Denominator in tbl_hierarchical tables with tbl_merge

Here is a simple example:

library(cards)
library(gtsummary)

adsl <- cards::ADSL %>%
  dplyr::mutate(
    SAFFL = dplyr::if_else(SAFFL == "Y", "Y", NA_character_),
    ITTFL = dplyr::if_else(ITTFL == "Y", "Y", NA_character_),
    EFFFL = dplyr::if_else(EFFFL == "Y", "Y", NA_character_),
    COMP8FL = dplyr::if_else(COMP8FL == "Y", "Y", NA_character_)
  )

group_vars <- c("SAFFL", "ITTFL", "EFFFL", "COMP8FL")

tbl_list <- lapply(group_vars, function(var) {
  gtsummary::tbl_hierarchical(
    data = adsl,
    denominator = adsl,
    id = USUBJID,
    by = .data[[var]],
    variables = c(SITEGR1, SITEID),
  )
})

tbl <- gtsummary::tbl_merge(tbls = tbl_list) |>
  gtsummary::modify_spanning_header(dplyr::everything() ~ NA_character_) |>
  gtsummary::modify_header(stat_1_1 = "**SAFFL**  \n(N = {n})",
                           stat_1_2 = "**ITTFL**  \n(N = {n})",
                           stat_1_3 = "**EFFFL**  \n(N = {n})",
                           stat_1_4 = "**COMP8FL**  \n(N = {n})") |> 
  remove_footnote_header(columns = everything())
tbl

However, I got all percentages equal to 100%:

Where did it go wrong?

Assuming you are after the percentage of each count relative to their group instead of 100%, you need to include the groups in denominator of tbl_hierarchical().

tbl_list <- lapply(group_vars, function(var) {
  gtsummary::tbl_hierarchical(
    data = adsl,
    denominator = adsl %>% 
      select(USUBJID, all_of(var)),
    id = USUBJID,
    by = .data[[var]],
    variables = c(SITEGR1, SITEID)
  )
})