I am performing a Kaplan–Meier estimator to determine the probability of remission time before a relapse. I have seen many examples, typically using cancer as an example. These focus on the singular time to event being a relapse or a death event over a group given a specific treatment. Everything else is considered as lost to follow-up aided by the fact that the observation usually has an end point.

My study differs in that I am performing a recurrent survival estimator (https://www.rdocumentation.org/packages/survrec/versions/1.2-2). My disease of interest can result in a pattern of remission into relapse into remission, etc. Each successive time to event is a relapse after the patient has remitted (a further dose of a particular drug of interest). Every event (relapse) is marked as a 1 in the R code, and all lost to follow-up are 0.

I would like to know how does one treat a "cure" status for survival analysis when the patient can still get the same disease a few years later? For example: if a patient has a migraine, takes meds, goes into remission and then relapses after two months, this could be considered as the same disease. Therefore the time till event would be two months between the two episodes (minus the time on the medication). However, if remission is two, three, or even more years apart, it is highly likely that these events are independent and therefore it would be inappropriate to take the time between each event as a remission time.

I would appreciate any thoughts on this. Thanks.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.