Re: Calculation of the 90% CI values from pop PK model estimates

From: "Gastonguay, Marc" marcg@metrumrg.com Subject: Re: [NMusers] Calculation of the 90% CI values from pop PK model estimates Date: Fri, 16 Sep 2005 07:24:39 -0400 Alan, Instead of writing your model as: CL=theta1*{1+theta2*DDI1+theta3*[(WT/WT0)**theta4-1]}*exp(eta1) You might consider re-estimating with a parameterization that is more closely suited to your purpose, such as: CL=(THETA(1) * THETA(2)**DDI1 * (WT/WT0)**THETA(3) ) * EXP(ETA(1)) ;where DDI1 is a 0,1 indicator variable ;THETA(2) is the DDI1 clearance ratio and is constrained with a lower bound of 0, initial estimate could be 1 Also include this code: CRW= (140/WT0)**THETA(3) )/(70/WT0)**THETA(3) ) ;clearance ratio for WT (140:70kg) in absence of DDI1 Run several hundred or more stratified nonparametric bootstrap estimation runs (shouldn't be too bad for a stable 90% CI, compared to the 95% CI we are usually interested in) to obtain empirical 90%CI on all parameters and derived variables of interest (specifically THETA(2) and CRW). You'd have to invert the clearance ratio to get the AUC ratio, but I think you get the idea... Marc --------------------------------- Marc R. Gastonguay, Ph.D. Metrum Research Group LLC www.metrumrg.com _______________________________________________________