gtsummmary::modify_header() with overall column

I would like to modify the column header formatting in a {gtsummary} table with a categorical varible (two levels) and an overall column, as in With gtsummary, is it possible to have N on a separate row to the column name?. I'm having trouble figuring out the correct gtsummary variable name to access the various column names. I currently have a workaround in which I first modify the categorical levels headers, and then add the overall and modify it. However, I'm wondering if there is a better way to do this in a single modify_header line. Reprex with various attempts below.

library(gtsummary)
library(dplyr)

# Shorten trial for examples
trial <- select(trial, trt, age)

# Modify headers
trial %>%
  tbl_summary(by = trt) %>%
  modify_header(update = all_stat_cols() ~ "**{level}**<br>N = {N}") %>%
  as_kable()

Characteristic	Drug A N = 200	Drug B N = 200
Age	46 (37, 59)	48 (39, 56)
Unknown	7	4

# Adding overall before modifying stats columns doesn't work with {level} or {label}
try(
  trial %>%
    tbl_summary(by = trt) %>%
    add_overall() %>%
    modify_header(update = all_stat_cols() ~ "**{level}**<br>N = {N}") %>%
    as_kable()
)
#> Error in eval(parse(text = text, keep.source = FALSE), envir) : 
#>   object 'level' not found

try(
  trial %>%
    tbl_summary(by = trt) %>%
    add_overall() %>%
    modify_header(update = all_stat_cols() ~ "**{label}**<br>N = {N}") %>%
    as_kable()
)
#> Error in eval(parse(text = text, keep.source = FALSE), envir) : 
#>   object 'label' not found

# Adding overall before modifying stats columns does work with plain text
trial %>%
  tbl_summary(by = trt) %>%
  add_overall() %>%
  modify_header(update = all_stat_cols() ~ "**THIS WORKS**<br>N = {N}") %>%
  as_kable()

Characteristic	THIS WORKS N = 200	THIS WORKS N = 200	THIS WORKS N = 200
Age	47 (38, 57)	46 (37, 59)	48 (39, 56)
Unknown	11	7	4

# And with {column} but then gives the gtsummary backend column name
trial %>%
  tbl_summary(by = trt) %>%
  add_overall() %>%
  modify_header(update = all_stat_cols() ~ "**{column}**<br>N = {N}") %>%
  as_kable()

Characteristic	stat_0 N = 200	stat_1 N = 200	stat_2 N = 200
Age	47 (38, 57)	46 (37, 59)	48 (39, 56)
Unknown	11	7	4

# Adding overall after modifying stats columns does work, but need to change label separately
trial %>%
  tbl_summary(by = trt) %>%
  modify_header(update = all_stat_cols() ~ "**{level}**<br>N = {N}") %>%
  add_overall(col_label = "**Overall**<br>N = {N}") %>%
  as_kable()

Characteristic	Overall N = 200	Drug A N = 200	Drug B N = 200
Age	47 (38, 57)	46 (37, 59)	48 (39, 56)
Unknown	11	7	4

^{Created on 2021-08-20 by the reprex package (v2.0.0)}

Solution

The issue you are running into is that all_stat_cols(), by default, selects the overall column and the other columns. Depending on whether you're assigning a label to the overall or split columns, you'll want to use slightly different syntax.

Example below!

library(gtsummary)

tbl <-
  trial %>%
  select(trt, age) %>%
  tbl_summary(by = trt, missing = "no") %>%
  add_overall() %>%
  modify_header(
    update = list(all_stat_cols(FALSE) ~ "**{level}**<br>N = {n}",
                  stat_0 ~ "**Overall**<br>N = {N}"))

show_header_names(tbl)
#> i As a usage guide, the code below re-creates the current column headers.
#> modify_header(update = list(
#>   label ~ "**Characteristic**",
#>   stat_0 ~ "**Overall**<br>N = 200",
#>   stat_1 ~ "**Drug A**<br>N = 98",
#>   stat_2 ~ "**Drug B**<br>N = 102"
#> ))
#> 
#> 
#> Column Name   Column Header          
#> ------------  -----------------------
#> label         **Characteristic**     
#> stat_0        **Overall**<br>N = 200 
#> stat_1        **Drug A**<br>N = 98   
#> stat_2        **Drug B**<br>N = 102

^{Created on 2021-08-20 by the reprex package (v2.0.1)}