I have a OneTouch Ultra meter system.  I understand that they claim meter
accuracy of +/- 20%.  My problem started when I started a new vial of strips
and it seemed like my BG's went up 5-10 points.  At one point I double
checked my BG on a contour meter and found 5point lower reading, which would
have been more my expected reading.  So, I did some analysis:

Code Number: 15
Range of Values with Check Solution: 99 - 132

The average value of this range is 115.5

So, I ran a check strip (actually ran 2 but didn't shake the liquid whic, as
it turns out, is very important) and got 123.  My conclusion: at a BG of
115.5 I will read 123, or 7.5 points higher (6.5% higher).

If I translated my readings by the 6.5% mean offset I discovered that my
numbers were just like the previous vial.  I didn't check strip the previous
vile but I did check strip one from the same box and got 118, very close to
the mean.

My conclusion is that I will have elevated readings for the next 75-99
readings due to this error and the fact that all strips in a box will most
likely be similar.

I know what you are thinking, there is error in the meter... yadda yadda
yadda.  Shure, but I say it is small because the range in values for the
check solution for a single code already represents a +/- 14.3% variation.
If the +/-20% is held it allows for a fairly small variation for the meter.
Also, the +/- 20% is held across all meters so I think that any 1 given
meter should be more accurate on a sample to sample basis than if you used a
different meter for each reading.

Another check I made was to use the OneTouch for a reading and got 116.  I
used the contour meter at the same time and got 111.  The computed value,
based upon the 116 should have been 108.  Pretty close, even with meter to
meter variation.

Why is this important?  2 reasons:
1. Anal Retentive data takers like me notice a rise of 5-7 points when we
spot it.  I hate the bigger numbers and am striving to get my averages down.
I am serious about this.
2. Low numbers... Mine have never made me hypo, yet!  My previous low was
85.  I had one number today (after missing lunch and a metformin pill) of
87.  87 doesn't bother me but if my theory is correct, an 87 computes to an
81, which would be an all-time low.

Note that, with a different vial of strips which creates low readings you
could get fooled as well.  What do you think?  Is my new "computed BG"
method based upon test strip performance and check solution worthy of
consideration?  I definitely think there is something to it.

I am compiling a spreadsheet of codes/ranges/readings and a conversion
table.  So far I am just starting.  In a few months I should have more data.
-Sax
I know I am nuts with this stuff, I'll get over it eventually.