Must Read Study on Meter Accuracy

http://www.journalofdst.org/September2012/PDF/VOL-6-5-ORG2-FRECKMANN.pdf

This is a comparison of 43 meters, and I think they did a great job. Please check out your meter. I think the biggest surprise on this list was that the Bayer Contour USB performed pretty poorly (it was by no means the worst on the list but it is surprising because it's one of the pricier ones on the list). Bayer Contour USB was within 5% only a third of the time. It was more than 10% off  about a third of the time. It was more than 20% off in 4% of tests- which I think is a lot. I mean, if my bg is actually 200, it only would guess between 190-210 a third of the time. And it would guess outside the 160-240 range for 1 in 25 tests. That's a couple times per week if you're a 7x per day tester. The iBGStar did similarly poorly.

Two meters were within 10% on all 200 tests. Those were the Accu Chek Active  and the Freestyle Lite. Of the two, the Freestyle Lite was within 5% a larger portion of the time. That means, if your blood sugar was 200, it should tell you a number between 180 and 220 every single time. At least 4 times out of 5, these two would give you a number between 190 and 210. That's pretty good.
The Accu Chek Active is so old that the strips change color depending on your blood sugar, and it takes a 1 microliter blood sample.
The Freestyle lite takes 0.3 microliters and has the modern standard of a 5 second wait time.

The way I see it, if you want to think about what accuracy is okay, corrections are what's most important. I'm gonna correct for the number I see. I need to be pretty sure that the number I see represents a narrow enough range that if I aim for a blood sugar of 110 with my correction, I'm not likely to go lower than 70- that means the range that the bg I see might actually represent would ideally be less than 40. At a blood sugar of 200, a 10% error means a 40 point range in blood sugar possibilities. The meter I use, the Accu Chek Aviva, missed this mark 9% of the time. That's 1 in 11 tests.  If you test 7x per day, that's about 4 times per week.

Actually, that's just what their stats suggest. If you look at the graphs, you can see it even better. The accu chek aviva was actually off by more than 40 mg/dl in a few of their tests, and all the tests they did where the Aviva was off by more than 40 mg/dl, the Accu Chek read lower than the real value. This means that there isn't really as much risk of over correcting as I might worry; there's a much bigger risk of undercorrecting.
Bayer Contour, on the other hand, did read more than 40 mg/dl higher than the actual value in multiple instances in the study. The Freestyle Lite had a tendency to read low and didn't once read more than 15 mg/dl higher than the actual value. At very high values it tended to really under read numbers. That's something I'd be comfortable correcting off of. At blood sugars like 300, the iBGStar read more than 60 mg/dl high. Not cool.

There's also a chart in this study about the percentage bias different meters tend to have.

Like I said, a must read. Please click on that link!