exercises.qmd

---
title: "Introducing {gtreg}: Exercises"
format: 
  html:
    code-fold: true
    code-summary: "Show the code solution"
---

# Instructions:

**Recommended:** Complete these exercises in the dedicated R in Pharma RStudio Cloud work space, which comes with 

1. all packages pre-installed, and

2. an Rmarkdown document to fill in.

It may still be helpful to peek here to verify that your tables match the 
desired output.

[Click here to enter R in Pharma RStudio Cloud work space for {gtreg}](https://rstudio.cloud/spaces/299322/join?access_code=Vnp96stsVIDLiEJI7AtHAQ4GB1gsptQgnOGb99_N){target="_blank"}

**Otherwise:** Follow along this document, work on your personal computer, 
and challenge yourself not to peek at the code solutions until you have completed
the exercise.

# Packages

```{r}
#| label: packages
#| message: false
#| echo: true
#| code-fold: false
library(tidyverse)
library(gtsummary)
library(gtreg)
library(labelled)
library(admiral)       # for subject data   
library(admiral.test)  # for ae data
```

# Output display setting

```{r}
#| label: output-setting
#| message: false
#| echo: true
#| code-fold: false
theme_gtsummary_compact()
```



# The Data

For these exercises we will be using STDM and ADaM datasets from the [{admiral.test}](https://github.com/pharmaverse/admiral.test) and [{admiral}](https://pharmaverse.github.io/admiral/index.html) and packages.


First, prepare a subject level data set named `subjects` from the `admiral_adsl` data set
with only the variables `USUBJID`, `ACTARM`, `AGE`, `SEX`, `RACE` and subjects who screen failed removed.

```{r}
#| label: prep-subjects
#| echo: true
#| code-fold: false
dat_subjects <- admiral_adsl %>% 
  filter(ACTARM != "Screen Failure" & ACTARM %in% c("Placebo", "Xanomeline High Dose") ) %>% 
  select(USUBJID, ACTARM, AGE, SEX, RACE) %>% 
  mutate(across(RACE, str_to_title)) %>% 
  mutate(
    # convert treatment arm and race to factors for table displays in a specific order
    ACTARM = fct_relevel(ACTARM, "Placebo", "Xanomeline High Dose"),
    RACE = fct_infreq(RACE)
  ) %>% 
  set_variable_labels(
    ACTARM = "Actual treatment arm",
    RACE = "Race"
  )
```

Next, prepare an adverse event data set named `ae` from `admiral_ae` with the 
only the variables `USUBJID`, `AESOC`, `AEDECOD`, `AESEV`, `AESER`, `AEREL`, `AESTDTC`.

Join actual treatment arm (`ACTARM`) from the `subjects` data frame to the `ae` data.



```{r}
#| label: prep-aes
#| echo: true
#| code-fold: false
dat_ae <- admiral_ae %>% 
  select(USUBJID, AESOC, AEDECOD, AESEV, AESER, AEREL, AESTDTC) %>% 
  left_join(dat_subjects %>% select(USUBJID, ACTARM), by = "USUBJID") %>% 
  # removing subjects  with missing treatment arm as these were the low dose 
  # subjects excluded from the subjects data set
  drop_na(ACTARM) %>% 
  mutate(
    AESEV = fct_relevel(AESEV, "MILD", "MODERATE", "SEVERE"),
    # converting treatment arm to factor
    # note that treatment arm must be stored the same way in the two data sets
    ACTARM = fct_relevel(ACTARM, "Placebo", "Xanomeline High Dose"),
    year = str_sub(AESTDTC, start = 1, end = 4) %>% as.numeric()
   ) %>%
  mutate(across(c(AESOC, AEDECOD, AESEV, AEREL), str_to_title)) %>% 
  # for brevity of output, only look at AEs from a single year and in
  # a few SOC categories
  filter(year == 2014) %>% 
  filter(AESOC %in% c("Eye Disorders", "Nervous System Disorders", "Skin And Subcutaneous Tissue Disorders")) %>%
  labelled::set_variable_labels(
     AESOC    = "Primary System Organ Class", 
     AEDECOD  = "Dictionary-Derived Term", 
     AESEV    = "Severity/Intensity",
     AEREL    = "Causality" 
  )
```


# Exercise 1

Create an adverse event table saved as `t1` that summarizes adverse events.

