Simulate competing risks survival data

Generates a dataset with two competing events under a sub-distribution hazard model parameterized by a four-level baseline covariate z1. Optionally adds independent or covariate-dependent censoring.

Usage

sim_data_cr(
  n = 100,
  censoring = "none",
  beta1 = log(1.5),
  beta2 = log(2.25),
  beta3 = log(3.4),
  p = 0.3
)

Arguments

n: Integer. Number of subjects to simulate. Default is 100.
censoring: Character string specifying the censoring mechanism. One of "none", "independent", or "baseline" (censoring rate depends on z1). Default is "none".
beta1: Numeric. Log sub-distribution hazard ratio for z1 = 1 vs z1 = 0 for event type 1. Default is log(1.5).
beta2: Numeric. Log sub-distribution hazard ratio for z1 = 2 vs z1 = 0 for event type 1. Default is log(2.25).
beta3: Numeric. Log sub-distribution hazard ratio for z1 = 3 vs z1 = 0 for event type 1. Default is log(3.4).
p: Numeric in (0, 1). Baseline probability of event type 1 in the reference group (z1 = 0). Default is 0.3.

Value

A data frame with columns:

z1: Four-level factor covariate (levels 0-3).
delta: Factor event indicator with levels "censor", "event_1", "event_2".
t: Event or censoring time.

Examples

set.seed(42)
dat <- sim_data_cr(n = 200, censoring = "baseline")
table(dat$delta)
#> 
#>  censor event_1 event_2 
#>      74      85      41