as_ped with CR cuts off times after the last event where status == 1L if using tdcs

[jonas-schropp]

If I try to use as_ped to build my competing risks data set with time dependent covariates, the follow up time is cut short at the last event time where status == 1L and any events for status > 1L (as well as censoring after that time, status == 0L) are ignored.

Example:

dat <- data.frame(
  id = factor(1:10),
  status = c(0L, 1L, 2L, 0L, 1L, 2L, 1L, 0L, 0L, 2L),
  time = seq(from = 10, to = 100, by = 10)
)

tdcs <- data.frame(
  id = factor(c(1:10, 1:10)),
  var = rbinom(20, 1, 0.3),
  tt = c(rep(0, 10), seq(5, 23, 2))
)


ped_dat <- as_ped(
  data = list(dat, tdcs),
  formula = Surv(time, status) ~ concurrent(var, tz_var = "tt"), 
  id = "id"
)

max(ped_dat$tend)

Returns [1] 70 when it should return 100.