January 21, 2010

Study: A1c Now Test is Crap and So Are Many Doctor Office A1c Tests

A study published in the January issue of the journal, Clinical Chemistry put eight brands of A1c test kits, including the A1c Now test sold in pharmacies and online, through rigorous testing. The other kits were the ones marketed to doctors for use in their offices.

You can read an abstract of this study here:

Six of Eight Hemoglobin A1c Point-of-Care Instruments Do Not Meet the General Accepted Analytical Performance Criteria. Erna Lenters-Westra1, and Robbert J. Slingerland. Clinical Chemistry 56: 44-52, 2010. First published November 19, 2009; 10.1373/clinchem.2009.130641

The study concluded only two of the eight kits produced clinically valid results. What is most telling, though, is that the study reports: Because of disappointing EP-10 results, 2 of the 8 manufacturers decided not to continue the evaluation. Their test kits were removed from the study after they completed only one of three CLSI protocols that were planned. CLSI stands for "Clinical and Lab Standards Institute."

Since four of the six kits that remained in the study ended up with unacceptable results, but must have had good enough results after the first protocol to encourage their manufacturers to keep them in the study, one can only conclude that the kits removed from the study did abysmally.

Which meters were withdrawn from the study after they did very poorly at the first level of testing? A1c Now and Quo-test.

Of the six remaining kits that completed all three laboratory test protocols,
Only the Afinion and the DCA Vantage met the acceptance criteria of having a total CV <3% in the clinically relevant range.
The CV is the "Coeeficient of variation." It is the statistic that represents the ratio of the standard deviation to the mean. It's a measure of how widely scattered data is. The larger it is, the more wide the range is that a set of data falls in. If a test kit produced a higher CV it tells you that using it repeatedly on the same sample produced an unacceptably broad range of results.

Even the kits that met the goals of the study had problems as the study concludes,
The EP-9 results and the calculations of the NGSP certification showed significant differences in analytical performance between different reagent lot numbers for all Hb A1c POC [point of care] instruments.
In other words, there was significant variation from batch to batch.

What does this mean for you?

1. Don't waste your money on a home A1c test. The accuracy is abysmal. This is actually disclosed in the package but in a way that confuses people. The packages says the results will fall within .5% of the actual lab result, but people don't realize this means that a 6.5% reading from this test could mean you'd get a reading of 6.0% or one of 7.0% at the lab. This is a significant difference. I used the A1c Now home kit twice several years ago and it was off by .5% both times.

An article about this published in the newsletter for endocrinologists that tipped me off to this added that lab results nowadays are usually very accurate.

2. If your doctor wants to test your A1c in the office and tells you it will give you an immediate result, demand to know which kit was used and if it is not the Afinion or the DCA Vantage, refuse the office test. It will be highly inaccurate, though you will be charged for it as if it were a lab test. Opt for a lab test instead. If your doctor gets snarky, direct him to the Clinical Chemistry article and suggest he use a meter that will give accurate results.

This is important because doctors often use nothing but the A1c to diagnose and gauge the progress of their patients. With the large variations from the true value you may get with an inaccurate test kit, you may think you are doing well when you are not, or vice versa.

3. If you are diagnosed with diabetes for the first time after being given an in office A1c test, be aware that this test is being administered for convenience. It is not a particularly accurate way of measuring your blood sugar. Get a blood sugar meter and test your blood sugar both first thing in the morning and one hour after each meal. If you see numbers over 200 mg/dl (11 mmol/L) at any time, you can be confident you are in fact diabetic. If you see only numbers between 140 mg/dl (7.7 mmol/L) and 199 mg/dl (11 mmol/L) after eating, you technically have "pre-diabetes", but your post-meal numbers are high enough to damage your heart, retinas, and nerves as this is the range where solid research demonstrates that "diabetic" complications begin to occur.

You can learn more here about Research Connecting Blood Sugar Levels With Complications.

You can learn how to lower your blood sugar levels to the safe range here: How To Get Your Blood Sugar Under Control

One final thought. You probably don't want to know what careful testing would find out about the CV of the obscenely expensive blood sugar test strips we rely on. The fact that we patients put up with $1.20 a piece strips that can vary by as much as 40 mg/dl(plus or minus) from the true reading at a reading in the mid 200s and by 20 mg/dl plus or minus at normal ranges is criminal.



Scott S said...

Excellent information; I was aware that Bayer's A1C Now test was very inaccurate, but I'm troubled that the in-office kits used by many doctors are equally error-prone. In the interest of making sure our healthcare dollars are maximized, I think we should demand lab tests, even if there are more convenient options; the proof will be in the results!

RLL said...

