Tree diagrams

library(litReview)
library(ggplot2)
data(studies)

reviewTree() draws a left-to-right hierarchical tree from a set of columns given in order: the first column forms the top-level branches, the next their children, and so on. Every level-1 branch gets its own colour, inherited by its descendants, and the leaves can attach a wrapped list of the studies that reach them. This article walks through every argument of

reviewTree(data, cols, study_id = StudyID, sep = "\r\n",
           show_members = TRUE, member_wrap = 36, label_wrap = 18,
           counts = "none", root_label = "All studies", colors = PALETTE,
           root_fill = "#F4F4C8", base_size = 11,
           na.rm = TRUE, na_label = "Not reported")

Default

Pass the data frame and the ordered vector of columns. Here the tree branches by InterventionType, then by Intervention, and lists the studies (by author) at each leaf:

reviewTree(studies, c("InterventionType", "Intervention"), study_id = Author)

cols: the hierarchy, in order

cols is the heart of the plot: a character vector naming the columns from root (first) to leaf (last). A single column gives a one-level fan-out:

reviewTree(studies, "Design", study_id = Author)

Add more columns to deepen the tree. Multi-value cells (like Outcome) are split, so a study can appear under several branches:

reviewTree(studies, c("InterventionType", "Intervention", "Outcome"),
           study_id = StudyID, member_wrap = 24)

study_id: what the leaves collect

study_id selects the column whose values are gathered into each leaf’s member box (default StudyID). Use Author for readable citations:

reviewTree(studies, c("Setting", "Design"), study_id = Author)

sep: multi-value separator

Cells that hold several values are split on sep (default "\r\n") at every level. Set it to match your data — here we rebuild a semicolon-separated column:

studies_semi <- studies
studies_semi$Outcome <- gsub("\r\n", "; ", studies_semi$Outcome)
reviewTree(studies_semi, c("Design", "Outcome"), sep = "; ", study_id = Author)

show_members: leaf study lists

By default each leaf attaches a box listing its studies. Turn it off for a bare taxonomy tree:

reviewTree(studies, c("InterventionType", "Intervention"), show_members = FALSE)

member_wrap: wrap the member lists

member_wrap controls how many characters fit per line in the member boxes. Narrower boxes are taller:

reviewTree(studies, c("InterventionType", "Intervention"),
           study_id = Author, member_wrap = 60)

label_wrap: wrap node labels

Long node labels wrap at label_wrap characters (default 18):

reviewTree(studies, c("Setting", "AnalysisApproach"),
           study_id = Author, label_wrap = 10, show_members = FALSE)

counts: number and/or percentage of papers

Annotate every node with how many studies it covers. counts = "count" adds the number of studies, "percent" the share of all studies, and "both" shows each node as count, percent. The root itself is left un-annotated. Because multi-value cells put a study in several branches, sibling percentages can exceed 100%.

reviewTree(studies, c("InterventionType", "Intervention"), study_id = Author,
           counts = "both", show_members = FALSE)

Counts and the member lists can be shown together:

reviewTree(studies, c("InterventionType", "Intervention"), study_id = Author,
           counts = "count")

root_label: name the root

reviewTree(studies, c("InterventionType", "Intervention"),
           study_id = Author, root_label = "Interventions")

colors: branch palette

colors supplies one colour per top-level branch (default PALETTE); descendants inherit it. Pass a custom vector:

reviewTree(studies, c("InterventionType", "Intervention"), study_id = Author,
           colors = c("#e15759", "#4e79a7", "#59a14f", "#b07aa1", "#f28e2b"))

root_fill: root node colour

reviewTree(studies, c("InterventionType", "Intervention"),
           study_id = Author, root_fill = "#d9d9d9")

base_size: overall scaling

A single knob scales the text and boxes proportionally:

reviewTree(studies, c("InterventionType", "Intervention"),
           study_id = Author, base_size = 14)

Missing data: na.rm and na_label

By default missing values are dropped. Keep them as an explicit branch with na.rm = FALSE; na_label names it:

reviewTree(studies, c("PubType", "FundingSource"), study_id = Author,
           na.rm = FALSE, na_label = "Not reported", show_members = FALSE)

Composing with ggplot2

reviewTree() returns a plain ggplot, so you can keep adding layers with +:

reviewTree(studies, c("InterventionType", "Intervention"), study_id = Author) +
  labs(title = "Intervention taxonomy",
       subtitle = "Studies grouped by intervention type")