Skip to contents

FG-weighted Fine-Gray sub-distribution hazard regression

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

Fits a Fine-Gray model on the Fine-Gray split dataset, weighting by the probability of remaining uncensored after the competing event (p_notcens_after_death), and uses a robust sandwich variance via cluster(id).

Usage

fg_weighted_cr(
  data_long_fg,
  covariate = "z1",
  extend = TRUE,
  event1_level = "event_1"
)

Arguments

data_long_fg

A data frame in Fine-Gray format with weights, as returned by add_fg_weights_cr().

covariate

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

extend

Logical. If FALSE, data are truncated at the minimum of the stratum-specific maximum follow-up times before fitting. Default is TRUE.

event1_level

Character string. Factor level in the delta column representing the primary event. Default is "event_1".

Value

A matrix with one row per term and two columns: the log sub-distribution hazard ratio and its robust standard error.

Examples

set.seed(42)
dat <- sim_data_cr(n = 300, censoring = "baseline")
dat_long    <- wide_to_long_cr(dat)
dat_long_fg <- fg_split_cr(dat_long)
dat_long_fg <- add_fg_weights_cr(dat_long_fg, strat = "no")
fg_weighted_cr(dat_long_fg)
#>          coef robust se
#> z11 0.9135669 0.3975405
#> z12 1.3004739 0.3955882
#> z13 1.4468114 0.3928682