From my survival analyses, I get a survival curve showing a very large number of censures between 4000 and 4500 days, and then a considerable drop due to the decrease in numbers (mostly drop outs).


You can see that very few participants are present after 5000 days. Is it normal to set an end date for the analysis (e.g. at day 4500)? How can I implement this in my R function?

survfit(Surv(event_days, event) ~ sex, data = data)

Visually, I can just shorten the survival curve using xlim = c(0,4500).


1 Answer 1


The drops in survival curves at late times are NOT due to dropouts/censoring.

Each drop in a survival curve represents the fraction of individuals still at risk who experience an event at that time. At late times there are fewer individuals at risk so that the precision of survival estimates decreases and single events lead to greater visual drops in the survival curve. But the drops in the survival curves are because of individuals experiencing events, not dropouts.

There is no need to set an end date. You might choose only to display results out to a certain time. You might choose some specific late time for calculating a restricted mean survival time. But there's no reason to throw out data before you do your analysis of differences between groups.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.