1. Summarize the adverse events by both treatment arm and severity.

2. Supply `id_df` to ensure the subject denominator is correct.

```{r ae-first}
t1 <- dat_ae %>%
  tbl_ae(
    id_df = dat_subjects,
    id = USUBJID,
    ae = AEDECOD,
    soc = AESOC,
    by = AESEV,
    strata = ACTARM
  ) 

t1
```

# Exercise 2

Create `t1_modify` that modifies `t1` such that:

1. Bold the system organ class rows.

2. Add an overall column to each treatment arm.

3. Label the overall column as `Total`.

4. In the header, place the number of subjects on a new line in each treatment arm. 

5. Add a table caption that states `2014 Disorders Adverse Events` (displays at the top of the table).

```{r ae-modify}
t1_modify <- t1 %>% 
  bold_labels() %>%
  add_overall(across = 'by') %>% 
  modify_header(all_overall_cols() ~ "**Total**") %>%
  modify_spanning_header(all_ae_cols(TRUE, TRUE) ~ "**{strata}**  \nN = {n}") %>%
  modify_caption("2014 Disorders Adverse Events")  

t1_modify
```

# Exercise 3

Create a descriptive statistics table from `subjects` named `t2` using `tbl_reg_summary`.

1. Summarize the variables `AGE`, `SEX` and `RACE` by treatment arm (`ACTARM`).

2. Add an overall column to summarize across both treatment arms.

3. Bold the variable labels.

4. In the header, place the number of subjects on a new line in each treatment arm. 

```{r summary-first}
t2 <- dat_subjects %>% 
  select(-USUBJID) %>% 
  tbl_reg_summary(
    by = ACTARM
  ) %>% 
  add_overall(last = TRUE) %>% 
  bold_labels() %>% 
  modify_header(
    list(
      all_stat_cols(stat_0 = FALSE) ~ "**{level}**  \nN = {n}",
      stat_0 ~ "**Overall**  \nN = {N}"
    ))

t2
```

# Exercise 4

Create a grouped table listing of adverse events named `t3`.

1. Display only events of `Probable` causality.

2. Sort events by system organ class and term.

3. Print subject id, term, and whether or not the event was serious.

4. Group listings by system organ class.

5. Bold system organ class labels.



```{r listing}
dat_list <- dat_ae %>% 
  filter(AEREL == "Probable") %>% 
  select(USUBJID, AESOC, AEDECOD, AESER) %>% 
  arrange(AESOC, AEDECOD) 

t3 <- dat_list %>% 
  tbl_listing(
    group_by = AESOC
  ) %>% 
  bold_labels() 

t3
```

# Exercise 5

Create `t2_shell`, a table shell from the `t2` table. This is the 
exact same format as `t2`, except with numeric values replaced by `xx`.

```{r summary-shell}
t2_shell <- dat_subjects %>% 
  select(-USUBJID) %>% 
  tbl_reg_summary(
    digits = everything() ~  style_xxx,
    by = ACTARM
  ) %>% 
  add_overall(last = TRUE) %>% 
  bold_labels() %>% 
  modify_header(
    list(
      all_stat_cols(stat_0 = FALSE) ~ "**{level}**  \nN = xx",
      stat_0 ~ "**Overall**  \nN = xx"
    ))

t2_shell
```


# Exercise 6

Export `t2_shell` to either word or excel.

```{r export-xlsx, eval=FALSE}
t2_shell %>%  
 as_hux_xlsx(
   file = "output/summary_shell.xlsx"
   )
```

```{r export-word, eval=FALSE}
t2_shell %>%  
  as_flex_table() %>% 
  flextable::save_as_docx(path = "output/summary_shell.docx")
```

# Session info

```{r}
#| label: session-info
#| eval: true
#| code-fold: false
devtools::session_info()
```