Overall Safety Summary

Overview of Adverse Events

Setup

See Prerequisites for installation instructions.

library(arframe)
library(pharmaverseadam)
library(dplyr, warn.conflicts = FALSE)
library(tidyr)
library(cards)

adsl_saf <- pharmaverseadam::adsl |>
  blank_to_na() |>
  filter(SAFFL == "Y", TRT01A != "Screen Failure")

adae <- pharmaverseadam::adae |>
  blank_to_na() |>
  filter(SAFFL == "Y", TRTEMFL == "Y")

arm_levels <- c("Placebo", "Xanomeline Low Dose", "Xanomeline High Dose")
arm_n <- setNames(
  vapply(arm_levels, function(a) sum(adsl_saf$TRT01A == a), integer(1L)),
  arm_levels
)
N_total <- nrow(adsl_saf)
n_vec   <- c(arm_n, Total = N_total)

Data Preparation

# ── Helper: count unique subjects with a condition, per arm + Total ──
ae_flag_row <- function(ae_data, adsl_data, condition, row_label,
                        arm_levels, N_total) {
  condition <- rlang::enquo(condition)

  flag_subjs <- ae_data |>
    filter(!!condition) |>
    distinct(USUBJID, TRT01A)

  by_arm <- flag_subjs |>
    count(TRT01A) |>
    complete(TRT01A = arm_levels, fill = list(n = 0L)) |>
    left_join(
      adsl_data |> count(TRT01A, name = "N"),
      by = "TRT01A"
    ) |>
    mutate(pct = sprintf("%d (%.1f)", n, n / N * 100)) |>
    select(TRT01A, pct) |>
    pivot_wider(names_from = TRT01A, values_from = pct)

  total_n <- n_distinct(flag_subjs$USUBJID)
  by_arm |>
    mutate(
      category   = "",
      stat_label = row_label,
      Total      = sprintf("%d (%.1f)", total_n, total_n / N_total * 100),
      .before    = 1
    )
}

# ── Severity helper: count by max severity per subject ──
severity_rows <- function(ae_data, adsl_data, arm_levels, N_total) {
  sev_order <- c("MILD", "MODERATE", "SEVERE")

  max_sev <- ae_data |>
    mutate(sev_n = match(AESEV, sev_order)) |>
    group_by(USUBJID, TRT01A) |>
    summarise(max_sev = sev_order[max(sev_n, na.rm = TRUE)], .groups = "drop")

  lapply(sev_order, function(s) {
    subjs <- max_sev |> filter(max_sev == s) |> distinct(USUBJID, TRT01A)

    by_arm <- subjs |>
      count(TRT01A) |>
      complete(TRT01A = arm_levels, fill = list(n = 0L)) |>
      left_join(
        adsl_data |> count(TRT01A, name = "N"),
        by = "TRT01A"
      ) |>
      mutate(pct = sprintf("%d (%.1f)", n, n / N * 100)) |>
      select(TRT01A, pct) |>
      pivot_wider(names_from = TRT01A, values_from = pct)

    total_n <- n_distinct(subjs$USUBJID)
    by_arm |>
      mutate(
        category   = "Maximum Severity",
        stat_label = tools::toTitleCase(tolower(s)),
        Total      = sprintf("%d (%.1f)", total_n, total_n / N_total * 100),
        .before    = 1
      )
  }) |> bind_rows()
}

# ── Build all rows matching FDA Table 6 ──
ae_wide <- bind_rows(
  ae_flag_row(adae, adsl_saf, TRUE,
              "Any TEAE", arm_levels, N_total),
  ae_flag_row(adae, adsl_saf, AESER == "Y",
              "Any Serious AE (SAE)", arm_levels, N_total),
  ae_flag_row(adae, adsl_saf, AESDTH == "Y",
              "Any AE Leading to Death", arm_levels, N_total),
  ae_flag_row(adae, adsl_saf, AEREL %in% c("POSSIBLE", "PROBABLE"),
              "Any AE Related to Study Drug", arm_levels, N_total),
  severity_rows(adae, adsl_saf, arm_levels, N_total)
) |>
  mutate(across(
    all_of(c(arm_levels, "Total")),
    ~ replace_na(.x, "0 (0.0)")
  ))

