Reply to J. Proost

摘要

On page 518, under simulation, I have described %CV added to Ka, volume and half-life. Concentration vs. time data were generated on Stella. Stella generates C vs. t data by changing PK parameters for individual subject (within the range of given %CV). In other words, each subject will have its own Ka, volume and half-life as seen in real-life situation. Once C vs. t data are generated in this way (where each subject has different pharmacokinetic parameters) it is not necessary (although it is commonly done) to add random error to the data. Therefore, both Tables 2 and 3 are realistic and meaningful. Addition of random errors to a data set generated by the addition of %CV on PK parameters may substantially increase the variability in the data which may not be realistic and true representative of real data. #