Hi, There are both pros and cons of using baseline observations as a covariate. We presented four different options for how to handle baseline and investigated estimation properties in the paper below. What we didn't look at, because it is a very bad idea, is to use an baseline observation both as a covariate and dependent variable.
Approaches to handling pharmacodynamic baseline responses. Dansirikul C, Silber HE, Karlsson MO. J Pharmacokinet Pharmacodyn. 2008 Jun;35(3):269-83. Best regards, Mats Mats Karlsson, PhD Professor of Pharmacometrics Dept of Pharmaceutical Biosciences Faculty of Pharmacy Uppsala University Box 591 75124 Uppsala Phone: +46 18 4714105 Fax + 46 18 4714003 www.farmbio.uu.se/research/researchgroups/pharmacometrics/<http://www.farmbio.uu.se/research/researchgroups/pharmacometrics/> From: owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com] On Behalf Of Peiming Ma Sent: Tuesday, July 28, 2015 3:31 AM To: Mark Sale; Zhao,Li; nmusers@globomaxnm.com Subject: [NMusers] RE: Is it reasonable to add covarite to PD parameters kin or kout? Dear Mark and Li, I think there is nothing wrong with using baseline as a covariate. Baseline observations contain more information than just some pre-treatment observations and thus have a lot of explanatory power; this is the reason they appear significant more often than not. Comparisons after treatments should try to use all information before treatments (thus including baselines). Statisticians habitually use baselines as covariates in their ANCOVA models for obviously good reasons. Regards, Peiming From: owner-nmus...@globomaxnm.com<mailto:owner-nmus...@globomaxnm.com> [mailto:owner-nmus...@globomaxnm.com] On Behalf Of Mark Sale Sent: Tuesday, July 28, 2015 3:34 AM To: Zhao,Li; nmusers@globomaxnm.com<mailto:nmusers@globomaxnm.com> Subject: [NMusers] RE: Is it reasonable to add covarite to PD parameters kin or kout? Li, I'm going to go out on a limb (since I haven't seen your data or understand the biology) and suggest that the baseline value is not a covariate, but is just another observation (presumably with drug concentration = 0). The baseline value is sort of by definition a function of kin. So, you might look for predictors of Kin (age? Disease stage?) rather than using the observed value to predict the parameters (which then predict the observed value). Also note that covariates are, by definition, measured without error, and a baseline value is likely measure with error (the same error as any other observation?). Mark From: owner-nmus...@globomaxnm.com<mailto:owner-nmus...@globomaxnm.com> [mailto:owner-nmus...@globomaxnm.com] On Behalf Of Zhao,Li Sent: Monday, July 27, 2015 2:39 PM To: nmusers@globomaxnm.com<mailto:nmusers@globomaxnm.com> Subject: [NMusers] Is it reasonable to add covarite to PD parameters kin or kout? Dear NMusers, Right now I am doing covarite analysis for an indirect response model. I tested a few potential covarites and found it's STATISTICALLY significant if I add observed baseline value to kin. But I am not sure if it makes sense to add the baseline value as a covarite to kin. Could you please help me if you have had similar experiences before? Thank you very much!
