Table shells

library(gtreg)
library(dplyr)
gtsummary::theme_gtsummary_compact()

Overview

Table shells are tables with generic xx place holders for numeric values. Table shells can be generated by:

Create dummy data or use your own data.
Pass the data to your function of choice.
Overwrite the statistic(s) shown to a fixed character string by implementing the style_xxx() function in the digits argument.

See more details about the digits argument in ?tbl_ae. Here are example specifications.

Integer Specification	Function Specification
`digits = 1`	`digits = function(x) style_number(x, digits = 1)`
`digits = c(0, 1)`	`digits = list(function(x) style_number(x, digits = 0), function(x) style_number(x, digits = 1))`

df_patient_characteristics %>%
  select(marker) %>%
  tbl_reg_summary(
    # integer specification
    digits = everything() ~ 1
    # function specification
    # digits = everything() ~ function(x) style_number(x, digits = 1)
  )
#> Setting theme "Compact"

Characteristic	N = 100
Biological Marker
N Non-missing	100.0
Mean (SD)	5.5 (1.0)
Median (Q1, Q3)	5.4 (4.8, 6.0)
Min, Max	3.5, 8.4
N Missing	0.0

`tbl_reg_summary()` uniform `xx`

Here, we specify the digits argument to uniformly apply to all summary statistics in the table. Note that we apply formatting to all columns using the column selector everything().

df_patient_characteristics %>%
  select(marker, trt) %>%
    tbl_reg_summary(
      digits = everything() ~  style_xxx
      ) %>%
    modify_header(stat_0 ~ "**N = xx**")
#> Setting theme "Compact"

Characteristic	N = xx¹
Biological Marker
N Non-missing	xx
Mean (SD)	xx (xx)
Median (Q1, Q3)	xx (xx, xx)
Min, Max	xx, xx
N Missing	xx
Treatment Group
Drug A	xx (xx%)
Drug B	xx (xx%)
¹ n (%)

`tbl_reg_summary()` custom `xx`

If you want custom formatting that differs between statistics, implement the digits argument with a list. Categorical variables are selected using all_categorical() and have two statistics to specify (n and %). The multi-line continuous variables are selected with all_continuous() and have nine statistics to specify.

df_patient_characteristics %>%
  select(marker, trt) %>%
    tbl_reg_summary(
      digits = list(
        all_categorical() ~ list(
          style_xxx,                                       # n 
          function(x) style_xxx(x, width = 4, digits = 1)  # %
          ),    
        all_continuous() ~ list(
          style_xxx,                                       # N
          function(x) style_xxx(x, width = 4, digits = 1), # Mean
          function(x) style_xxx(x, width = 4, digits = 1), # SD
          style_xxx,                                       # Median
          style_xxx,                                       # IQR lower
          style_xxx,                                       # IQR upper
          style_xxx,                                       # Range lower
          style_xxx,                                       # Range upper
          style_xxx                                        # N missing
    ))) %>%
    modify_header(stat_0 ~ "**N = xx**")
#> Setting theme "Compact"

Characteristic	N = xx¹
Biological Marker
N Non-missing	xx
Mean (SD)	xx.x (xx.x)
Median (Q1, Q3)	xx (xx, xx)
Min, Max	xx, xx
N Missing	xx
Treatment Group
Drug A	xx (xx.x%)
Drug B	xx (xx.x%)
¹ n (%)

`tbl_ae()` uniform `xx`

First, we create some dummy data to fill in our table shell. Note that not all values of grade need to be observed as factor levels populate the tables.

dat_shell <- tribble(
  ~ id,   ~ soc,   ~ ae, ~ grade,
     1, "SOC 1",  "ae1",       1,  
     2, "SOC 1",  "ae2",       1, 
     3, "SOC 2",  "ae3",       1, 
     4, "SOC 2",  "ae4",       1
  ) |> 
  mutate(
    grade = factor(grade, levels = 1:5)
  )

dat_shell
#> # A tibble: 4 × 4
#>      id soc   ae    grade
#>   <dbl> <chr> <chr> <fct>
#> 1     1 SOC 1 ae1   1    
#> 2     2 SOC 1 ae2   1    
#> 3     3 SOC 2 ae3   1    
#> 4     4 SOC 2 ae4   1

Here, we specify the digits argument to uniformly apply to all summary statistics in the table; in this case this is both n and %. As tbl_ae() provides the same summary statistic(s) across all variables, column selectors are not required.

Be sure to use the zero_symbol = NULL option in tbl_ae* functions to print xx’s rather than dashes for zero cells.

dat_shell %>% 
  tbl_ae(
    id = id,
    ae = ae,
    soc = soc,
    by = grade,
    zero_symbol = NULL,
    digits = style_xxx
    ) %>%
  modify_header(all_ae_cols() ~ "**Grade {by}**") %>%
  modify_spanning_header(all_ae_cols() ~ "**N = xx**")

Adverse Event	N = xx
Adverse Event	Grade 1	Grade 2	Grade 3	Grade 4	Grade 5
SOC 1	xx (xx)	xx (xx)	xx (xx)	xx (xx)	xx (xx)
ae1	xx (xx)	xx (xx)	xx (xx)	xx (xx)	xx (xx)
ae2	xx (xx)	xx (xx)	xx (xx)	xx (xx)	xx (xx)
SOC 2	xx (xx)	xx (xx)	xx (xx)	xx (xx)	xx (xx)
ae3	xx (xx)	xx (xx)	xx (xx)	xx (xx)	xx (xx)
ae4	xx (xx)	xx (xx)	xx (xx)	xx (xx)	xx (xx)

`tbl_ae()` custom `xx`

If you want custom formatting that differs between n and %, specify the digits argument with a list.

dat_shell %>% 
  tbl_ae(
    id = id,
    ae = ae,
    soc = soc,
    by = grade,
    zero_symbol = NULL,
    digits = list(
      style_xxx, #  n
      function(x) style_xxx(x, width = 4, digits = 1) # %
    )) %>%
    modify_header(all_ae_cols() ~ "**Grade {by}**") %>%
    modify_spanning_header(all_ae_cols() ~ "**N = xx**")

Adverse Event	N = xx
Adverse Event	Grade 1	Grade 2	Grade 3	Grade 4	Grade 5
SOC 1	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)
ae1	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)
ae2	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)
SOC 2	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)
ae3	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)
ae4	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)	xx (xx.x)

Overview

tbl_reg_summary() uniform xx

tbl_reg_summary() custom xx

tbl_ae() uniform xx

tbl_ae() custom xx

`tbl_reg_summary()` uniform `xx`

`tbl_reg_summary()` custom `xx`

`tbl_ae()` uniform `xx`

`tbl_ae()` custom `xx`