Is refrigerator storage safe for insulin?

I’ve written before about my purchase of Med Angel refrigerator bluetooth temperature monitors. I continue to use them today and have found them useful. When I returned home after a few four days of travel, I was able to get all the temperature data from my time away and see that my insulin safely stayed in the recommended temperature range.

I received an email today from the company that sells the Med Angel products and found this info interesting. Refrigerators do not keep the entire volume of the fridge at the exact same temperature. Turns out that the temperature ranges from warmer in the butter compartment to colder in the bottom drawers. That didn’t surprise me but what I didn’t know is that the warmer and colder zones can exceed the safe temperature zone recommended for storing insulin.

I’m aware of the marketing dimension within this linked piece but I feel the other information it contains makes it useful to us. I have no connection to this company other than as a satisfied customer.

Im not concerned about my insulin maintaining the exact temp for as long as I keep it in there. Im skeptical of the graph shown in the link, especially the “middle shelf” chart. Always below freezing? Looks to me someone need to adjust their fridge temp.

Ive stored my insulin in the top door storage area for…I dunno…52 years? Ive used ice chests, and even stored it outside before. I dont ever remember having a bad bottle because of this.

Honestly, for me, this is close to the bottom of things I worry about.


The middle shelf in the linked chart varies above and below 8 degrees Celsius, well above freezing.

I agree with your sentiment. In the grand scheme of things that demand our everyday attention, the temperature of our refrigerator doesn’t rank very high.

On the other hand (there always seems to be another hand!), have you ever wondered post meal, after you’ve delivered a carefully calculated dose of meal-time insulin and also gave it the perfect pre-bolus time, yet the diabetes deities decided to mess with you and you have totally unexplainable high blood sugar? I have and I’ve just adopted the attitude that I can’t explain everything that I observe about my blood sugar control, but I still feel this nagging sense that there must be an empirical explanation, even if I don’t know it.

I think insulin storage temperature might explain some of those mysteries. Would you detect, say, a 10% degradation of your insulin potency? I don’t think I would and would likely just adjust my pump settings and move on. But this is one factor that we, as people trying to live and function on our life-sustaining insulin, should not have to deal with.

That’s a problem with standard US refrigerators; the ones that have the blower fan. If you put something next to the blower it will freeze and that can kill insulin (see below, toward the end).

I think the blower outlet is always in the top. I’ve seen freezing when the top has stuff crammed into it so that the freezing air coming out of the outlet isn’t able to pass all the way across the top before it hits something. That something freezes.

I store insulin in the door. There’s more temperature variation there but, realistically, it doesn’t matter that much. After all the instructions for kwikpens basically say to let them come to room temperature before use. My wife ships an emergency kwikpen in her voluminous hand bag, when I need a new one I take that one and swap it for one from the 'fridge - once a week.

The kwikpen box/leaflet itself doesn’t seem to have any information beyond “do not freeze” however I found this web page:

That temperature range (for ‘room temperature’) is 60F-80F, but that is for 28 days and that’s all they tested.

The high temperature limit is 70C (168F):

Translation: if your insulin hits 70C it is dead, pretty much instantly. Between 25C and 70C the process will get exponentially faster - it takes longer than 28 days at 25C, next to no time (seconds, if that) at 70C. Avoid the parcel shelf of an automobile at all costs, but bear in mind that the human body runs at 37C/98F approximately and insulin lasts 8 hours inside us; I don’t know whether that limit is because of heat denaturation or chemical but it is one or other (probably chemical.)

What about the low temperatures? Freezing stuff, particularly complicated glassware like insulin, frequently destroys it. Being T1Ds someone had to test it:

Oops. So the freezing did denature the insulin but the result was not what you might expect; it apparently changed the adsorption characteristic but did not prevent the hormonal behavior. That’s just one experiment but my conclusion is that if it freezes throw it away.

The “normal” temperature of a fridge is 2C/5F, some people (I will not name names but she is married to me) think that winding the temperature down because parts of the fridge seem warmer is a good idea. Freezing both non-homogenized milk and fizzy beer ruins them.

I can’t see any evidence that zero C damages insulin and I would not expect it to; the problem is the freezing. My intuition is that that problem is instantaneous, just like a frozen pipe bursting: it happens at the instant of freezing, it doesn’t matter how long you leave it frozen after that, so just throw it away.

1 Like

We just noticed that the veggies stored in the drawer below the one with my insulin in it were frozen. Middle (insulin) drawer seems ok, but I was just thinking about maybe I need to get a monitor before losing like 9 vials (zombie apocalypse stash in there too). So it’s a pretty timely post for me!

