• Guest, the forum is undergoing some upgrades and so the usual themes will be unavailable for a few days. In the meantime, you can use the forum like normal. We'd love to know what you think about the forum! Take the 2025 Survey »

How reliable are HbA1c tests?

L

Lupf

Well-Known Member
I am type 2 diabetes. Thanks to losing 10 kg my Hb1Ac reduced to 42mmol/mol on 28 Nov 2019.
I agreed with my doctor to come off metformin (2x 1g /day) in mid-Dec 2019.
It was a bit of a shock that my retest on 18 Feb 2020 came back with Hb1Ac at 55 mmol/mol.
My weight average over the 3 months before each test was the same
as were the blood sugar tests (prick finger and strips). I test once or twice a week in the morning before breakfast.
My three months averages were
Date weight BS Hb1Ac
[kg] [mmol/l] [mmol/mol]
28 Nov 2019 81.2 6.2 42
18 Feb 2020 81.1 6.1 55
From these data there should be no change, apart from the effect of coming of metformin.
If I look at my mos recent 10 blood sugar tests these were at 6.5 mmol/l,
which is ~0.5 mmol/l higher than the 3 month average.
From this I only expected an increase of +3 for my Hb1Ac,
which is much less than the measured rise of +13.

My questions:
1) What is the margin of uncertainty in Hb1Ac tests?
2) Is this increase normal when you come off metformin?
3) What is wrong in my argument that you should be able to predict Hb1Ac lever from your blood sugar tests?
This worked during last years' weight loss.
4) What should I do now?
My plan is to continue to fast twice a week when possible. I hope to reduce weight even more.
What is annoying is that my shoulder hurts and since Sept 2019, I have not been able to do judo anymore,
it also bothers me for cycling. I am doing physio, but that hasn;t helped so far.

PS
1) I note that many members have a signature at the bottom of their posts,
but I couldn't find where to add this info.
2) I gave more info on myself in another post "T2 reversed?" in Greetings and Introductions,
but it says I am not allowed to post links.
 
Last edited:
Well a hba1c tests the average of your blood sugar over a full 3 month period, a few tests a week aren't going to match a hba1c as levels fluctuate 24/7

What is your diet like?

Sorry as I'm Type 1 I cant give insight to stopping metformin
xx
 
Lupf said:
1) I note that many members have a signature at the bottom of their posts,
but I couldn't find where to add this info.
Click to expand...
On a pc, hover over your name at the top right and choose signature. On phone, tap on your name to find it. I don't know how to do it if you use the app.
Lupf said:
2) I gave more info on myself in another post "T2 reversed?" in Greetings and Introductions,
but it says I am not allowed to post links.
Click to expand...
You can only post links after you've made some more posts. Here's the link to your post on that thread: https://www.diabetes.co.uk/forum/threads/t2-reversed.170844/
 
HbA1c results are the standard way of diagnosing and monitoring blood sugar control in the UK. As far as I know they are pretty reliable if you have standard red blood cells. If you are anaemic or have other disorders of the red cells they may be inaccurate.
As for comparing your own finger pricks with your HbA1c it can never be totally accurate especially if you are only testing your fasting sugars. Doing this you will miss any spikes after meals. Metformin will have only a minimal effect on your levels, I dropped one tablet for three months once and my HbA1c only went up by one.

Here’s where to click for editing your signature. First on your screen name circled in blue, then on signature circled
 
Whilst weight loss is beneficial in many cases it’s not the be all and end all of control. Yes you can infer hb1ac from fingerprick tests but you simply are not doing anywhere near enough to do so. Your levels will go up and down all day, largely but not solely, dependant on what you’ve eaten.

The tests are not going to be out by 10 or more mmol. A couple maybe. You dropped the metformin which helps by reducing liver dumps (look up dawn phenomenon) and reduced exercise. Both of these may account for some of the change. Did your food intake change in anyway? Did you double down on “healthy” wholgrains or fruit for example?
 

