Relationship between A1c and number of lows

I was wondering if anyone has come across research that looks at the A1c and the number of lows experienced. The question being, if I go up by an A1c value of X, I can reduce the number or severity of lows experienced?

I actually think the research shows the opposite: Those with higher A1Cs are also more prone to lows. They’ve just got poorer control overall.

But I think* what you’re asking about is basically the math – like clearly there’s some (approximate) differential equation that could roughly model blood sugar and the amplitude of oscillations in either direction could be damped or enhanced by varying those parameters. My guess is that for a given kid, if you increase the target range by X you can then reduce how much time you spend below Y pretty reliably. But I don’t know the ratio or the calculation.

I just think this method doesn’t work so well once you get higher into the BG range because insulin sensitivity is nonlinear across BG range. So if you get into the highs, you need more insulin to bring it down, but there’s more uncertainty about how any extra insulin will affect BG once it’s closer in range.

That is what I am looking for. My thoughts are that when you get to a lower A1c, there might be a tradeoff in not being super aggressive with your corrections. i.e. many of the adults with great control on the board are throwing insulin at numbers and trends I would normally ignore. If these aren’t increasing the number of lows, then it is a good strategy and I should be trying to reinforce this behavior with my son.

If instead, aiming for a slightly higher range, you can demonstrably reduce the number of lows, but still have an acceptable A1c then for many this would be a better approach. I was just wondering if someone had seen something on this. I am willing to ignore the data for people that mostly live above the physiologic normal range, since that doesn’t apply to us.

I hadn’t seen any but found a paper that seems relevant to your question: Glycated hemoglobin A1c as a risk factor for severe hypoglycemia in pediatric type 1 diabetes

It might be interesting for you to look up the original paper to delve into all the nitty-gritty details, but from the abstract I get the following:

There is indeed an increased risk of serious hypoglycemia, measured by relative risk greater than one for every 1% decrease in A1c. But the relative risk has come down significantly in recent years, from 1.22 in 1995-2003, to only 1.06 in 2004-2012 - presumably due to increased use of insulin pumps, short- and long-acting insulin analogs, and glucose monitoring increased (CGM or otherwise).

By the same token, increasing one’s A1c by 1% would seem to (on average) only confer a decreased risk of serious hypoglycemia by 94%. Unless serious hypoglycemia has an issue for you, then it seems that this relatively small risk would argue for decreasing A1c to the extent possible and practical. But I would add my personal belief that quality of life is of paramount importance, so any increase or decrease of A1c needs to be judged by the impact it has on quality of life.


I think that, while you can theoretically predict this or empirically derive this for a population using a lot of data, you should just do an experiment: Raise your son’s target by 5 points for two weeks, then lower it by 5 points for two weeks, and see if you get a) lower average BG b) more time spent low/high.

What we have found is that if we lower our target too much, we wind up correcting lows so often that we are still hovering at about the same average BG and yet fighting or experiencing more lows AND highs. This probably has to do with the fact that insulin sensitvity is higher at lower BGs, so once you are aiming at a low enough target, all of a sudden maybe half your target range requires one ISF and the other half requires another, and you can’t really combine them very well.

So just above that point is the target BG we aim for.

I have long thought that you could really simplify this process if you could analyze a person’s settings – some ratio of ISF, carbF, Basal rate and TDD is probably enough to tell you how tight a level of control you can aim for, because on some level they probably do literally plug into an ODE as different parameters and collectively can tell you how damped the BG system is or how good the level of control can theoretically be given certain “forcing functions” (a.k.a. meals). For instance, people are always saying that toddlers have more erratic BGs – but this just seems to be a function of the parameters you enter (some ratio of carbF:ISF) which is why as they age out of those parameters their BGs stabilize.

Figuring this out would require me breaking open a 15-year-old ODE textbook, digging through the research papers to see if someone has ever attempted this, and then just guessing on my own to derive this theoretically. Too much work and I have no confidence I could do it correctly.

Another thing you can do, which probably is a little easier, is try to find a good probability distribution function that represents BG readings (it would probably be a skewed normal distribution), and then look at your standard deviation or calculate a different version that accounts for the skew. From there you can figure out P(Y|X) – What fraction of the distribution lies below y if the mean is X, for instance.

It gets trickier when it comes to things like correcting at 120 or even 100, etc… I’m not sure how you could figure out the optimal threshold for issuing corrections. That’s so context dependent that I doubt you could come up with hard and fast rules that your son could easily follow.

1 Like

i am wondering, now that i have been swimming, how going high with my bgs, in order to get into he pool at a “safe” number, will effect my overall A1c . i am only high for a couple of hours before i swim, and then for the remainder of the day, i am mostly in target range. my last 2 A1cs were 5.8 and 6.2 . i know those are great #s but these were taken before i began swimming. and during periods that i was having a number of lows.

does anyone out there experience a change in A1c % due to trying to stay a little bit higher b/c of exercise, or does the exercise compensate in some way or another ? since i have been swimming, btw, i have not had as many lows as i used to have. (i know that this will not apply with eric, but i am certain that eric will have some intelligent answer to this question :wink: .)

Your A1C is probably a bit higher because of what you have to do for swimming.

