Introduction
The STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement encourages the transparent reporting of observational studies. One important element is a flow diagram that documents how the study population was selected.
The strobe package provides tools to generate reproducible STROBE-style diagrams from your cohort-building process. This vignette shows how to use plot_strobe_diagram() to visualize cohort selection, and how to adjust visual parameters for publication-ready output.
The plot_strobe_diagram function assumes you’ve already built a STROBE derivation log using strobe_initialize() and strobe_filter(), and converts the derivation log into a STROBE figure using the DiagrammeR package in R.
Setup
We start by creating a STROBE-compatible log, again using the
cytomegalovirus data from the medicaldata package:
library(strobe)
library(dplyr)
#>
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#>
#> filter, lag
#> The following objects are masked from 'package:base':
#>
#> intersect, setdiff, setequal, union
library(medicaldata)
data(cytomegalovirus)
cytomegalovirus_df<-cytomegalovirus %>%
strobe_initialize("All transplant recipients") %>%
strobe_filter("age >= 18 & age <= 65", "Age 18–65", "Excluded: Outside 18–65") %>%
strobe_filter("recipient.cmv == 1", "CMV positive", "Excluded: CMV negative") %>%
strobe_filter("donor.cmv == 1", "Donor CMV positive", "Excluded: Donor CMV negative") %>%
strobe_filter("prior.transplant == 0", "No prior transplant", "Excluded: Prior transplant")Basic Usage
The most basic STROBE diagram can be produced by simply calling the
plot_strobe_diagram function in interactive mode:
While this will work in interactive mode, it typically will not work well when knitting an R Markdown file. In that case, we recommend exporting to a file and then directly incorporating the exported file in the R markdown file.
svg_file_2_1 <-plot_strobe_diagram(export_file = "../man/figures/strobe-diagram_vignette2_1.svg")
knitr::include_graphics("../man/figures/strobe-diagram_vignette2_1.svg")
The strobe package supports .png and .svg formats. The
SVG format can easily be edited in an SVG editor such as SVG-Edit or a
downloadable SVG editor such as Inkscape.
Controlling Box Size
If your text is getting truncated or looks cramped, adjust the box dimensions.
Wider boxes
We can use the incl_width_min option to set the minimum
width of the inclusion boxes, and the excl_width_min box to
set the minimum width of the exclusion boxes.
We first demonstrate the use of the minimum width setting:
svg_file_2_2a <-plot_strobe_diagram(incl_width_min = 10, excl_width_min = 10,
export_file = "../man/figures/strobe-diagram_vignette2_2a.svg")
knitr::include_graphics("../man/figures/strobe-diagram_vignette2_2a.svg")
Taller boxes
We can use the incl_height option to set the height of
the inclusion boxes, and the excl_height box to set the
minimum height of the exclusion boxes.
svg_file_2_2b <-plot_strobe_diagram(incl_height = 5, excl_height = 5,
export_file = "../man/figures/strobe-diagram_vignette2_2b.svg")
knitr::include_graphics("../man/figures/strobe-diagram_vignette2_2b.svg")
Locking Size Across Boxes
The DiagrammeR package, by default, takes input sizes as
suggestions and may choose to ignore them. If you want to force the use
of the sizes that you provide, use the lock_width_min or
lock_height options:
svg_file_2_3 <-plot_strobe_diagram(lock_width_min = TRUE,
lock_height = TRUE,
export_file = "../man/figures/strobe-diagram_vignette2_3.svg")
knitr::include_graphics("../man/figures/strobe-diagram_vignette2_3.svg")
While not helpful in this case, it may be helpful when you have multi-line text in some nodes but not others, but want to ensure uniformity of size.
Adjusting Font Size
You can separately control font sizes for inclusion and exclusion boxes:
svg_file_2_4 <-plot_strobe_diagram(incl_fontsize = 16,
excl_fontsize = 14,
export_file = "../man/figures/strobe-diagram_vignette2_4.svg")
knitr::include_graphics("../man/figures/strobe-diagram_vignette2_4.svg")
Use larger fonts for presentation or slide decks:
svg_file_2_5a <-plot_strobe_diagram(incl_fontsize = 150,
excl_fontsize = 150,
export_file = "../man/figures/strobe-diagram_vignette2_5a.svg")
knitr::include_graphics("../man/figures/strobe-diagram_vignette2_5a.svg")
The use of larger fonts will likely require some trial and error with box sizing for optimal results:
svg_file_2_5b <-plot_strobe_diagram(incl_fontsize = 150,
excl_fontsize = 150,
incl_width_min = 20, excl_width_min = 30,
export_file = "../man/figures/strobe-diagram_vignette2_5b.svg")
knitr::include_graphics("../man/figures/strobe-diagram_vignette2_5b.svg")
Combining Size and Font Adjustments
Here’s an example with many settings tuned:
svg_file_2_6 <-plot_strobe_diagram(
incl_width_min = 5,
excl_width_min = 4,
incl_height = 1.5,
excl_height = 1.2,
incl_fontsize = 18,
excl_fontsize = 16,
lock_width_min = TRUE,
lock_height = TRUE,
export_file = "../man/figures/strobe-diagram_vignette2_6.svg"
)
knitr::include_graphics("../man/figures/strobe-diagram_vignette2_6.svg")
Accounting for terminal branching
You don’t have to do anything special to account for terminal
branching when plotting–the plot_strobe_diagram function is
able to handle the details based on the structure of the strobe
flow:
#Add a terminal branch
cytomegalovirus_df<-create_terminal_branch(cytomegalovirus_df, variable = "cgvhd", label_prefix="CGVHD value:")
svg_file_2_7 <- plot_strobe_diagram(export_file = "../man/figures/strobe-diagram_vignette2_7.svg",
incl_fontsize = 90, excl_fontsize = 90,
lock_width_min = TRUE,
incl_width_min = 20, excl_width_min = 20)
knitr::include_graphics("../man/figures/strobe-diagram_vignette2_7.svg")
Troubleshooting Layout
If boxes are overlapping or lines look incorrect:
Try Locking Width/Height
The plot_strobe_diagram’s underlying dependence on DiagrammeR means
that the suggested widths and heights may not be followed precisely
unless the lock_width_min or lock_height
options are used:
svg_file_2_8 <-plot_strobe_diagram(lock_width_min = TRUE,
incl_width_min = 15,
excl_width_min = 10,
export_file = "../man/figures/strobe-diagram_vignette2_8.svg")
knitr::include_graphics("../man/figures/strobe-diagram_vignette2_8.svg")
Use Fewer Words or Manual Line Breaks
Use in labels to manually split long lines. Example:
strobe_filter(
condition = "...",
inclusion_label = "Eligible recipients\nwith CMV+ donors",
exclusion_reason = "Excluded:\nCMV- donors"
)This often improves appearance more than tweaking sizes alone.
Summary
plot_strobe_diagram() gives you fine-grained control
over the appearance of STROBE cohort diagrams. You can:
- Tune box dimensions
- Control font sizes
- Lock box sizes for uniform appearance
- Export to PNG or SVG for publication use or, in the case of SVG, further editing
These tools help make your derivation logic transparent, reproducible, and ready for review.