# ETA vs. covariate plot

**URL:** <https://forum.certara.com/t/eta-vs-covariate-plot/3375>\
**Category:** Modelling and Simulation\
**Created:** [December 10, 2018, 10:48pm UTC](https://forum.certara.com/t/eta-vs-covariate-plot/3375 "2018-12-10T22:48:12Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![charvi](https://avatars.discourse-cdn.com/v4/letter/c/e47c2d/32.png) [@charvi](https://forum.certara.com/u/charvi)\
**Post date:** [December 10, 2018, 10:48pm UTC](https://forum.certara.com/t/eta-vs-covariate-plot/3375/1 "2018-12-10T22:48:12Z")

</div>

Hello All,

I have a PK parameter that exhibits a negative trend with a particular covariate that can be been seen for the ETA vs covariate plot (shows a downward trend). In cases like these, it seems Phoenix NLME gives a negative estimate for the exponent along with a negative CV%.

Please see the hypothetical example below that illustrates the issue:

For example, for CL = tvCL\*(BW/median BW)^dCLBW\*exp(nCL), the estimate and CV% for dCLBW comes out to be negative.

Has anyone observed this before? Should I change the covariate model structure or it is ok this way?

Thanks in advance for your help.

Regards,

Charvi

---

<div class="post-metadata">

**Author:** ![Samer\_Mouksassi](https://avatars.discourse-cdn.com/v4/letter/s/ce7236/32.png) [@Samer\_Mouksassi](https://forum.certara.com/u/Samer_Mouksassi)\
**Post date:** [December 11, 2018, 7:01am UTC](https://forum.certara.com/t/eta-vs-covariate-plot/3375/2 "2018-12-11T07:01:11Z")

</div>

Hi Charvi:

with this equation

CL = tvCL\*(BW/median BW)^dCLBW\*exp(nCL)

it is a power model of BW effect on tvCL if the exponent is negative that means CL decrease when BW increase which is not intuitive. Note that negative CV% is know issue just ignore the sign as phoenix does not know that for negative parameters we need to take the absolute value.

There is many things to consider to understand why the CL parameter seems to be decreasing with BW but without the data and a reproducible example there is little I can do to help you.

Samer

---

<div class="post-metadata">

**Author:** ![charvi](https://avatars.discourse-cdn.com/v4/letter/c/e47c2d/32.png) [@charvi](https://forum.certara.com/u/charvi)\
**Post date:** [December 11, 2018, 5:44pm UTC](https://forum.certara.com/t/eta-vs-covariate-plot/3375/3 "2018-12-11T17:44:43Z")

</div>

Hi Samer,

Thank you for your response.

I totally understand that this is a non-intuitive relationship. As I mentioned, this is only a hypothetical example that I gave to better illustrate the situation I am facing. This is not what I am actually observing, my relationship is for another PK parameter and some other covariate that I could not disclose in an open forum (hence the hypothetical example)

I was more concerned about the negative CV% and was not aware that it is a known issue in Phoenix NLME. Thank you for clarifying that for me.

Thank you once again for your prompt response, as always very helpful !

Regards,

Charvi