Back to the Top
The following message was posted to: PharmPK
Dear all,
I would like to have a clarity on identification of a
statistical outlier.
Is it mandatory to run a test like Lund's test to identify a
statistical outlier before conducting a statistical analysis
of PK parameters in a BE study?
Few of the expertees in this field have recommended to run
the tool to identify it in the begining and then discuss the
data with and without statistical outlier.
According to others, identifying a statistical outlier
should be discouraged and without identifying a statistical
outlier by any means, the results should be discussed.
Regards,
Dr Suhas Khandave
Back to the Top
Dear Dr. Suhas,
Before statistical analysis, PK parameter values are to be observed
carefully and should check whether any subject showing significantly
small
or large bioavailability by means of PK parameter. Once if any
observation
or subject is suspected to be outlier then only outlier test is to be
performed.
It is sure that on the basis of statistical test of outlier alone
subject is
not considered outlier and excluded from the stat analysis. A strong
scientific justification should be given for excluding the subject
from stat
analysis. results are to be presented with and without outlier.
u please ensure that the Lund test is not appropriate though it is
recommended by USFDA as our observations for different period are
correlated
with each other.
there are published literature on identifying outlier in BE studies.
you can
also go through a book:
"Chow & Lie - Design and Analysis of Bioavailability and bioequivalence
studies, 2nd ed."
Regards,
Dipesh Jayswal
Biostatistician,
BioArc Research Solution,
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)