HbA1c Precision

[SamJB]

Anyone know how precise HbA1c results are? I.e. what are the error margins. Do GPs have this information on a particular test? For example, my last result was 6.5%, does my GP know this to be 6.5% +/- 0.2% or something similar?

 

[robert72]

Probably depends on the lab and the equipment they use.

And of course there are lots variables.
http://www.ngsp.org/factors.asp

 

[phoenix]

Precision?

We can conclude that, at least whenHbA1c is measured
in an accredited laboratory, physicians can be reasonably
(95%) certain that a difference of 0.5% Hb A1c or
greater between successive patient samples represents a
statistically significant change in glycemic control. If the
laboratory is using a method that demonstrates a within laboratory
CV
2%, as is still the case with some assay
methods, there is less confidence that a difference of 0.5%

I think that all UK labs are part of the NGSP Process for standardisation. The paper shows how it works, no idea how it works at the bottom of the chain .
http://www.clinchem.org/content/57/2/205.full.pdf

So basically what it shows is individual . If your HbA1c goes up or down more than 0.5% then it is probably significant for you but your 6% may not reflect the same average blood glucose levels as someone else's 6%. (depend how average you are :lol: )

That is why the UK authorities don't report the estimated average glucose level.

The Eag study writers said

The results of the A1c-Derived Average Glucose (ADAG) study support the notion of a close relationship between A1C levels and AG for both type 1 and type 2 diabetes. .

http://care.diabetesjournals.org/content/31/8/1473.full

R. David G. Leslie and Eric S. Kilpatrick, (both UK diabetologists/researchers) replied to this paper suggesting

1) that the conversion factor was probably flawed because the study group was non randomly selected and included non diabetics.
2)The formula wasn't validated on specific groups and did not take into account the variations brought about by haemoglobin variants, pregnancy, renal problems, ethnicity etc
3) and probably for most of us the most important given the amount of discussion on what a 'good' Hb1c level is.

Third, there is uncertainty associated with the eAG value itself, acknowledged by the authors but not evident in the formula promulgated on the ADA Web site. ( I think that is the same as the one this site now)That uncertainty is such that two patients with the same true mean glucose level of 170 mg/dl could have an A1C value anywhere from 6.5% (which is as low as the A1C values of the intervention groups of the VA and ACCORD [Action to Control Cardiovascular Risk in Diabetes] studies) to 9.0% (which was adversely high in the Diabetes Control and Complications Trial)

 

[SamJB]

Thanks for the answers guys, been very useful. Phoenix, I'm always impressed by the papers you dig out for various questions! Do you work in science? Where do you get them all from?

The CVs of 2-3% in that paper mean that the error bars are around +/- 0.1% for 1 standard deviation if your HbA1c is in the 6s, which is good enough for me, an ex-astrophysicist! For 2 standard deviations, your error bars are +/- 0.5%. Doesn't sound very precise!

 

[Yorksman]

Anyone know how precise HbA1c results are? I.e. what are the error margins. Do GPs have this information on a particular test? For example, my last result was 6.5%, does my GP know this to be 6.5% +/- 0.2% or something similar?

Labs have different methods of measuring HBA1c and have different equipment. They are supposed to run controls throughout the day and recalibrate at regular intervals. Different methods of measuring HBA1c require some of the data to be converted and some of these have their own error margins. Below is a statement from an article with its link:

"Today, most methods for measuring HbA1c are NGSP-certified. According to the NGSP manufacturer certification criteria, this means that for a panel 40 samples analyzed in duplicate under ideal conditions, the 95% confidence interval of the differences between the method being tested and a NGSP SRL is within +/–0.75% HbA1c."

Your +/- 0.2% is optimistic. Labs rarely achieve the ideal conditions required.

HbA1c
An Overview of Current Analytical Testing Issues
http://www.aacc.org/publications/cln/20 ... bA1c.aspx#