Measure development: sensitivity to change analysis

Hello! I am trying to design a study to evaluate evidence for sensitivity to change of an instrument. We know that the trait to be measured (children's neurological development, essentially) changes with maturation. Unfortunately, we don't have a gold-standard intervention to measure pre- and post-scores on the instrument. Therefore, we will be collecting data from the same children at two (perhaps 3) time points, spaced by 1 year each. We will also have test-retest data that we can use to estimate the day-to-day error. My statistics knowledge is still fairly basic, but this is what I think I need to do. Thoughts?

I think I need to propose a mixed effects model where time is a fixed effect and original score is a random effect. I also suspect that change slows as children get older, so I'd like to include an interaction effect for age. I think we need to somehow include the expected error from test-retest reliability, but I'm not exactly sure how to phrase that. Does any of this sound right?

Thanks so much for your help!