Fingerstick meter accuracy -- how I use the lab to verify my meter

I use my Accu-Chek Aviva Connect fingerstick blood glucose meter as a fundamental component of my open-source artificial pancreas system (APS). I regularly calibrate my Dexcom continuous glucose monitor (CGM) with this meter. The CGM, in turn, passes its values every five minutes to the brains of the APS. This value then guides the every-five-minutes decision about whether to add, subtract, or stand pat with the pump’s programmed basal rate.

You’re likely aware that our consumer-grade BG-meter is not nearly as accurate and precise as a professional medical lab. Yet many of us make several dosing decisions every day regarding a potent drug, insulin. Insulin can easily be under- or over-dosed with serious safety and health effects. Knowing the BG number combined with confidence in its accuracy and precision is important to us.

So how do I check my BG meter? Whenever I see my endocrinologist I request a lab blood draw for a plasma glucose number. At the same time I do three consecutive fingerstick measurements and record them on my phone. I literally do this routine as I sit in the chair waiting for the blood draw. I take extra care to wash my hands with soap and hot water and thoroughly dry them before I do these checks.

Late last week I had my blood drawn for the plasma glucose test and my fingerstick meter gave me this series of numbers: 80, 85, and 86 mg/dL (4.4, 4.7, 4.8 mmol/L). This averages out to 83.66 mg/dL (4.65 mmol/L). My plasma glucose number came back today with an 82 mg/dL result (4.6 mmol/L). My fingerstick meter compares favorably with the lab result, within 2%. I now have renewed confidence that my fingerstick meter can safely guide my APS system to make reliable decisions.

Does anyone else go through a similar routine?

9 Likes

I sort of do the same, and will again this Friday when I get my pre-Endo lab-work. I’ll do the three fingersticks at the time I have my blood drawn for the fasting, and then I’ll compare my tested A1c against the 90 day average from my meter. So far, my Contour Next One has been within 1 mg/dL of my A1c. This will be the first time I’ve matched my fasting BG to this meter.

I’ve found my lab A1c corresponds poorly to my blood glucose average, either CGM or meter. My actual A1c comes in at 0.5% to 1.0% higher than the BG numbers indicate. Did you realize that any A1c actually corresponds to a range of blood glucose?

One study I read, for instance, showed that a 6.0% A1c can mean a corresponding average range of 100-152 mg/dL? For many years I thought that the A1c actually tracked to a narrow and discrete BG number, like A1c 6.0% = 126 mg/dL.

1 Like

Terry - The amount of excellent advice and great practices you have are continually astounding. You raise an important point and provide an easy approach to verification.

Awesome !!!

1 Like

I am aware of the A1c “range” issue, and also some of the underlying things that can make it inaccurate for some people. Strangely enough, I’m in the small group of people that A1c (translated to “average” blood glucose) directly corresponds to average BG from my meters. All four times I’ve gotten it tested since diagnosis, it’s been an exact match (within 1 mg/dL).

2 Likes

I’m a bit obsessive compulsive at times… My doctor was kind enough to give in to my desire for tight control as well as better understanding of my correlation of daily finger stick numbers and A1c and ordered monthly lipid and a1c tests. I have 3 meters that I bring with me, 2 Freestyle freedom lite and a bayer counter next. All three have been within 2%and my average 14 day readings seem to track as well. 112 and 5.4 this month, and 114 and 5.6 last month. Ive compared when my serum glucose was anywhere from 85-117.

The only time I’ve seen big errors on the contour next was when i took it out of a warm jacket pocket and used it outdoors in 30 degree (F) temperatures. It read way low, 62 when in reality I was closer to 100.

3 Likes

I got my first ever “obviously” wrong number last week after I’d been out walking in San Diego. Nice hot day, tested before supper, reading was 39. Screen on my phone starts flashing “CALL YOUR WIFE NOW” in all red text (the bonus or drawback of having a bluetooth enabled tester I suppose). Re-tested and it was 72 mg/dL. Not sure if it was the hot day or if it just did a partial-drop read for some reason (usually asks for more blood).

1 Like

I always do a single fingerstick when I have my A1C blood draw. I have yet to see a greater than 4mg/dl difference between the lab result and my Accu-Chek Aviva Combo meter, or previously my Accu-Chek Aviva meter.

