Data set:

  • 50000 participants
  • Assessments of various risk factors at baseline
  • Dates when participants were included in the study
  • Dates when participants died or were censored (after 2 years)

Calculating time differences in days between inclusion and death/censored results in a vector [0,1,2,3, ... , 370.5]

About 50 participants had a time difference of 0. Logically, assessments and inclusion must have happened before participant death. Performing survival analyses with event times = 0 presents technical difficulties. Would you exclude those individuals (my initial thought), or set their event times at 0.1 (or similar low value) to be able to include them? Or maybe set event times to 0.5?

  • 1
    $\begingroup$ I would be very careful (an understatement) with just eliminating them! Those patients could be dying under the operation, say, which would certainly be important information! and elimination of those could really give you problems later! So maybe just replace 0 days with 0.1 days? $\endgroup$ – kjetil b halvorsen May 1 at 13:33

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.