# Re: [IP] One Touch Ultra and varying consecutive readings - comments?

```Curtis Lomax quoted someone's web site thusly:
>...it's important to understand how accuracy is defined.  A home blood
>glucose meter's test result is considered accurate if it falls within 120%
>of the lab test. Comparing a meter's test result with a lab test is the
>only valid method of determining it's accuracy

Hi, Curtis.

PRECISION, which is (according to another web site) "the ability [of a
device] to produce the same value or result, given the same input
conditions and operating in the same environment."

Precision estimates are shown on the One Touch Ultra strip insert. Here's
what I see in the version distributed here in Europe:

- on blood samples having a mean bg of 220 mg/dl, the CV (coefficient of
variation = standard deviation/mean) = 1.8%.

- on blood samples with a mean bg of 364 mg/dl, the CV was 1.6%.

I'll conservatively assume, then, that at the bg's Catherine was measuring,
the CV is 2%.

If the measurements are normally distributed about the mean (not a very bad
assumption), then the Ultra will measure +/- 3 std. deviations or +/- 6%
about the mean over 99.7% of the time. For a bg of 300, 6% is 18 mg/dl. We
could thus expect the analyzer to measure between 282 and 318 mg/dl over
99% of the time. Over 95% of the time, the analyzer would report +/- two
std. deviations, or between 288 and 312 mg/dl. Over two thirds the time,
the analyzer would report +/- one std. deviation, or between 294 and 306 mg/dl.

Catherine reported results of 280, 310 and 340 for what _she_ assumes to be
the same bg. Thus, the results Catherine reported are NOT typical of this
analyzer IFF (if and only if) samples with the same bg were indeed being
measured via an accredited procedure.

The cause of the variation she saw can be a combination of problems with
the samples (what was measured), the measurement procedure (how it was