Titles, Footers, and Referential Footnotes
In rtables: Reporting Tables

knitr::opts_chunk$set(comment = "#")
suppressPackageStartupMessages(library(rtables))
suppressPackageStartupMessages(library(dplyr))

```{css, echo=FALSE} .reveal .r code { white-space: pre; }

# Titles and Non-Referential Footer Materials

An `rtables` table can be annotated with three types of header (title)
information, as well as three types of footer information.

Header information comes in two forms that are specified directly
(main title and subtitles), as well as one that is populated
automatically as necessary (page title, which we will see in the next
section).

Similarly, footer materials come with two directly specified
components: main footer and provenance footer, in addition to one that
is computed when necessary: referential footnotes.

`basic_table()` accepts the values for each static title and footer
element during layout construction:

```r
library(rtables)
library(dplyr)
lyt <- basic_table(title = "Study XXXXXXXX",
                   subtitles = c("subtitle YYYYYYYYYY", "subtitle2 ZZZZZZZZZ"),
                   main_footer = "Analysis was done using cool methods that are correct",
                   prov_footer = "file: /path/to/stuff/that/lives/there HASH:1ac41b242a") %>%
    split_cols_by("ARM") %>%
    split_rows_by("SEX", split_fun = drop_split_levels) %>%
    split_rows_by("STRATA1") %>%
    analyze("AGE", mean, format = "xx.x")

tbl <- build_table(lyt, DM)
cat(export_as_txt(tbl, paginate = TRUE, page_break = "\n\n\n"))

Page-by splitting

We often want to split tables based on the values of one or more variables (e.g., lab measurement) and then paginate separately within each of those table subsections. In rtables we do this via page by row splits.

Row splits can be declared page by splits by setting page_by = TRUE in the split_rows_by*() call, as below.

When page by splits are present, page titles are generated automatically by appending the split value (typically a factor level, though it need not be), to the page_prefix, separated by a :. By default, page_prefix is name of the variable being split.

lyt2 <- basic_table(title = "Study XXXXXXXX",
                   subtitles = c("subtitle YYYYYYYYYY", "subtitle2 ZZZZZZZZZ"),
                   main_footer = "Analysis was done using cool methods that are correct",
                   prov_footer = "file: /path/to/stuff/that/lives/there HASH:1ac41b242a") %>%
    split_cols_by("ARM") %>%
    split_rows_by("SEX", page_by = TRUE, page_prefix = "Patient Subset - Gender", split_fun = drop_split_levels) %>%
    split_rows_by("STRATA1") %>%
    analyze("AGE", mean, format = "xx.x")

tbl2 <- build_table(lyt2, DM)
cat(export_as_txt(tbl2, paginate = TRUE, page_break = "\n\n~~~~ Page Break ~~~~\n\n"))

Page by row splits can be nested, but only within other page_by splits, they cannot be nested within traditional row splits. In this case, a page title for each page by split will be present on every resulting page, as seen below:

lyt3 <- basic_table(title = "Study XXXXXXXX",
                   subtitles = c("subtitle YYYYYYYYYY", "subtitle2 ZZZZZZZZZ"),
                   main_footer = "Analysis was done using cool methods that are correct",
                   prov_footer = "file: /path/to/stuff/that/lives/there HASH:1ac41b242a") %>%
    split_cols_by("ARM") %>%
    split_rows_by("SEX", page_by = TRUE, page_prefix = "Patient Subset - Gender", split_fun = drop_split_levels) %>%
    split_rows_by("STRATA1", page_by = TRUE, page_prefix = "Stratification - Strata") %>%
    analyze("AGE", mean, format = "xx.x")

tbl3 <- build_table(lyt3, DM)
cat(export_as_txt(tbl3, paginate = TRUE, page_break = "\n\n~~~~ Page Break ~~~~\n\n"))

Referential Footnotes

Referential footnotes are footnotes associated with a particular component of a table: a column, a row, or a cell. They can be added during tabulation via analysis functions, but they can also be added post-hoc once a table is created.

They are rendered as a number within curly braces within the table body, row, or column labels, followed by a message associated with that number printed below the table during rendering.

