Time to multiple medication discontinuations

[kc]

Hello,

I am using the following recurrent event model (counting process style input) to plot cumulative mean frequency of multiple medication initiations. Just like patients in the study can initiate multiple medications during an observation period of 2 years, they can discontinue multiple medications as well. 

I have the following questions - 

1. If one wants to summarize the discontinuations  during the 2 year period, what is it that you are trying to answer - Is it similar to initiations? If so, how would the code below change?

2. If not, what are some of the relevant questions  wrt. this type of data.

I am not very familiar with the medication data, so any pointers be it papers/code/links would be helpful.

Thanks!

proc phreg data=meds covs(aggregate) covm;
class trt(ref='B');
model (TSTART, TSTOP)*STATUS(0) = trt / rl;
hazardratio "A vs. B" trt / diff=ref;
baseline covariates=covd out=outd cmf=_all_/nomean;
id subjid;
run;