tlf-efficacy-km.qmd

# Efficacy figure

```{r, include=FALSE}
source("_common.R")
```

Following the [ICH E3 guidance](https://database.ich.org/sites/default/files/E3_Guideline.pdf),
primary and secondary efficacy endpoints need to be summarized
in Section 11.4, Efficacy Results and Tabulations of Individual Participant.

```{r}
library(haven) # Read SAS data
library(dplyr) # Manipulate data
library(r2rtf) # Reporting in RTF format
library(survival) # Fit survival model
```

In this chapter, we illustrate how to create a simplified Kaplan-Meier plot in a study.
For the survival analysis in efficacy, time to dermatologic event (TTDE) will be analyzed.

::: {.callout-note}
R packages such as
[visR](https://cran.r-project.org/package=visR) and
[survminer](https://cran.r-project.org/package=survminer)
can create more informative Kaplan-Meier plots.
Interested readers can find examples on their websites.
:::

## Analysis dataset

To prepare the analysis, the `adtte` dataset is required.

```{r}
adtte <- read_sas("data-adam/adtte.sas7bdat")
```

First, to prepare the analysis ready data,
filter all records for the efficacy endpoint of time to event of interest (`TTDE`)
using `PARAMCD` (or `PARAM`, `PRAMN`), then select the survival analysis related variables:

- `TRTP`: treatment arm (using corresponding numeric code `TRTAN` to re-order the levels, "Placebo" will be the reference level)
- `AVAL`: time-to-event analysis value
- `CNSR`: event (censoring) status

```{r}
adtte_ttde <- adtte %>%
  filter(PARAMCD == "TTDE") %>%
  select(TRTP, TRTAN, AVAL, CNSR) %>%
  mutate(
    TRTP = forcats::fct_reorder(TRTP, TRTAN), # Recorder levels
    AVAL_m = AVAL / 30.4367 # Convert Day to Month
  )
```

## Create Kaplan-Meier curve

The survival package is used to obtain the K-M estimate.

```{r}
# Fit survival model, convert the time value from Days to Month
fit <- survfit(Surv(AVAL_m, 1 - CNSR) ~ TRTP, data = adtte_ttde)
```

We save the simplified K-M plot into a `.png` file using code below.

```{r, results = FALSE}
# Save as a PNG file
png(
  file = "tlf/fig_km.png",
  width = 3000,
  height = 2000,
  res = 300
)

plot(
  fit,
  xlab = "Time in Months",
  ylab = "Survival probability",
  mark.time = TRUE,
  lwd = 2,
  col = c(2, 3, 4),
  lty = c(1, 2, 3)
)

dev.off()
```

Now, we can use the r2rtf package to create a formatted RTF figure.
More details can be found on the [r2rtf website](https://merck.github.io/r2rtf/articles/example-figure.html).

```{r}
# Create RTF figure
rtf_read_figure("tlf/fig_km.png") %>% # Read the PNG file from the file path
  rtf_title(
    "Kaplan-Meier Plot for Time to First Dermatologic Event by Treatment Group",
    "All Participants"
  ) %>% # Add title or subtitle
  rtf_footnote("footnote") %>% # Add footnote
  rtf_source("[datasource: adam-adtte]") %>% # Add data source
  rtf_figure(fig_width = 6, fig_height = 4) %>% # Set proportional figure size to the original PNG figure size
  rtf_encode(doc_type = "figure") %>% # Encode figure as rtf
  write_rtf(file = "tlf/tlf_km.rtf")
```

```{r, include=FALSE}
rtf2pdf("tlf/tlf_km.rtf")
```

```{r, out.width = "100%", out.height = if (knitr::is_html_output()) "400px", echo = FALSE, fig.align = "center"}
knitr::include_graphics("tlf/tlf_km.pdf")
```

In conclusion, the steps to create a K-M plot are as follows.

- Step 1: Read the data `adtte` into R.
- Step 2: Define the analysis-ready dataset. In this example, we define the analysis dataset for the TTDE endpoint `adtte_ttde`.
- Step 3: Save figures into `png` files based on required analysis specification.
- Step 4: Create RTF output using the r2rtf package.