Hands on with Cardinal: Harmonizing Clinical Reporting

Open-Source TLGs with {gtsummary}

Clinical Reporting

TLGs

Beginner

Authors

Abinaya Yogasekaram (Amaris Consulting)

Emily De La Rua (Roche)

Overview

Beginner Friendly Clinical Reporting TLGs

Cardinal is an open-source collection of standardized table, listing, and graph (TLG) templates designed to streamline the process of clinical output review, comparison, and meta-analysis—promoting efficient communication to stakeholders while aligning with CDISC’s ARD/ARM efforts.

What You’ll Learn

📊 {gtsummary} fundamentals for table creation
📚 Cardinal’s TLG catalog and template library
🔧 Hands-on practice generating key clinical outputs
🤝 Contributing to the Cardinal project
🎯 Integration into your own workflows

Prerequisites

Required Knowledge:

Basic R programming
Familiarity with clinical trial data (CDISC preferred)
No prior experience with {gtsummary} required

Recommended:

Understanding of common TLG types
Experience with clinical study reports

Key Packages & Tools

{gtsummary}

{cardinal}

{gt}

{cards}

CDISC ADaM

Why Cardinal?

The Problem

🔄 Inconsistent formatting across studies and organizations
⏱️ Time-consuming manual table creation
🔍 Difficult to compare outputs across trials
📝 Repetitive coding for standard analyses

The Solution

Cardinal provides:

✅ Standardized templates for common TLGs
✅ Quality-controlled outputs aligned with CDISC
✅ Reusable code reducing development time
✅ Harmonized formats for meta-analysis

Introduction to {gtsummary}

{gtsummary} is the foundation for creating publication-ready tables:

library(gtsummary)
library(dplyr)

# Simple demographic table
trial %>%
  select(age, grade, trt) %>%
  tbl_summary(
    by = trt,
    statistic = list(
      all_continuous() ~ "{mean} ({sd})",
      all_categorical() ~ "{n} ({p}%)"
    )
  ) %>%
  add_p() %>%
  add_overall()

Key Features

Intuitive syntax for common analyses
Flexible customization options
Built-in statistical tests
Export to multiple formats (Word, HTML, LaTeX)

Cardinal TLG Catalog

Tables

1. Demographics and Baseline Characteristics

Subject disposition
Demographics (age, sex, race, ethnicity)
Baseline disease characteristics
Medical history

2. Efficacy Tables

Primary/secondary endpoints
Subgroup analyses
Time-to-event summaries
Response rates

3. Safety Tables

Adverse events (by SOC, PT, severity)
Laboratory values
Vital signs
Deaths and serious AEs

4. Disposition Tables

Study completion
Protocol deviations
Exposure duration

Listings

Patient profiles
Adverse event listings
Laboratory abnormalities
Concomitant medications

Graphs

Kaplan-Meier survival curves
Forest plots (subgroup analysis)
Swimmer plots (treatment duration)
Waterfall plots (tumor response)

Workshop Content

Session 1: Getting Started with {gtsummary}

Basic Table Creation:

library(gtsummary)
library(cardinal)

# Load example CDISC data
data("adsl", package = "cardinal")

# Demographic table
demo_table <- adsl %>%
  select(AGE, SEX, RACE, ARM) %>%
  tbl_summary(
    by = ARM,
    label = list(
      AGE ~ "Age (years)",
      SEX ~ "Sex",
      RACE ~ "Race"
    )
  ) %>%
  add_p() %>%
  add_overall() %>%
  modify_header(label ~ "**Characteristic**") %>%
  bold_labels()

Session 2: Cardinal Templates

Using Pre-built Templates:

# Cardinal provides ready-to-use templates
template_demographics(
  data = adsl,
  treatment_var = "ARM"
)

template_ae_summary(
  data = adae,
  treatment_var = "ARM",
  ae_vars = c("AESOC", "AEDECOD")
)

Session 3: Customization

Adapting Templates to Your Needs:

Modifying statistical summaries
Custom formatting rules
Adding footnotes and titles
Styling for different audiences (clinical vs. regulatory)

Session 4: Advanced Topics

Complex Tables:

Multi-level headers
Nested statistics
Cross-tabulations
Survival analysis tables

Session 5: Contributing to Cardinal

How to Get Involved:

Submit new template ideas
Contribute code improvements
Report issues and bugs
Share use cases and examples

Hands-On Exercises

Exercise 1: Basic Demographics Table

Create a standard demographics table using your own data or provided examples.

