I have a group of subjects on supplement A for 3 years. Their blood pressure (BP) and kidney function, as measured by creatinine clearance (eGFR) were monitored at 1-, 2- and 3-year. In unadjusted models, BP decreased and eGFR increased significantly over time. It is well regonised that those who has a better kidney function (higer eGFR) will have a lower BP.

I would like to assess whether the changes in BP remained significant after controlling eGFR (to evaluate whether supplement A has direct effect on BP reduction or indirectly through eGFR improvement). Is it appropriate to use LMM with time-varying covariate for such analysis?

Thank you.

#edit: part of my text went missing earlier