Factors/levels + dplyr::arrange()

[ariannaa7]

In this situation, we needed to display rows in a customized order based on the variable_level (med names) column.

Here, I created multiple vectors based on the category of medication and set the desired order of the medications within the vectors. I then mutated the variable_name column as a factor and specified the vectors as the level (again in the desired order).

# Vectors with med names in the order in which they should appear for that category of meds
med_group_A <- c("med_A2", "med_A4", "med_A1", "med_A3")
med_group_B <- c("med_B5", "med_B4", "med_B2", "med_B3", "med_B1")
med_group_C <- c("med_C1", "med_C3", "med_C4", "med_C2")

# Vectors for the order in which rows of variable_name column should appear
variable_name_levels <- c("example1", "example2", "example3", "example4")

meds_summarised_result <- meds_summarised_result %>%
  omopgenerics::splitGroup() %>%
      dplyr::mutate(
        variable_name = factor(
          variable_name,
          levels = variable_name_levels
        ),
        variable_level = factor(
          variable_level,
          levels = c(med_group_B, med_group_A, med_group_C,) # we first want all the meds in groupB, then groupA, then groupC so we specify this with the order in which we list the vectors
        )
      ) %>%
      dplyr::arrange(cdm_name, strata_name, variable_name, variable_level) %>% # arrange from zoomed out to zoomed in
      dplyr::mutate(variable_name = as.character(variable_name), # remove factors
                    variable_level = as.character(variable_level)) %>%
      omopgenerics::uniteGroup(cols = c("cohort_name")) # unite once more
 

Using omopgenerics::splitGroup() and omopgenerics::uniteGroup maintains the integrity of the summarised_results object! Attempting to arrange without it or with just "variable_level" (instead cdm_name, strata_name, ..., variable_level) for example, completely reformatted the table.