1 Like

I’ll do a meter check whenever I’m sitting in the endo’s lab chair as well. When I was using two different meters I’d check both but it soon became clear the Contour Next One was the closest match. So, that’s my only meter at this point. Now, my A1c is always a little higher than my meter’s average but I get why as they are not using the same data. But when it comes to blood glucose meters I highly suspect that, like with everything else, best practice and best meter is subject to individual quirks. But I’ll continue to check my meter with each laboratory serum specimen because otherwise I have no way of knowing if my meter accuracy is slipping.

1 Like

Actually I would expect to see very large discrepancy between A1C and our meter average. We often use the meter when we suspect a BG problem. Which obviously will significantly skew the meter average into the “meaningless” category.

CGM average on the other hand I would expect to see much closer to the A1C - assuming no significant “gaps” in CGM data.

But I still like Terry’s approach of asking for a lab BG and getting a meter BG at the same time then comparing those values later when the labwork is returned. No A1C involved there just actual BG at a given moment in time.

1 Like

Great post @Terry4, thanks! I’ve done such tests over the past couple of years, which is how I eliminated LifeScan’s One Touch Ultra in favor of Bayer’s Contour Next One. The One Touch Ultra not only had much wider variability in consecutive fingersticks, but also had a noticeable systematic offset error with respect to the lab. Contour Next One has been spot on. With Dex CGM carefully calibrated by Counter Next One, my lab A1c also happens to perfectly match my CGM average bg according to the commonly used linear regression formula from this 2009 paper:

Estimated average bg [mg/dL] = 28.7 x A1c - 46.7

where A1C is expressed in %. For example, A1c = 6.5% corresponds to estimated average bg of 28.7 x 6.5 - 46.7 = 139.9, rounded to 140 mg/dL. It is important to understand that this formula applies on average, while individual results may deviate from it for any number of reasons.

1 Like

I do essentially the same thing. I have a blood draw every 90 days and I do a fingerstick while I’m sitting there, then compare the number to what comes back from the lab. The most it’s ever been off is about 5 points.

1 Like

You know, I inadvertently stumbled on a solution to the “biased readings by testing when feeling poorly” issue which has worked pretty well for me so far. When first diagnosed (as Type 2), I wanted to get a handle on how my BG worked through the day and week while only being prescribed 3 strips per day for a 60 day period (after which they’d drop it to 1 per day for, presumably, life or the beginning of insulin-dependency). I read up about “bracketing” BG readings (in a similar manner to doing repeated measures structured in such a way as to simulate large-scale random sampling). So… fasting every day; before and after lunch on Day 1; before and after breakfast on Day 2; before and after supper on Day 3; before and after exercise on Day 4; after snack and before bed on Day 5; etc. Mix up the timing of the “brackets” every week.

After being re-diagnosed as LADA and getting my testing strip prescription extended to a reasonable allotment, I kept up the routine, in addition to testing when I felt out of whack (to catch lows or highs). In continuing this simulated random sampling, I’ve managed to maintain a pretty close correlation between fingersticks and A1c. Either that or I’m just plain lucky.

1 Like

My last A1c was 6.1, done 4/5/17. My average blood glucose for 90 days prior to this appointment, as calculated using my Contour One meter readings, was 114 . The lab’s serum glucose came back 115. The point of care finger stick by the office staff was 136. My finger stick done on my meter was 116. All of these tests were drawn within a couple of minutes of each other. I’m going to throw out the office POC finger stick as I know they cleaned my finger with alcohol and didn’t let it dry and used a jackhammer to produce a stream of blood which they smeared all over the tip of the test strip. Only slight exaggeration there. Back in April I didn’t have a CGM but I do now.

My takeaway is my meter is reasonably accurate and can be trusted. My testing 6-8 times a day still wasn’t enough to capture sufficient data to know what my blood glucose was doing between sticks. I’m doing a fairly good job of controlling this thing. Let the office staff do their POC stick but pay no attention to the result, be it high, low or perfect. Did I get that right? :wink:

2 Likes

:thumbsup: I think you got it! Could not help but giggle at your description of office staff performing an utterly incompetent fingerstick :slight_smile: Had very similar experience some years back, and have politely refused the service ever since.

1 Like