In article <37uv7v$nrt at www.interramp.com>, <pp000165 at interramp.com> wrote:
>>Brian: But it was you who originally said, unless I am
>mistaken, that the elisa hiv test can give false +s 98%
>of the time? in the context of testing donor blood from
>your blood bank?
>Kevin.
So? This is, contrary to what one might conclude from
a casual glance, completely compatible with the HIV
test being one of the best tests ever invented. This
concept is not very easily transmitted with a few sentences
on the Net, and requires some relatively advanced
understanding of the theory behind diagnostic testing.
In an exceedingly low-prevalence setting (such as repeat
blood donors to a blood bank from a relatively low
HIV-prevalence area, who are also extensively screened for
high-risk behaviors), a test with 99+% sensitivity
and 99%+ specificity can *still* produce many more false
positives than true positives. The same is true of
any diagnostic test that is not perfect (in other words,
all of them).
Brian