Thanks a lot. I know that Hb1Ac tests are pretty reliable, but the scientist in my would like to have this quantified. Is there a doctor/researcher out there who can put a numerical uncertainty on Hb1Ac tests?
Is it +-1 or +-3mmol/mol or larger?
I know that finger pricks are not totally accurate. When I started measuring last year they were all over,
but since last summer they vary only by +-0.8 (standard deviation). Thus the average of 3 months should be meaningful.
It is also my understanding - that Metformin should not have a large impact, thanks for confirming.
 
Last edited:
Interestingly, my DSN wants to keep me on metformin and not even drop dose until I've lost more weight and attempted to get hba1c even lower. My results were 39 this time and I've lost 2 stone.

I'm ok with that as I know it has helped me. May not be the main reason, but with LC has had an impact I'm certain.
 
Hi @Lupf,

I've done some research into the topic and yes, HbA1c does not always reflect actual blood sugar levels for individuals. In fact, it can be off quite signficantly. Below is a graphic from the following article, which compares average glucose levels using CGM in mg/dl (divide by 18 to get mmol) and HbA1c: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5521971/



Based on this chart, for an HbA1c of 6.3%, for example, average blood glucose readings based on CGM can range from about 110 mg/dl to about 180 mg/dl.

Another set of researchers found that HbA1c is out by 7.2 mmol in 40% of individuals. Here's the link to this article: https://academic.oup.com/edrv/article/40/4/988/5487357).

However, although researchers have observed that between individuals HbA1c can vary quite significantly, they also state that within individuals HbA1cs tend to be quite stable. In other words, if your Hba1c is off, it tends to be off by a very similar amount by a similar amount for each HbA1c test.

So, in your case if 55 mmol was too high, the previous test of 42 mmol should have been too high too.

Why not use a CGM for two weeks to see what your blood sugars are doing throughout the day (and night)?
 
What caught my attention was that you are not testing after eating - for a type two that is the significant test time - as the amount of carbs in a meal will be what pushes your BG level up.
I keep my two hour tests to 7mmol/l - perhaps a point or two higher if I have a dessert after dinner, but that is usually only once a week. These days I rarely need to test, having settled what I can eat and when, but from this I know that my Hba1c will be around the top end of normal, or to be more precise, 42.
This is actually higher than predicted but my blood corpuscles might live longer than normal - it is a figure everyone seems happy with.
Weightloss, steady weight and fasting blood glucose level seem to have no effect at all on Hba1c for me.
 
Hi ziggy_w

thanks a lot. This is very useful.
Do I understand that CGMs are the metres, e.g. Dexcom G6, which measure just beneath your skin?
For illustration Hb1Ac measurement of 53 mmol/l corresponds to a CGM of 160 mg/dl on average.
The plot contains 95% intervals, but even if I eyeball a 67% interval (1 standard deviation)
you can get a variation of Hb1Ac values about 6 values up to 59 and down to 47 which also will
give CGM of 160. This is quite scary for two reasons.
1) When people fret about being below the diabetes threshold of 42, they could be off by a huge margin.
2) I would be surprised if all GPs know that there can be such large discrepancies between Hb1Ac and CGM.

I also read that for individuals the Hb1Ac variation is much more stable.
Thus I have to accept that for me the rise from 42 to 55 is real. Maybe Christmas where I did not resist sweets (which I usually do) and also not fast had a bigger impact on my Hb1Ac.
 
When I started measuring blood sugar, I didn't have much guidance, the nurse just showed me how to do it. As a scientist, I know that you have to do this systematically and since I was measuring my weight in the mornings, I decided to do the same with the blood sugar. The values started dropping along with the weight until I reached about 6.0 mmol/l on average. This was enough for me. Basically I am testing my engine on empty. I could test after eating, but then I would need to carry the test kit to work, which I don't really want. I understand that carbs are the problem. On fast days I don't eat any food which is high on carbs.
 

Thanks for the info.
Maybe Christmas where I did not resist sweets (which I usually do) and also not fast had a bigger impact on my Hb1Ac.
Breakfast is typically home-made muesli based on oats, some fruit and full-fat yoghurt. I now can't eat whole grains (more like whole sugar) anymore.
 
I think Hba1c is not that accurate. I had 2 hba1c tests done on the same day last year and the results were quite different
 
Well it sounds like you have your answer.

