> I am not convinced by Tim Reynolds argument about CV and number of
> analytes, although since he is a much better mathematician than I
> am, perhaps I could be persuaded. I agree that CV rises with the
> number of analytes, but this does not necessarily mean ...
> Brian Senior, Royal Bolton Hospital.
The "rule" that CV rises as more anaytes are measured is incorrect.
It depends HOW the measures are combined.
1) The SD of a combination of the form A+B+C
is = sqrt (var(A) + var(B) + var(C))
The CV (SD/mean) of A+B+C will be lower than the CV of A, B or C for
POSITIVE values of all but can be higher, lower or unchanged if
some values are negative.
2) For a combination of the form AB/C or A.B.C
the CV (note CV here) = sqrt (CV2(A) + CV2(B) + CV2(C))
where CV2 is the square of the CV
here the CV of the combination will usually be higher.
In the case of Downs screening the mathematics is vastly more complex
but I would have thought the CV would be lower if anything.
Aubrey Blumsohn
Aubrey Blumsohn
Directorate of Biochemical Medicine
Ninewells University Hospital, Dundee
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|