NONMEM Users Network Archive

Hosted by Cognigen

Re: Time-varing covariate

From: Alison Boeckmann <alisonboeckmann>
Date: Wed, 28 Aug 2013 11:26:01 -0700

Nick,

Thanks for the suggestions.

Stuart supplied another example in NONMEM Uses Guide VI (PREDPP) p. 57:


A value of 2 indicates that the event is neither an observation nor
dose event. The corresponding event is referred to as an other-type
event. Examples of other-type events are: the time a urine
collection begins, the time a urine collection ends, and the time a
change in a covariable (such as glomelular filtration rate) is noted.
The user may create an other-type event for whatever reasons he wishes;
he need only mark an occurence of this type of event with an event
record containing an EVID data item equal to 2.


On Wed, Aug 28, 2013, at 12:53 AM, Nick Holford wrote:

Alison,
I think the problem with the on-line help arose because a relatively
inexperienced nmuser was searching through the help to find some clues
on what to do. The use of the "physiological variable changes"
expression to describe an EVID=2 event seems to have been interpreted
as something special within NONMEM that knew about physiological
changes. Of course, this was a misunderstanding.
To avoid the misunderstanding I suggest you make it clearer that the
change in a physiological variable is just an example of a covariate
change at a non-dose and non-observation event time e.g.
"Examples of other-type events are: A compartment is turned on or off
(CMT specifies which compartment is to be turned on or off); a
prediction is obtained at a specified time so that it may be displayed
in a table or scatterplot ; some event occurs at a different time
than any observation or dose event e.g. a covariate such as weight
changes, an intervention such as hemodialysis is started or stopped."
Adding more specific examples of the use of EVID=2 would perhaps be
useful. Does anyone have any other examples?
I also suggest removing reference to PCMT "(PCMT specifies the
compartment from which the prediction is obtained)" because it is not
directly relevant to EVID=2. An inexperienced user might interpret the
remark about PCMT to imply that PCMT is required for use with EVID=2.
In my own experience I have never found the need to use PCMT. I usually
do not rely on the default compartment with complex models but use the
compartment explicitly in $ERROR to define the prediction I want to
output.
Best wishes,
Nick
On 27/08/2013 10:45 p.m., Alison Boeckmann wrote:

There have been a number of interesting comments.
The original issue has to do with the way this is described in on-line
help for EVID.
Would it be more clear if this said:

  a physiological variable changes (and this is at a different time
  than any observation or dose event).


Or can someone suggest a better wording that would not add to the
confusion?

On Fri, Aug 23, 2013, at 10:51 AM, siwei Dai wrote:

Hi, Nick:



Thank you for the response.



I meant to say EVID = 2 but not '4', my mistake. In the user guide, it
says:

 2 Other-type event. The DV data item is ignored. Dose-related
      data items must be zero. Examples of other-type events are: A
      compartment is turned on or off (CMT specifies which compartment
      is to be turned on or off); a prediction is obtained at a speci-
      fied time so that it may be displayed in a table or scatterplot
      (PCMT specifies the compartment from which the prediction is
      obtained); a physiological variable changes.



I am asking the question because I thought that usually the covariates
stay the same, but I want to add a covariate that changes during the
day, so every observation line will have a different covariate value.



If I understand your email correctly, I don't need to do anything
special to treat this type covariates then?



Thanks!



Best regards,



Siwei



On Fri, Aug 23, 2013 at 1:10 PM, Nick Holford
<[1]n.holford

  Siwei,
  I don't know why you think this complicated. Suppose you have age
  (AGE) as a covariate. This must of course be a time varying
  covariate if it is intended to be the current age. And you might
  have weight (WT) or creatinine clearance (CLCR) as covariates which
  typically change with time. So just code the $INPUT data items and
  use them as you wish e.g.
  $INPUT ID TIME AGE WT CLCR etc
  ...
  $PK
  ; CL=(CLnon-renal*f(age) + CLrenal*f(renal_function)) * allometric
  WT
  CL=(THETA(1)*EXP(THETA(2)*(AGE-40)) +
  THETA(3)*CLCR/100)*(WT/70)**0.75
  EVID=4 has nothing to do with using time varying covariates.
  Perhaps you could explain more clearly what your problem is and why
  you think it is complicated to use time varying covariates?
  Best wishes,
  Nick

On 23/08/2013 6:00 p.m., siwei Dai wrote:

