Baseline Interaction

Hi Group,

I have a question that I’m hoping someone may be able to help with. A colleague has asked me for assistance and I am not able to provide a confident answer.

They have observational data for an eye parameter (outcome) measured at 3 subject ages (20, 27 and 28 yrs). There is only one group and they are interested in the change in the parameter over time. They plan to include additional covariates (otherwise you could just run a simple paired t-test).

My initial suggestion was to use a mixed model with all 3 measures (at each of the ages) as outcomes (in long format) and then run planned contrasts for differences at paired time points following this.

They now theorise that in fact the value of the parameter at baseline (20 yrs) is a potential modifier in the rate of change of the parameter over time (higher values at baseline resulting in faster reductions over time). They want to know if it’s appropriate to include an interaction term between the baseline measure and age (time). I can’t see how to do this without duplicating the baseline as a covariate also as the first outcome measure.

Is there something simple I’m missing in how to assess the question of baseline interaction with time? I presume the simple inclusion of the baseline as a covariate and using change scores as the outcome does not address this?

Thanks for any help,

Paul

1 Like