Just how accurate is a1c anyway???

[mids1999]

After receiving my last batch of blood work which was interestingly done on two separate occasions 3 days apart my dr told me the results which made me wonder.
How accurate is an a1c test done in a lab anyway?
result 1. 5.9
result 2. 5.2

I had my typical consistently good control of my sugars that having a pump and cgms allows me during the months leading up to this test and during the test.

Interesting thing is that if I were to jump up almost a point during a 3 month period my drs would get concerned that I was slipping.

If this is supposed to be a 3 month average how could my a1c change so much during 3 days of good range (80-140 consistently mostly 90-110) numbers along with being steady prior?

[REDLAN]

The variation in the test is around 11%, which as I understand it is fairly typical.

I was reading an article about cholesterol testing. Apparently a variation of around 10% is expected between labs, testing the same blood.

It is very likely that your blood tests were done using different samples of blood, at different labs, by different technicians, using different batches of testing solutions on different machines. (or a combination of the above)

Unfortunately this is the nature of the beast - there is an error rate with lab testing. It's generally not publicised, but it's there nonetheless.

10% variation is significantly better than the expected 20% variation between meters used for home testing.

puts A1C tests in a different light doesn't it?

It means that a change in A1C of less than 10% is not significant.

[xMenace]

The variation in the test is around 11%, which as I understand it is fairly typical.

I was reading an article about cholesterol testing. Apparently a variation of around 10% is expected between labs, testing the same blood.

It is very likely that your blood tests were done using different samples of blood, at different labs, by different technicians, using different batches of testing solutions on different machines. (or a combination of the above)

Unfortunately this is the nature of the beast - there is an error rate with lab testing. It's generally not publicised, but it's there nonetheless.

10% variation is significantly better than the expected 20% variation between meters used for home testing.

puts A1C tests in a different light doesn't it?

It means that a change in A1C of less than 10% is not significant.

Thanks. You made my day. :eek:

[REDLAN]

I'm off today, and decided to do a little research...

So I looked up some stuff on accuracy of HbA1c and it's accuracy.

The gold standard is done on something known as an HPLC analyser - this is what your lab should have.

the accuracy of these beasts should have a coefficient of variation better than 2% - for an HBA1c of 5.9, this would mean that you could have an HbA1c as high as 6.2 or as low as 5.6 (I think that's right - if someone is better at maths than me, wouldn't mind checking my assumptions)

This is the accuracy you can expect if your Hba1C is always done by the same machine at the same lab. (you will find that manufacturers promoting their new machine will achieve much lower coefficient of variations than this, but hey that's advertising for you :) )

Desirable performance standards for HbA1c analysis ? precision, accuracy and standardisation Consensus statement of the Australasian Association of Clinical Biochemists (AACB), the Australian Diabetes Society (ADS), the Royal College of Pathologist (http://www.atypon-link.com/WDG/doi/abs/10.1515/CCLM.2007.158?journalCode=cclm)

the sticky stuff comes when your sample is done by different labs, on different machines, and this is where it gets interesting...

linked below is an article on standardisation of Hba1c in the UK

Standardization of HbA1c measurements (http://www.d4pro.com/IDM/site/standardization_of_hba1c_measurements.htm)

They tested a variety of different assay methods, and then reconciled them to the values found in the DCCT.

so....

an A1c of 7.0% for instance could actually mean that your A1C is anything from 7.2% to 8.03% - depending on where and how it is done.

seems that the best way of ensuring that your HbA1c is as accurate as it can be, is to ensure that the collection method is identical, and that your test is performed on the same machine each time.

this link I found interesting - it discusses the variability in A1c tests and average BG.

glycosylated haemoglobin (HbA1c) and accuracy as a measure of glycaemia - General Practice Notebook (http://www.gpnotebook.co.uk/simplepage.cfm?ID=x20070614192625295600)

[Jan B]

Wow, that is a huge difference in 3 days. Good info here. Sounds like we need to pay closer attention to the trends and how we are living/feeling than worry so much about the exact numbers. (That also applies to weight, bg readings, etc.)

I also hear from a lot of folks that the 3-month A1c test more closely reflects the last 6 weeks. Our red blood cells have a life of approx 120 days, but new ones are continually being made, others recirculating, while still older ones are dying off. I've also heard the A1c test trutly reflects 3-4 months.

Thank you -- this also helps me prepare for my next A1c that I've been anticipating. I know the same lab always does my bloodwork. However, these days I get my A1c results in 5 minutes in the office. I want to learn the percentage variation on that little tester.

WOW -- thanks REDLAN!!

[Evermont]

An amount of blood was taken for the first test. The body then produces additional NEW red blood cells to replace them. Normally about 2 million red blood cells per second are produced, each cell living about 120 days. This rate of production is increased to compensate for cells lost. When the new cells are factored into the "average" in measuring a new A1c a few days later, the result will be lower.

But by how much? We're taking maybe 25ml of the total 2.7-3.0 Liters, what's that like 1%? After donating a pint of blood, about 450ml, plasma volumes will return to normal in around 24 hours, while red blood cells are replaced by bone marrow into the circulatory system within about 3-5 weeks.

