So yesterday I saw my doc and did the usual download from my Aviva meter
to the Accuchek Compass software (yes, I'm running it with a USB
adapter on Win7) the night before. One of the many features of the
software is an average BG - I ran a 3 month report (about 500 points of
data) and then took that number and plugged it into the calculator linked below and it said A1C=6.99 - the next day at the docs, the test
said 6.9 -- that's a pretty good match I think. While I probably would
not try to convince him to stop running the real test, it was cool to
find this new tool, know that I can "check" my A1C anytime I want for
free, and get confirmation that my meter is functioning spot on. Free - no registration required - Accu-chek meter not required - check
it out.
This calculator uses the current ADAG formula. My estimated HbA1c according to this formula differs from my measured HbA1c by 30%. You can access the same calculation at the ADA site http://professional.diabetes.org/GlucoseCalculator.aspx.
My measured average blood sugars have always varied by 20-30% from what would be expected based on HbA1c. Some of it is probably due to the selected times that I test not being representative and there are some significant variations in how much people’s HbA1c varies with glucose. It actually turns out the the HbA1c is not that great a measure of average blood sugar.
Interesting … when I used the ADA calculator (which is the reverse of the one I posted) it estimated my average to be 151 but my actual is 154 - that’s only a 2% difference
I’m sure that you’re right about about how often and when you test will affect the correlation between the tests - in my case, I test frequently and so an average of nearly 500 points over a 90 day period is statistically significant enough to coincide with the A1C test.
The idea of the A1c is to have a seperate control figure. The blood glucose is highly dynamic and only CGMS users will get the full picture of the BG progression. What happens between 1 and 6 in the morning? I do not know but the A1c will tell me. Thus it is highly optimistic to derive the A1c from 4 to 10 glucose tests per day. This is why we still have no A1c calculator in our Glucosurfer project. It can me misleading.
Right, the idea of a separate control test is great for the doctor who has trouble getting patients to test, let alone log, their BG. What’s interesting is that most of the calculators on the web only give you an estimated average BG based on an A1C result - not the other way - this calc allows you to input an average BG (which many meters automatically provide) and predict an A1C. Really, the A1C tells you nothing about your BG “between 1 and 6” am - based on my very limited experience, I’d say this calc is accurate enough for me and I don’t feel it’s “optimistic” - it’s just another tool and more information - and like I originally said, I don’t think it’s a substitute for the real test.
It is more than just having a representative sample. If you compare the results of the fructosamine test (another glycated serum protein test) to the HbA1c, some 40% of the HbA1c appear to be more 1% off (http://care.diabetesjournals.org/content/30/10/2756.long). That is a huge difference. If I walk into a doctors office, and my measured HbA1c is 6% then there is a 40% change that my control is actually either < 5% or > 7%. That that is a lousy measure. That’s all I’m sayin, don’t put too much faith in the HbA1c. My doctor also orders a fructosamine test.
I’ve had varying luck when comparing the average of my meter with my HbA1c. Sometimes its a perfect match sometimes there has been a slight variation.
The reason for this is that red blood cells have a half life of 30 days. This skews the HbA1c toward the most recent recent blood glucose values. When my sugars have been relatively stable the average has matched quite nicely. But on my last test my sugars went haywire the last month preceding my test. My test average was 5.9 but my Hb1Ac came back 6.1
I still always look at the averages when I download my meter data and compare various periods to gain insight into how well I’m doing. HbA1c and meter averages are both flawed in their own ways and yet still valuable. I guess in a perfect world we would all have a non invasive CGMS
One last point to throw another monkey wrench in the works, different meters seem to run higher or lower than one another.
But most people preserve their eating habits for long periods of time. So I think the higher glycation effect by fructose is not that problematic. It may be difficult to compare individuals but for the same person a trend in A1c numbers is certainly important. An increased A1c despite good meter readings will give me an incentive to question my control. This why the independent measurement is still important.
There’s the concept of “high glycators” and “low glycators” e.g. the relationship between A1C and average bg actually depends on other things than just the bg.
The A1C measurement has matured a lot since I first had it run probably 25 years ago (back then my docs told me it was a “new experimental measurement and hard to interpret”) but A1C measurement standardization between labs is still pretty poor, with a lot of variance between labs.
IMHO the “artificial A1C” (computed from average of individual measurements) is probably superior to the actual A1C if what you actually want is average bg. Because (no big surprise) the average of individual measurements is the average of individual measurements!
Traditionally doctors have had a hard time getting patients to check their bg more than a few times a day, or even to check their bg at all. So for many (maybe the majority of patients) the A1C always had to be the stand-in for the actual bg numbers the doctors wanted.
There’s some good graphs in the paper at http://diabetes.diabetesjournals.org/content/56/7/1913.full showing the large scatter. If the “formula” was perfect all the points would fall on the straight line. But you see enormous scatter around the straight line, because some are high glycators, others are low glycators, lab differences, bg meter differences, etc.
Great post Tim. I agree with you and it is for this reason that I bring all my meter readings and remind my endo that the HbA1c is only one measure and not necessarily a good one at that. He usually orders a fructosamine test for me, which work similarly to a HbA1c, but only corresponds to the blood sugar over the last two weeks. Unfortunately, he usually won’t take the time to look at my meter readings or meter statistics.
Some key things I have learned. Anemia can make your HbA1c artificially low, real low. And the opposite is true, high hematocrit (polycythemia) can cause your HbA1c to be artificially high.
If you observe an ongoing and major discrepency between your average meter readings and your HbA1c discuss this with your doctor. If you HbA1c is 20% lower than would be expected from your meter (like 5.6% when you are actually 7%), don’t let your doctor treat you to your HbA1c. Have him order confirming tests like the fructosamine and look at your readings.
I think that most practicing real-world docs have a hard time dealing with large quantities of numbers. When I remind them that I have a PhD in physics and took a lot of math they mostly lapse into horror stories about how they struggled with math (and probably organic chemistry too - I’m no whiz kid in organic chemistry!).
So yes, getting them to look at anything other than the A1C is hard. I find the most effective thing is to bring a chart or a graph (again I love graphs but some docs look at a graph or a thousand data points like they’re getting hit in the face with a brick) and then point at a couple of individual numbers and start the conversation that way. Otherwise they feel overwhelmed by the quantity of data and fall back on the A1C as the final arbiter.
Historically one of the reasons why I probably don’t like the “low A1C” as a judge of my control, is that in the past I had my A1C in the 5’s (even in the low 5’s) and had a LOT OF HYPOS. My docs would lecture me about the dangers of hypos but I rarely listened. I ended up with a couple of 911 calls and ER trips before I straightened up and flew right, but today my A1C is actually a little higher and I feel far better about my overall control (although I still struggle with individual numbers a lot).
I agree with you that most doctors can’t handle it. I have given up showing graphs and all. I list average fasting numbers and then give percentages of times I am within target, like 40% of my postprandial readings are within target. That almost seems understandable to them. What I actually do involves excel and computing things like standard deviation.
You give more than I do! I don’t like giving them actual numbers because 99% of the numbers can be really good and they manage to focus on the 1% that isn’t. I won’t give the chart anymore either showing what percent I am in range because the software I have also shows the lowest & highest number in addition to the percentages.
