Dear Roger,
The "parametricist's" view of this would be to do the log distribution
and then antilog. In fact much of the predictive work that has gone on
using various distribution rely upon this "classical" statistical ploy
(?fudge) for sorting skewed distributions. The results tend to be fairly
similar to the simple non-parametric route. We have been implementing
Reference Change Values which assumes sd's rather than percentiles (or
other "iles"). We just gritted our teeth and got on with it.
Yours aye,
Henry Chandler
In message <[log in to unmask]>,
"Bertholf, Roger" <[log in to unmask]> writes
>Dear Brian:
>
>For our in-house reference range study involving 98 healthy volunteers, the
>97.5th percentile for ALT was 53 U/L, and the highest value was 69. The mean
>and median, however, were 18 and 14, respectively. Because the distribution
>of ALT measurements in healthy individuals is severely skewed,
>parametrically-determined reference intervals underestimate the upper limit
>of normal. I don't know whether Roche, or your neighboring laboratories,
>used the mean +/- 2 SD to establish their ALT reference ranges, but this
>could be one reason that you're seeing an unexpectedly large number of
>mildly elevated results.
>
>Kind regards,
>
>Roger
>
>Roger L. Bertholf, Ph.D.
>Associate Professor of Pathology
>Director of Clincal Chemistry & Toxicology
>University of Florida Health Science Center/Jacksonville
>
>
--
Dr Henry Chandler
formerly Chem Path St George's.
------ACB discussion List Information--------
This is an open discussion list for the academic and clinical
community working in clinical biochemistry.
Please note, archived messages are public and can be viewed
via the internet. Views expressed are those of the individual and
they are responsible for all message content.
ACB Web Site
http://www.acb.org.uk
List Archives
http://www.jiscmail.ac.uk/lists/ACB-CLIN-CHEM-GEN.html
List Instructions (How to leave etc.)
http://www.jiscmail.ac.uk/
|