How to deal with glucometer variability?

[Amberyl]

I'm trying to achieve as tight control over my BG as possible, but I don't trust either of my glucometers (Ascensia Contour and a Freestyle Lite).

The potential range of variability is huge. For instance, I talked to customer support for the Freestyle Lite for two readings taken one after another, to test my brand-new glucometer, at 79 and 131, and was told that this big of a spread is normal and indicates no error.

I'm trying to eat and exercise to the meter. I'm going to draw pretty different conclusions from a BG reading of 79 vs. 131!

How do those of you who routinely live by your meter readings deal with the uncertain readings? What do you do to make the readings as accurate as possible, and figure out how to interpret fuzzy results?

[xMenace]

This is a very good question.

First though, those numbers are not within an acceptable variance. 20% is the maximum allowed.

As a type 1 I have to be very cognizant of the potential for errors. If I have a very high reading and need a large correction, I've learned the hard way that I should verify that reading. Likewise with lower numbers. Of course I hardly ever verify them.

Follow-up testing is also often critical. What I do is mentally set my expectation for future readings 1, 2, or 4 hours down the road. "I'm 7.5 now, so with a 1 unit correction, I should be about 6.1 in two hours." If I'm near that number, I relax my guard a bit. If not, I usually escalate my alertness level.

I will also tell myself to be careful with certain numbers and even change my behavior. Many times I'm mid-4's when I leave work in the afternoon. Experience says it won't go lower, but I'm constantly on guard. I pick up DW first, so I usually do a second test then. I may even make my wife drive home; though I'm not sure which is more dangerous :eek:

[Amberyl]

According to the customer service reps, as long as two numbers are within the A and B ranges of the error grid, the glucometer accuracy is considered fine. :mad:

At the time, I also tested with my other meter, which read 96.

Right now, I'm relying on the glucometer to teach myself about what happens when I eat and when I exercise. The variable readings are driving my engineer brain up the wall. :cool:

[butterflykisses]

According to the customer service reps, as long as two numbers are within the A and B ranges of the error grid, the glucometer accuracy is considered fine. :mad:

At the time, I also tested with my other meter, which read 96.

Right now, I'm relying on the glucometer to teach myself about what happens when I eat and when I exercise. The variable readings are driving my engineer brain up the wall. :cool:

Interesting, so exactly what are the A and B ranges of the error grid? They're supposed to be within 20%+- of what your glucose actual is. Even if you consider that the range is based on what your glucose actually is, those readings can not possibly be correct, or acceptable. However, this is how it goes with meters. I have gotten readings with much greater separation. One time in the 400's and then in the 200's!!! No reason, clean hands, within seconds of each other. But then they'll go back to be at least within 20 % of each, usually much closer. It is reality. It sucks, but it's the best we have to work with right now. Did you do a control solution check?

That said, if you continue to get that kind of range I would call and give them a little math lesson and ask for a new meter. Those are not acceptable error ranges.

[Amberyl]

They entered the readings into their a computer system, they claimed, and the computer said the samples all fell into the A and B ranges. (Clarke's error grid, I assume.)

I didn't do a control solution check, because the Freestlye doesn't include control solution with the meter. I ordered the control solution as soon as I got the meter, but it takes 7 to 10 days to arrive. The 79 and 131 were the very first readings I'd ever taken with that meter.

[sumi]

Astonishing answer from their help line! I will usually get numbers which are 20 points apart if I test twice in a row, but your numbers are terribly off. Perhaps it is already time to switch meters. These variations will drive you up the wall particulary if you are newly diagnosed. When my last A1C exactly matched that predicted by my average bg level, and because I am not on anything that would cause a low (nor do I go low), I was able to relax and now just use readings to establish any trends.

[butterflykisses]

They entered the readings into their a computer system, they claimed, and the computer said the samples all fell into the A and B ranges. (Clarke's error grid, I assume.)

I didn't do a control solution check, because the Freestlye doesn't include control solution with the meter. I ordered the control solution as soon as I got the meter, but it takes 7 to 10 days to arrive. The 79 and 131 were the very first readings I'd ever taken with that meter.

Ahhh...Clarke's didn't pop in my head for some reason. So, since the results probably wouldn't kill ya trying to fix them they're perfectly fine. Frustrating!