Exercise 2: Adverse Events Summary

Generate a treatment-emergent adverse events (TEAE) table by system organ class.

Exercise 3: Efficacy Analysis

Build a primary endpoint analysis table with statistical comparisons.

Exercise 4: Custom Template

Modify an existing Cardinal template to match your organization’s standards.

Practical Applications

Clinical Study Reports (CSRs)

Consistent formatting across studies
Rapid generation of standard tables
Easy updates during review cycles

Meta-Analysis

Harmonized outputs from multiple trials
Simplified data extraction
Comparable statistics across studies

Regulatory Submissions

CDISC-aligned outputs
Traceable and reproducible code
Audit-ready documentation

Cardinal vs. Other Approaches

Feature	Cardinal + {gtsummary}	Traditional Approach	Other Tools
Standardization	✅ High	❌ Low	⚠️ Medium
Learning Curve	✅ Gentle	❌ Steep	⚠️ Moderate
Flexibility	✅ High	✅ High	❌ Limited
Open Source	✅ Yes	⚠️ Varies	❌ Often proprietary
CDISC Alignment	✅ Built-in	❌ Manual	⚠️ Varies
Cost	✅ Free	⚠️ Varies	❌ Often expensive

ARD/ARM Integration

Cardinal aligns with CDISC Analysis Results Data (ARD) and Analysis Results Metadata (ARM) standards:

Structured output metadata
Machine-readable results
Facilitates automation
Supports regulatory review

Learning Outcomes

By the end of this workshop, you will be able to:

✅ Create professional clinical tables with {gtsummary}
✅ Use Cardinal templates for standard TLGs
✅ Customize outputs to match your organization’s needs
✅ Generate CDISC-aligned, reproducible tables
✅ Contribute to the Cardinal project
✅ Accelerate your clinical reporting workflows

Workshop Materials

Resources

Cardinal Website: https://pharmaverse.github.io/cardinal/

GitHub: https://github.com/pharmaverse/cardinal

{gtsummary} Documentation: https://www.danieldsjoberg.com/gtsummary/

Example: Complete Workflow

library(cardinal)
library(gtsummary)
library(gt)

# 1. Load data
data("adsl")
data("adae")

# 2. Create demographics table
demo <- template_demographics(adsl, treatment_var = "ARM")

# 3. Create AE summary
ae_summary <- template_ae_summary(
  adae,
  treatment_var = "ARM",
  soc_var = "AESOC",
  pt_var = "AEDECOD"
)

# 4. Export to Word
demo %>%
  as_gt() %>%
  gt::gtsave("demographics.docx")

ae_summary %>%
  as_gt() %>%
  gt::gtsave("adverse_events.docx")

Future of Cardinal

Roadmap:

Expanded template library
More graph types
Enhanced CDISC integration
Integration with {teal} and other pharmaverse tools
Machine learning-assisted table suggestions

Getting Help

Documentation: Comprehensive guides on the website
GitHub Issues: Report bugs and request features
Slack: pharmaverse Slack channel
Email: Contact maintainers directly

Next Steps

After this workshop:

Explore the full Cardinal template catalog
Try Cardinal in your next project
Join the pharmaverse community
Consider contributing a template

Pro Tips

Start with templates - Don’t reinvent the wheel
Customize gradually - Master basics before advanced features
Document your modifications - Make it reproducible
Share back - Your improvements can help others
Stay updated - Cardinal is actively developed

Similar Workshops

Advanced Clinical Reporting - Word documents with officer/flextable
Polished Documents with Quarto - Multi-format publishing

Tools & Resources

{gtsummary} - Most mentioned package
{cardinal} - Full catalog entry

Next Steps

For Word output: Try officer/flextable workshop
Career skills: Clinical Reporting Automation

