reason to keep new users in the dark. Unhelpful in-box instructions fixated on unit conversions to injected insulin should instead introduce basic dynamic management techniques. I hope to develop a simplified set of these principals for those starting Afrezza.


Sugar Surfing with injected insulin brings you closer to normal glucose levels, but further from a normal life. (Source: Stephen Ponder)

From the sample chapter of Sugar Surfing:

“Did you know that when you use your insulin pump to bolus 1 unit of insulin it can be off by a range of plus or minus 30% – 200%? “

“Personally, I found this obsession with ultra-precision in insulin dosing a bit unnerving since I didn’t usually see any improved control in patients who practiced it. “

“Stuck on the idea that you are only going to inject X number of times per day and no more? As I said above, the pancreas of a non-diabetic is known to unleash 11 internal insulin bolus events per day How can you expect to have the blood sugar profile of a non-diabetic if you’re not even open to the possibility of giving yourself half that many injections? “

“the human body produces on average 11-12 waves of insulin from the pancreas each day. This is actually consistent across all persons without diabetes. I’m not advocating 11-12 insulin doses a day, but if non-diabetics produce insulin between meals but if non-diabetics produce insulin between meals, then maybe learning how to do that properly isn’t a bad idea “

Afrezza & Tresiba Update

This post reflects on the positive impacts on my day-to-day life since switching to Afrezza and Tresiba, along with some new challenges. A simple mistake led to three weeks of terrible glucose results, and the cause took a while to track down.

The day after I have my best ever 24hr glucose result (see charts at bottom), my glucose levels began going out of control. Unexpected highs between meals led me to try increasing my basal dose. Then major lows began, so I started changing the time of dosing basal. My total daily dose of Afrezza dramatically increased. I even switched back to injectable bolus for two days to see if that would help.

When my basal cartridge ran out, I discovered what the problem was. I had accidentally refilled the pen with a Levemir cartridge. Both cartridges are green, and are in blank foil blister packs. The pen can also completely cover the label, depending on how the cartridge is rotated. I certainly won’t be making that mistake again.

Both cartridges have green rings and are in blank foil packs.

Both cartridges have green rings and are in blank foil packs.

When I restarted Tresiba, I dosed twice in the first 24 hours to try and speed up the transition. After the third dose, levels were returning to normal.

As troubleshooting this issue has consumed about three weeks, in the video I reflect on the positive impact that switching to Afrezza and Tresiba have had on my life so far. In addition, there are some new challenges that these treatments present.


All of the impracticalities of dosing insulin before a meal are gone. When I’ve eaten enough and I feel full, I now don’t have to force myself to finish the meal. If I want to share food I’m eating with someone else, I can offer it to them without having to explain that I have already had the insulin for everything that’s on my plate.

Cooking is a pleasure now I don’t need to weigh ingredients, or weigh dishes while serving. I no longer have to copy numbers from nutrition panels and look up web sites in order to perform a series of complex mathematical equations before every meal.

I am no longer constantly worrying about encountering another company like HelloFresh again, who provide diabetics with potentially lethal falsified carbohydrate data.

Previously, visiting friends for dinner required them listing all the ingredients and quantities for the meal before it was served. I avoided going to restaurants wherever possible to avoid the impossible task of dosing for unknown meals of unknown sizes being served at an unknown time. It’s an amazing feeling being relaxed for the first time in these settings.

However there are new challenges. Afrezza acts for a shorter length of time than injected insulins. Some fatty meals like pizza require split doses of injectable insulin. After about a year of trial and error I settled on injecting two doses five hours apart. Without a CGM, I don’t have alerts for when my glucose is going out of range so identifying the optimum timings using Afrezza will take some time. I hope to systematically develop some simple guidelines for dosing to help out others that are not using CGMs.


For the first time in my life, I can sleep in on weekends without glucose levels suffering. Many years ago I dosed Lantus once daily in the evening which allowed me to sleep in, however it was unable to last 24 hours so levels were rising in the evenings.

With Lantus and Levemir twice daily I had to return home at the same time every evening for my second dose, ruling out spontaneous social events. It’s liberating to be able to chose when I want to go home, rather than have my basal always counting down the minutes.

