Date: Fri, 28 Jul 2000 10:46:44 +0200
From: "Thomas Kerbusch" <Aptkr@slz.nl>
Subject: test of differences between posthocs
Dear NM-users,
We have a problem in the statistical analysis of posthocs. We are using a population model which describes autoinducible clearance using differential equations (ADVAN6). A dummy compartment is used to obtain posthoc estimations for the AUC. This population has infusion duration (groups 1, 2 and 3) as covariable. We would like to compare the posthocs of group 1, 2 and 3. An independent samples t-test is not allowed. Can one test differences of posthocs derived from independent populations? Are there any nonparametric tests we could use for comparison of posthocs?
With friendly regards,
Thomas Kerbusch
Ron Mathôt
Slotervaart Hospital/The Netherlands Cancer Institute
Louwesweg 6, 1066 EC
Amsterdam, The Netherlands
test of differences between posthocs
Date: Fri, 28 Jul 2000 08:53:49 -0400 (EDT)
From: "Jogarao Gobburu 301-594-5354 FAX 301-480-3212" <GOBBURUJ@cder.fda.gov>
Subject: Re: test of differences between posthocs
Hello,
1. If you believe that there is autoinduction, treating infusion duration as a covariate will limit the use of the model. It might be more mechanistic to model clearance as concentration dependent.
2. While it is not clear to me, as to the purpose of the whole modeling exercise (probably to prefer one duration over the others rather than deriving optimal duration/input-rate that might not have been studied) - you can use the nonparametric Kolmogorov-Smirnov's test to compare distributions of a test statistic (eg; AUC corrected for dose if necessary) from two populations. If you do have duration as a significant (by some standard) covariate, you probably already know that these groups are different.
Regards,
Joga Gobburu
Pharmacometrics,
CDER, FDA.