I'm trying to figure out this problem from Applied Survival Analysis by Hosmer, Lemeshow, and May, 2nd edition. Chapter 7, #3
In the GBCS data, does hormone use improve survival after cancer recurrence, controlling for tumor grade and size?
The data information can be seen here
I am confused about how to handle the data. There are two censoring variables: recurrence and death. I am not sure how to handle the subjects who do not have recurrence.
The example defines TWO separate time-to-event distributions: 1. Time to recurrence 2. Time to death (i.e. survival)
Assuming #2 starts at the same time as #1, you should first subset the data so only patients who have a recurrence are included and then define a new time-to-event variable whose time is survtime-rectime and whose flag is simply censdead. This gives you time from recurrence to death.