About a year ago they did an office A1C on me. I was expecting 4.8-4.9, and it came out 4.1. The doctor insisted it was an utterly accurate result. I loved it for bragging rights, but assured people I did not believe it.

Jenny said...

Raf sent in a comment about the Wal-mart Relion mail away A1c test which got accidentally deleted. Unfortunately, I don't know anything about the accuracy of that test. Has anyone here tried it?

Jack said...

> Has anyone here tried... the Wal-mart Relion mail away A1c test...

I have, once, after a VA administered
test returned a 6.5 when I was expecting
a 5.3, based on BG test data. I should add that I test 5-8 times daily, & that
previous VA lab tests had supported my
statistical estimates...

The Reli-On test returned a 5.2 - I'm about to do it again, this time prior to my next scheduled VA test.


Jack said...

> The Reli-On test returned a 5.2 ...

Forgot to mention - The VA did a re-test
about a month later - 5.2 also.

This is'nt just number-jiggery. My docs
want to muck around with my insulin &
& metformin regime which, so far, has worked well. I'm not willing, being of the opinion that safety & efficacy of
alternative therapies have yet to be demonstrated.


Carl H said...

David Mendosa likes the test that I buy at Walmart as a Relion A1c kit (it's marketed under a few different names), here's a link to his write up:


Good enough for me to give it a try. The comments at his site were interesting. The testing lab (Olathe KS) is just a few miles from me, I know a bunch of local Med-Tech's from college, maybe some of them work over there. I'll ask around next chance I get.

ShottleBop said...

I participate in a couple of online diabetes discussion forums. The question about the ReliOn mail-in A1c test comes up periodically. I've never seen anyone report more than a 0.1 difference between the ReliOn test and their lab test.

I've tried the A1cNow test, myself, and agree with the new assessment. Three months after a lab-tested 5.3, and without any significant changes in my meter average, the A1cNow reported a 5.8. Three months later, the second test in the box reported a 5.9, the day before my lab test produced a 5.4. Since the test was being advertised with assistance from Steve Freed, of "Diabetes in Control," I asked him to check with the manufacturer. The Bayer marketing rep he inquired with told him that a variance of 0.5 was within the design tolerance:

"There is not a tremendous difference between 5.3 and 5.8, particularly with 50 percent of the number is captured in the last 30 days. A lab method (Tosoh or Biorad) can differ just the same. The product is NGSP certified yearly. No kits go out the door before they are compared to a boronate affinity analyzer. Sunnyvale is an NGSP lab that compares themselves to other prestigious labs."

Personally, I think that there IS a tremendous difference between two readings 0.5 apart, especially in an environment where the establishment is pushing towards using the A1c as the sole indicator of one's "control," and where it is sometimes (often?) to get the support of one's doctor/insurer for frequent self-testing of BGs.

Rishara said...

I knew the A1C Now test was crap when it reported that my A1C went up from 7.9% (lab test from doc) to 8.0% after 3 months of very faithful low-carb dieting. Just out of curiosity, I used the 2nd A1C just before my last doctor's visit. It reported 7.0% A1C and my lab results reported 6.5%. My doctor doesn't do an office test but sends the tests to the lab, so I believe those readings are accurate. I was at least relieved that the results were skewed in the wrong direction in my case...

The Anon Guy said...

I'm not so sure about the Relion test's accuracy.

I was diagnosed with an 8.6 from LabCorp on April Fools day. By the time I had my followup and received a meter (about three weeks) I had already made dietary and exercise changes, so I decided to try a Relion before I started taking Metformin. It came back an 8.1, which I thought was inline with the past month of changes I had taken.

But, in anticipation of my three-month LabCorp test I went ahead and tried another Relion to get a heads-up. I was hoping a sub 6, which would mean my Precision Xtra was actually accurate (it seemed low), but was stunned when it came back a 4.8. When I took the LabCorp a week later is was 5.3.

I guess being low by 0.5 is just at their range of being off, but let's say it said 5.9 and I was happy but, in reality, I was at 6.4. It makes me question doing another one. Actually maybe I should buy two kits and do them both at the same time and see what the variance is.

By the way, loved your book. It's anti-carb tilt helped me steer away from the oatmeal and wheat toast I had adopted in those three weeks. Probably why I'm at 5.3 now.

Jenny said...

I do not advise using ANY over the counter A1c test. They're all highly innaccurate.

The Relion meter I recommend is the blood glucose meter which should be used after meals to see what foods raise blood sugar to unacceptable levels. All the blood sugar meters are inaccurate too, but give or take 10-15 mg/dl at near normal readings they still give useful information.

The Anon Guy said...

