Carbs, excercise, but INSULIN?

We all know how difficult it is to accurately measure carbs (not to mention protein and fat) and then to adjust for the amount of exercise we have been getting; all recent, continuous, topics of discussion. Is the insulin we inject any less difficult to measure? A recent piece of research suggests that it might be the biggest error:

That’s not something ‘off the internet’; it’s a properly researched piece of science that produced what appears to be a certain amount of excitement [I can’t tell for sure because I’m not subscribed to the relevant publication.]

The bottom line is that it suggests that when we inject 1 unit of insulin we are getting something between 0.4 and 0.9 units. The corollary is that every time we inject insulin we don’t actually know how much we injected within +/- 50%, and that’s assuming we know the figures from the paper.

Yet we obsess about the accuracy of our blood glucose meters; +/-15% (at best). But so what if our insulin is +/-50%?

So far as I can tell this is some inconvenient truth that at this moment is being buried where no one can find it, so I figured I would create a topic so that we could all watch what happens.

John Bowler

Thanks to Terry4 for the original link, I’m posting it as a new topic because it is so much more important that anything else raised in the last month, year, millennium.


Thanks for posting this…it’s pretty shocking. Are there any ideas for us to attempt our own quality assurance (other than cussing at the CGM and wondering if we have a bad batch)? I have usually ruled out insulin degradation when I am mentally searching for reasons for a high BG…but not any more.

I have never doubted that there is a +/- error amount in the delivered insulin.

BUT I do believe the impact will vary based on the actual size of a bolus injection. So an injection of 10 units may deliver between 9.5 and 10.5. Not quite as scary. Plus I don’ t remember, ever attempting a 1 unit injection.

I use a pump which I believe will provide a more consistant amount from 1 injection to the next. Meaning it will always be a little over or under. And assuming a fairly consistant regime adusting the I/C ratio and reasonable carb amount per meal will minimize problems.

But that isn’t news. We have known for a long time that pumps provide better results than MDI. ( for many of us) This may explain part of the reason why. Another reason is that before being approved for a pump you must demomstrate reasonable control using MDI.

Diabetes treatment has always been a huge balancing act. And this is just one of the many variables.

I find my biggest variable is the carb calculation.

The intact insulin concentration ranged from 13.9 to 94.2 U/ml, mean 40.2 U/ml. No vial met the minimum standard of 95 U/ml.

I suppose one way to look at it is that if you’ve never actually gotten a full-strength vial, you’ve been compensating for the deficit all along by setting your I:C and correction ratios accordingly without knowing it. If your supply chain has been relatively consistent, not too big a deal, but I’m sure there’s a lot variability in there. So yeah, kind of a big fudge factor in the whole thing, but I’ve always had a big SWAG factor in my own calculations in addition to whatever my endos have tried to tell me.

ETA: the other big question is whether this also pertains to pre-filled injector insulin, which they didn’t analyze. If that stuff does better it could certainly explain why sometimes doing a correction with one of my Novolog pens can seem to be much more effective than doing it from the pump.

I forgot to post the link to the Diabetes Mine article which, was actually what Terry4 posted. The link I posted is to the original research (from December 2017). The Diabetes Mine article:

from February 2018 (originally) includes the information (at the end) that a larger study has been launched in October 2018 with an expected report date mid-2019. The original study was done on Humalin vials, so it is quite possible that the mimalog insulins in pens may well show different variability - different chemical, different packaging.

One of the reasons I posted this was that I had just changed my Omnipod and, therefore, changed my insulin pen - one pen fills two Omnipods. It so happens the new pen was from a new box and, therefore, probably a new batch (I don’t have the old box). That was at 4:34AM yesterday. After that, during the day, my BG was spiking above 180 most of the time despite repeated corrections. This continued during the night at which point it started to exceed 250mg/dl - something I never want to happen.

