Re-add data pre-processing prior to add_difference() and add_p()?

[cforne]

Dear Daniel:

After updating the gtsummary package, add_difference of a tbl_summary with dichotomous variables is not working. Next error is obtained:

For variable XXXX (YYYY) and "estimate", "statistic", "p.value", "parameter", "conf.low", and "conf.high" statistics: Expecting variable to be either or <numeric/integer> coded as 0 and 1

The dichotomous variable is a factor, coded as "Yes", "No".

In the tbl_summary, the level of the dichotomous variable is displayed on a single row. What is wrong?

[ddsjoberg]

Thank you for the post @cforne . It was a conscience decision to no longer perform pre-processing on the data before calculating tests (and other statistical computations) and leave it to the user to pass data that was compatible with the tests they are requesting.

But it seems I did not give enough consideration to situations like these. I'll think on it, but it feels like this functionality should be added back to the package. Sorry for the inconvenience.

library(gtsummary)

trial |> 
  dplyr::mutate(response = factor(response, levels = c(0, 1), labels = c("no", "yes"))) |> 
  tbl_summary(
    by = trt, 
    include = response
  ) |> 
  add_difference() |> 
  as_kable()
#> The following errors were returned during `as_kable()`:
#> ✖ For variable `response` (`trt`) and "estimate", "statistic", "p.value",
#>   "parameter", "conf.low", and "conf.high" statistics: Expecting `variable` to
#>   be either <logical> or <numeric/integer> coded as 0 and 1.

Characteristic	Drug A N = 98	Drug B N = 102	Difference	95% CI	p-value
response	28 (29%)	33 (34%)
Unknown	3	4

trial |> 
  tbl_summary(
    by = trt, 
    include = grade,
    value = grade ~ "I"
  ) |> 
  add_difference() |> 
  as_kable()
#> The following errors were returned during `as_kable()`:
#> ✖ For variable `grade` (`trt`) and "estimate", "statistic", "p.value",
#>   "parameter", "conf.low", and "conf.high" statistics: Expecting `variable` to
#>   be either <logical> or <numeric/integer> coded as 0 and 1.

Characteristic	Drug A N = 98	Drug B N = 102	Difference	95% CI	p-value
Grade	35 (36%)	33 (32%)

^{Created on 2025-02-22 with reprex v2.1.1}