i have two measurement times for a creatinine score, at recruitment and 2 months later. i have 330 participants: 80 die before 2 months. 230 were able to complete the second measurement. in total, 120 of my 330 participants died.
When I carry out my dependent cox time analysis, the result gives me 80 deaths out of 540 and not 330.
is this normal? it's as if it considers each time interval as an ID, so what's the point of the cluster=ID?
Here's my code:
summary(coxph(Surv(tStart, tStop, DEATH) ~ Creat_score ,data =df, cluster = ID)
(this is an example but i obtain the same problem)