Yeah, I've had some days where I tested twice just to see and there was a 10 point difference. That said, I tested just before going into LabCorp and had an 86 and 89. My blood test came back an 83, so I guess the Precision is relatively on-track. I think they make the Relion one and strips too.

On a side note, I was in Target yesterday and noticed they had a meter and strips with their Up and Up house brand. Strips were $40 for 100 and the meter $6.95. Seems to be competitive with Relion. I think Wavesense, though, made the Target product. Not sure of their rep.

Costas said...

My A1CNow result was 7.0. Repeated it the next day and got 7.7! At the doctor's office, using also the Bayer technology it was 6.2.
My conclusion: Those results are much too variable and therefore not useful.
In a telephone conversation with Bayer's customer rep, I pointed out that a 7.7 result is indeed ~10% higher than a 7.0 result. I requested to send me their explanation(s) in writing, since their oral ones were confusing to me.

Jenny said...

Costas, Test with your meter after meals every so often. Those numbers are a much better gauge of whether you are damaging your body or not. The A1c doesn't predict neuropathy, the post-meal blood sugar does.

That's really all people need to know.

Rob said...

Jenny's last comment (11/10/10) stated after meal tests were all we need to know. I question this (info seek).
We test at all the WORST possible times - after meals, 1st AM, etc. How does this correlate with potential pathology? Can't we "safely" spike but be generally quite well? Highs are probably never good, but isn't the A1c the best determinant of overall success?
I was diagnosed by a lab A1c of 11.5% (former physician only used urine dip-stick - no problems). Found Bayer test 6mos later. Got a 5.9 - way too low for me to believe. Re-test 2 wks later said 6.6. Still really nice in so short a time.
Have tested ~every 2 mos. Latest was 5.7%. I WAS exctatic until I read these comments.
Next Dr visit in Jan, I may request a lab test and do the A1cNow immediately after. If all goes according to plan, I will report results.
<$15 per test is a lot better than lab costs if at all reliable. I'd LOVE it to be 0.5% lower! Not all of us have insurance - or jobs.
I might add that I am a biologist/biochemist and can follow testing instructions - most of us can as we need to test so often.

Dr Rob

Jenny said...

Sorry but the correlation of A1c to complications is VERY poor. It's a useful statistic for large studies but fails at the individual level.

Go review the studies looking at what correlates to complications which you'll find HERE. As you can see there is no relationship between A1c and neuropathy. There is one with heart disease, but there too the blood sugar after a glucose challenge is a better predictor.

Read this study to understand why the A1c doesn't map to glucose levels well:

<a href="http://diabetesupdate.blogspot.com/2008/03/more-insight-into-why-a1c-doesnt-match.html>http://diabetesupdate.blogspot.com/2008/03/more-insight-into-why-a1c-doesnt-match.html</a>

It's the glucose levels that damage your organs, so you want to track them, not a secondary measurement that can be misleading if you have one of many varients of red blood cells that causes them to live long or if you are anemic.

Unknown said...

Bayer's a1c customer service reps told me that their A1C NOW self check test is accurate to the extend of + or - 0.9%. They also said that the A1c performed at a doctor's office using the Bayer technology has the same accuracy. So, a 6.0% result means that the "real" value is somewhere between 5.1 and 6.9%. I think that patients should be aware of these "details".

Jenny said...

The reps may tell you all they want--they're notorious for stretching the truth. The fact that their meters were withdrawn early from the study because of disappointing results tells you all you need to know.

My own experience comparing their test kits to lab results confirmed that they were a waste of money. I've heard the same from quite a few other people posting on discussion forums.

Anonymous said...

My A1C is 5.7 with frequent urination (10 times a day drinking 2 litres of water).
I did low carb diet since 2 months ago and have lost at least 12lbs with average blood glucose 80-90mg/dl and post meals (2hours) below 120mg/dl. However, my urination is still frequent.
Can frequency urination become an indicator of our actual glucose level?

Jenny said...

Frequent urination is not a symptom of blood sugar problems in the range you report. It usually occurs as a diabetic symptom only when people are running sugars in the 300 mg/dl range or higher.

It's possible you have a subclinical urinary tract or urethral infection, which can cause frequency. The urethral infections can be hard to diagnose as they don't always culture and many doctors aren't aware of them. A course of doxycycline or metronidazole can some times clear up these nonculturing infections.

Harold said...

I am a physician and am diabetic. I really don't care if my blood sugar meter is 20% high or low when reading 200 because 180, 200, or 220 is far to high for me and I will actively pursue getting it down to an acceptable level. I also never get in that range following a very low carb life style with insulin. My guess is that to make a meter that is very accurate at the level of 200 would cost a lot of money and I don't see that it would be valuable to me. I do want my meter to be accurate at the normal blood sugar reading (ie 83) and have not had problems getting reproducible glucose readings with Walmarts relion or Walgrens Truetests both inexpensive meters and test strips.
A1c testing is less accurate than knowing what your blood glucose reading is throughout a typical day and I'm not sure it is worth doing. It is certainly more expensive.