# Build one row per subject with dichotomous AE flags
ae_subj <- adsl_saf |>
  select(USUBJID, TRT01A) |>
  left_join(
    adae |>
      group_by(USUBJID) |>
      summarise(
        any_teae    = any(TRTEMFL == "Y", na.rm = TRUE),
        any_sae     = any(AESER == "Y", na.rm = TRUE),
        any_death   = any(AESDTH == "Y", na.rm = TRUE),
        any_related = any(AEREL %in% c("POSSIBLE", "PROBABLE"), na.rm = TRUE),
        max_sev     = if (all(is.na(AESEV))) NA_character_
                      else c("MILD", "MODERATE", "SEVERE")[
                        max(match(AESEV, c("MILD", "MODERATE", "SEVERE")),
                            na.rm = TRUE)],
        .groups = "drop"
      ),
    by = "USUBJID"
  ) |>
  mutate(
    across(c(any_teae, any_sae, any_death, any_related),
           ~ replace_na(.x, FALSE)),
    max_sev = factor(max_sev, levels = c("MILD", "MODERATE", "SEVERE"))
  )

ae_ard <- ard_stack(
  data = ae_subj,
  .by  = "TRT01A",
  ard_dichotomous(
    variables = c(any_teae, any_sae, any_death, any_related),
    value     = list(
      any_teae    = TRUE,
      any_sae     = TRUE,
      any_death   = TRUE,
      any_related = TRUE
    )
  ),
  ard_categorical(variables = "max_sev"),
  .overall = TRUE
)

ae_wide_cards <- fr_wide_ard(
  ae_ard,
  statistic = list(
    dichotomous = "{n} ({p}%)",
    categorical = "{n} ({p}%)"
  ),
  decimals = c(p = 1),
  label = c(
    any_teae    = "Any TEAE",
    any_sae     = "Any Serious AE (SAE)",
    any_death   = "Any AE Leading to Death",
    any_related = "Any AE Related to Study Drug",
    max_sev     = "Maximum Severity"
  )
)

arframe Pipeline

The rendered table below uses the dplyr data prep (ae_wide). The cards tab produces an equivalent ae_wide_cards — swap it in to use the cards path instead.

ae_wide |>
  fr_table() |>
  fr_titles(
    "Table 14.3.1",
    "Overview of Adverse Events",
    "Safety Population"
  ) |>
  fr_cols(
    category   = fr_col(visible = FALSE),
    stat_label = fr_col("Event", width = 3.2),
    !!!setNames(
      lapply(arm_levels, function(a) fr_col(a, align = "decimal")),
      arm_levels
    ),
    Total = fr_col("Total", align = "decimal"),
    .n = n_vec
  ) |>
  fr_header(bold = TRUE, align = "center") |>
  fr_rows(
    group_by    = list(cols = "category", label = "stat_label"),
    blank_after = "category",
    group_style = list(bold = TRUE)
  ) |>
  fr_styles(
    fr_row_style(rows = 1L, bold = TRUE)
  ) |>
  fr_footnotes(
    "TEAE = Treatment-Emergent Adverse Event (onset on or after first dose of study drug).",
    "Related = Investigator-assessed relationship of Possible or Probable.",
    "Maximum severity = worst severity across all TEAEs per subject.",
    "Subjects counted once per category. Percentages based on N per treatment group.",
    "CDISCPILOT01 Safety Population."
  )

Rendered Table

Table 14.3.1

Overview of Adverse Events

Safety Population

Event	Total (N=254)	Placebo (N=86)	Xanomeline High Dose (N=72)	Xanomeline Low Dose (N=96)

Any TEAE	217 (85.4)	65 (75.6)	68 (94.4)	84 (87.5)
Any Serious AE (SAE)	3 ( 1.2)	0	1 ( 1.4)	2 ( 2.1)
Any AE Leading to Death	3 ( 1.2)	2 ( 2.3)	0	1 ( 1.0)
Any AE Related to Study Drug	184 (72.4)	43 (50.0)	64 (88.9)	77 (80.2)

Maximum Severity
Mild	77 (30.3)	36 (41.9)	20 (27.8)	21 (21.9)
Moderate	111 (43.7)	24 (27.9)	40 (55.6)	47 (49.0)
Severe	29 (11.4)	5 ( 5.8)	8 (11.1)	16 (16.7)

TEAE = Treatment-Emergent Adverse Event (onset on or after first dose of study drug).

