Is a lower HbA1c always better?



Please don’t misunderstand me.

I’m not saying that anyone with a low A1c must have these conditions.

However, if someone has a lower A1c than you’d expect, then I think ensuring they don’t is a good idea.


Yes, let me try to clarify that a bit.

Specifically, a low A1C in a diabetic can be from aggressive blood sugar control.

But in a non-diabetic, it is obviously not from aggressive blood sugar control!! It can be caused from other things. In a non-diabetic, it could just be genetics. But it could also be from problems, such as anemia or liver problems, like in the list @Katers87 gave.

So in the diabetic, a low A1C might be the result of a certain type of control or management, whereas in the non-D that low A1C might be from problems.

And the same for the example @MM2 gave! That low (4.6) A1C was from something not related to blood sugar.

So for those reasons, I think A1C’s for D’s versus non-D’s are not necessarily the same thing.

In reading through these comments, it seems like we are all on the same page here, are we not?




I’m coming to this discussion a little late, but found a recent paper that is clearly relevant, and that supports oni’s contention that mortality is increased for both low and high A1c. I can’t vouch for whether this paper is better than any others. And parenthetically, I’m not going to change my behavior based on what it seems to suggest - I work very hard to keep my A1c where it is, and based on this one study I am not interested in raising it back into the high-6 to 7 range because frankly I feel better doing what I’m doing now.

But it does seem relevant to the question being considered, as it is a paper looking at all cause mortality in a group of 2764 T1 diabetics, segregated by A1c. It raises the point for serious consideration, that maybe it isn’t always best for a T1 (or T2 - see references) to push their A1c to the lowest possible value. Here are a couple relevant quotes (note the final caveat - a sign of a good researcher in my opinion):

“Results from our study in type 1 diabetes patients from the EURODIAB PCS point toward a nonlinear relationship between HbA1c and all-cause mortality risk. HbA1c levels below and above our selected reference HbA1c (median value) were associated with increased all-cause mortality risk after adjustment for age and sex. This U-shaped relationship also remained after additional adjustments for diabetes duration, physical activity, BMI, systolic BP and total to HDL cholesterol ratio.”
“Type 1 diabetes patients generally have a longer glycemic exposure than those with type 2 diabetes because of its onset at a younger age and are therefore at increased risk for micro- and macrovascular complications. This risk factor profile could indicate that when considering optimum target HbA1c for type 1 diabetes patients, the lower the better might not be appropriate.”
“Our findings confirm the reported increased risk of death associated with low HbA1c shown in observational studies among type 2 diabetes patients.”
" The potential mechanism underlying these findings remains unclear. Lower HbA1c targets are mechanically associated with higher risk of hypoglycemia, a common complication in intensive glucose treatment. … In our study, severe hypoglycemia events more often occurred in the lower compared with the higher HbA1c quintiles ( P for trend < 0001);"
“We recognize that HbA1c levels below 5.6% in individuals with type 1 diabetes may be related to anemia, renal insufficiency, infection, or other factors not available in our database. Findings on this relationship from observational and intervention studies among type 1 and 2 diabetes patients are inconsistent, and our data will require confirmation by other groups. If confirmed, the potential mechanisms underlying this increased mortality risk among those with low HbA1c will need further study.”

Association between HbA1c and all-cause mortality risk in type 1 diabetes patients in the EURODIAB PCS. Restricted cubic spline regression with five knots is located at the fifth, 25th, 50th, 75th, and 95th percentiles (HbA1c 5.6%, 7.1%, 8.1%, 9.5%, and 11.8%). The y-axis represents adjusted HRs for all-cause mortality for any value of HbA1c compared with a reference HbA1c of 8.1%. HRs are adjusted for age and sex ( P for overall effect of HbA1c = .04; P for nonlinearity = .11). Dashed lines indicate 95% CIs.



Yes, my initial assumption would be that the people with the lowest A1cs had another condition. I’ll have to look more thoroughly later though.
The study did not control for anemia or renal insufficiency, and it occurred in 2014 before cgms were as commonplace as they are now.


I am not sure why we are surprised that there is such a thing as a too low. A1c. Just like most things medical, there is a normal range and most people should fall within the range. When they don’t, you and your physician should be ruling out the “bad” things and deciding what your optimal target is. Just like there are outliers in non-diabetics, there will be outliers in diabetics, although the risks may be different.

We should all be striving for individualized medicine, i.e. guided by the norms, but not beholden to them. If I remember correctly, @docslotnick’s physician asked him to raise his A1c due to individualized concerns. That should be the gold standard we are all asking for in our medical care.


I think some of these studies do not really change anything in what we are doing. I honestly do not know how someone targets a specific A1C, as opposed to a BG target range.

For example, let’s take two different cases. One person, let’s call him “Test Subject A”, and the second person, let’s call him “Eric”.