So I junked the Omnipod at 6:35AM this morning; the canula looked fine but it is very hard to see on an installed pod. The new Omnipod is in a new site but with insulin from the same pen. At this point I do not know what was going wrong or, indeed, whether I have fixed the problem! If I haven’t I’ll start using a new batch of insulin, but even if that suddenly fixes the problem it doesn’t actually prove anything.

Ok, that’s a long story to demonstrate a simply point: I have always relied on the insulin being the right strength, within say +/-20% all the time, and have expected to compensate for bad carb counts, unpredictable delays from protein/fat, excercise, stress, inactivity (long car drives) by accurate correction boluses and accurate basals. If the insulin is, however, +/-50% this doesn’t work because it takes four hours to know what the effect really is.

It does have the potential to work in a system with a consistent (not accurate, but consistent) CGM because the rate of change of blood glucose allows a feedback system to increase or decrease the insulin delivery without needing to know the strength of the insulin. My understanding is that the Tandem systems have the basis of this; using rate of change, not relying on absolute values.

As for actually measuring insulin strength; how about a home test kit, medical entrepreneurs? It doesn’t have to be that accurate - I want to know if the stuff is half strength or double strength.


Indeed - that’s what I assume. That’s where my +/-50% figure comes from too. I assumed a mean (arithmetic?) of about 50IU/ml and a variability of +/-20IU, so 95% of insulin pens/vials will be in the range (about) 10IU to 90IU. If we set up our ratios with stuff that averages out as half strength (50, not 100IU/ml) then it’s the variability that hurts, not the absolute strength.

We should also know about this critique of the study:

1 Like

It should be noted that the writer of this critique is the Chief Scientific, Medical and Mission Officer of the American Diabetes Association. How is the ADA funded?

From the ADA website.

Our ADA, as much as I like the idea of an organization that champions the cause of people with diabetes, is hopelessly conflicted.

1 Like

At the very least the OP report raises enough of a question that it calls for a fuller study with more robust methods, not just a critique of the methods it did use.

1 Like

Interesting. It may explain some of my crazy results when doing the same things etc. there are so many factors it is hard to say, but digestion is another really important aspect for me which is often ignored. Dex is often 50-100 points off for me so I dont treat from it unless I have 2-3 close finger sticks.

I asked my endo about this and he said that reasonable control was not a requirement for the pump. But i have heard from others that such a state is desirable.

