gtsummary - Wilcoxon on ordered factor

Question

By updating gtsummary, I ran into an issue. It used to be possible to ask for a wilcoxon-test for an ordered-factor variable. That should be possible, as the documentation says:

wilcox.test(
  as.numeric(variable) ~ as.factor(by), 
  data = data, 
  conf.int = TRUE, 
  conf.level = conf.level, 
  ...
)

After the update (2.0.4) that no longer seems to work.

testdf <- data.frame(
  out = sample(LETTERS[1:4], 100, replace = TRUE),
  group = sample(LETTERS[1:2],100, replace = TRUE)
) %>%
mutate(
  orderedfac = factor(out, levels = LETTERS[1:4], ordered = TRUE),
  outnum = as.numeric(orderedfac)
)

tbl_summary(
  testdf %>% select(group, orderedfac), by = group) %>%
    add_p(test = list(orderedfac = "wilcox.test")
)

results in error : 'x' must be numeric

Anyone knows a workaround?

Answer 1

In the 2.0 release of gtsummary, we took a step back from doing pre-processing on user data before calculating any test results, so you will need to convert your column to a numeric.

If that is, for some reason, not an option for you, you can write a custom Wilcox test. Example below!

library(gtsummary)

# create my own wilcox test function
my_wilcox <- function(data, by, variable, ...) {
  data[[variable]] <- as.numeric(data[[variable]])
  cardx::ard_stats_wilcox_test(data = data, by = all_of(by), variables = all_of(variable))
}

trial |> 
  dplyr::mutate(trt = factor(trt, ordered = TRUE)) |> 
  tbl_summary(
    by = trt, 
    include = age
  ) |> 
  add_p(test = all_continuous() ~ my_wilcox) |> 
  as_kable() # convert to kable to display on SO

Characteristic	Drug A N = 98	Drug B N = 102	p-value
Age	46 (37, 60)	48 (39, 56)	0.7
Unknown	7	4

^{Created on 2025-02-03 with reprex v2.1.1}