Re: Question on comparing lab results


Richard Ulrich wrote:
> First recommendation: Visit a lab similar to your own, and
> see what they do. The process that you describe seems to
> hint at a slew of pitfalls.

I do plan on visiting some labs similar to mine to hopefully get some
training in several areas. As usual I've gotten myself into a
situation I don't have the background for! :)

The comparison method I described is not something I would recommend,
but is representative of that sort of thing that gets done here.

> For instance: You have "several tests" on each sample.
> It *seems* that you count the result as "correct" if-and-only-if
> each of those tests are within 10%.

I need to clarify what I meant here. Blood from an individual is sent
to this contract lab where 6 separate tests are performed on it to
determine the occurance of 6 separate disease states. Each test yields
a numerical value which is then converted into a "negative" (healthy),
"positive", or "indeterminate" result. I believe we would want to
examine the results of each test separately. We would then count a
result "correct" for Disease A if the numerical value behind it agreed
(according to some rule) with the numberical value we get when testing
the same blood sample for Disease A. From there I need to decide how
many samples the contractor must report "correctly" for us to consider
them proficient at testing for Disease A.

>Are the assays
> - equally variable (Does one *need* a 1% error limit?)
> - equally suspect (Does a complicated procedure lead to
> occasionally-huge errors?)
> - equally important?

While I know how each of the assays are being performed, I don't know
alot about how to evaluate the error associated with each. I'd doubt
they are all equally variable. At this point I do consider them all
equally suspect simply because of my lack of understanding of possible
error involved. Since they all test for serious in-born errors of
metabolism, I can say they're all equally important.

It definately sounds like I need more information but I think I'm even
more confused than I was before. This really bothers me because unless
I come up with a better plan, we're going to go with numbers pulled out
of the air and dedicate several full-time staff and $100K's in
equipment to retesting these samples in a way that may not tell us
anything of value!

Tom

.

Relevant Pages

  • New lab will probe bioterror tools
    ... New lab will probe bioterror tools ... Denver Post Staff Writer ... Disease Control and Prevention's Vector-Borne Disease Laboratory in Fort ... the old buildings - where vaccines for West Nile virus are ...
    (sci.med.diseases.lyme)
  • Re: Mystery Illness Continued (Perhaps Endocarditis)
    ... and strange red lesions on both forearms. ... It sounds like the tick borne disease that went undiagnosed in my case long enough to lead to lasting disability. ... it might be useful to have the leading edge biopsied and cultured by a lab with experience/success in culturing B.b. ... your Lyme serology may not have been as negative as you think; labs don't report all the bands specific for B.b unless your doctor demands that they do so. ...
    (sci.med.cardiology)
  • Plum island Mistake not to be repeated on Mainland?????
    ... Disease Center ... Montana Sen. Tester Said Animal Center Too Risky ... Lab 257: the Disturbing Story of the Government's Secret Plum Island ...
    (misc.news.internet.discuss)
  • Rural Montana prepares for major bio defense lab
    ... Rural Montana prepares for major bio defense lab ... Rocky Mountain spotted fever. ... some in the local community sued to prevent the building of ... disease and prion diseases, a type of disease that includes so-called ...
    (sci.med.diseases.lyme)
  • FEATURE-Rural Montana prepares for major bio defense lab (Lyme disease work done there)
    ... FEATURE-Rural Montana prepares for major bio defense lab ... Rocky Mountain spotted fever. ... some in the local community sued to prevent the building of ... disease and prion diseases, a type of disease that includes so-called ...
    (sci.med.diseases.lyme)