Get group effect over time at a specific level of a covariable and overall in mixed effects linear regression

[Martin Mueller]

Dear all

Two groups consisting each of multiple patients (pid) were tested on an ergometer on different performance scales over time. The continuous outcome was recorded over time.

* Example generated by -dataex-. For more info, type help dataex
clear
input float(pid group performance) double outcome
1 1 200  30
1 1 200  82
1 1 200 128
1 1 200  68
1 1 200 132
1 1 200 158
1 1 200 157
1 1 200 122
1 1 200 168
1 1 200 173
end
label values group group
label def group 1 "Group 1" 0 "Group 2", modify

I ran a mixed effects linear regression to incorporate the repeated measurement nature of the data and allowed the outcome to vary as a restricted cubic spline over time at each performance level.

This is my Stata code (Stata MP version 18.0):

mkspline timespl = time, cubic
mixed outcome i.group##c.timespl* i.performance  || pid: timespl*

margins i.group, over(time)
marginsplot, xdim(time) recast(line)  recastci(rarea) ciopt(color(%30)) legend(position(6) col(2))

I now like to answer the following questions:
1.) Is there an overall significant difference of the two curves, i.e. a group effect over time regarding the outcome?
2.) If there is a group effect over time: Are the curves different at the first performance level already, or at the second, or third, ...

How can I test this in Stata (margin command with dydx and at option? lincom command? test command?)? I am not sure.
Thank you for your help in advance.

Martin