Drop Labels from a Table

library(tidyREDCap)

The Problem

The tidyREDCap package creates data sets with labelled columns.

tidyREDCap::import_instruments(
  url = "https://bbmc.ouhsc.edu/redcap/api/",
  token = keyring::key_get("REDCapR_test")
)

If you would like to see the labels on the data set demographics, you can use the RStudio function View(), as shown below.

View(demographics)

Demographics preview with labels

However, some functions do not work well with labeled variables. For example:

Show two numeric variables with labels

demographics |> 
  rowwise() |> 
  mutate(x = sum(c_across(c(age, days))))
#> Error in `mutate()`:
#> ℹ In argument: `x = sum(c_across(c(age, days)))`.
#> ℹ In row 1.
#> Caused by error in `vec_cast.labelled.labelled()`:
#> ✖ You are trying to combine variables with different labels
#> You can use tidyREDCap::drop_label() to erase one.

So you need a way to drop the label off of a variable or to drop all the labels from all the variables in a dataset.

The Solution

Drop a single label

You can drop the label from a single variable with the drop_label() function. For example:

new_demographics_table <- drop_label(demographics, "name_first")
# Or
new_demographics_table <- drop_label(demographics, name_first)
# Or
new_demographics_table <- demographics |> drop_label("name_first")

Drop multiple labels

If you need to drop labels from multiple variables, you can drop them individually or using helper methods (i.e., across()).

demographics |> 
  mutate(age = drop_label(age)) |> 
  mutate(days = drop_label(days)) |> 
  rowwise() |> 
  mutate(x = sum(c_across(c(age, days))))
# Or
demographics |> 
  mutate(across(c(age, days), drop_label)) |> 
  rowwise() |> 
  mutate(x = sum(c_across(c(age, days))))

Show three numeric variables without labels

You can use tidyselect helper methods (i.e., contains() or starts_with()) to include more than one variable or list them. The following code produces the same result:

demographics_changed_2 <- drop_label(demographics, contains("name"))
# Same as:
demographics_changed_3 <- drop_label(demographics, name_first, name_last)
# Verifying:
identical(demographics_changed_2, demographics_changed_3)

NOTE: You do not normally need to enclose the variable names in quotations outside of tidyselect helpers (i.e., contains()) though the function still operates if you choose to.

Use inside a `mutate`

You can now use drop_label() inside a mutate pipe like this:

demographics_from_mutate <- demographics |> 
  mutate(name_first = drop_label(name_first))
# Or
demographics_from_mutate <- demographics |> 
  mutate(across(starts_with('name'), drop_label))

Drop all dataset variable labels

You can drop all the labels using the drop_label() function. For example:

demographics_without_labels <- drop_label(demographics)

NOTE: tidyREDCap versions prior to 1.2.0 handled dropping all variable labels from a dataset by using drop_labels(). This function can still be used, but we added a polite message to use drop_label() instead.