Dear NMusers,
I would like to present you with a problem that some of you might have faced
before. I am currently building a PK model for a drug with once daily oral
administration. The drug has a long half-life and can be taken either in the
morning or at night. Unfortunately due to lack of consistency in the
recording of the administration times by the study team part of the patients
flagged to have taken the medication in the mornings were in fact taking it
at night (i.e. 08:25 corresponds to 20:25 the previous day). At this stage
there is no way of tracking back what the correct intake times were.
The data consists of 2 sparse measurements per patient and indeed looking at
individual profiles you can clearly see that some samples are from the
elimination phase, indicating the drug was taken more than 12h prior. I
could correct the dataset based on an empirical judgement backed up by the
visual analysis but I was hoping there would be a more robust way.
Is there any way to put a THETA on the dosage time for those patients to let
NONMEM estimate a corrected administration time?
I tried simply adding an extra parameter on the intake time for those
patients (placed the code in $PK):
IF(FLAG_AM.EQ.1) TIME=TIME-THETA(17)
But received the following error message:
AN ERROR WAS FOUND IN THE CONTROL STATEMENTS.
AN ERROR WAS FOUND ON LINE 30 AT THE APPROXIMATE POSITION NOTED:
IF(FLAG_AM.EQ.1) TIME=TIME-THETA(17)
X
THE CHARACTERS IN ERROR ARE: TIME
236 DATA ITEM MAY NOT BE CHANGED WITH THIS VALUE OF ICALL.
I would be grateful for some suggestions on how to deal with this issue.
Regards,
Andrzej
Andrzej Bienczak
MSc, MPharm, DiplPharm
Pharmacometrics Group
Division of Clinical Pharmacology
Department of Medicine
University of Cape Town
Estimation of Intake Time
Hi Andrzej,
you could add 12h to each of your observation times (leave dosing times as
is).
In NONMEM you'd do something like
TimeShift=12
IF(FLAG_AM.EQ.1) TimeShift=12 * 1/(1+EXP(THETA(17))) ;ensure TimeShift is
in between 0 and 12
ALAG1=TimeShift
If you want to go all the way, you could extend this with IOV or similar...
Kind regards
Sven Mensing
Quoted reply history
2015-06-18 15:35 GMT+02:00 Andrzej Bienczak <[email protected]
>:
> Dear NMusers,
>
>
>
> I would like to present you with a problem that some of you might have
> faced before. I am currently building a PK model for a drug with once daily
> oral administration. The drug has a long half-life and can be taken either
> in the morning or at night. Unfortunately due to lack of consistency in the
> recording of the administration times by the study team part of the
> patients flagged to have taken the medication in the mornings were in fact
> taking it at night (i.e. 08:25 corresponds to 20:25 the previous day). At
> this stage there is no way of tracking back what the correct intake times
> were.
>
>
>
> The data consists of 2 sparse measurements per patient and indeed looking
> at individual profiles you can clearly see that some samples are from the
> elimination phase, indicating the drug was taken more than 12h prior. I
> could correct the dataset based on an empirical judgement backed up by the
> visual analysis but I was hoping there would be a more robust way.
>
>
>
> Is there any way to put a THETA on the dosage time for those patients to
> let NONMEM estimate a corrected administration time?
>
>
>
> I tried simply adding an extra parameter on the intake time for those
> patients (placed the code in $PK):
>
> IF(FLAG_AM.EQ.1) TIME=TIME-THETA(17)
>
>
>
> But received the following error message:
>
>
>
> AN ERROR WAS FOUND IN THE CONTROL STATEMENTS.
>
> AN ERROR WAS FOUND ON LINE 30 AT THE APPROXIMATE POSITION NOTED:
>
> IF(FLAG_AM.EQ.1) TIME=TIME-THETA(17)
>
> X
>
> THE CHARACTERS IN ERROR ARE: TIME
>
> 236 DATA ITEM MAY NOT BE CHANGED WITH THIS VALUE OF ICALL.
>
>
>
>
>
> I would be grateful for some suggestions on how to deal with this issue.
>
>
>
> Regards,
>
>
>
> Andrzej
>
>
>
>
>
> Andrzej Bienczak
>
> MSc, MPharm, DiplPharm
>
> Pharmacometrics Group
>
> Division of Clinical Pharmacology
>
> Department of Medicine
>
> University of Cape Town
>
>
>
>
>
Hi Andrej, You can try something along the following lines:-subtract 12 hours
from the time of all doses in the dataset (create new time column)-add a fixed
ALAG of 12 hours and some variability on it. You could also fix the variability
to a big number. You can also test covariates (patient subgroups) on this ALAG.
