2
$\begingroup$
library(cmprsk)   
library(MASS)
data(Melanoma)
Melanoma$event <- NA
Melanoma$event[Melanoma$status == 2] <- 0
Melanoma$event[Melanoma$status == 1] <- 1
Melanoma$event[Melanoma$status == 3] <- 2

cif=cuminc(ftime=Melanoma$time, fstatus=Melanoma$event,cencode=0)


cifd <- data.frame(
  cause = factor(rep(c(2, 1),c(length(cif[[1]]$time), length(cif[[2]]$time)))),
  time  = c(cif[[1]]$time, cif[[2]]$time),
  cif   = c(cif[[1]]$est,  cif[[2]]$est)) 

this above sample code estimates cumulative incidence for competing risk (cif) and then makes a data frame (cifd) to house cumulative incidence values.

my questions are-

1, Is the risk over the complete study of event 1 = 0.3387175 in the presence of competing event 2 ? As indicated by cif?

2, Why is it that the cif duplicates each time in the list cif[[1]]$time for event 1 and event 2 in cif[[2]]$time ?

$\endgroup$
1
$\begingroup$

Hey @bvowe in terms of your first question, the result is

print(cif)
Estimates and Variances:
  $`est`
1000       2000       3000      4000      5000
1 1 0.12745714 0.23013963 0.30962017 0.3387175 0.3387175
1 2 0.03426709 0.05045644 0.05811143 0.1059471 0.1059471

$var
1000         2000         3000        4000        5000
1 1 0.0005481186 0.0009001172 0.0013789328 0.001690760 0.001690760
1 2 0.0001628354 0.0002451319 0.0002998642 0.001040155 0.001040155

which presents there is just one group with 2 events, where from est you can tell that cumulative incidence function for event 1 after 5000 days is 0.3387175. It is not improper to say it's a ``risk of event 1 in the presence of event 2` since that's what the cumulative incidence function shows https://clincancerres.aacrjournals.org/content/13/2/559

For you second question: estimates for both events are present for all the observed times, that's why the times are the same for 2 events.

| cite | improve this answer | |
$\endgroup$

Your Answer

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

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