Guidence regarding scm

Dear NM Users, I have developed a model and want to perform SCM by using CRCL, AGE and WT as continuous covariates and SEX as categorical covariate on CL, V1 and V2. Below is my control stream which I wrote for running the model. Can someone guide me is it Ok to use this control stream for performing stepwise covariate modelling. If there is mistake (which I hope there is with TVV1 = THETA(2) * WT and TVV2 = THETA(3) * WT) then how can I resolve this problem because when I omit influence of WT on V1 and V2 in model the results are worrisome. $PROBLEM PK $INPUT ID TIME DV AMT CMT MDV EVID WT CRCL AGE SEX $DATA Vanco_016_iv_1.csv IGNORE=@ $SUBROUTINES ADVAN3 TRANS4 $PK TVCL = THETA(1) * (1+THETA(6)*CRCL) TVV1 = THETA(2) * WT TVV2 = THETA(3) * WT TVQ = THETA(4) CL = TVCL*EXP(ETA(1)) V1 = TVV1 V2 = TVV2 Q = TVQ S1 = V1 $ERROR IPRED = F W=F*THETA(5) IRES=DV-IPRED IWRES=IRES/W Y=F+W*EPS(1) ; Combined residual random error $THETA (0,0.5,7) ; CL (0,0.3,0.9) ; V1 0.732 FIX ; V2 0.5 ; Q (0,0.44) ; Proportional error 0.01 FIX ; Proportional change in CL $OMEGA 0.07 ; IIV CL $SIGMA 1 FIX $EST SIG=4 MAX=9999 PRINT=5 NOABORT METHOD=1 INTER MSFO=MSFB1 $TABLE ID TIME DV AMT CMT MDV EVID WT CRCL AGE SEX IPRED IWRES ETA1 CWRES PRED NOPRINT ONEHEADER NOAPPEND FILE= mytab3 ;Xpose $COV Best regards -- *Muhammad Usman* PhD Scholar Westfälische Wilhelms-Universität Münster Institut für Pharmazeutische und Medizinische Chemie - Klinische Pharmazie - Corrensstrasse 48 48149 Münster Room No. A120.209 Phone: +49 251/8332171 Mob # +49 176 70962067

Muhammad, Based on your earlier email, I may suggest you to focus again on your results. Make sure, your parameter estimates, error estimates, RSE, CV, OFV, GOF are in accord. rRUN GAM once. When you are perform forward addition, add one covariate at a time on CL, based on the results add the covariates combination. Do the same for V1 (most likely wt only), sometimes you may not get WT too. But when you apply WT as a covariate on CL & Vd you may get better model. Your data may dictate WT as a covariate on V2. Bakcward elimination, p<0.001 wthi chi sq 10.83 will be good. Take a look at Nonlinear Mixed Effects Models:Theory chapter in Bonate's book.( https://books.google.com/books/about/Pharmacokinetic_Pharmacodynamic_Modeling.html?id=kfS5oLko95kC&hl=en) else chapter 5 section 5.7 in Jill Fideler-Kelly's book.

Dear NMUsers, I am using AGE, CRCL and Body weight as continuous covariates and Gender as categorical covariate for scm. When I start scm by using PsN the scm starts but there is a warning message stating "Warning: Individuals were found to have multiple values in the CRCL column, but CRCL was not set as time varying in the configuration file. Results may not be as expected". In my data file the values for CRCL are not similar for all the samples of few patients because there is a time difference of almost 72 hours between 1st and last sample. Can someone guide me is this a normal Situation or should I define CRCL in time_varying command in my configuration file. If so how can i write this command in configuration file....? Waiting for Kind suggestions.... regards -- *Muhammad Usman* PhD Scholar Westfälische Wilhelms-Universität Münster Institut für Pharmazeutische und Medizinische Chemie - Klinische Pharmazie - Corrensstrasse 48 48149 Münster Room No. A120.209 Phone: +49 251/8332171 Mob # +49 176 70962067

Dear Muhammad, The time_varying option is described on pages 8-9 in the scm userguide, see http://psn.sourceforge.net/pdfdocs/scm_userguide.pdf . Flagging a covariate as time varying will affect the computation of the median value of that covariate, and hence the centering of that covariate and the break-point of the hockey-stick parameterization. If the values of CRCL at different time points for the same individual do not differ much from each other, or if you do not use parameterizations where the median is important, then you can ignore the warning. If you have further questions regarding scm, I suggest you post them on the PsN-general mailing list. To subscribe please go to http://psn.sourceforge.net/list.php . Best regards, Kajsa On 06/01/2015 11:12 AM, Muhammad Usman wrote: Dear NMUsers, I am using AGE, CRCL and Body weight as continuous covariates and Gender as categorical covariate for scm. When I start scm by using PsN the scm starts but there is a warning message stating "Warning: Individuals were found to have multiple values in the CRCL column, but CRCL was not set as time varying in the configuration file. Results may not be as expected". In my data file the values for CRCL are not similar for all the samples of few patients because there is a time difference of almost 72 hours between 1st and last sample. Can someone guide me is this a normal Situation or should I define CRCL in time_varying command in my configuration file. If so how can i write this command in configuration file....? Waiting for Kind suggestions.... regards -- Muhammad Usman PhD Scholar Westfälische Wilhelms-Universität Münster Institut für Pharmazeutische und Medizinische Chemie - Klinische Pharmazie - Corrensstrasse 48 48149 Münster Room No. A120.209 Phone: +49 251/8332171 Mob # +49 176 70962067 -- ----------------------------------------------------------------- Kajsa Harling, PhD System Developer Department of Pharmaceutical Biosciences Uppsala University [email protected] +46-(0)18-471 4308 http://www.farmbio.uu.se/research/researchgroups/pharmacometrics/