From: "Ziad Hussein" <Z.Hussein@Medeval.Com>
Subject: NONMEM Analysis Strategy
Date: Fri, 17 Sep 1999 16:46:32 +0100

Dear Users,

As you know there are few strategies to follow when doing population PK/PD analysis. For example, the strategy I have been using for years is establishing the basic model, followed by univariate analysis, multiple regression analysis with forward selection and finally multiple regression analysis with backward elimination leading to a final optimal model. All this I have been doing using NONMEM.

I would appreciate any comments on this strategy and any recommendations.

I apologies if this topic was raised before and would be grateful referring me to such correspondence.

Yours sincerely,
Ziad Hussein

Dr Ziad Hussein
Senior Pharmacokineticist
Department of Pharmacokinetics & Data Management
Medevla Ltd
Skeleton House
Manchester Science Park
Lloyd Street North
Manchester M15 6SH
Tel: xx-44-(0)161-226 6525 ext. 297






Date: Fri, 17 Sep 1999 10:12:31 -0700
From: LSheiner <>
Subject: Re: NONMEM Analysis Strategy

You might want to look at Mandema JW, Verotta D, Sheiner, LB. Building population pharmacokinetic-dynamic models. I. Models for covariate effects, J Pharmacokin Biopharmaceut, 20:511-528, 1992., and papers since then by EN Jonsson and MO Karlsson.

Lewis B Sheiner, MD Professor: Lab. Med., Biopharm. Sci., Med.
Box 0626 voice: 415 476 1965
UCSF, SF, CA fax: 415 476 2796
94143-0626 email:






Date: 17 Sep 1999 14:49:13 -0400
From: "Ette" <>
Subject: RE: NONMEM Analysis Strategy

You may want to look up the following reference in addition to the ones provided by Dr. Sheiner:

Ette EI, Ludden TM. Population pharmacokinetic modeling: importance of informative graphics. Pharm. Res. 1995; 12: 1845 -1855.

Ene Ette.
Vertex Pharmaceuticals






From: NIclas Jonsson <>
Date: Thu, 23 Sep 1999 08:53:53 +0100 (IST)
Subject: NONMEM Analysis Strategy

One of the papers by EN Jonsson and MO Karlsson that Lewis talks about is "Automated covariate model building within NONMEM" (PharmRes 15(9) 1998, pp. 1463-1468). It describes a method with which the covariate model can be built with as much automation as possible (basically forward inclusion and backwards elimination for both liear and non-linear covariate relationships).

The procedure (written in perl) for doing this is available by anonymous ftp from in the files psn0.071.tar.gz and depending on you plattform.

Best regards,

Niclas Jonsson