1 Like

Middle draw is probably safest in the main compartment. Leafy vegetables are pretty good freeze detectors - they don’t have much volume/surface area so they freeze very fast and they turn black so you always notice.

I had a quick look on amazon because I was sure I remembered some kind of ‘freeze patch’ - something that can be applied to the outside of a vial (or box) and turns black (or red, etc) if it ever hits zero C. It’s not the same as freezing - a vial of insulin takes a while to freeze and, indeed, slowly freezing it is likely to produce different results to a fast freeze - but it would be really helpful to have some kind of patch.

John Bowler

1 Like

Tangential to this issue and an interesting sidebar is the issue of whether the insulin directly issued from the manufacturing source is always at 100% potency. From a DiabetesMine column published in the last year or so:

They randomly picked up 18 vials of insulin and stored them in a research lab at the appropriate temperatures, before discovering that the concentrations weren’t adequate. Insulin vials and cartridges are required by the FDA to contain 95 U/ml of insulin when they’re sent out by the manufacturer. Regulators also require proper handling and temperatures in order to maintain the quality.

But when Carter and his team examined the insulin they’d purchased, the vials ranged from 13.9 to 94.2 U/ml, with a mean of 40.2 U/ml. No vial was at the minimum FDA-required standard of 95 U/ml.

As if we who try to do this incredible metabolic balancing act don’t have a hard enough job and then find out that the central tool we’re using and thinking contains 100 units per mL only contains 40.2 units per mL! That’s not funny. :angry:

According to the article, the Diabetes Technology Society was supposed to do some follow up but I’ve not seen it covered anywhere. As expected, the Pharma-financed ADA dismissed this news.

I wonder how careful the insulin supply chain manages the temperature of insulin?

1 Like

That’s a mean of 40IU/ml - an arithmetic mean I assume - and that means that it wasn’t the odd bottle that was at only 13IU/ml! Uugh. So our insulin is actually less than half the strength we think it is.

I wonder if this is new - I calibrated myself back when I started on the Omnipod and I got something like 64mg/dl correction for 1IU and 12g/IU carb coverage. Those match the “standard” rates pretty closely. Now I’ve been battling all day with a sticky high. Am I using a 13IU/ml pen?

I assume the 95IU/ml figure is the minimum, but the fact that they didn’t get anything even at that number implies that the original manufacturer is cutting the insulin to make more money. That’s standard drug supplier procedure I know, but it’s still annoying.

Reply to my own post; I found the stuff I was looking for:

So they sell 24 labels for $13.20 and the above example changes color, permanently, at 90F (or above). They don’t seem to have low temperature detection, but for me living in Oregon (115F peaks pretty much every year) spending 50c/kwikpen to be sure it hasn’t been fried seems worth it. I’m thinking I would stick one of these to the pen when I took it from the fridge.

I prefer chemical detection because what I’m trying to detect is, itself, an irreversible chemical change (to the insulin). Also if I buy any more electronic stuff I won’t have to worry about being a T1D any longer because my wife will kill me :slight_smile:


I remember that one of the criticisms of this report was that 18 vials of insulin don’t represent much statistical heft. I’d really like to see some credible agency take this on and verify whether this is an actual problem or not. Seems to me that our $300+/vial should easily finance some independent quality verification/control.

1 Like

I couldn’t get the original paper (well, ok, I wasn’t prepared to spend $40 to get it from Sage Publishing) however the idea that 18 tests are not statistically significant when every single one was below the regulatory minimum and the average is so far below seems bogus. I saw the criticism too, it seemed to come via the ADA.

I can’t evaluate the criticism without the original data - specifically I need to know both the mean and standard deviation (the abstract published the mean). I do know that the SD can’t be more than 54 and even at that outside estimate the chance of the result not being significant - being random - would be less than 1 in 4 (less than .25).

I suspect that the researchers would not have published if the chance of it being a random result was greater than 1 in 20 and it keeps coming back to me that all the numbers were less than the minimum. The chance of that is about 1 in a quarter of a million unless there is decay in the supply chain.

Pity, they should have got 20 vials, then we would have hit the 1-in-a-million magic publicity number.

Seems to me that our $300+/vial should easily finance some independent quality verification/control.

I agree, but that is a fake price; the whole insurance/pharmacy/manufacturer price reporting scheme is set up to inflate the prices charged to Medicare. My current best estimate of the actual money that changes hand is that, on a $300 vial, it is at most $120 of which $60 would be the price the manufacturer charges. Unlike the statistics I wrote elsewhere this is speculation; they sure as certain-is-certain are not going to tell me.

