R for Tables, Listings, and Graphs (TLG)

Introduction

Transitioning from Clinical SAS to R can feel daunting, especially when working with Tables, Listings, and Graphs (TLG)—the backbone of clinical study reporting. While SAS offers PROC REPORT and ODS for TLGs, R provides flexible and powerful alternatives like dplyr, tables, gt, and ggplot2.

This chapter will guide you through R’s ecosystem for generating tables, listings, and visualizations, mirroring common SAS workflows. By the end, you’ll be able to:
- Create summary tables (like PROC TABULATE or PROC REPORT).
- Generate listings (similar to PROC PRINT).
- Build publication-ready graphs (akin to SGPLOT or GTL).

Let’s dive in!

Tables in R

Clinical trials often require descriptive tables (demographics, adverse events) or summary tables (efficacy endpoints). Below are R alternatives to SAS procedures.

1. Descriptive Tables with dplyr and tables

In SAS, you might use PROC MEANS or PROC FREQ for summaries. In R, dplyr (for data wrangling) and tables (for structured output) are your go-to tools.

Example: Demographics Table

library(dplyr)
library(tidyr)

# Simulated demo data
demo_data <- data.frame(
  SubjectID = 1:50,
  Age = rnorm(50, mean = 45, sd = 10),
  Sex = sample(c("Male", "Female"), 50, replace = TRUE),
  Treatment = sample(c("Placebo", "Drug"), 50, replace = TRUE)
)

# Summary using dplyr
demo_summary <- demo_data %>%
  group_by(Treatment, Sex) %>%
  summarise(
    N = n(),
    Mean_Age = mean(Age, na.rm = TRUE),
    SD_Age = sd(Age, na.rm = TRUE)
  )

print(demo_summary)

Formatted Tables with gt

For polished tables (like SAS ODS), use gt:

library(gt)

demo_summary %>%
  gt() %>%
  tab_header(title = "Demographics Summary") %>%
  fmt_number(columns = c(Mean_Age, SD_Age), decimals = 2) %>%
  cols_label(
    Mean_Age = "Mean Age",
    SD_Age = "SD Age"
  )

Key Functions:
- group_by(): Groups data (like BY in SAS).
- summarise(): Aggregates data (like PROC MEANS).
- gt(): Formats tables professionally.

2. Listings (Like PROC PRINT)

Listings display raw data (e.g., adverse events, lab results). In R, DT (interactive tables) or knitr::kable (static tables) work well.

Example: Adverse Events Listing

library(DT)

# Simulated AE data
ae_data <- data.frame(
  SubjectID = sample(1:50, 30, replace = TRUE),
  Term = sample(c("Headache", "Fatigue", "Nausea"), 30, replace = TRUE),
  Severity = sample(c("Mild", "Moderate", "Severe"), 30, replace = TRUE)
)

# Interactive table
datatable(ae_data, 
          filter = "top",
          options = list(pageLength = 5))

Tip: Use kable for static reports:

knitr::kable(ae_data, caption = "Adverse Events Listing")

Graphs in R (Replacing SGPLOT/GTL)

R’s ggplot2 is the gold standard for graphs. Below are clinical trial examples.

1. Basic Plot: Adverse Events by Severity

library(ggplot2)

ae_data %>%
  count(Term, Severity) %>%
  ggplot(aes(x = Term, y = n, fill = Severity)) +
  geom_bar(stat = "identity", position = "dodge") +
  labs(title = "Adverse Events by Severity", x = "Term", y = "Count") +
  theme_minimal()

Key Layers:
- aes(): Maps variables to aesthetics (like SAS SGPLOT’s x= and y=).
- geom_bar(): Creates bar charts.
- labs(): Adds labels.

2. Survival Plot (Kaplan-Meier)

library(survival)
library(survminer)

# Simulated survival data
surv_data <- data.frame(
  Time = rnorm(50, mean = 365, sd = 100),
  Event = sample(0:1, 50, replace = TRUE),
  Treatment = sample(c("Placebo", "Drug"), 50, replace = TRUE)
)

fit <- survfit(Surv(Time, Event) ~ Treatment, data = surv_data)
ggsurvplot(fit, data = surv_data, risk.table = TRUE)

Comparison to SAS:
- survfit(): Equivalent to PROC LIFETEST.
- ggsurvplot(): Like PROC SGPLOT’s survival curves.

Advanced: Automated TLGs with rtables

For regulatory submissions, rtables (from Roche) provides a framework akin to SAS TLGs:

library(rtables)

lyt <- basic_table() %>%
  split_cols_by("Treatment") %>%
  analyze_vars("Age", format = "xx.xx")

tbl <- build_table(lyt, demo_data)
tbl

Why Use rtables?
- Reproducible, structured outputs.
- Alignment with CDISC standards.

Key Takeaways

1. Tables:
2. Use dplyr for summaries, gt for formatting.

3. Replace PROC PRINT with DT or kable.

4. Graphs:

5. ggplot2 replaces SGPLOT/GTL.

6. survminer handles survival plots.

7. Regulatory-Grade Outputs:

8. rtables for structured TLGs.

By mastering these tools, you’ll confidently transition from SAS to R for clinical reporting!

Next Steps:
- Practice with real-world datasets (e.g., ADaM-like data in R).
- Explore Tplyr (an R package mimicking SAS PROC TABULATE).

Happy coding! 🚀