Sobering – I only read abstract. Reported range of concentration is obviously unacceptable.
However, I do feel compelled to comment on the other two factors – carb’s and exercise.
Firstly – carb’s. All diabetics, regardless of type (with the only exception of which I am aware being MODY2 in which there is no insulin secretory deficiency), should be aware that carb’s CANNOT be compensated for using peripheral (i.e. injected) insulin. This is fundamental to understanding the basics of diabetic hyperglycemia.
Those in engineering and other technical fields may know how regulation is controlled via negative feedback in man-built systems. In natural systems, including our planet and our bodies, this negative-feedback regulation is also present, evolved via natural selection.
Islet insulin (within endocrine pancraeas blood) is the primary regulator of alpha-cell secretion of glucagon, and is counterregulatory. Glucagon is the master hormone for hepatic fuel regulation – simply the most powerful hormone of the body. Insulin, the less powerful hepatic hormone, derives its important through its role in the islets in limiting glucagon secretion.
In a non-diabetic this relationship absolutely shuts down glucagon secretion while portal glucose (i.e. absorption of a meal with carb’s) stimulates the islets. In a diabetic (excepting MODY2) glucagon RISES instead of falling with glucose-stimulated islets. This generates POSITIVE feedback – the inverse of negative feedback, which is intrinsically unstable and oscillatory. It cannot be predicted in phase (i.e. lag or delay), magnitude, etc. It is unstable by its very nature.
Meanwhile, the portal and islet blood concentrations of hormones (e.g. insulin and glucagon) are much higher, especially prandially, than peripheral blood concentrations. Injected insulin does not have any measureable “central” effect (on the liver), and only works by increasing the rate of absorption of peripheral blood glucose into peripheral insulin-sensitive large-mass tissues (e.g. muscle and adipose). A diabetic who wishes to optimize BG control using exogenous/peripheral insulin must eat only meals DOMINATED by dietary protein. Portal amino acids must dominate the stimulation of the islets rather than portal glucose, during the meal absorption interval.
These principles were developed as hypotheses, demonstrated by experiment, and further underpinned by the successful disproof of ensuing challenges (to the hypotheses) by the mid 1970s. Continuing research, to this day, has all been consistent and only expanded the understanding of the islets-liver endocrinology and metabolism.
The work of labs such as Alan Cherrington’s (Vanderbilt Med.), Roger Unger’s (UTSW Med.), Gerald Shulman’s (Yale Med.) and many others continue to perform the related basic research.
For example, it appears that the first known event (likely causal) in the progression of T2DM is the development of insulin resistance of the alpha cell, thus disrupting the primary regulatory role of islet insulin in (counter)regulation of glucagon secretion. This then elevates the secondary regulator of portal blood glucose concentration to the primary role. In prediabetes (aka metabolic syndrome – i.e. earliest T2DM progression) there is already evidence of alpha-cell IR, and of hyperglucagonemia. The subcutaneous adipose is already saturated, disabling the ability of the normal adipose tissues to perform de novo lipogenesis and to store both dietary carb’s and dietary triglyceride as body fat through the normal pathways.
Hence, both basal (non-prandial) and bolus (i.e. prandial) regulation of BG is disrupted and crippled, and significant hyperglycemia is present in prediabetes. The only remaining capability preventing frank (clinical) onset is the preservation of insulin granulation (i.e. storage) within the beta cell population. Once this is lost, when overall beta-cell population capacity/function (to secrete insulin in response to a glucose stimulus) falls to 20% (of normal), regardless of the type of diabetes, acute (severe) insulin dysregulation occurs and diagnosis quickly follows.
To summarize, any diabetic who thinks he can compensate reliably for a significant carbohydrate content in a meal is fooling himself.

