Back to the Top
Dear all,
could anyone give me a piece of advice how the problem of unblinding is tackled
when concentration data are to be analysed during the course of a study (to be
used for Population Pharmacokinetic Modelling)?
Thanks a lot in advance.
Harry Mager
Back to the Top
Dear Harry,
One way is to have the analytical staff seperated from the blinded
trial, and giving them a copy of the randomisation code list. This
way, they will not have to analyse the placebo samples. If this is
possible, the rest of the study team will still be blinded. Then, the
PK analysist doing the POP PK modelling will probably also have to be
unblinded. The question is really if you need to know anything about
the plasma levels DURING the study or if you can wait untill the
effect analysis has been finalised.
Best regards,
Thomas Senderovitz
Back to the Top
Harry,
I am also interested in this issue. I don't have your answer, but I would
like it if you could send me the responses that you get from the group.
A suggestion: Look at the recent population PK guidance by FDA (on their
website) and there is a discussion of real time data analysis.
Richard Lalonde
Back to the Top
I guess a lot depends on the design and the objectives (i.e., number of
co-variables, treatment groups, etc.) of the study as set forth in the
protocol. However, data cane be analyzed without unblinding the study.
Without going in to details, one option could be as follows:
1) analyze plasma samples from patients randomized to active and placebo
treatment groups (i.e., all patients), without knowing the randomization
code.
2) create the NONMEM dataset containing a MDV (missing dependent variable)
data item and treat data from patients showing no plasma levels as a non
observation record
3) confirm results when the study is eventually unblinded for efficacy and
safety reasons (never let the PK be the cause for the statistical penalty)
In past the above approach helped us. However, we had defined the
objectives a priori and they were simple. We also relied heavily on data-rich
Phase I and II results. Further, we only had three treatment groups.
Also, while doing these analysis, we had to be ultra-careful to preserve the
blindness of the study.
My 2 cents
Nik Singh
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)