Skip to contents

Convert wide competing risks data to long (counting-process) format

Source: R/competing-risks.R
wide_to_long_CR.Rd

Splits the wide dataset at every censoring time, creating one or more rows per subject. The resulting long dataset is suitable for fitting Cox models for the censoring distribution and for use with add_ipcw_weights_cr().

Usage

wide_to_long_cr(
  dat,
  time_var = "t",
  event_var = "delta",
  covariate = "z1",
  cens_level = "censor",
  event2_level = "event_2"
)

Arguments

dat

A wide-format competing risks data frame. Must contain the columns specified by time_var, event_var, and covariate.

time_var

Character string. Name of the event/censoring time column. Default is "t".

event_var

Character string. Name of the event indicator column (factor with levels for censoring and the two event types). Default is "delta".

covariate

Character string. Name of the covariate column. Default is "z1".

cens_level

Character string. Factor level in event_var representing censoring. Default is "censor".

event2_level

Character string. Factor level in event_var representing the competing event (event type 2). Default is "event_2".

Value

A data frame in long format with columns id, delta, censor, event2_time, tstart, tstop, and the covariate column.

Examples

set.seed(42)
dat <- sim_data_cr(n = 100, censoring = "baseline")
dat_long <- wide_to_long_cr(dat)
head(dat_long)
#>   z1 id censor event2_time       tstart        tstop  delta
#> 1  0  1      0          NA 0.0000000000 0.0008870787 censor
#> 2  0  1      0          NA 0.0008870787 0.0236331037 censor
#> 3  0  1      0          NA 0.0236331037 0.0295500713 censor
#> 4  0  1      0          NA 0.0295500713 0.0724101320 censor
#> 5  0  1      0          NA 0.0724101320 0.0812450008 censor
#> 6  0  1      0          NA 0.0812450008 0.1824410379 censor