NONMEM Users Network Archive

Hosted by Cognigen

RE: Interoccation variation question

From: TKT <tkt>
Date: Mon, 6 Apr 2009 15:25:18 +0200

Dear Nick,

Thank you for this which answers the question re. how to parameterize and i=
s in line with what
I have been doing so far. Basically I was wondering if the fact that all su=
bjects in trial 1 only have one occasion could
in any way represent a problem for separation and estimation of the two ETA=
s for those subjects, and if it would make any sense to
only let subjects with data from >1 occasion contribute to the estimation o=
f Eta2 and Eta3 (in your version of the code).


From: Nick Holford [mailto:n.holford
Sent: 6. april 2009 14:50
To: TKT (Thomas Klitgaard)
Subject: Re: [NMusers] Interoccation variation question


In trial 1 I would give all subjects an OCC=1. In trial 2 give each subje=
ct an OCC=1 for the first week of PK sampling and OCC=2 for the third w=
eek i.e. 2nd PK sampling occasion.

I would use this code (with KA as an example):

0.5 ; BSVKA
0.5 ; BOVKA1


You can use this code for estimation and simulation.

I don't really understand your questions so perhaps you could try asking th=
em again based on the above advice.


TKT (Thomas Klitgaard) wrote:
Dear colleagues,

I'm currently working on a population PK analysis where I probably need to =
incorporate interoccation variation on one or more parameters. However,
I'm wondering about one issue. To put it short, here's the problem:

Trial 1: 1 dosing occation, PK observed for 1 week
Trial 2: 3 dosing occation (1 x week), not full washout between, PK observe=
d for 1 week after dose 1 and after dose 3. PK varies betw. wk 1 and 3. wit=
hin subject.

Question: Should

1) occasion flag OCC=1 for all subject in trial 1 and OCC=1 (week1) and=
 2 (week 3) in trial 2? Or should
2) occasion flag OCC be omitted for trial 1 subjects and set to 1 or 2 as i=
n 1) for trial 2.

On one hand, 1) would strictly speaking seem correct as these subjects have=
 only a dose 1 occation. On the other hand, my feeling is that NONMEM would=
put in trouble with seemingly always two inseparable individual ETAs in tri=
al 1 subjects for parameters with IOV on. (I'm using FOCEI).

IF (OOC.EQ.1) Ka = Ka*EXP(ETA(Occ_1))
IF (OOC.EQ.2) Ka = Ka*EXP(ETA(Occ_2))

If 2) is the better approach of the two, would the simulation model not sti=
ll be Ka=TVKA*EXP(ETA(IIV_Ka)+ETA(Occ)) for any subject, regards of no. o=
f occasions?
(Ie. above manoeuvre would be only to avoid an estimation problem).


Thx. for any input



Thomas Klitgaard
Principal Scientist

Novo Nordisk A/S
Krogshøjvej 53 A
DK-2880 Bagsværd
+45 4442 4960 (direct)
+45 3079 4960 (mobile)

Changing possibilities in haemophilia
Novo Nordisk is focused on the quality of life of people living with haemop=
hilia; for this reason our pipeline continues to evolve, it is why we conti=
nue to strengthen our existing formula. Learn more<http://www.novonordisk.c=

This e-mail (including any attachments) is intended for the addressee(s) st=
ated above only and may contain confidential information protected by law. =
You are hereby notified that any unauthorized reading, disclosure, copying =
or distribution of this e-mail or use of information contained herein is st=
rictly prohibited and may violate rights to proprietary information. If you=
 are not an intended recipient, please return this e-mail to the sender and=
 delete it immediately hereafter. Thank you.

Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New Zealan=
0 fax:+64(9)373-7090

Received on Mon Apr 06 2009 - 09:25:18 EDT

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to:

Once subscribed, you may contribute to the discussion by emailing: