In this trial there are 100 patients split into 4 groups, they are all given treatment A at time 0 and at different times, depending on various conditions, they are given treatment B. All patients receive A and B sequentially. There are some covariates.
I am happy analysing the effect of group on both the total time to treat and the time to change to B with stcox.
My concern is how to analyse the time from B to the end of the trial as this may be affected by the A time.
Is it acceptable to do a separate Cox regression on this time period, do I need to split the records (stsplit) or do I need to use the multistate (package multistate from http://fmwww.bc.edu/repec/bocode/m) or something similar?
Julie
I am happy analysing the effect of group on both the total time to treat and the time to change to B with stcox.
My concern is how to analyse the time from B to the end of the trial as this may be affected by the A time.
Is it acceptable to do a separate Cox regression on this time period, do I need to split the records (stsplit) or do I need to use the multistate (package multistate from http://fmwww.bc.edu/repec/bocode/m) or something similar?
Julie
