Example data

First lets get an example clinical trial dataset that we might want to summarise, and some packages that we will need.

library(knitr)
library(survival) #pbc dataset lives here
data(pbc)
suppressPackageStartupMessages(library(tidyverse)) 
suppressPackageStartupMessages(library(janitor))
suppressPackageStartupMessages(library(knitr)) # to kable
suppressPackageStartupMessages(library(flextable)) #to flextable

Summary table

OK. So lets start with creating our summary table. If you havent used the tabyl function in janitor, get on it!

#glimpse(pbc)

pbc %>%
  tabyl(stage, trt) %>%
    adorn_totals() %>%
  adorn_percentages("col") %>%
  adorn_pct_formatting(digits = 1)  %>%
  adorn_ns("front") ->t1

Tables in word

Now, typically if you want to print this nicely in Rmarkdown, you would use kable(or maybe kableextra). That would look like this:

t1 %>%
  select(stage, `1`, `2`) %>%
  kable(col.names = c("Stage", "Placebo", "Treatment"))

So while there are plenty of ways to beautify this in html or latex (pdf), there aren’t a lot of options for anyone to render in word.

Enter, flextable. Thankyou David Gohel! His also the author of officer, another amazing package for writing Rmarkdown reports in word.

So how do we use flextable?

t1 %>%
  select(stage, `1`, `2`) %>%
  filter(is.na(stage)==FALSE) %>%
  flextable() %>%
  set_header_labels() %>%
  knit_print()

Its that easy. But now lets beautify with all of the functions that you might want to use, and the ones that will work in rendering to word.

t1 %>%
  select(stage, `1`, `2`) %>%
  filter(is.na(stage)==FALSE) %>%
  add_row(stage="Stage", .before = 1) %>%
  flextable() %>%
  set_caption("Stage of disease by treatment") %>%
   set_header_labels(stage=" ",`1`="Placebo", `2`="Treatment" ) %>%
   align(j=1, align = "left", part = "all") %>%
   align(j=2:3, align = "center", part = "all") %>%  
   bold( i=c(1)) %>%
   font(fontname = "Calibri", part="all") %>%
   padding(i=c(2:5), j=1, padding.left = 20) %>%
   set_table_properties(width = 1, layout = "autofit") %>%
  knit_print()

Now, this has rendered a bit wierdly in my blog, due to the Blog formatting etc. Please trust me this is lovely in word.

Lets unpack some of these functions.

• set_caption is the table caption in word.
• set_header_labels are the column names
• align aligns the columns, and part="all" means we want to align the header as well as the table data. j specifies the columns that you wish to align.
• bold is used to make a row bold (row 1 as i=1). You can also use italics
• font sets the font of the table data
• padding - my absolute favourite - adds some space to indent the column/row specified. You can also pad right if you want to (or top or bottom)
• set_table_properties fits the table to the width of the word document. Its not great when I render it here in html but trust me, its great in word.

Some further notes when you are setting up the Rmarkdown file.

library(officer)
library(flextable)

autonum <- run_autonum() #to number the tables

t1 %>%
  flextable() %>%
   set_caption("Table 1", style = "Table Caption", autonum = autonum)