Those who managed to read and finish my first reply post above might be wondering … OK, so what happens with a meal dominated by dietary protein? If so, good question.
The ability of peripheral/exogenous insulin to reliably and accurately compensate for an adequately low-carb (how few carb’s is adequately low depends upon the individual diabetic, but less than 20g daily is a good guideline for most – my diet contains closer to zero) is limited. Amino acids are the other stimulant of both insulin secretion and of glucagon secretion by the islets (i.e. beta cells and alpha cells, respectively).
Any T1, advanced T2, or HNF-MODY (such as myself) diabetic who thinks he need not use bolus insulin injections to compensate for the hyperglycemia induced by dietary protein is not controlling his BG in any serious manner and probably has not been able to calibrate his hyperglycemic prandial response to dietary protein only because it is masked by that of excessive carb’s. Carb’s (i.e. portal glucose) are a more potent stimulus of insulin secretion than the amino acids – probably by a factor of ~4 on a gram for gram basis, using the typical ratios of amino acids present in animal protein.
It is an interesting, fortunate, and originally (back in the 1970s) well-observed characteristic of early or less advanced T2DM that the normal (i.e. non-diabetic) response to dietary protein is still intact. Nuttall and Gannon (in more recent decades) have carried out studies underpinning this already known fact. To my knowledge the cellular mechanisms are not well understood. This phenomenon is no longer well studied within the basic research area, to my knowledge, and this is unfortunate.
In the non-diabetic (and evidently also in the early T2D), dietary protein (and certain amino acids more than others, as individual contributors) stimulates an acute elevation of both insulin and glucagon secretion by their respective secretory cells. But the other key factor is the powerful counterregulatory role of insulin upon the alpha cells within the islets.
This results in a balanced effect upon the liver, with offsetting amounts of portal insulin and portal glucagon preventing both a rise in peripheral BG (via glucagon-stimulated glycogenolysis and hepatic artery) and a fall in peripheral BG (via insulin-stimulated hepatic and peripheral absorption of BG). This results in a sharp rise in portal and peripheral blood insulin to drive the amino acids quickly into tissues such as liver and muscle, which is vital – indeed, IMO this is the primary role of insulin as a hormone – amino acids have no long-term storage depot within the body, and must be absorbed into the tissues quickly. The role of insulin in enhancing fuel storage (in adipose) is secondary IMO. Insulin probably also plays a special role in neural tissues (including brain).
In a diabetic the critical counterregulatory role of islet insulin is severely reduced, resulting in a portal-hormones response to dietary protein which is hypoinsulinemic and hyperglucagonemic. This upsets the balance in favor of excessive hepatic glucose production (HGP), raising BG (unless the liver is in more of a fasted state, or at least a semi-fasted state, which can only be achieved temporarily with prolonged fasting or, to the lesser extent but with longer duration, employing a “ketogenic” diet).
This unbalance nevertheless can be pretty well calibrated with a consistent diet made up of well-known amounts of the amino acids and insignificant quantities of carb’s. Hence, a consistent bolus dose of peripheral insulin coupled with consistent meals (both in size and content) can result in fairly good BG control. There is no positive feedback with respect to amino-acid stimulated insulin secretion (AASIS) as there is with glucose-stimulated insulin secretion (GSIS) in a diabetic.
This is key to understand for almost all diabetics. Many famous diabetologists do not understand these basics AT ALL – in fact, I am unaware of any who do. Many explain insulin dosing and the role of glucagon and insulin in response to meals COMPLETELY incorrectly. Those who state that this has something to do with gluconeogenesis are WRONG, and completely ignorant of the basic research. Gluconeogenetic rates do not measurably respond to any meal, nor within the time frame of hours – these respond over periods of weeks or months to changes in the environment, such as seasonal changes in the food supply, in mammals. In the modern hominid, whose evolutionary diet for the past 2.5 million years has been essentially that of a supercarnivore (i.e. meats rich in fat from large game), even this medium-term adaptive response is quite weak because it has not been necessary shortly after our ancestors came down from the trees and grew huge brains, lost the digestive system required for significant reliance upon plant foods, and became “human” (i.e. homo).
Those who state that glucagon plays a role secondary to insulin in controlling liver metabolism and BG regulation are similarly wrong and ignorant. It is the opposite – glucagon is the primary portal hormone and insulin is secondary in its effect upon liver. However, insullin “appears” to be dominant under certain conditions only because of its primary role in regulating the amount of glucagon secreted into the portal vein. Most IR refers to as “hepatic” is NOT – it is within the islets where the effect takes place, and likely where the IR actually is (within the alpha cells themselves). The liver is generally responding normally to the primary portal hormones of glucagon and insulin. Excessive serum fatty-acid levels, fatty liver, fatty islets, fatty muscle, and many other consequences of the prediabetic state are clearly well downstream of, and secondary to, the primary/upstream factors taking place in the islets and adipose tissues in T2DM and its precursor, metabolic syndrome.
And T1Ds and monogenic diabetics also can have metabolic syndrome, exacerbating and confounding their therapy and BG regulation. Many do.
Hopefully this explains, albeit briefly and simplistically, the role of dietary protein in the hyperglycemia of the diabetic.
One final practical tip – use ONLY human insulin (i.e. insulin Regular, such as Novolin-R or Humilin-R) as a bolus for a meal dominated by dietary protein. These insulins have the best time profile for absorption of protein, which is steady and slow and predictable. Do NOT use ANY fast-acting insulin analog – these are a failed attempt to compensate for dietary carb’s, and totally inappropriate for dietary protein. They will result in initial hypoglycemia followed by later hyperglycemia during the prandial/absorptive period, at best.