[MCS]

Meters do work better than that. The 20% error listed is to accomadate a correct reading within a 99.99% group. Which means the manufacture can garuntee a reading from that meter will be correct +/- 20% 99.99% of the time. You have to alleviate all other factors, clean hands etc. One thing that I found that helps me is to let the puncture site bleed, wipe it off several times before reading the last drop. I work in lab which reports % error on a lot of our results. There will always be a % error with any instrument that you use.

[kamkazmoto]

Try testing the same drop of blood twice, not two different drops. My meter is a one touch ultra and is usually within 5 points for two consecutive tests on the same drop of blood. That tells me that the meter is consistent and that the variations are probably with the way I draw the blood (how hard you squeeze your finger) or with contamination on my hand.

[arrow]

If it helps, take a look at the products from WaveSense.

The website has all the clinical studies about the high accuracy of their meter products. There is a new Jazz meter that has Glycemic Variability tracking that uses a Standard Deviation for better pattern management.

Check it out at WaveSense - Home (http://www.wavesense.info). :)

[Amberyl]

Thanks! I hadn't seen the Wavesense meters before. I've ordered a Jazz and now hope I'm not going to run out of my current batch of test strips before it arrives...

[flowerbabe68]

I use a Freestyle Lite meter and love it. Take your meter with you the next time you have blood drawn at a lab and do a meter reading from the same blood drawing site as the lab draws blood from. My meter compares fairly close to the lab results the 3 times I have done that. It is within 5-10 points and I feel that is the best I can get. It would be great if meters were right on the nose. With the expertise that is out there, cannot understand why it isn't. Doing a control check also gives you information of how close your meter is to the 20% variance. Hope this helps. Drives me nuts.

[mosaic]

Just got my latest labs back today. I took a reading with my Jazz a few minutes after the blood draw.

Lab BG: 109
Jazz BG: 117

Not bad.

[pappyiii]

Glucometer accuracy is my concern right now. 12 years of testing has not left me cozy with results. Adjusting insulin levels based on the accuracy of present meters is unnerving. Right now, stated accuracy of all meters is +/-20% (A range). At 200, that is 40 points up or down. B range is over 50% error. In addition, sometimes it is unclear wether the number is a whole blood or plasma report. Lab tests are always plasma after removing red cells. Glucometer readings are whole blood, sometimes corrected to a plasma reading (+12%), another 24 points.

I'm not injecting fast acting insulin before bedtime based on these measurements. i've been hypo before and I do not want to be there when I'm asleep.

Many of you have given some very good advice to achieve better accuracy. 1) wipe the first drop away and get a fresh drop by squeezeing a new drop. No one ever told me that in all these years. Most test solutions say to do that! Do all solutions register correctly in all meters??
2) Compare your meter with the lab simultaneously. When is the lab measurement actually done-sometime after the draw?
3) Compare several meters. Not cool! Perhaps just guess about the reading without bothering to stick!
4) Take several measurements! This is excellent advice. Lab classes at Drexel University promote several measurements which are averaged. This is like a Student Distribution in statistics.
5) Go for a WaveSensor meter. It appears that the Presto and Jazz meters are actually better than 20% for 99.5% samples. And rejects other sugars like Maltose. Personally, I plan to use only waveSense meters (I get a toaster for each referral).

What does continuous measurement offer?

Comments please! I feel it is almost useless to rely on the meter to adjust insulin when accuracy is so far off.

Those of you who feel your meter is spot on might be well advised to question further.

[mazea]

I have a second glucometer which is more accurate but gives me a good idea of how accurate my meter is. I use my second glucometer sometimes when I don't want to take another blood sugar reading. I go to bed at night without testing for lows sometimes because I have never been able to sleep once between 1 and 4.3 mmol ever. If I can't sleep I test and 100% of the time I'll be between 3 and 4.3mmol, so I know if the glucose machine is pretty accurate. Being out by at 1mmol at the most which is pretty accurate. It's a handy tool.

I tried doing 10 readings at 3.00pm one after the other once to see what the meter variability was. It was a funny result. 7 readings were about 6.3 mmol, 2 readings were around 7mmol and 1 was 5.4mmol. From the larger variability of higher blood glucose readings I wondered if maybe the readings less accurate on a meter at higher levels of blood sugar(6-10) than lower levels(2-4.5). Do you think the accuracy of the readings vary also depending on how high or low they are?

[Amberyl]

Yes, accuracy is range-dependent.

By the way, I switched to a WaveSense Jazz, and am much happier now. Absent my own user error, my readings are much more consistent now. (Re-testing with a second sample of blood from the same lancing consistently produces a result within 3 to 7 points of the original.)