A big part of what I do day-to-day is to provide a report of summary statistics. The researchers I work with would usually want this report in the format of a Word document (.docx) as it is (unfortunately) still the most widely used word processing software.

I much prefer markdown and TeX files for reports, but you have to work to your audience. This is how I went down the rabbit hole of trying to find an R package that outputs summary tables into Word documents. So far I’ve found that gtsummary and table1 combined with the flextable package does the job best.

Another package that’s been brought to my attention is stargazer, but alas it only outputs to html. So I’ll walk you through a reproducible example of how to use the gtsummary and table1 packages to create aesthetically pleasing summary tables in Word, supported by the flextable package.

Using `gtsummary`

Let’s create a table for the mtcars data, grouping by engine type (V-shaped vs straight). Here’s how we’d do it using the gtsummary package:

library(gtsummary)

mtcars %>%
  select(mpg, cyl, disp, am, vs) %>% 
  tbl_summary(by = vs)

Characteristic	0, N = 18¹	1, N = 14¹
mpg	15.7 (14.8, 19.1)	22.8 (21.4, 29.6)
cyl
4	1 (5.6%)	10 (71%)
6	3 (17%)	4 (29%)
8	14 (78%)	0 (0%)
disp	311 (276, 360)	121 (83, 162)
am	6 (33%)	7 (50%)
¹ Median (IQR); n (%)

Not bad–and it plays well with the tidyverse. I also like that gtsummary tends to be smart about how the results should be displayed. For example, the cyl and am variables were automatically treated as discrete instead of continuous variables (thus giving us counts and percentages instead of median and IQR). You’ll notice that if you have the levels of a categorical variable as 0 and 1 (as in am), gtsummary provides only the count and percentage for the values coded as 1, which I’d argue eliminates redundancy.

Using `table1`

Now, let’s use the table1 package:

library(table1)

table1(~ mpg + factor(cyl) + disp + factor(am) | factor(vs),
       data = mtcars)

	0 (N=18)	1 (N=14)	Overall (N=32)
mpg
Mean (SD)	16.6 (3.86)	24.6 (5.38)	20.1 (6.03)
Median [Min, Max]	15.7 [10.4, 26.0]	22.8 [17.8, 33.9]	19.2 [10.4, 33.9]
factor(cyl)
4	1 (5.6%)	10 (71.4%)	11 (34.4%)
6	3 (16.7%)	4 (28.6%)	7 (21.9%)
8	14 (77.8%)	0 (0%)	14 (43.8%)
disp
Mean (SD)	307 (107)	132 (56.9)	231 (124)
Median [Min, Max]	311 [120, 472]	121 [71.1, 258]	196 [71.1, 472]
factor(am)
0	12 (66.7%)	7 (50.0%)	19 (59.4%)
1	6 (33.3%)	7 (50.0%)	13 (40.6%)

One could argue that the formula format is a bit tedious as you can’t select the variables (for example, from a list or vector), but I find that you usually need to be cognisant of which variables you want included in a summary table anyways so this isn’t much of a problem for me. I also prefer how it looks as it fits more into the format of a ‘Table 1’ for scientific publishing (hence the package name).

I did however have to wrap the cyl and am with factor() as table1 does not automatically treat these numeric variables as categorical variables.

Exporting using `flextable`

Now let’s export these tables into Word. First you have to save your table as a flextable object. You do this by either using the as_flex_table() function for a gtsummary object, or the t1flex() function for a table1 object. Once you’ve done that, save_as_docx() saves the table into a Word document which you can copy and paste. Here’s how it would look:

# For gtsummary
table <- mtcars %>%
  select(mpg, cyl, disp, am, vs) %>% 
  tbl_summary(by = vs)

as_flex_table(table) %>% 
  save_as_docx(path = "./gtsummary.docx")

# For table1
table <- table1(~ mpg + factor(cyl) + disp + factor(am) | factor(vs),
       data = mtcars)

t1flex(table) %>% 
  save_as_docx(path = "./table1.docx")

flextable also supports exporting to HTML, PDF, and PowerPoint documents; it also supports rendering tables as graphic files (such as png and jpeg).

Additional notes

You can add content to Word and PowerPoint documents using the officer package, but I haven’t used it before–my end users usually just copy-and-paste the tables. This is a package I’d like to look into.

Using gtsummary

Using table1

Exporting using flextable

Additional notes

Using `gtsummary`

Using `table1`

Exporting using `flextable`