Laboratory of Clinical Pharmacokinetic, Bordeaux II University, France
The former investigation made the modeling of pharmacokinetic observations from 16 samples/patient possible. What are the parameter mean values if the number of samples decreases from 16 to 5, then to 4 and finally to 3 per patient? The same models with and without covariable have been studied. Optimizing the sampling is done via the D optimization process which is offered in the ADAPT II software (D.Z. D'Argenio and A. Schumitsky).
When pharmacokinetic information is produced by highly informative points (D optimization theory), the non linear model only requires 3 points per patient to be explained. This is completely validated by refering to the full model without covariable. However, the conclusion is less significant for the full model with covariables, as for this model 5 points per patient seem to be the right compromise.