Both of them treat a low BG. Neither one would want to leave a 60 alone! So in that way they are the same! In both instance, they try to correct low BG’s.

However, where they differ is that maybe if they are both at 130, Test Subject A would leave it alone, and Eric would try to correct it down to 90.

So the end result of that different approach might be an A1C of 6.0 for Test Subject A, and an A1C of 5.0 for Eric.

So what I would like to see is a study that shows that correcting a BG of 130 is less safe than leaving it alone. I think that is where we fundamentally disagree. There is no way I would leave a 130 alone, and the end result is that I have a lower A1C than someone who is not as aggressive.

But how is that damaging? I am not targeting a BG of 50 or 60.

And also, my lower target (90 versus 130) in no way makes me more likely to have hypos. I would have the same chance of a hypo if I were to correct a 200 as if I were to correct a 130. But I have been doing this too long, I know how to apply the brakes to a dropping BG, no matter where it is coming from.

So the bottom line for all of this for me, is I would need to see how a lower (but still safe) BG target like 90 is more damaging than a higher target, such as 120 or 130.

And I do not think anyone has ever demonstrated that.


There is absolutely no reason to make that assumption. Find me any other study that controls for those things. The authors were simply being thorough in trying to find other possible reasons for the results they observed. I’ve had T1 for 43 years - to me 2014 is really not very long ago.

Take the contributors on this forum - would you say that all of us here with T1 and A1c’s less than 6 must have anemia or renal insufficiency? Clearly that isn’t the case.

Jeez, everyone was saying “oh but if we only had a study of T1 diabetics, but since we don’t we’ll ignore the results”. Well here is a study of T1 diabetics, so why not consider that there may be some validity in the results? Don’t change your behavior - just open your mind to the possibility.



Actually quite a few of the other studies I looked at did control for these things. Renal insufficiency in T1s is not unexpected and it does lower A1c considerably.

I’m sorry I can’t look at the study more right now. I’m about to have company over. I will look it over though by the end of the day tomorrow.

Also, I have an A1c of 5.1%, but my predicted A1c is around 5.8%. This thread is particularly relevant to me because I want to understand why these don’t align to ensure I don’t have another condition that is creating a lower A1c and could be affecting my long term health.


I think Katie is raising a valid point, because one of Oni’s posts was a study on non-diabetics!

I don’t think she is saying that the only ones with low A1C’s are diabetics with those conditions.

I do believe she is saying that a study that uses low A1C’s for non-diabetics is not very relevant to us!

I don’t want to put words in her mouth, so @Katers87, please let me know if I am wrong in my assumption.


I can’t speak to what Oni posted, since I joined this topic only recently. This is a different study, which I did not see discussed above, done on a group of 2764 T1 diabetics. I contributed here because I think this is a topic that is worthy of consideration - is it REALLY the case that a lower HbA1c is always better? As Chris alluded to above, I think the answer is not always yes, regardless of whether or you accept the results from the study I cited or not.


I would not say always to anything diabetes related.

But I think there are so many possibilities when they do all-cause mortality.

How about the question I posed earlier. Has 130 been proven to be a safer BG target than 90? In any study ever? Has there ever been a definitive answer for why a 130 would be healthier than a 90?

Another thing to consider - as a problem with all-cause mortality and cohort studies - maybe the diabetics with the lower A1C’s were more likely to die from a hypo reaction. It isn’t the low A1C that is killing them, it is the hypo. That certainly seems like a reasonable possibility.

But nobody is targeting a hypo! It is just a bad result that might happen for some who are targeting more aggressive control.

So again, the rule would just be, try to target a safe BG number, and try to reduce both lows and highs.

If you are not having bad lows, why would you need to raise your BG target, just to meet some arbitrary A1C number?

I don’t think it needs to be any more controversial than that.


I second that.


It took me a while to go through this thread, because of the sheer number of references, many of which had significant and complex analysis. I read every single article listed here, and I have them open in different tabs as I am writing this post.

The data presented so far
Many of the studies listed are not applicable to us, since we only look at how diabetic control impacts the odds of complication and early death. In order of appearance:

  • @oni’s original article which started this thread analyzes A1c levels for non-diabetics only, and bunches all diabetics together. It is not directly relevant to our concerns, yet it is useful (see below).

  • @TiaG has several fascinating references. The most interesting, to me, is a meta-study that compares CVD and All-cause mortality across both diabetic and non-diabetic populations, analyzed against A1c. The conclusion for all-mortality was a J curve in both cases (elevated at both ends). The non-diabetic population showed elevated mortality below 5%, and optimal between 5 and 6%. The diabetic population showed the same curve, but with an elevated mortality below 6%, optimal between 6 and 8%. @TiaG also references several articles for glucose normals that display the same J curves as the original article referenced by @oni.