So far, I have not observed any detectable difference if I dose one morning and then wait until the next night. As summer approaches here, I am sure I will have more opportunities to test out the limits of Tresiba’s flexible dose spacing.

The only downside so far is managing the different basal requirements I have on the weekend. I imagine this was a problem before, but is highly visible now I’m using the FreeStyle Libre.

During the week, I eat three times a day and am less active that I am on weekends where I tend to eat twice daily. My solution so far is to eat or drink something sugary while doing extended physical activity like working in the garden.

23 Sep - Best Results. Note the Interquartile Range.

23 Sep – Best Ever 24hr Period. Note the tight interquartile range.

FreeStyle Libre Data

23 Sep – FreeStyle Libre Data

Afrezza: Timing is Everything

This video covers my experiences with Afrezza dose timing, and why I think timing is the most important aspect of dosing – even more important than the size.

I have previously mentioned the importance of dosage timing when taking Afrezza, but have not gone into much detail. So I thought it would be worthwhile talking about what I have learned so far about timing.

As a general rule, I dose about 10 minutes after I start eating, which is before my glucose levels start to rise from the meal. As I mention in the video, the best time to dose seems to depend on the fat content of the meal. And for some high fat meals, a follow up dose of Afrezza is neccesary.

I have found the same rule also applies if a follow-up dose is required. It is important to have the follow-up dose before the levels begin to rise out of range. If I had a CGM with alerts, I would use this to notify me as soon as it levels began to increase. That way I wouldn’t miss the optimum time for the follow-up.

It is worth remembering that much, much more Afrezza is needed to correct high glucose levels than to cover meals. As the video details, if I wait too long to dose, the dose required may be four times higher!

– Matt

Update: Although timing is important when using Afrezza, the above high in the graph and the hypo referred to in the video were the result of unintentionally replacing a Tresiba cartridge with a Levemir cartridge.

The hypo wouldn’t have occurred had I been taking the correct dose of either basal. Similarly, the high in the graph is during a period of no basal coverage. See this post for more details of what happened.

Afrezza with Tresiba: A Perfect Match?

This video discusses my switch to the new Tresiba basal insulin for use with Afrezza. Is this ultralong-acting basal the perfect match for the ultrarapid-acting Afrezza?

After testing 18 different dose permutations of Lantus, I remained disappointed with my glucose levels between meals. Switching to Levemir improved things greatly, but introduced new problems each day.

A new basal called Toujeo contains identical ingredients to Lantus and still seems to be inconsistent in its release.  This study observed 50 people for 2 days on Toujeo. This figure shows that Toujeo’s release sometimes varied considerably between the two days and did not always lower glucose steadily throughout the day – its activity sometimes spiked or ran out.

That study was by the manufacturer of Toujeo and Lantus. Another study by its competitor showed that Lantus was over 300% more variable in its release over 24 hours than the new ultralong acting basal Tresiba.

They calculated the risk of experiencing more than double the usual maximum effect on any given day (potential hypoglycaemia) was <0.1% for Tresiba and 11% for Lantus. That’s a random hypo every 9 days on average! Plus random highs every 5.8 days on average. That’s the average for the whole group of people – some were far, far worse.

So I switched to Tresiba to see how it works with Afrezza.  It comes with many other advantages, and eliminates some routines that have ruled my life for 24 years!

The number of doses of Afrezza I use each day is reduced, including the dose I previously needed as soon as I woke up. Additionally, all of the problems introduced with Levemir are gone. It’s early days, but here’s a picture of where I’m at now.

Now that my basal is almost fully optimised, I can soon work on finding the optimal strategy for timing and dosing of Afrezza.

UPDATE: Tresiba has just received FDA approval in the US. Still waiting for Australian TGA approval.

Afrezza + Tresiba

It’s early days for Tresiba, but so far I’m very impressed.

FreeStyle Libre: 8 Weeks On

Here’s a video update about the FreeStyle Libre glucose monitoring device that I’ve now been using for about eight weeks.

Unfortunately three sensors have failed early. One gave some odd readings the day before and then gave an error saying the sensor needed to be changed.

