Creating Publication-Ready Tables with gtsummary

advanced

tables

statistics

Learn how to create professional statistical tables for manuscripts and reports

Author

Unijos R Users Team

Published

January 15, 2025

Introduction

One of the most time-consuming tasks in research is creating well-formatted statistical tables. The gtsummary package makes this incredibly easy while producing publication-ready output.

Why gtsummary?

Creates APA-style tables automatically
Supports complex statistical models
Exports to Word, HTML, PDF, and LaTeX
Minimal code required
Perfect for epidemiological studies

Installation

install.packages("gtsummary")
install.packages("flextable")  # For Word export

Basic Summary Tables

Descriptive Statistics

library(gtsummary)
library(tidyverse)

# Create a summary table stratified by exposure status
outbreak_data %>%
  select(age, sex, symptoms, exposed) %>%
  tbl_summary(
    by = exposed,
    statistic = list(
      all_continuous() ~ "{mean} ({sd})",
      all_categorical() ~ "{n} ({p}%)"
    ),
    label = list(
      age ~ "Age (years)",
      sex ~ "Sex",
      symptoms ~ "Symptoms Present"
    )
  ) %>%
  add_p() %>%  # Add p-values
  add_overall() %>%  # Add overall column
  bold_labels()

Regression Tables

Logistic Regression for Outbreak Investigation

# Fit logistic regression model
model <- glm(
  case_status ~ age + sex + water_source + food_type,
  data = outbreak_data,
  family = binomial
)

# Create publication-ready table
tbl_regression(
  model,
  exponentiate = TRUE,  # Show odds ratios
  label = list(
    age ~ "Age (years)",
    sex ~ "Sex",
    water_source ~ "Water Source",
    food_type ~ "Food Type"
  )
) %>%
  add_global_p() %>%  # Add overall p-values
  bold_p(t = 0.05) %>%  # Bold significant p-values
  bold_labels()

Exporting Tables

To Microsoft Word

my_table %>%
  as_flex_table() %>%
  flextable::save_as_docx(path = "table1.docx")

To HTML

my_table %>%
  as_gt() %>%
  gt::gtsave("table1.html")

Advanced Tips

Compare Multiple Models

# Compare unadjusted vs adjusted models
tbl_merge(
  list(
    tbl_regression(unadjusted_model, exponentiate = TRUE),
    tbl_regression(adjusted_model, exponentiate = TRUE)
  ),
  tab_spanner = c("**Unadjusted**", "**Adjusted**")
)

Customize Themes

# Set a journal-specific theme
theme_gtsummary_journal("jama")

# Or create custom theme
theme_gtsummary_compact()

Practical Example: Diphtheria Outbreak

Here’s a complete example analyzing outbreak data:

# Load outbreak data
diphtheria_data <- read_csv("diphtheria_outbreak.csv")

# Create comprehensive table
diphtheria_data %>%
  select(age_group, sex, vaccination_status, outcome, district) %>%
  tbl_summary(
    by = outcome,
    statistic = list(
      all_categorical() ~ "{n} ({p}%)"
    ),
    label = list(
      age_group ~ "Age Group",
      sex ~ "Sex",
      vaccination_status ~ "Vaccination Status",
      district ~ "District"
    )
  ) %>%
  add_p(test = all_categorical() ~ "fisher.test") %>%
  add_overall() %>%
  modify_header(label = "**Characteristic**") %>%
  bold_labels() %>%
  bold_p(t = 0.05)

Resources

Join our next session to learn more advanced table techniques! RSVP on Meetup

--- title: "Creating Publication-Ready Tables with gtsummary" author: "Unijos R Users Team" date: "2025-01-15" categories: [advanced, tables, statistics] description: "Learn how to create professional statistical tables for manuscripts and reports" --- ## Introduction One of the most time-consuming tasks in research is creating well-formatted statistical tables. The `gtsummary` package makes this incredibly easy while producing publication-ready output. ## Why gtsummary? - Creates APA-style tables automatically - Supports complex statistical models - Exports to Word, HTML, PDF, and LaTeX - Minimal code required - Perfect for epidemiological studies ## Installation ```{r} #| eval: false install.packages("gtsummary") install.packages("flextable") # For Word export ``` ## Basic Summary Tables ### Descriptive Statistics ```{r} #| eval: false library(gtsummary) library(tidyverse) # Create a summary table stratified by exposure status outbreak_data %>% select(age, sex, symptoms, exposed) %>% tbl_summary( by = exposed, statistic = list( all_continuous() ~ "{mean} ({sd})", all_categorical() ~ "{n} ({p}%)" ), label = list( age ~ "Age (years)", sex ~ "Sex", symptoms ~ "Symptoms Present" ) ) %>% add_p() %>% # Add p-values add_overall() %>% # Add overall column bold_labels() ``` ## Regression Tables ### Logistic Regression for Outbreak Investigation ```{r} #| eval: false # Fit logistic regression model model <- glm( case_status ~ age + sex + water_source + food_type, data = outbreak_data, family = binomial ) # Create publication-ready table tbl_regression( model, exponentiate = TRUE, # Show odds ratios label = list( age ~ "Age (years)", sex ~ "Sex", water_source ~ "Water Source", food_type ~ "Food Type" ) ) %>% add_global_p() %>% # Add overall p-values bold_p(t = 0.05) %>% # Bold significant p-values bold_labels() ``` ## Exporting Tables ### To Microsoft Word ```{r} #| eval: false my_table %>% as_flex_table() %>% flextable::save_as_docx(path = "table1.docx") ``` ### To HTML ```{r} #| eval: false my_table %>% as_gt() %>% gt::gtsave("table1.html") ``` ## Advanced Tips ### Compare Multiple Models ```{r} #| eval: false # Compare unadjusted vs adjusted models tbl_merge( list( tbl_regression(unadjusted_model, exponentiate = TRUE), tbl_regression(adjusted_model, exponentiate = TRUE) ), tab_spanner = c("**Unadjusted**", "**Adjusted**") ) ``` ### Customize Themes ```{r} #| eval: false # Set a journal-specific theme theme_gtsummary_journal("jama") # Or create custom theme theme_gtsummary_compact() ``` ## Practical Example: Diphtheria Outbreak Here's a complete example analyzing outbreak data: ```{r} #| eval: false # Load outbreak data diphtheria_data <- read_csv("diphtheria_outbreak.csv") # Create comprehensive table diphtheria_data %>% select(age_group, sex, vaccination_status, outcome, district) %>% tbl_summary( by = outcome, statistic = list( all_categorical() ~ "{n} ({p}%)" ), label = list( age_group ~ "Age Group", sex ~ "Sex", vaccination_status ~ "Vaccination Status", district ~ "District" ) ) %>% add_p(test = all_categorical() ~ "fisher.test") %>% add_overall() %>% modify_header(label = "**Characteristic**") %>% bold_labels() %>% bold_p(t = 0.05) ``` ## Resources - [gtsummary website](https://www.danieldsjoberg.com/gtsummary/) - [Gallery of examples](https://www.danieldsjoberg.com/gtsummary/articles/gallery.html) - [Customization vignette](https://www.danieldsjoberg.com/gtsummary/articles/themes.html) --- **Join our next session** to learn more advanced table techniques! [RSVP on Meetup](https://www.meetup.com/unijos-r-users/)