  • However, another of @TiaG’s articles (EPIC-UK) showed a curve going continuously up for CVD starting at 5%, indicating 5% as the optimal A1c for diabetics and non-diabetics alike.

  • On a side note, @TiaG referenced an outstanding, brand new article (Oct 2018) out of DCCT that shows the advantage of 30 years good glycemic control (A1c , 7.2%). The results are amazingly good:

Thirty years of excellent vs. poor glycemic control substantially reduced the incidence of retinopathy requiring laser therapy (5% vs. 45%), end-stage renal disease (0% vs. 5%), clinical neuropathy (15% vs. 50%), myocardial infarction (3% vs. 5%), stroke (0.4% vs. 2%), and death (6% vs. 20%).

  • Like @TiaG, @oni referenced several articles showing a J curve for mortality for general populations (not selected for diabetes), indicating increased mortality at 5% A1c and below. @oni also re-references the interesting study earlier quoted by @TiaG.

  • On a side note, meta-studies are not “the most rigorous, high quality study type that exists on the scientific hierarchy.” That place is occupied by tightly controlled interventional studies or trials, called randomized double blind placebo control studies:

  • @Katers87 quotes an interesting discussion of causes of hypoglycemia in glucose normals

  • Finally, @jag1 references what may be the most relevant of all studies, focused on T1s (EURODIAB), although unfortunately still a cohort study. The hazard ratio for this population as a function of A1c was again a J-shaped curve, whose lowest point was located between 6.5% and 8% depending upon the spline curve used. Mortality data was after 7 years. However, the study is very careful to qualify its result:

Results from our study in type 1 diabetes patients suggest that target HbA1c below a certain threshold may not be appropriate in this population. We recognize that these low HbA1c levels may be related to anemia, renal insufficiency, infection, or other factors not available in our database. If our data are confirmed, the potential mechanisms underlying this increased mortality risk among those with low HbA1c will need further study.

My interpretation

  1. It appears clear to all of us, I think, that glucose normals, when displaying an A1c below 5%, have an increased mortality. I think we all agree that this increased mortality is due to outside (medical) reasons, which cause the lower A1c, and the additional mortality: the lower A1c is not what causes the additional mortality.

  2. These very same reasons also exist for PWDs in the very same way as they exist for glucose normals. Therefore, we should expect to see a J-curve for PWDs at low A1c, unless the decrease in mortality due to the effect of lower average glucose counteracts this increased mortality.

  3. But I think we also all agree that the beneficial effect of going down from 5.5% to 5% is minimal at best (almost all curves I have seen show increased all-causes risk kickstarting at 5.5%). Therefore we should not expect a significant decrease in mortality among PWDs due to more normal glucose levels when we go from 5.5% to 5% or below.

  4. In turn, this means that we should see that same increase in mortality in the low part of the curve for PWDs that we see in glucose normals, and for the same reason.

  5. However, when affected by medical problems which cause an A1c below 5% for glucose normals, regular PWDs should see the effect of those medical problems superimposed to their baseline A1c, which is, in general, higher than glucose normals. Therefore, we should see, for PWDs, a J curve inflection when moving towards lower A1c at a higher level than 5%. For instance, if chemo causes a 1% drop of A1c, a PWD with a 7.5% A1c baseline will go down to 6.5% with chemo, not below 5%.

  6. Therefore, when looking at the effect of the regular mortality J curve for glucose normals upon PWDs, we should expect another J curve, but with a higher minimum than for glucose normals, probably around 6.5-7.5%. That is exactly what we are seeing.

My conclusion
The regular mortality J curve against A1c for glucose normals explains the J curve for PWDs, without the need for an additional hypothesis. Occam’s razor tells us that this is the explanation we should prefer.

So, until we see proof that a lower A1c actually causes (through true causality, not cohort studies) increased mortality (as opposed to a third party reason, such a medical conditions that increase mortality and lower A1c), I do not believe that it is necessary to assume anything else, such as complicated and unproven assumptions about the “dangerousness of a low A1c per se.”

At the same time, we all know that YDMV. We (my family) have always tried to aim for a glucose-normal average glucose level as a goal, whether we reach it or not for any given 3-month period. I certainly don’t see anything here that would make me change my mind. But others may differ :slight_smile:

Imho, however, there is nothing so far in the literature presented that logically requires any other explanation than that given by the regular J curve for glucose normals.


Thanks for writing out your interpretation @Michel. I think you’ve summed everything up very well.

I can’t really think of anything to add.


Thank you, I think that is an excellent summary. I would, however, add one thing to your conclusion that is worthy of consideration. It is certainly true that the J curve for T1 diabetics COULD be solely due to the same reasons that glucose normals have a J curve. But that is not necessarily the case, and I think it is worth investigating this further, as the study I included suggested.