Last updated: November 2025 | R/Pharma 2025 Conference

--- title: "Hands on with Cardinal: Harmonizing Clinical Reporting" subtitle: "Open-Source TLGs with {gtsummary}" author: - "Abinaya Yogasekaram (Amaris Consulting)" - "Emily De La Rua (Roche)" categories: [Clinical Reporting, TLGs, Beginner] --- ## Overview [Beginner Friendly]{.badge .badge-beginner} [Clinical Reporting]{.badge .badge-category} [TLGs]{.badge .badge-category} **Cardinal** is an open-source collection of standardized table, listing, and graph (TLG) templates designed to streamline the process of clinical output review, comparison, and meta-analysis—promoting efficient communication to stakeholders while aligning with CDISC's ARD/ARM efforts. ### What You'll Learn - 📊 **{gtsummary} fundamentals** for table creation - 📚 **Cardinal's TLG catalog** and template library - 🔧 **Hands-on practice** generating key clinical outputs - 🤝 **Contributing** to the Cardinal project - 🎯 **Integration** into your own workflows ## Prerequisites ::: requirements **Required Knowledge:** - Basic R programming - Familiarity with clinical trial data (CDISC preferred) - No prior experience with {gtsummary} required **Recommended:** - Understanding of common TLG types - Experience with clinical study reports ::: ## Key Packages & Tools ::: tool-tag {gtsummary} ::: ::: tool-tag {cardinal} ::: ::: tool-tag {gt} ::: ::: tool-tag {cards} ::: ::: tool-tag CDISC ADaM ::: ## Why Cardinal? ### The Problem - 🔄 **Inconsistent formatting** across studies and organizations - ⏱️ **Time-consuming** manual table creation - 🔍 **Difficult to compare** outputs across trials - 📝 **Repetitive coding** for standard analyses ### The Solution Cardinal provides: - ✅ **Standardized templates** for common TLGs - ✅ **Quality-controlled** outputs aligned with CDISC - ✅ **Reusable code** reducing development time - ✅ **Harmonized formats** for meta-analysis ## Introduction to {gtsummary} **{gtsummary}** is the foundation for creating publication-ready tables: ``` r library(gtsummary) library(dplyr) # Simple demographic table trial %>% select(age, grade, trt) %>% tbl_summary( by = trt, statistic = list( all_continuous() ~ "{mean} ({sd})", all_categorical() ~ "{n} ({p}%)" ) ) %>% add_p() %>% add_overall() ``` ### Key Features - Intuitive syntax for common analyses - Flexible customization options - Built-in statistical tests - Export to multiple formats (Word, HTML, LaTeX) ## Cardinal TLG Catalog ### Tables #### 1. **Demographics and Baseline Characteristics** - Subject disposition - Demographics (age, sex, race, ethnicity) - Baseline disease characteristics - Medical history #### 2. **Efficacy Tables** - Primary/secondary endpoints - Subgroup analyses - Time-to-event summaries - Response rates #### 3. **Safety Tables** - Adverse events (by SOC, PT, severity) - Laboratory values - Vital signs - Deaths and serious AEs #### 4. **Disposition Tables** - Study completion - Protocol deviations - Exposure duration ### Listings - Patient profiles - Adverse event listings - Laboratory abnormalities - Concomitant medications ### Graphs - Kaplan-Meier survival curves - Forest plots (subgroup analysis) - Swimmer plots (treatment duration) - Waterfall plots (tumor response) ## Workshop Content ### Session 1: Getting Started with {gtsummary} **Basic Table Creation:** ``` r library(gtsummary) library(cardinal) # Load example CDISC data data("adsl", package = "cardinal") # Demographic table demo_table <- adsl %>% select(AGE, SEX, RACE, ARM) %>% tbl_summary( by = ARM, label = list( AGE ~ "Age (years)", SEX ~ "Sex", RACE ~ "Race" ) ) %>% add_p() %>% add_overall() %>% modify_header(label ~ "**Characteristic**") %>% bold_labels() ``` ### Session 2: Cardinal Templates **Using Pre-built Templates:** ``` r # Cardinal provides ready-to-use templates template_demographics( data = adsl, treatment_var = "ARM" ) template_ae_summary( data = adae, treatment_var = "ARM", ae_vars = c("AESOC", "AEDECOD") ) ``` ### Session 3: Customization **Adapting Templates to Your Needs:** - Modifying statistical summaries - Custom formatting rules - Adding footnotes and titles - Styling for different audiences (clinical vs. regulatory) ### Session 4: Advanced Topics **Complex Tables:** - Multi-level headers - Nested statistics - Cross-tabulations - Survival analysis tables ### Session 5: Contributing to Cardinal **How to Get Involved:** 1. Submit new template ideas 2. Contribute code improvements 3. Report issues and bugs 4. Share use cases and examples ## Hands-On Exercises ### Exercise 1: Basic Demographics Table Create a standard demographics table using your own