Harold said...

Diabetic frequent urination is do to glucose being excreted in the urine. A urine glucose strip will tell you right away if your problem is high blood sugar and I doubt that yours is. Check your urine specific gravity as that will give clue to diabetes insipidus (urine won't concentrate).

CW said...

Interesting comments....I have a lab background and have been familiar with various HbA1c assays (test products) for years. The only POC test that measures up to gold standard testing is the A-S Afinion! The NGSP results prove it and the analyzer utilizes the same methodology as the NGSP's reference method (Boronate Affinity) which has no Hb variant interference which is huge and why most others fail are all over the board with results and reproducibility. The Afinion is a very high qulity analyzer and being used more and more by Endo's and Diabetes Ctr’s everywhere!

Mary B said...

Four months ago I didn't know what an A1C test was. Then my fourteen year old son who is a bit overweight and not very active (into computers though) got a blood test and had a 5.95. I also discovered I had had the A1C a year and a half ago and gotten a 5.79. (I hadn't even noticed, it was part of a health fair blood draw) Which launched me into a huge amount of research: Blood Sugar 101, Gary Taubes book, Protein Power, diabetes blogs, the Fathead movie, etc. I also got blood glucose meters and started using them. Not surprisingly, we have cut WAY back on our carbs in this family.

Here's the weird thing regarding the A1C: a couple of days ago we took my 14-year old and his 12-year old lean, fit brother who is constantly running around into the local hospital lab for the A1C test. The results were 5.84 for the 14 year old and 6.04 for the 12 year old. I decided to do the post meal blood sugar test on my 12 year old the next morning. I used the Ultra Onetouch mini AND the Walmart Relion (which generally has been running about 10 points higher). His a.m. fasting numbers on the Ultra were 81 and 89 (tested twice), a bit higher on the Relion. Then I fed him 65 grams of pancakes, syrup, and orange juice. One hour after that his glucose was 71 and 76 on the Ultra, 80 on the Relion! At two hours, he was at 89 on the Relion. Also, he has tested himself quite a few times in the last three months at various times just because he thought it was a cool gadget and he was interested in seeing me doing it. He says his results are never very high, not over 120. So how the heck could he have an A1C of 6.04?? Any ideas?

Also - our pediatrician seems a bit too wedded to the mainstream view - low fat is good, etc. He wants me to bring both boys in for an appointment and let him handle it and give them 'the talk'. I don't think so because he already did that once with my older son and after all my reading I disagreed with some of what he said. I'm going to tell him what happened on the two hour test, see what he has to say about that.

Jenny said...

Mary, I have blogged about this in another post: HERE and HERE.

The very short version: genetic variations in red blood cells, anemia, and unusual length of life of red blood cells are just some of the known factors that make the A1c unreliable.

Doctors like it because it's an easy test to administer and they don't have the time--or in many cases expertise--to interpret meter readings. But meter readings are far more reliable.

Mary B said...

Thanks, Jenny. I'm glad I've been reading your website, otherwise I would have been clueless and not known to do the two hour post meal test. I hope I can get the doctor to read all this.

Abraham said...

I'm suspicious about A1c and blood glucose ranges we are told about. On the back of the Bayer Act Now box it states that 6.0 is normal and that 5 and below is for non diabetics. It also places 7.0 on the border line.
I lab tested with a 7.0 result.
I have read that post meal number should be below 180. Dr. Neal Barnard has this in his book.
I was diagnosed two years ago and have been working to keep it all under control. from what I have read glucose ranges became much more stringent in the past 10 years, so I am wondering how much truth there is to what we are told. Diabetes is big business and keeping people sick is making money for alot of people.
I suppose if I use the Bayer A1c Now I would err on the side of the higher number, which would force me to stay focused and disciplined. I know lab tests can be shady, so the home test can't be any worse.

Jenny said...

Dr. Barnard's advice is flawed because he isn't trained in treating diabetes and is just parrotting the outdated ADA line. Blood sugars of 180 are high enough to cause all diabetic complications.

The research data pointing to safe blood sugars is presented in great detail on these two pages on the main Blood Sugar 101 site.

Research Connecting Blood Sugar with Organ Damage

A1c and High Post-Meal Blood Sugars Predict Heart Attack

Read them and draw your own conclusions. Drug companies like higher targets because their oral drugs will get NOT patients anywhere near the safe levels without low carb diets (which the establishment refuses to study or endorse) or insulin.