Cox Model on Quantile Bins with an Explicit Missing Level

Fits `Surv(time, status) ~ bins` where the bins come from [cut_with_missing_category()] and reports each level against the reference (the lowest bin).

Usage

tbl_cox_binned_missing(
  data,
  time_col,
  status_col,
  feature,
  n_bins = 3,
  labels = NULL,
  missing_label = "Missing",
  reference = NULL,
  subset_mask = NULL,
  conf_level = 0.95
)

Arguments

data

A data frame.

time_col, status_col

Survival time and event indicator.

feature

Numeric feature column.

n_bins, labels, missing_label, reference, subset_mask

See [cut_with_missing_category()].

conf_level

Confidence level.

Value

A tibble with one row per level: `level`, `n`, `events`, `hr` (1 for the reference), `conf.low`, `conf.high`, `p.value` (`NA` for the reference) and `reference`. The Cox fit is in the attribute `fit`.

See also

[cut_with_missing_category()], [cox_single()]

Examples

set.seed(1)
d <- data.frame(time = rexp(200, 0.1), status = rbinom(200, 1, 0.6), x = rnorm(200))
d$x[sample(200, 60)] <- NA
tbl_cox_binned_missing(d, "time", "status", "x")
#> # A tibble: 4 × 8
#>   level       n events    hr conf.low conf.high p.value reference
#>   <chr>   <int>  <int> <dbl>    <dbl>     <dbl>   <dbl> <lgl>    
#> 1 Low        47     25 1       NA         NA     NA     TRUE     
#> 2 Med        46     27 0.935    0.541      1.62   0.810 FALSE    
#> 3 High       47     25 0.951    0.545      1.66   0.859 FALSE    
#> 4 Missing    60     42 0.966    0.586      1.59   0.892 FALSE