  Hi, Dear NMusers:
  I want to add a time-varing covariate in my model. For example,
  blood pressure or blood flow as covariates. But I am not sure how to
  do it. I see some earlier threads to discuss it but they all use
  complicated methods.
  I am wondering if there are any new way to do it in NM 7.2? I see
  in the user guide that EVID=4 can indicate physiological change. Is
  this what I should use?
  Thank you very much for any suggestions.
  Best regards,
  Siwei

  --
  Nick Holford, Professor Clinical Pharmacology
  Dept Pharmacology & Clinical Pharmacology, Bldg 503 Room 302A
  University of Auckland,85 Park Rd,Private Bag 92019,Auckland,New
  Zealand
  office:[2]+64(9)923-6730 mobile:NZ [3]+64(21)46 23 53 FR [4]+33(7)85
  36 84 99
  email: [5]n.holford
  [6]http://holford.fmhs.auckland.ac.nz/
  Holford NHG. Disease progression and neuroscience. Journal of
  Pharmacokinetics and Pharmacodynamics. 2013;40:369-76
  [7]http://link.springer.com/article/10.1007/s10928-013-9316-2
  Holford N, Heo Y-A, Anderson B. A pharmacokinetic standard for
  babies and adults. J Pharm Sci. 2013:
  [8]http://onlinelibrary.wiley.com/doi/10.1002/jps.23574/abstract
  Holford N. A time to event tutorial for pharmacometricians. CPT:PSP.
  2013;2:
  [9]http://www.nature.com/psp/journal/v2/n5/full/psp201318a.html
  Holford NHG. Clinical pharmacology = disease progression + drug
  action. British Journal of Clinical Pharmacology. 2013:
  [10]http://onlinelibrary.wiley.com/doi/10.1111/bcp.12170/abstract



--

Alison Boeckmann

[11]alisonboeckmann

--
Nick Holford, Professor Clinical Pharmacology
Dept Pharmacology & Clinical Pharmacology, Bldg 503 Room 302A
University of Auckland,85 Park Rd,Private Bag 92019,Auckland,New Zealand
office:+64(9)923-6730 mobile:NZ +64(21)46 23 53 FR +33(7)85 36 84 99
email: [12]n.holford
[13]http://holford.fmhs.auckland.ac.nz/

Holford NHG. Disease progression and neuroscience. Journal of Pharmacokinetics a
nd Pharmacodynamics. 2013;40:369-76 [14]http://link.springer.com/article/10.1007
/s10928-013-9316-2
Holford N, Heo Y-A, Anderson B. A pharmacokinetic standard for babies and adults
. J Pharm Sci. 2013: [15]http://onlinelibrary.wiley.com/doi/10.1002/jps.23574/ab
stract
Holford N. A time to event tutorial for pharmacometricians. CPT:PSP. 2013;2: [16
]http://www.nature.com/psp/journal/v2/n5/full/psp201318a.html
Holford NHG. Clinical pharmacology = disease progression + drug action. British
Journal of Clinical Pharmacology. 2013: [17]http://onlinelibrary.wiley.com/doi/1
0.1111/bcp.12170/abstract

References

1. mailto:n.holford
2. tel:%2B64%289%29923-6730
3. tel:%2B64%2821%2946%2023%2053
4. tel:%2B33%287%2985%2036%2084%2099
5. mailto:n.holford
6. http://holford.fmhs.auckland.ac.nz/
7. http://link.springer.com/article/10.1007/s10928-013-9316-2
8. http://onlinelibrary.wiley.com/doi/10.1002/jps.23574/abstract
9. http://www.nature.com/psp/journal/v2/n5/full/psp201318a.html
  10. http://onlinelibrary.wiley.com/doi/10.1111/bcp.12170/abstract
  11. mailto:alisonboeckmann
  12. mailto:n.holford
  13. http://holford.fmhs.auckland.ac.nz/
  14. http://link.springer.com/article/10.1007/s10928-013-9316-2
  15. http://onlinelibrary.wiley.com/doi/10.1002/jps.23574/abstract
  16. http://www.nature.com/psp/journal/v2/n5/full/psp201318a.html
  17. http://onlinelibrary.wiley.com/doi/10.1111/bcp.12170/abstract
--
  Alison Boeckmann
  alisonboeckmann


Received on Wed Aug 28 2013 - 14:26:01 EDT

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to: nmusers-request@iconplc.com.

Once subscribed, you may contribute to the discussion by emailing: nmusers@globomaxnm.com.