The other two sensors kept saying I was dangerously low when I felt fine, but the system’s self-diagnostics didn’t detect the malfunction. I applied a second sensor and used a second reader, which showed a higher reading by about 2.0 mmol/L by the time I removed the sensor. It may have had something to do with the adhesive coming loose on the sensor – which hadn’t happened before.

Since recording the video, a third sensor has failed early, and underneath there was pronounced skin irritation (pictured below). This may be responsible for it becoming unstuck. Other users online have also found that their initial sensors worked fine, but their bodies seem to attack sensors applied later on. It’s happening in almost half of FreeStyle Libre users, and some are experimenting with a Compeed barrier between the skin and sensor. I might try this next time.

It’s also worth noting that the sensor applied most recently caused bleeding out of the hole immediately after application (also pictured below), but it hasn’t caused any issues so far.

Notice the skin irritation that may have caused the sensor to become unstuck. Also note the bleeding from the newly-installed sensor.

The skin irritation that might be linked to the sensor malfunction. Also note the bleeding from the newly-installed sensor, which did not seem to affect its performance.

Fortunately the incorrect Libre results are in a safe direction, and wouldn’t have resulted in an unsafe insulin dosage decision. However for the first time in six weeks, I have had to supplement the Libre with finger testing, as I need to be confident in my results while switching my basal to Tresiba.

Overall I still really like the Libre, but it would be amazing if it had the ability to transmit a high glucose alert (say, at 6.0 mmol/L) like a CGM. The Dexcom G6 coming in 2017 sounds like it might be a great (but expensive) alternative to the Libre. However, I’ve already got high hopes for a future Libre that might transmit via Bluetooth to an Apple Watch.

Also check out my video of applying the sensor.

FreeStyle Libre Sensor Application

If you’re wondering how the Libre sensor is applied, I made this video showing how it works.

The applicator is like a stamp that fires it into the skin and sticks the adhesive in one action. The application looks much friendlier than the Dexcom, and isn’t particularly painful.

Also check out this post which sums up my experience using the Libre for 8 weeks.

Afrezza and Levemir

This video update covers my switch to Levemir, and its nifty pen. Scroll down for my charts comparing the two.

The high variability of Lantus prevented me from further optimising my basal dose, and the issue makes some consider it dangerous.

So I decided to switch to Levemir. My experience with Levemir mirrors that of this user in the UK, who finds “the duration of the Levemir is frustrating, as the amount taken in each of the two injections is not quite enough to get me through a twelve hour duration“. Further, I need to be cautious during its peak.

As you can see in the video, the NovoPen Echo is very cool. However my third pen has already failed once already, and I hope it’s the top display that’s wrong and not the one on the side. Novo has asked me to send in the pens to investigate.

If I can get a pen I can trust, I would like to pair the NovoPen Echo with the longer acting, flat profile Tresiba basal (not yet available in Australia).

Update: I have since changed basal insulin to Tresiba. Novo has sent two replacement pens, and I await the results of their investigation into the faults.

The goalposts have shifted a lot in a couple of weeks.

The goalposts have certainly shifted a lot in a couple of weeks.

Afrezza Units & Insulin:Carb Ratios

This video explains why insulin:carb ratios don’t work with Afrezza, no matter how hard you try.

Afrezza works so differently to previous diabetes treatments, and yet there is nothing in the box explaining how or why. It works so differently that for a while I wondered if I had a bad batch.

This video explains why the labelling on the box is unhelpful and confusing. The units just can’t be compared, and it takes a while to work out that Afrezza has two independent effects.

The first phase begins immediately but it doesn’t lower blood glucose, instead it temporarily stops it rising. This effect seems to be responsible for all the weird and wonderful properties of Afrezza, the first phase does not happen with previous insulins.

The second phase, however, works like a small dose of injectable insulin. It lowers blood glucose, and takes longer to start working. This is the only phase of Humalog or Apidra.

Previous treatments work by flooding the body with an unnaturally high level of insulin for hours, in order to slowly lower high blood glucose levels. Afrezza stops blood glucose rising in the first place using the natural first phase signal to the liver instead. This means that for the first time much lower, more natural levels of insulin can now be used in Type 1 diabetics.

– Matt