Erythropoietin is bound by circulating red blood cells. Low numbers of circulating red blood cells leads to a relatively high level of unbound erythropoietin, which then stimulates production of new red blood cells in the bone marrow resulting in a lower average glycated hemoglobin and thus a lower A1c.

It appears negligible but even if this effect is slight, combined with the other variables already mentioned, it is a contributing factor.

[Jan B]

An amount of blood was taken for the first test. The body then produces additional NEW red blood cells to replace them. Normally about 2 million red blood cells per second are produced, each cell living about 120 days. This rate of production is increased to compensate for cells lost. When the new cells are factored into the "average" in measuring a new A1c a few days later, the result will be lower.



But if you are eating lots of carbs and not taking enough insulin (or exercising, etc.) can the second test really be lower? Because you'd also be factoring in more lack of carb-control with the new cells.

[Evermont]

But if you are eating lots of carbs and not taking enough insulin (or exercising, etc.) can the second test really be lower? Because you'd also be factoring in more lack of carb-control with the new cells.

As I understand how the A1c test works, recent meals, which would certainly have an effect on a blood glucose test with a portable meter, would not have time to significantly increase the glycation of hemoglobin which is what A1c measures. This process takes weeks or months to have an effect on A1c. That's why A1c is an "average" and not a measure of current BG levels.

[fgummett]

Quite apart from the variation between labs and samples (which is very interesting... many thanks for the research folks) it is worth remembering that the A1c is an average and should be used alongside your own testing: for example someone who has highs every morning and lows every afternoon could still end up with an apparently good A1c.

As I understand it, the glycation (?) takes place all the time but the reason A1c gives a picture of the past 3 months is all to do with the lifespan of the blood cells.

[Jan B]

Quite apart from the variation between labs and samples (which is very interesting... many thanks for the research folks) it is worth remembering that the A1c is an average and should be used alongside your own testing: for example someone who has highs every morning and lows every afternoon could still end up with an apparently good A1c.

Exactly. Good point to keep in mind. We could both have an A1c of 5, but if you were steady, and I was low half the time and high half the time, you'd be in a lot better shape.

[fgummett]

A1c of 5 :love: now that's something to dream of :)

[Emm]

Does anyone know if there's truth to the idea that your A1c will come out lower if your current BG is low, and higher if the current BG is high?

I know that the A1c is heavily weighted towards your recent BGs, and perhaps that gets a lot stronger as you get into the last weeks / hours / seconds BG results... but mathematically I can't see how it could make a big enough difference.

[REDLAN]

there is possibly some truth in this.

It was always thought that once glucose had stuck to haemaglobin then the bond was permanent. Hence the claim that hba1c is an average of blood glucose over 120 days - the average lifespan of a red blood cell.

More recently however they have discovered that an enzyme Fructosanine 3 kinase (FN3K) can deglycate haemaglobin.

Enzymatic deglycation--a new paradigm or an epiphe...[Biochem Soc Trans. 2003] - PubMed Result (http://www.ncbi.nlm.nih.gov/pubmed/14641081?ordinalpos=10&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsP anel.Pubmed_RVDocSum)

the above is the only referenced study I can find to this phenomenon. The authors state in their conclusions that they don't know whether this phenomenon is significant, and also that FN3K can not explain all the deglycation that occurs, so there must be other deglycation mechanisms.

so nobody really knows whether what they demonstrated in the test tube actually happens for real in the human body.

so the hypothesis goes....

if Haemaglobin is slowly deglycated by the body, then your blood glucose nearer the time of the test will have a greater influence on your HbA1c, than your average glucose further away from the test - hence the rule of thumb that the last 6 weeks has a greater influence than first 10 weeks.

nobody knows for sure how true this is. what is known is that there is considerable variation of hba1c in humans, and it is believed that this is down to differences in deglycation rates.

so if your one of those lucky people who have good deglycation rates and a very low hba1c, then you could conceivable "cheat" on your hba1c by making sure your control was extra good close to the test.

[Scrabblechick]

Why is the thought of cheating the A1C such an enticing one? LOL.

[Blue Dog]

... Low numbers of circulating red blood cells leads to a relatively high level of unbound erythropoietin, which then stimulates production of new red blood cells in the bone marrow resulting in a lower average glycated hemoglobin and thus a lower A1c.

It appears negligible but even if this effect is slight, combined with the other variables already mentioned, it is a contributing factor.

From what I've read, A1c is accurate enough to give you a ball park snap shot of how you're doing in terms of sugar control. I wouldn't get to hung up on the difference between 5.2 and 5.9. To me the two numbers taken together indicate reasonably good control.

Unfortunately for me, my A1c tends to be in the 5.2 range regardless of my control. I suffer from chronic anemia due to a blood disease that results in a shortened lifespan for my red blood cells. So I test a lot in order to give my doctor a better image of what my control is actually like. On the semi-upside, my total cholesterol is 120 for the same reason (chronic anemia results in lower cholesterol numbers, something I didn't know until recently).

Congratulations on the good A1c numbers.