Is it reasonable to adjust for baseline covariates that will change with treatment?


I’ve seen the previous post (and many discussions out there) about the problems with adjusting for post-baseline covariates in a pre-post ANCOVA analysis of treatment groups. Are there any issues or considerations for adjusting for baseline variables that will be effected by the treatment?

For example, say we are testing the effect of a low-dose vs high-dose statin on carotid intima media thickness (CIMT). We measure CIMT before treatment, and CIMT one year later. We also measure LDL-C at baseline. Would there be any issues in testing post-CIMT between groups, adjusted for pre-CIMT AND pre-LDL-C?

Thank you in advance!



you mean something like this: Changing characteristics of statin-related cIMT trials from 1988 to 2006 They included baseline ldl and % reduction in ldl as covariates?: " The relationship between annualized mean mean change in cIMT and various specified variables including … baseline LDL value, percent LDL reduction, … was explored using a simple meta-regression model". It does not seem to make much sense to me. A joint modelling approach might be of interest and if these things are measured together at repeated visits then it could make a lot of sense…?

edit: i think i misunderstood your Q. Your approach sounds fine ie tenable ie adjust for baseline ldl