If there is any soul who managed to read and finish BOTH of my posts above – congrat’s.
I wrote earlier that I would comment upon EXERCISE. I will do so in this post.
A diabetic can manipulate his hepatic glycogen levels using SIGNIFICANT amounts of aerobic exercise. I do. This is not likely to be usable for everyone, and will require excellent fitness as a starting point. But I believe it is a significant health-enhancement/maintenance tool.
It is important to understand the regulatory role of hepatic glycogen and the fed/fasted/semi-fasted states first. The research behind this knowledge goes back to the 1960s, even before that of the islet hormones. In modern society, and especially in industrialized countries, almost all of us are walking around with fully repleted liver glycogen (let’s call this 100% HGS – hepatic glycogen stores) at all times. This means there is no further buffering capacity in-liver to store incoming portal glucose from dietary carb’s, for one thing.
If we engage in a five-to-seven day fast (water only – no caloric food) and we do not have metabolic syndrome (in which case it may take several weeks to achieve the fasted state) our HGS will drop to ~25%. By this time the liver will have switched over to fuel its own operations completely using beta oxidation (i.e. fatty acids), and throwing off large amounts of the by-product (unique to liver metabolism) ketones to primarily fuel the brain.
We cannot go without nutrition indefinitely (without death due to starvation), although achieving the fully fasted state once or twice a year is enormously beneficial. This would have occurred naturally even more often, because there is no need or pressure to eat every day in our normal evolutionary state (which we mostly are not in at all now – constant hunger is an abnormal/unhealthy status due to a non-evolutionary diet and lack of adaptation to this modern diet). The cells are in the fully “catabolic” (rather than “anabolic”) state when we truly fast for at least five days continuously, and only in this state do they execute certain repair, maintenance and cleanup processes to the degree that is normal from an evolutionary point of view. But since the opportunity to absorb key nutrients was only intermittent during the evolution of the modern hominid (homo erectus and homo sapien) we and other mammals have evolved such that anabolic stimuli (i.e. primarily meals) always override. Now that we eat almost all the time, day and night, our cells have hardly any time to undergo maintenance and repair. All of the modern chronic tissue-specific conditions are the consequence. The constant overload of excessive anabolic stimulus is a significant secondary insult IMO, but the highly evolved and potent capacity for repair of the cells, and its diminution/prevention by too-frequent anabolic stimuli, is primary.
So-called “intermittent” fasting, and a “ketogenic” diet, and so forth, are forms of dietary practice in which ALL key nutrients are continuously supplied. The fully fasted state will never be achieved using these tools alone. However, to varying degrees in different individuals, hepatic glycogen stores BETWEEN 100% and 25% will be achieved. This will induce a moderate level of BK (blood ketones), and a moderately increased abililty of the cells to repair themselves.
I run, typically once per week or so, from 8 to 20 miles in the woods. I do it in the winter with tungsten-carbide studded, cleated shoes. So I do it year-round. If I run 12 miles or more I will typically be in the fully-fasted state by the time I get back home. That is because a duration of several hours of heavy aerobic exertion will deplete muscle and liver glycogen, through a combination of fermentative glycolytic and respirative ATP production, down to that of the fully fasted state. Weightlifting is good for health, but won’t have any measureable effect in this regard – it simply does not significantly deplete whole-body glycogen stores. And hepatic glycogen stores are the primary signal, directly (in liver) or indirectly (for peripheral tissues) regulating the catabolic/anabolic balance of cellular activity.
Any physical activity requiring heavy breathing (eventually) for a few hours will have the same result. For me, as a diabetic, it often takes two days to get back to my normal partially repleted levels of hepatic glycogen, while continuing to eat normally. For a non-diabetic on a ketogenic diet it will be quicker, because I ALWAYS am in a state of hyperglycagonemia which will slow the repletion. For anyone eating a diet not extremely low in carb’s this practice of hours-long aerobic exercise will be a waste of time, and hepatic glycogen will be immediately restored with the first meal – carb’s from the meal will be absorbed first by the liver until it has restored 100% HGS.
Gluconeogenesis is a very expensive and demanding (in terms of gluconeogenic substrate and of energy and of time) process. It only occurs at a certain rate. Without dietary carb’s the HGSs can be repleted only via this process. That is why it takes quite a while.
While I wrote that a medium-term fast to induce the fasted state requires at least five days for a heathy individual (more for those with metabolic syndrome), most will require the first three days just to first achieve the fasted state. Then at least two days in this state are required to achieve the benefits. These benefits not only include the repair taking place before the fast is ended, but also an exceptional acute cellular regeneration and renewal phase that is primed (epigenetically, by extracellular signaling molecules including ketones and by the intracellular sensing of lack of incoming nutrients available in blood) by the fast but takes place only in the first few days of refeeding.
I do sometimes go for long runs when I have already been fasting for many days. I can achieve BHB blood levels of ~10mmol/L, and BG levels of ~30mg/dL, when I do so. A non-diabetic would probably achieve somewhat lower BHB but equivalent BG levels under the same conditions. During a prolonged fast I will discontinue insulin injections, and so the low BG is normally physiological – not iatrogenic.
The kind of physical exercise regimen described is what is required to accomplish a signficant effect in terms of BG and BK. Weightlifting (aka resistance training), HITT, heavy aerobic (i.e. cardio) and light aerobic exercise is all beneficial to health in many ways. They will burn a small amount of carb’s, but this is not a significant benefit or tool for any diabetic. These forms of exercise cannot be used to compensate in any significant way to compensate for dietary carb’s. Nor can the hours-long (running) regimen, for that matter, unless someone does this EVERY day. Even then, postprandial BG transients would lead eventually to the complications of diabetes. If I am fully fasted and my BG is starts at 50 mg/dL and I eat a bowl of beans my BG will rise to 170mg/dL over the course of many hours, and then slowly return back down to between 50 and 60 mg/dL over the course of many more hours. I have done this and watched. The starches in the beans will generate a period of many hours of (deficient) GSIS and consequent hyperglucagonemia because they are slow to be digested and absorbed – I defy any diabetic to try to compensate reliably and accurately for this response with bolus insulin injection(s) and/or exercise.
So these are some thoughts about exercise. It has NO PLACE in an insulin regimen IMO. The insulin regimen must be composed of careful diet/meal timing and composition and careful timing and sizing and selection of insulin bolus/type (# units and time profile).

