Suppose, a list of targeted population is invited to participate in a health program. Invitation date could vary for every individual based on eligible criteria (this criterion is the same for all). Everybody will be followed at the highest 365 days from the invitation date. My event of interest is early participation, that means who participated within 180 days from the date of the invitation. This creates the concept of time-to-event (e.g. time-to-participate) with a follow-up period of 180 days. Besides, there are three types of cases:

  1. participate within 180 days (early participation)
  2. participate between 181 days and 365 days (overdue)
  3. non-participate within 365 days (non-participation)

Following table is giving the snapshot data

enter image description here

Event date = end date for those individuals who participated with the period.

Data are right truncated? Or right censored-right truncated? Would it be possible to fit Cox regression or proportional hazard model for this problem? Do we have an appropriate statistical package in R/STATA?

  • 1
    $\begingroup$ It is right-censored data. you fit Cox proportional hazard model. If you want to use variable status as censoring indicator, overdue and early participation should use the same code because both of them are participant, i.e., status = 1 for overdue. I think you can use R to do it. $\endgroup$ – user158565 Jun 8 '19 at 6:48
  • $\begingroup$ Thank you. But my event of interest is "early participation within 180 days", but "overdue (>180 days)" is not a part of this. In this situation, would it be good to consider "overdue" and "non-participation" in a single group (i.e. status = 0)? $\endgroup$ – JRK Jun 8 '19 at 7:24

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.