data or provided examples. ### Exercise 2: Adverse Events Summary Generate a treatment-emergent adverse events (TEAE) table by system organ class. ### Exercise 3: Efficacy Analysis Build a primary endpoint analysis table with statistical comparisons. ### Exercise 4: Custom Template Modify an existing Cardinal template to match your organization's standards. ## Practical Applications ### Clinical Study Reports (CSRs) - Consistent formatting across studies - Rapid generation of standard tables - Easy updates during review cycles ### Meta-Analysis - Harmonized outputs from multiple trials - Simplified data extraction - Comparable statistics across studies ### Regulatory Submissions - CDISC-aligned outputs - Traceable and reproducible code - Audit-ready documentation ## Cardinal vs. Other Approaches | Feature | Cardinal + {gtsummary} | Traditional Approach | Other Tools | |-----------------|--------------------|-------------------|-----------------| | **Standardization** | ✅ High | ❌ Low | ⚠️ Medium | | **Learning Curve** | ✅ Gentle | ❌ Steep | ⚠️ Moderate | | **Flexibility** | ✅ High | ✅ High | ❌ Limited | | **Open Source** | ✅ Yes | ⚠️ Varies | ❌ Often proprietary | | **CDISC Alignment** | ✅ Built-in | ❌ Manual | ⚠️ Varies | | **Cost** | ✅ Free | ⚠️ Varies | ❌ Often expensive | ## ARD/ARM Integration Cardinal aligns with **CDISC Analysis Results Data (ARD)** and **Analysis Results Metadata (ARM)** standards: - Structured output metadata - Machine-readable results - Facilitates automation - Supports regulatory review ## Learning Outcomes By the end of this workshop, you will be able to: ✅ Create professional clinical tables with {gtsummary}\ ✅ Use Cardinal templates for standard TLGs\ ✅ Customize outputs to match your organization's needs\ ✅ Generate CDISC-aligned, reproducible tables\ ✅ Contribute to the Cardinal project\ ✅ Accelerate your clinical reporting workflows ## Workshop Materials ::: callout-note ## Resources **Cardinal Website:** <https://pharmaverse.github.io/cardinal/> **GitHub:** <https://github.com/pharmaverse/cardinal> **{gtsummary} Documentation:** <https://www.danieldsjoberg.com/gtsummary/> ::: ## Example: Complete Workflow ``` r library(cardinal) library(gtsummary) library(gt) # 1. Load data data("adsl") data("adae") # 2. Create demographics table demo <- template_demographics(adsl, treatment_var = "ARM") # 3. Create AE summary ae_summary <- template_ae_summary( adae, treatment_var = "ARM", soc_var = "AESOC", pt_var = "AEDECOD" ) # 4. Export to Word demo %>% as_gt() %>% gt::gtsave("demographics.docx") ae_summary %>% as_gt() %>% gt::gtsave("adverse_events.docx") ``` ## Future of Cardinal **Roadmap:** - Expanded template library - More graph types - Enhanced CDISC integration - Integration with {teal} and other pharmaverse tools - Machine learning-assisted table suggestions ## Getting Help - **Documentation:** Comprehensive guides on the website - **GitHub Issues:** Report bugs and request features - **Slack:** pharmaverse Slack channel - **Email:** Contact maintainers directly ## Next Steps After this workshop: - Explore the full Cardinal template catalog - Try Cardinal in your next project - Join the pharmaverse community - Consider contributing a template ::: callout-tip ## Pro Tips 1. **Start with templates** - Don't reinvent the wheel 2. **Customize gradually** - Master basics before advanced features 3. **Document your modifications** - Make it reproducible 4. **Share back** - Your improvements can help others 5. **Stay updated** - Cardinal is actively developed ::: ------------------------------------------------------------------------ ### Similar Workshops - [Advanced Clinical Reporting](officer-flextable.qmd) - Word documents with officer/flextable - [Polished Documents with Quarto](branded-quarto.qmd) - Multi-format publishing ### Related Presentations - [Beyond {gtsummary}: {crane}](../presentations/europe-us-sessions.qmd#beyond-gtsummary-the-crane-package) - Extension for pharma - [LLM-Powered {gtsummary}](../presentations/europe-us-sessions.qmd#llm-powered-gtsummary-qc-ready-tables) - AI-enhanced tables ### Tools & Resources - [{gtsummary}](../summary/tools-catalog.qmd#gtsummary) - Most mentioned package - [{cardinal}](../summary/tools-catalog.qmd#cardinal) - Full catalog entry ### Next Steps - **For Word output:** Try [officer/flextable workshop](officer-flextable.qmd) - **Career skills:** [Clinical Reporting Automation](../summary/career-insights.qmd#2-clinical-reporting-automation) ------------------------------------------------------------------------ *Last updated: November 2025 \| R/Pharma 2025 Conference*