Back to the Top
I have recently been asked to analyse some toxicokinetic data (involving the
pooling of data from a number of animals to generate a series of mean conc v.
time values, which are then subsequently analysed in a non-compartmental
fashion) and have been faced with triplicate determinations of which one or
even two of the triplicates are below the LOQ for the assay, without being
"not-detected" per se. It has been suggested to me that, rather than leaving
the mean value for these points undetermined (thereby losing that point from
the graph altogether, and analysing the remaining points only) I instead assign
a value of half the LOQ to these determinations, and thus derive a mean for
inclusion in the graph and analysis. I am instinctively uneasy about this, but
in the context of TK analysis, where sampling is very limited, it would be of
enormous benefit to be able to ascribe a value to all the time points
available.
Does anyone have any comments on this approach?
Thanks in advance,
Ray French.
Back to the Top
It depends on what the LOQ of the assay is? Personally I prefer to use
that one particular concentration for which you have a value rather than
guessing for the samples with LOQ. Other words use one value that
have a real number and don't use take the value two LOQ samples LOQ
= ? of the lower limit of quantitation (assuming two out of three
determinations are LOQ).
This is my two penny thought on this subject.
Merry Christmas
Prasad Tata, Ph.D.
Otsuka America Pharmaceuticals, Inc.
>>> RAYF by way of David_Bourne12/19/96 11:59am >>>
PharmPK - Discussions about Pharmacokinetics
Pharmacodynamics and related topics
I have recently been asked to analyse some toxicokinetic data (involving
the pooling of data from a number of animals to generate a series of
mean conc v. time values, which are then subsequently analysed in a
non-compartmental fashion) and have been faced with triplicate
determinations of which one or even two of the triplicates are below the
LOQ for the assay, without being
"not-detected" per se. It has been suggested to me that, rather than
leaving the mean value for these points undetermined (thereby losing that
point from the graph altogether, and analysing the remaining points only) I
instead assign a value of half the LOQ to these determinations, and thus
derive a mean for inclusion in the graph and analysis. I am instinctively
uneasy about this, but in the context of TK analysis, where sampling is
very limited, it would be of enormous benefit to be able to ascribe a value
to all the time points available.
Does anyone have any comments on this approach?
Thanks in advance,
Ray French.
Also see: http://www.cpb.uokhsc.edu/pkin/pkin.html
Back to the Top
Certainly these points provide some information, an approximate
upper-bound if nothing else. Some statistical software will allow
their use as constraints. If you choose to assign a specific value,
recognize that the statistical weighting given this value must
reflect the uncertainty. - R. Scheyer
Richard Scheyer, M.D.
Dept. Neurology, Yale School of Med.
P.O. Box 208018, New Haven, CT 06520-8018 USA
Richard.Scheyer.aaa.yale.edu
Back to the Top
Mr. French,
In similar situations with clinical PK studies, I have ascribed zero to those
values.
Lester Gibbs
Penederm, Inc.
Foster City, CA
This message sent using the FirstClass SMTP/NNTP Gateway for Mac OS.
Back to the Top
An old problem. Our arbitrary solution is if at least one data point in the
set is above the LOD, we use the LOD for the remaining points. If the data
falls between the LOD and the LOQ we use the data. In either case it must be
clear that these points are estimates.
Want to post a follow-up message on this topic?
If this link does not work with your browser send a follow-up message to PharmPK@lists.ucdenver.edu with "TK data and LOQs." as the subject | Support PharmPK by considering my eBooks |
Copyright 1995-2014 David W. A. Bourne (david@boomer.org)