Oats and fruit are not good for my bgl. (I make a granola with seeds, nuts and coconut instead). Test it for yourself. Before and 2 hrs later and it should rise less than 2mmol. Less is even better. Weekends if doing it at work is really an issue. Same for evening meals.
 
HbA1c does not measure blood glucose levels but an effect of the levels, glycated haemoglobin, as a proxy. The relationship between HbA1c and average blood glucose levels was determined empirically by measuring HbA1c and average blood glucose for a sample of people and deriving a "formula". So, it is not a question of the HbA1c test being inaccurate but the relationship to blood glucose is a generalisation and the "formula" will apply more accurately to some individuals than others.
 
Lupf said:
1) How reliable are Hb1Ac tests?

2) Is this increase normal when you come off metformin?
Click to expand...
1: I get my hba1c tested every three months from the pathology shop at a private hospital near where I live. I completely trust the results for being accurate when I get them back.

2: I came off metformin after I had a string of hba1c readings in the mid 30 mmol/mol range and I have now been off of it for a year this month.

I also test before and after meals three times a day.

The pathology lab.



My graph of my hba1c readings from the time I first started metformin, till the last three tests upto December

 
I have done a lot of testing using about 6 blood glucose monitors (2 synchronised with my Doctors Surgery), A1CNow kits and FreeStyle Libres. Long story short for me was that, my HbA1c appeared to get slightly worse despite testing, and lowering my fbg by a full point and not getting spikes. So much testing that it has damaged my ability to use certain fingers for biometric security systems.

Like a dog with a bone I would not let the "discordance" between my fasting (circa 3.8 - 4,4) and post prandial (no more than 6, but usually circa 5.4) readings go unchallenged. I needed to understand why I could be a HbA1c of 39 with such low meter readings across multiple different devices, and the A1CNow multiple tests at circa 31, which would have appeared closer to what my meters were saying. In order to see if I was getting spikes I have tested for a couple of years many times with the Libre FreeStyle, where I only found 1 reading of I think a 6.5 post prandial, which was due to either sweetened sausages and or the mash made of swede, butternut squash and parsnips.

After some questions on this site @Bluetit1802 said she tested on another assay, which produced a lower result. I did 2 tests at Medichecks, one assay was the same type my hospital uses and the other was another type @Bluetit1802 mentioned. The first test was the same result as my Doctors test, the other was 4.5 mmol/L lower. I believe for some eating more red meat / other protein potentially extends the life time of the red blood cells. This I have heard several times in podcasts, coincidentally this one earlier today.


I now test less often, but notice no matter what device I use, my average is always 4.4 mmol/L



So now I just rely of fbg and post prandial readings mainly, as my diet is quite settled. So what I have learned in my n of 1 is that even with a full point drop in fbg, other "things" could keep HbA1c's high out of step "discordance" with accurate meter readings; so the important thing is what is the variance from a known accurate point.
 
Hi,

I'm a type 2. My last 2 HbA1c results were in the low 30s so my doctor considers my blood sugar levels to be well controlled. I still test myself at least 3 times a day. I'm still on metformin as I have a lot of weight to lose and I want to reduce the liver dumps while I do that.

I have a science degree but am not a scientist. My understanding is that the HbA1c is an average of your levels for the previous 3 months- if you don't test during the day you don't know how high or low you are going. The theory seems to be that to avoid complications you should keep your spikes below 7.8. I rarely get into the 7s now and am mostly in the low 5s with some 4s and the occasional 6. I think this is because I test regularly. I have gone very low carb and intend to stay there for the forseeable future. Initially I did have the idea that one day maybe I could have more carbs but I'm not sure that is the case. I'm also not sure what that amount would look like. If I can go up to say 50 grams of carbs that would still not have oats and fruit.

I think what struck me was that you have one HbA1c within the normal range and the next one was up. I would really suggest that you test regularly until you have at least two or more within the normal range and then keep eating the foods which got you in the normal range. I also don't like the regular testing but they keep me focussed on my readings which I consider to be really important. The HbA1c is really for me a confirmation- I rely on my daily testing.

Good luck.
 
You must log in or register to reply here.
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn More.…