I am absolutely a big proponent of Occam’s razor. But if we are going to use Occam’s razor, I think we need to use it carefully. Namely, we should not excise the fact that there ARE risks to diabetics from having severe low blood sugars. So it is not reasonable to just disregard the additional risk that those of us using insulin experience. Simply put, we ARE at greater risk than the population at large since we can have severe low blood sugars, and they can not. And studies have shown that the risk of severe low blood sugars is greater when we lower our A1c’s, though this is improving with the improved technology we have available these days.

It is not just cardiovascular risk (which is being considered in another discussion). The most obvious risk besides “dead in bed” is the risk of injury from falling or from heavy machinery (auto) accidents. There are many anecdotal reports of this that can be cited, and I would imagine that most long term T1’s who have had diabetes for many years could give their own examples of times when they were at increased danger and/or hurt themselves in some way because of having a significantly low BG.

Insulin is a dangerous drug. A tiny amount too much can be catastrophic. I’m afraid there is just no way around that. Please note that I am not saying that we should therefore all give up and accept average blood sugars in the 200’s. But I am saying that when we tighten control and push our average BG and A1c down, we need to use care, and we need to be aware of the risks. And at a certain point we should consider the risks/benefits of always pushing BG lower and lower.


So no one has done a study proving that targeting a 130 will result in fewer hypos than targeting a 90. But when you factor in a) math (aka if you’re dropping X/sec on a CGM and you start from a lower number, you’re almost guaranteed to hit a hypo range more frequently) b) experience (most people are not like you, and do not know how much carb/insulin to take to stop a low in its tracks, or are not able to be on top of it as quickly as you) c) CGM lag – unless you’re testing every half hour there’s a high chance that you’ll miss oncoming lows because CGM lags by 15 minutes d) anecdotal data from both endocrinologists and DOC (those who are more aggressive have more hypos) and e) insulin: it’s slow, sometimes unreliable, and works differently depending on what’s going on in your body…

It just stands to reason that an average person with diabetes, average in their knowledge, ability and disease course, will likely have more hypos if they target a lower BG number than they will if they target a 130. 130 provides a cushion for those who don’t know how to slam on the brakes hard enough, so to speak. That says nothing about individuals on this site, who I believe are often above average in their knowledge, experience and willingness to use the tools at hand to achieve a low A1C.

As an aside, and in all my time in the DOC I have yet to see someone who was not low-carb who didn’t have significant time spent below 70 mg/DL if they had a low A1C. With Samson’s diagnosis we were told to aim for less than 2% below 70 mg/DL, less than 0.5% below 55 mg/DL – those with A1Cs below 5.6 – do you think it would be realistic/feasible for you to reach those targets while maintaining your A1C? IF you were the parent of a child and that was your goal – is there any way you could achieve those hypoglycemia targets with an A1c in the 5 range?


My initial summary was this:

I also recognize the risks of severe hypoglycemia, and I agree that a higher A1c would be much better than frequent time spent in this range with a lower A1c.


I am not saying anyone should target any particular A1C number. I certainly would not advise parents to try to hit a particular number.

But my point is simply that while being hypo can be dangerous if it is not treated, or you are driving a car or sleeping, being 90 is NOT dangerous. If someone can hit 90 and not have excessive hypos, and they have an A1C in the 5’s, I think it is wrong for a doctor to suggest their A1C is too low.

A1C’s do not indicate lows! They indicate highs, or absence of highs.

Here is the comparison I have used previously. Saying a low A1C is from too many low BG’s is apples and oranges.

You give me 20 apples at the beginning of the week, and at the end of the week, I have 18 apples. So you yell at me, “You are eating too many oranges!!”

But you don’t know how many oranges I ate. You don’t know anything about oranges at all. You only know that I ate 2 apples.

A low A1C does not say one thing about low BG’s. It only tells you how much your RBC’s were glycated. From high’s!

So looking at a study of people from around the world - who are not me, and I have no idea about their level of control and experience - and saying, “because a low A1C is more dangerous for those people, it must be dangerous for you, Eric…”, that is flawed thinking.

I said in previous posts, don’t target an A1C. Target a safe BG. Target a BG that you can manage without excessive lows or highs. If it is 90, great. If it is 120, great. Target what is safe for you.

But don’t target a higher A1C because some other people could not obtain it safely. If your BG is in a good zone, trying to raise your A1C because of a study of others is foolish.

That is all I am trying to say. It doesn’t seem like that should be controversial at all.


Great point.

Prior to CGMS, it makes sense that the A1C was used to “in theory” associate a low A1C with conclusion there were (too) many lows.

And for many that don’t have CGMS, I think this continues. Even with meter downloads, they could be missing lows. When I FEEL really low, I’m likely to treat first without BG test. So may not even show in meter download.