Hello everyone,
I'm running a data analysis to evaluate the association between a time-varying predictor (knee pain) and incident falls in the following year using mixed-effects complementary log-log regression models because the falls were assessed yearly. For a specific person, fall may occur at different years.
My question is: should I use the first outcome occurrence (and drop data thereafter) or allow repeated outcomes? And if the later is allowed in mecloglog, should I just assign incident outcomes as 1 at some timepoints and 0 at others? (see example below, kneepain is assessed prior to falls: e.g. the first row means kneepain at year 0)

ID falls kneepain year
1 0 1 1
2 1 1 2
3 0 0 3
4 1 1 4
... ... ... ...

Any idea? Thank you