Good luck, Mannie
Quoted reply history
From: Andrzej Bienczak <[email protected]>
To: [email protected]
Sent: Thursday, June 18, 2015 9:35 AM
Subject: [NMusers] Estimation of Intake Time
<!--#yiv0048360651 _filtered #yiv0048360651 {font-family:"Cambria
Math";panose-1:2 4 5 3 5 4 6 3 2 4;} _filtered #yiv0048360651
{font-family:Calibri;panose-1:2 15 5 2 2 2 4 3 2 4;}#yiv0048360651
#yiv0048360651 p.yiv0048360651MsoNormal, #yiv0048360651
li.yiv0048360651MsoNormal, #yiv0048360651 div.yiv0048360651MsoNormal
{margin:0cm;margin-bottom:.0001pt;font-size:11.0pt;font-family:"Calibri",
"sans-serif";}#yiv0048360651 a:link, #yiv0048360651
span.yiv0048360651MsoHyperlink
{color:#0563C1;text-decoration:underline;}#yiv0048360651 a:visited,
#yiv0048360651 span.yiv0048360651MsoHyperlinkFollowed
{color:#954F72;text-decoration:underline;}#yiv0048360651
span.yiv0048360651EmailStyle17 {font-family:"Calibri",
"sans-serif";color:windowtext;}#yiv0048360651 .yiv0048360651MsoChpDefault
{font-family:"Calibri", "sans-serif";} _filtered #yiv0048360651 {margin:72.0pt
72.0pt 72.0pt 72.0pt;}#yiv0048360651 div.yiv0048360651WordSection1 {}-->Dear
NMusers, I would like to present you with a problem that some of you might
have faced before. I am currently building a PK model for a drug with once
daily oral administration. The drug has a long half-life and can be taken
either in the morning or at night. Unfortunately due to lack of consistency in
the recording of the administration times by the study team part of the
patients flagged to have taken the medication in the mornings were in fact
taking it at night (i.e. 08:25 corresponds to 20:25 the previous day). At this
stage there is no way of tracking back what the correct intake times were. The
data consists of 2 sparse measurements per patient and indeed looking at
individual profiles you can clearly see that some samples are from the
elimination phase, indicating the drug was taken more than 12h prior. I could
correct the dataset based on an empirical judgement backed up by the visual
analysis but I was hoping there would be a more robust way. Is there any way
to put a THETA on the dosage time for those patients to let NONMEM estimate a
corrected administration time? I tried simply adding an extra parameter on the
intake time for those patients (placed the code in $PK): IF(FLAG_AM.EQ.1)
TIME=TIME-THETA(17) But received the following error message: AN ERROR WAS
FOUND IN THE CONTROL STATEMENTS. AN ERROR WAS FOUND ON LINE 30 AT THE
APPROXIMATE POSITION NOTED: IF(FLAG_AM.EQ.1) TIME=TIME-THETA(17)
X THE CHARACTERS IN ERROR ARE: TIME 236 DATA ITEM MAY NOT BE
CHANGED WITH THIS VALUE OF ICALL. I would be grateful for some suggestions
on how to deal with this issue. Regards, Andrzej Andrzej Bienczak MSc,
MPharm, DiplPharmPharmacometrics GroupDivision of Clinical
PharmacologyDepartment of MedicineUniversity of Cape Town
Dear Andrezj,
You could consider using a shifted lag time parameter and then estimating the
between subject variability (BSV) on this lag time parameter to account for the
uncertainty in the dosing time.
You then, would have to shift the current dosing times and time of drug
cocnentrations by 12 hrs.
Thanks.
Mathangi Gopalakrishnan, MS, PhD
Center fro Translational Medicine,
UMB
http://www.ctm.umaryland.edu
Quoted reply history
From: [email protected] [mailto:[email protected]] On
Behalf Of Andrzej Bienczak
Sent: Thursday, June 18, 2015 9:36 AM
To: [email protected]
Subject: [NMusers] Estimation of Intake Time
Dear NMusers,
I would like to present you with a problem that some of you might have faced
before. I am currently building a PK model for a drug with once daily oral
administration. The drug has a long half-life and can be taken either in the
morning or at night. Unfortunately due to lack of consistency in the recording
of the administration times by the study team part of the patients flagged to
have taken the medication in the mornings were in fact taking it at night (i.e.
08:25 corresponds to 20:25 the previous day). At this stage there is no way of
tracking back what the correct intake times were.
The data consists of 2 sparse measurements per patient and indeed looking at
individual profiles you can clearly see that some samples are from the
elimination phase, indicating the drug was taken more than 12h prior. I could
correct the dataset based on an empirical judgement backed up by the visual
analysis but I was hoping there would be a more robust way.
Is there any way to put a THETA on the dosage time for those patients to let
NONMEM estimate a corrected administration time?
I tried simply adding an extra parameter on the intake time for those patients
(placed the code in $PK):
IF(FLAG_AM.EQ.1) TIME=TIME-THETA(17)
But received the following error message:
AN ERROR WAS FOUND IN THE CONTROL STATEMENTS.
AN ERROR WAS FOUND ON LINE 30 AT THE APPROXIMATE POSITION NOTED:
IF(FLAG_AM.EQ.1) TIME=TIME-THETA(17)
X
THE CHARACTERS IN ERROR ARE: TIME
236 DATA ITEM MAY NOT BE CHANGED WITH THIS VALUE OF ICALL.
I would be grateful for some suggestions on how to deal with this issue.
Regards,
Andrzej
Andrzej Bienczak
MSc, MPharm, DiplPharm
Pharmacometrics Group
Division of Clinical Pharmacology
Department of Medicine
University of Cape Town