supplementary_code.R

# rtables workshop R in Medicine
# June 6, 2023
# Gabriel Becker and Adrian Waddell


# Install rtables packages and dependencies ----
# install.packages(c("rtables", "remotes"))
# remotes::install_github("insightsengineering/rtables") # for workshop


# Getting started ----

# Data
library(rtables)

# synthetic adsl data
head(ex_adsl)
names(ex_adsl)

# R base table function
table(ex_adsl$ARM)
table(ex_adsl$SEX, ex_adsl$ARM)

# Now with qtable in rtables

library(rtables)
qtable(ex_adsl, "ARM")
qtable(ex_adsl, "SEX", "ARM")

# nested frequency tables
qtable(ex_adsl, c("COUNTRY", "SEX"), "ARM")

# using a aggregation function other than count
qtable(ex_adsl, c("COUNTRY", "SEX"), "ARM", avar = "AGE", afun = mean)

# multiple analysis values
fivenum2 <- \(x) setNames(as.list(fivenum(x)), c("min", "Q1", "MED", "Q3", "max"))
qtable(ex_adsl, "SEX", "ARM", avar = "AGE", afun = fivenum2)

# now using rtable's core way of defining a table

lyt <- basic_table() |>
  split_cols_by("ARM") |>
  split_rows_by("SEX") |>
  analyze("AGE", afun = fivenum2, format = "xx.xx")

build_table(lyt, ex_adsl)


# Basics of rtables ----

lyt <- basic_table() |>
  split_cols_by("ARM") |>
  analyze("AGE", mean, format = "xx.xx")

tbl <- build_table(lyt, ex_adsl)

print(tbl)


#  Deriving cell values ----

lyt <- basic_table() |>
  analyze("AGE")

build_table(lyt, ex_adsl)


# in_rows use and formatting options
fivenum_afun <- function(x) {
  in_rows(n = sum(!is.na(x)),
          "mean (sd)" = c(mean(x), sd(x)),
          median = median(x),
          "min - max" = range(x),
          .formats = c(n = "xx",
                       "mean (sd)" = "xx.x (xx.x)",
                       median = "xx.x",
                       "min - max" = "xx.x - xx.x"))
}

lyt2 <- basic_table() %>% analyze("AGE", fivenum_afun)

build_table(lyt2, ex_adsl)


# Cell Value Formatting  ----

rcell(1.234567, "xx.xx")
rcell(c(1, 0.1), "xx.xx (xx.xx%)")

formatters::list_valid_format_labels()

example(table_shell)



# Percentages ----
pct_afun <- function(x, .N_col) {
  rcell(
    sum(!is.na(x)) * c(1, 1/.N_col),
    format = "xx (xx.x%)"
  )
}

lyt <- basic_table() |>
  analyze("AGE", pct_afun)

build_table(lyt, ex_adsl)


# Faceting ----

## Column Faceting
library(ggplot2)
ggplot(ex_adsl, mapping = aes(x = AGE)) +
  geom_boxplot() +
  facet_grid(cols = vars(ARM))

lyt <- basic_table() |>
  split_cols_by("ARM") |>
  analyze("AGE", range, format = "xx.xx - xx.xx")

build_table(lyt, ex_adsl)

## Row Faceting
ggplot(ex_adsl, mapping = aes(x = AGE)) +
  geom_boxplot() +
  facet_grid(rows = vars(SEX))

lyt2 <- basic_table() |>
  split_rows_by("SEX") |>
  analyze("AGE", range,
          format = "xx.xx - xx.xx")

build_table(lyt2, ex_adsl)

## Grid Faceting

ggplot(ex_adsl, mapping = aes(x = AGE)) +
  geom_boxplot() +
  facet_grid(rows = vars(SEX),
             cols = vars(ARM))

lyt3 <- basic_table() |>
  split_cols_by("ARM") |>
  split_rows_by("SEX") |>
  analyze("AGE", range, format = "xx.xx - xx.xx")

build_table(lyt3, ex_adsl)

# Nested Splits ----
library(dplyr)

ex_adsl2 <- ex_adsl |>
  filter(SEX %in% c("M", "F"))

ggplot(ex_adsl2, mapping = aes(x = AGE)) +
  geom_boxplot() +
  facet_grid(cols = vars(ARM, SEX))

lyt <- basic_table() |>
  split_cols_by("ARM") |>
  split_cols_by("SEX", split_fun = drop_split_levels) |>
  analyze("AGE", mean, format = "xx.xx")

tbl <- build_table(lyt, ex_adsl2)

print(tbl)


# Group Summaries
ex_adsl3 <- ex_adsl |>
  mutate(B1HL = factor(ifelse(BMRKR1 > mean(BMRKR1), "H", "L"), levels = c("L", "H")))

lyt <- basic_table() |>
  split_cols_by("ARM") |>
  split_rows_by("SEX") |>
  split_rows_by("B1HL") |>
  analyze("AGE", \(x) list(B = "a"))

build_table(lyt, ex_adsl3)

lyt <- basic_table() |>
  split_cols_by("ARM") |>
  summarize_row_groups() |>
  split_rows_by("SEX") |>
  summarize_row_groups() |>
  split_rows_by("B1HL") |>
  summarize_row_groups() |>
  analyze("AGE", \(x) list(B = "a"))

build_table(lyt, ex_adsl3)


# Group Summaries in qtable ----
qtable(ex_adsl3, c("SEX", "B1HL"), "ARM", avar = "AGE", afun = \(x) list(B = "a"))

qtable(ex_adsl3, c("SEX", "B1HL"), "ARM", avar = "AGE", afun = \(x) list(B = "a"), summarize_groups = TRUE)

# Complex Tables ----
# code can be found here:
# https://github.com/insightsengineering/adv_rtables_training/blob/main/training2.Rmd#L627


# Analysis Reporting Datasets (ARDs) ----