Putting either pharmacy benefit managers or the low-cost distribution networks that the successful pharmacies (Walgreens, Walmart) use in the supply chain inevitably means problems will occur. It’s difficult to believe that the manufacturer QC has gone down the tubes so far that numbers like this can happen but it’s entirely possible to believe that they ship 95IU/ml because they know that is permitted. However if they do that then their QC is within a fraction of a percent of that value.

I still haven’t got the Federal standard. Based on experience I would guess that it is that 95% of manufactured units have between 95IU/ml and 105IU/ml.

You think it’s possible some of us might be getting U105 insulin when dosing U100? That’s twist I hadn’t even considered!

You think it’s possible some of us might be getting U105 insulin when dosing U100? That’s twist I hadn’t even considered!

It’s what most engineers would describe as “within limits”. It’s what I expect; I don’t care at all about +/-5% because everything else I put into my body as a T1D is less accurate.

I do care, a great deal, about 40%, but I just want to know: if I know the problems with the technology I can deal with them, they become my problem and I really don’t want to rely on someone else to solve my problems. I’ve been dealing with my problems all my life. If I’m not told, if the truth is hidden, then I can’t deal with it.

I buy about 6-9 months of insulin cartridges at a time because I live 2 1/4 hour drive from Canada border and a box of 5 Humalog cartridges in Canada sells at retail for USD $48 vs list price of $498 in US for same item from same factory. Canada does not require a prescription so you can buy as much as you want but do need a prescription to show US customs if they request it. The supply then goes on top refrigerator shelf and a cheap $4 room thermometer is propped against the dozen or so boxes of insulin. Refrigerator temperature control is set to keep the insulin below freezing as shown on thermometer and never have had a problem using this system for Humalog/Lantus/Novolog and Levemir.

I keep my insulin in the butter dish on side of fridge. It is a new fridge and digital reading of 37 degrees F. I have had trouble in hotels with insulin freezing in bad old refrigerators when I have moved and had to stay in hotels. I read if the insulin freezes you must throw it out.

Good to know! I used to keep my insulin in the door of the fridge, but the last few months have had troubles. Insulin won’t last a day out of the fridge before it becomes useless and it is supposed to last 28 days! So I got a mini fridge on my counter, just for the insulin. Hoping this works better… so much for stashing some in case of a zombie apocalypse. Only time will tell.

Here’s the storage information, verbatim, including capitalization errors, from the leaflet included with an Eli Lilly HUMALOG KwikPen:

Unused Pens
• Store unused Pens in the refrigerator at 36°F to 46°F (2°C to 8°C).
Do not freeze your insulin. Do not use if it has been frozen.
• Unused Pens may be used until the expiration date printed on the
Label, if the Pen has been kept in the refrigerator.

The “expiration date” on the boxes I have means that the insulin can be stored in that temperature range for at least 2 years and 9 months. Then for pens that are in use:

In-use Pen
• Store the Pen you are currently using at room temperature [up to 86°F
(30°C)]. Keep away from heat and light.
• Throw away the HUMALOG Pen you are using after 28 days, even if it
still has insulin left in it.

The upper figure here, 86F/30C, is significantly higher than the 25C quoted elsewhere as a general rule and this is only for pens that are in use. Immediately after this section of the leaflet:

• Always carry an extra Pen in case yours is lost or damaged.

So clearly given the 28 day limit for an “in-use” pen storage at temperatures up to 30C is safe for at least 56 days; otherwise it would be necessary to throw away every other pen without use for people who don’t use a complete pen in 28 days. The wording of the 28 day limit is clearly written to preclude this - it applies to the “in-use” pen, not the extra pen.

This may be because part of the issue is contamination of the previously sterile insulin container from airborne bacteria. Anyone who has done any airbourne fermentation (as in lambic beer, but you can do it with bread and, I guess anything) knows how easy it is to introduce yeast just by waiting and bacteria are a lot smaller and more readily mobile than eukaryota like yeast and, for that matter, us.

I keep my insulin in the bottom drawer of the fridge and have one thermometer in the front of it and one in the back. The temps range from 32/34 to 40 degrees Fahrenheit with the lower temp in the back. It has never frozen or had any problem. There is a blower on one side of that compartment but I keep the cover on it mostly closed so it isn’t blowing much into the drawer. That drawer is full of insulin, and some vials of B12 injectable, as I have about a two year supply of both in there and have never had any problem with any of it. The B12 injectable vials are in the back of the drawer, which is the coldest part, and it is on the side that has the blower. It would be really easy to see if that has frozen and it has never frozen at all either. The strength of all insulin has seemed pretty consistent.