I agree. Pretty close to lab testing.
I'm pretty sure my doctor uses bayer in his clinic each visit. The results I get there match the relion (walmart) kits I've used at home.
The National Glycohemoglobin Standardization Program (NGSP) was started to calibrate A1cs against a reference so that we have consistency across labs as well as report independently on the accuracy of different test methods/tools. The latest survey results show that the Bayer A1c Now still has pretty poor performance, having as much as an average error of 0.42%.
Another recent study of so-called point of care systems (such as those used in doctors offices), including the Bayer A1c Now also found that they didn't have very good accuracy. The Bayer A1c Now performed so poorly, a complete evaluation using all the protocols was not performed. On the other hand, David Mendosa looked into the Relion A1c available quite affordably at Walmart and had very good things to say about Heritage the lab used (you have to mail off the test). Heritiage/Apprise uses the Roche Cobas Integra 800 CTS one of the best performing methods out there, assessed by the NGSP as having a maximum error of 0.11% and low variability.
My conclusion, if need to depend on the home A1c test then the Relion is the best choice right now.
Not to get pick, Brian. But unless my math is faulty a .42% error rate is a 99.58 success rate! That sounds pretty ok to me!
I'm not sure he meant error rate. I think it might meant the average error is a deviation of .42% from the A1C as calculated by a lab, but again, I'm not sure. For example, lab says 6, but the Bayer home kit says 5.6
It's kinda of confusing what the .46% for the A1C Now means.
Looking at the chart from the link, if the NGSP reference value is 5.6, a Bayer A1C Now would give you an average of 5.14 from 28 different labs. If the NGSP reference value is 9.4 then the A1C Now average from the 28 labs was 8.49. An NGSP reference value of 7.2 would give an average reading of 6.59 from the 28 labs using an A1C Now kit.
Whatever it means, like Brian says, an A1C Now kit didn't give a very accurate reading across a range of reference A1Cs and generally the A1C Now underestimates the A1C.
Shawnmarie is correct I was quoting an absolute error. But it turns out I quoted wrong and it is even worse: The test had each lab (there were 28 labs) do three measurements and computed the average absolute error. Here is the date:
Actual A1c Measured A1c Error
5.6 5.14 -0.46
9.4 8.49 -0.91
7.2 6.59 -0.61
I read the wrong line. In fact they found the Bayer A1c was off by nearly 1% at times. That is a relative error rate of nearly 10%. That is almost as bad as a glucose meter. That is really bad.
But I suppose its "badness" is relative. For example, let's say we have a person who has been completely ignoring their D management and has an A1C of 12. If they then start to pay attention and really try to get their BG under control, the home test could assist them in determining whether their actions are having an impact. That person might use the home test and get a result of 7. Even if the actual number 7.9, that person still knows the work is paying off and it could really encourage them to continue. In a perfect world, said person would be going to the doc to get official A1C results, but we all know this world is far from perfect!
Yeah, no doubt, and I think if you are interested in A1C trends, as long as the A1C Now error is consistent, you can get a good idea what's happeing from month to month by keeping tabs on your A1C using the home kit.
FWIW, when CVS was doing their free A1C screening offer, they were using the A1C Now kit. They had me at 5.4 when my offical lab reading was actually 5.8. That seems within the range the study posted by Brian had.
Ah, that is different, and all good points. I also wonder what the error rate is for the lab test? I assume the "actual A1C" means lab tested A1C. I believe the reason I thought pretty positively about the home test is because the lab results also had an error rate and so the difference was smaller.