strange simulation results

Dear all, I have a big problem understanding some simulation results and hope that somebody can shed some light on what is happening. I have simulated a PK study using a previously developed PopPK model (very standard, 3-cmt model with some IIV and also inter-occasion variability on KA and F1). I wanted to use the model to design a new PK study. The aim of the new study design would be to be able to reliably estimate CL, V2, KA, and F1 when combining the data set with the data that is already available. I did some initial investigations with WinPOPT to obtain good sampling time points, and then later on simulated two different study designs: · Parallel design, 8 subjects with single dose of dose strength A and 8 subjects with single dose of dose strength B. Sampling time until 32 h after dosing. · Crossover design with 24 h interval between doses; 8 subjects with single dose of dose strength A and then single dose of B after 24 h, and the other 8 subjects with single dose of dose strength B and then single dose of A after 24 h. Sampling time points for each dose exactly like in the parallel design, with the exception that the first dose can only be sampled until 24 h after the dose, because then the next dose is administered. My problem: The parallel design looks great for estimation of CL and F1, whereas for V2 and KA it is imprecise (while still accurate). Now I thought that I might switch to the crossover design, because this contains a lot more information (basically each subject with two doses instead of only one, subject number the same). In theory, I expected a better estimation of KA, i.e. higher precision, because the design would enable the estimation of the IOV component as well. But what it seems like is that CL is estimated less precisely and for F1, a bias is introduced (simulated F1 of 1, estimated around 0.9). How can the parameter estimates become worse in the simulations when I only add information that should fit the model? Thanks and best regards Nele ______________________________________________________________ Dr. Nele Käßner PK/PD Sciences Modelling and Simulation (RDP/MS) Nycomed GmbH Byk-Gulden-Str. 2 D-78467 Konstanz, Germany Fon: (+49) 7531 / 84 - 4759 Fax: (+49) 7531 / 84 - 94759 mailto: [email protected]<mailto:[email protected]> http://www.nycomed.com/ County Court: Freiburg, Commercial Register HRB 701257 Chairman Supervisory Board: Charles Depasse Management Board: Stefan Brinkmann, Dr. Frederik Bendiksen, Gilbert Rademacher, Dr. Rainer Wiartalla -------------------------------------------------------------------- The content of this email and of any files transmitted may contain confidential, proprietary or legally privileged information and is intended solely for the use of the person/s or entity/ies to whom it is addressed. If you have received this email in error you have no permission whatsoever to use, copy, disclose or forward all or any of its contents. Please immediately notify the sender and thereafter delete this email and any attachments.