Adding Cell- and Analysis-row Referential Footnotes At Tabulation Time

afun <- function(df, .var, .spl_context) {
    val <- .spl_context$value[NROW(.spl_context)]
    rw_fnotes <-  if(val == "C") list("This is strata level C for these patients") else list()
    cl_fnotes <- if(val == "B" && df[1, "ARM", drop = TRUE] == "C: Combination")
                     list("these Strata B patients got the drug combination") else list()

    in_rows(mean = mean(df[[.var]]),
            .row_footnotes = rw_fnotes,
            .cell_footnotes = cl_fnotes,
            .formats = c(mean = "xx.x"))
}

lyt <- basic_table(title = "Study XXXXXXXX",
                   subtitles = c("subtitle YYYYYYYYYY", "subtitle2 ZZZZZZZZZ"),
                   main_footer = "Analysis was done using cool methods that are correct",
                   prov_footer = "file: /path/to/stuff/that/lives/there HASH:1ac41b242a") %>%
    split_cols_by("ARM") %>%
    split_rows_by("SEX", page_by = TRUE, page_prefix = "Patient Subset - Gender", split_fun = drop_split_levels) %>%
    split_rows_by("STRATA1") %>%
    analyze("AGE", afun, format = "xx.x")

tbl <- build_table(lyt, DM)
cat(export_as_txt(tbl, paginate = TRUE, page_break = "\n\n\n"))

We note that typically the type of footnote added within the analysis function would be dependent on the computations done to calculate the cell value(s), e.g., a model not converging. We simply use context information as an illustrative proxy for that.

The procedure for adding footnotes to content (summary row) rows or cells is identical to the above, when done within a content function.

Annotating an Existing Table with Referential Footnotes

In addition to inserting referential footnotes at tabulation time within our analysis functions, we can also annotate our tables with them post-hoc.

This is also the only way to add footnotes to column labels, as those cannot be controlled within an analysis or content function.

## from ?tolower example slightly modified
.simpleCap <- function(x) {
    if(length(x) > 1)
        return(sapply(x, .simpleCap))
    s <- strsplit(tolower(x), " ")[[1]]
    paste(toupper(substring(s, 1, 1)), substring(s, 2),
          sep = "", collapse = " ")
}

adsl2 <- ex_adsl %>%
    filter(SEX %in% c("M", "F") &
           RACE %in% (levels(RACE)[1:3])) %>%
    ## we trim the level names here solely due to space considerations
    mutate(ethnicity = .simpleCap(gsub("(.*)OR.*", "\\1", RACE)),
           RACE = factor(RACE))

lyt2 <- basic_table() %>%
  split_cols_by("ARM") %>%
    split_cols_by("SEX", split_fun = drop_split_levels) %>%
    split_rows_by("RACE", labels_var = "ethnicity",
                  split_fun = drop_split_levels) %>%
    summarize_row_groups() %>%
    analyze(c("AGE", "STRATA1"))

tbl2 <- build_table(lyt2, adsl2)
tbl2

We do this with the fnotes_at_path<- function which accepts a row path, a column path, and a value for the full set of footnotes for the defined locations (NULL or a character vector).

A non-NULL row path with a NULL column path specifies the footnote(s) should be attached to the row, while NULL row path with non-NULL column path indicates they go with the column. Both being non-NULL indicates a cell (and must resolve to an individual cell).

fnotes_at_path(tbl2, c("RACE", "ASIAN")) <- c("hi", "there")
tbl2

fnotes_at_path(tbl2, rowpath = NULL, c("ARM", "B: Placebo")) <- c("this is a placebo")
tbl2

Note to step into a content row we must add that to the path, even though we didn't need it to put a footnote on the full row.

Currently, content rows by default are named with the label rather than name of the corresponding facet. This is reflected in the output of, e.g., row_paths_summary.

row_paths_summary(tbl2)

So we can add our footnotes to the cell like so:

fnotes_at_path(tbl2,
               rowpath = c("RACE", "ASIAN", "@content", "Asian"),
               colpath = c("ARM", "B: Placebo", "SEX", "F")) <- "These asian women got placebo treatments"
tbl2