Determination of PK parameters Mixed-effects population models

Mixed-effects population models 11 were fitted to the log-transformed propofol blood and plasma concentration versus time data. The program NONMEM version 6 12 was used. Models were fitted using the first order conditional estimation method with interaction between the interindividual error terms (ETAs) and the random residual error term allowed. Random residual error was described using an additive error model. A Compaq Digital FORTRAN V6.6 compiler was used with an Intel vPro 1.9GHz CPU (Intel, Santa Clara, USA) under Microsoft Windows XP Professional (Microsoft, Seattle, WA, USA). The data were first applied to twoand three-compartment mamillary models. A two-compartment model provided the best description of the basic model structure. This model was developed by adding ETA parameters to each structural parameter until no further model improvement could be justified, according to the goodness-of-fit criteria described below. A proportional variance model was used to describe the inter-individual variability.