If my insulin strength varied +/- 50%, I would have to redo carb ratios at least every 3 months (3 month Rxs). I can at least say insulin strength has been consistent. That’s not to say that it can’t be a problem in some supply chains.

I do think people should focus on taking less insulin. Yes insulin is very difficult to measure.

This whole discussion just reaffirms what most long-time diabetics already know: controlling diabetes is doing the best we can with a continuously changing set of variables (i.e. diet, exercise, hormones, emotions, medications, individual physiologies, etc.) My husband, who is a detail person, was expressing frustration that my readings from my PDM (I use an OmniPod) did not match my readings from my new G6 CGM. I told him that he was seeking precise target practice with an antique rifle filled with buckshot. He wants precision when the reality is that the buckshot will hit – or miss – parts of the target goal. We all GUESS at the precise amount of carbs in any given meal. Our bodies all react differently to the same amount of exercise. We all react differently to a particular dose of insulin. Emotions and hormones affect some while those same two components have very little effect on others. Bottom line: accept that if you are new to this, you have lots of trial and error as you learn your own body’s reactions. If you are an “old pro” like some of us, rely on what you know works, analyze the situation when something doesn’t work, and keep on learning about yourself as you age. Mathematically trying to figure out diabetes isn’t wrong, but I personally think that it will not give anyone conclusive answers as far as better control is concerned. Use your knowledge to do the best you can, and please remember to live a fulfilling life.