I could probably be in the low 4’s if I did nothing but sit on the couch all day and watch my BG and not eat any carbs. But for me, that would be a crappy life. I’d rather run like an animal and eat like one too. That’s my preference.

There are soooo many benefits to exercise. Read this, please:

Even if your A1C is a bit higher because of what you have to do for exercise, you are getting so many more benefits from it. Mentally and physically.

And given enough practice, your BG’s for exercise will be even better. We still have some work to do with it! :wink: And a few more tricks to work into the practice.

So a few thoughts on the A1C…

First of all, what is the A1C? When you have high BG, the glucose molecules attach to the hemoglobin in your red blood cells (“red blood cells” which I will be referring to as “RBC” in this post). The A1C is a measure of how much glucose has attached itself to your red blood cells (RBCs).

  1. Longer and higher periods of high blood sugar cause more glucose to bind to the hemoglobin in the RBCs, and thus results in higher levels of glycated hemoglobin, and a higher A1C test. So longer highs and higher highs will raise your A1C more than the shorter and lesser high BGs.

  2. Frequent or severe lows can lower your A1C a bit. A low BG will cause your body to strip away the glucose that has bound to the RBCs in an attempt to use it. So that will lower your A1C result.

  3. Your RBCs have a limited life span. Some of your RBCs from 2 months ago will be dead, which means the BG you had 2 months ago will not be reflected as much as the more recent ones. Your more recent BG levels will be weighted more in the test result than the ones from several months ago (because there are more of them alive from a few days ago than from months ago). So recent BGs will always weight the result more.

A simpler way of saying it, your BG in the few weeks right before your A1C test will affect it more than the what you did 2 months ago.

  1. The lowest A1Cs I have had were always at the end of intense training periods. The reason for that was because my RBCs were not living as long (there is a test that tells you how old your RBCs are). The training stress shortened the life span of my RBCs, so my A1C tests were not showing BG over 3 months, but possibly only a few weeks. There is a limit to how long you can subject your body to that type of stress. Eventually you blow up.

Bottom line, keep doing what you are doing, and keep getting better. Don’t worry about a few 1/10ths of a point on your A1C.

A lot of stuff here, DaisyMae. Let me know if I need to explain any of this better!

Maybe the mid 5’s. And it’s not such a crappy life :grin:

1 Like

Really? I’ve never heard this - clearly when your BG is hypoglycemic you won’t be adding as much glucose as you would at a higher (normal) BG, but that your body can actually strip away glucose in an attempt to use it? Do you know where you read that?

I have heard it from a number of different sources, including my Endo, and a researcher at a company working on artificial pancreas therapy. I don’t have any online resources to share but I can look.

I’m also curious—I was under the impression that once glycosylated, hemoglobin stays that way.

Very interested in reading something on this, because I have heard both things from Endo’s and am interested.

My opinion is when a CGM is involved, the A1c is overused.

An accurate CGM delivers far greater and better information than an A1c test every 3 months.

gotta disagree somewhat. No long-term, well-designed studies have tied Dexcom CGM data with long-term complications. The A1C is basically the only measure that has been robustly tied to complication risk.

1 Like

The A1c has been around for a long time so there would naturally be many studies on it.

If you want a single number then certainly the A1c is your best bet.

It takes more work to use the data generated from the CGM. However the CGM clearly has so much more data that the A1c obscures into a single data point for 3 months.

But complication risk hasn’t been proven to be correlated to blood sugar variability (which is all he cgm really shows the user) to any meaningful extent… it has certainly been proven time and time again to be associated with a1c… so I think it’s not entirely on point to say that cgm data is really more valuable than a1c measurement… the cgm is a management tool, the a1c is the real measure of how effective the overall management is…

So if I had to chose to have an a1c that was on target even with variability, or minimal variation as indicated by a cgm but an elevated a1c, I’d take the a1c that was at goal level every time

I totally agree that the CGM data is more useful day-to-day, but I would argue it’s useful specifically because it gives people the information needed to dramatically lower A1C.
In other words, in a situation where CGM data and A1C consistently disagree, I would use the A1C as a better predictor of risk, at least until the necessary studies are done to tie CGM trends to complications and health.

1 Like

I get a pretty accurate A1c from my CGM. Three months ago xDrip+ said it would be 5.7, it was actually 5.7. The other day at my endo appointment it estimated 5.6, and by golly it was 5.6.

I’ll bet many people used to get similar results when Clarity still offered the A1c estimate.

In addition a CGM allows a person to more easily treat highs and avoid lows. My son’s endo, just asked him to teach part of a peer to peer class, because his (our) techniques allow you to identify and treat highs, and once you figure out the puzzle avoid lows.

At our last appointment they said he shouldn’t lower his A1c at all since he would be having more lows, but he managed to lower his A1c and reduce the percentage of time he was low at the same time. I know we are blessed to have this technology available to help learn how to manage this.


yes absolutely. IF the question was: you have to either ditch your CGM or ditch your 3-monthly endo appt, complete with A1C measure, I would ditch the latter in a heartbeat. The information you get is so valuable for basically getting a step ahead of your diabetes and the feeling of day-to-day life is just radically different. But that isn’t the same as saying the A1C and the CGM data are truly synonymous.