Related = Investigator-assessed relationship of Possible or Probable.

Maximum severity = worst severity across all TEAEs per subject.

Subjects counted once per category. Percentages based on N per treatment group.

CDISCPILOT01 Safety Population.

/opt/quarto/share/rmd/rmd.R 01APR2026 09:52:11

--- title: "Overall Safety Summary" subtitle: "Overview of Adverse Events" execute: echo: true eval: true --- ```{r} #| label: prereqs #| include: false library(arframe) fr_theme(hlines = "header", font_family = "Courier New") blank_to_na <- function(df) { df[] <- lapply(df, function(x) { if (is.character(x)) x[x == ""] <- NA_character_ x }) df } ``` ## Setup See [Prerequisites](../install.qmd) for installation instructions. ```{r} #| label: setup library(arframe) library(pharmaverseadam) library(dplyr, warn.conflicts = FALSE) library(tidyr) library(cards) adsl_saf <- pharmaverseadam::adsl |> blank_to_na() |> filter(SAFFL == "Y", TRT01A != "Screen Failure") adae <- pharmaverseadam::adae |> blank_to_na() |> filter(SAFFL == "Y", TRTEMFL == "Y") arm_levels <- c("Placebo", "Xanomeline Low Dose", "Xanomeline High Dose") arm_n <- setNames( vapply(arm_levels, function(a) sum(adsl_saf$TRT01A == a), integer(1L)), arm_levels ) N_total <- nrow(adsl_saf) n_vec <- c(arm_n, Total = N_total) ``` ## Data Preparation ::: {.panel-tabset} ### dplyr ```{r} #| label: dplyr-code # ── Helper: count unique subjects with a condition, per arm + Total ── ae_flag_row <- function(ae_data, adsl_data, condition, row_label, arm_levels, N_total) { condition <- rlang::enquo(condition) flag_subjs <- ae_data |> filter(!!condition) |> distinct(USUBJID, TRT01A) by_arm <- flag_subjs |> count(TRT01A) |> complete(TRT01A = arm_levels, fill = list(n = 0L)) |> left_join( adsl_data |> count(TRT01A, name = "N"), by = "TRT01A" ) |> mutate(pct = sprintf("%d (%.1f)", n, n / N * 100)) |> select(TRT01A, pct) |> pivot_wider(names_from = TRT01A, values_from = pct) total_n <- n_distinct(flag_subjs$USUBJID) by_arm |> mutate( category = "", stat_label = row_label, Total = sprintf("%d (%.1f)", total_n, total_n / N_total * 100), .before = 1 ) } # ── Severity helper: count by max severity per subject ── severity_rows <- function(ae_data, adsl_data, arm_levels, N_total) { sev_order <- c("MILD", "MODERATE", "SEVERE") max_sev <- ae_data |> mutate(sev_n = match(AESEV, sev_order)) |> group_by(USUBJID, TRT01A) |> summarise(max_sev = sev_order[max(sev_n, na.rm = TRUE)], .groups = "drop") lapply(sev_order, function(s) { subjs <- max_sev |> filter(max_sev == s) |> distinct(USUBJID, TRT01A) by_arm <- subjs |> count(TRT01A) |> complete(TRT01A = arm_levels, fill = list(n = 0L)) |> left_join( adsl_data |> count(TRT01A, name = "N"), by = "TRT01A" ) |> mutate(pct = sprintf("%d (%.1f)", n, n / N * 100)) |> select(TRT01A, pct) |> pivot_wider(names_from = TRT01A, values_from = pct) total_n <- n_distinct(subjs$USUBJID) by_arm |> mutate( category = "Maximum Severity", stat_label = tools::toTitleCase(tolower(s)), Total = sprintf("%d (%.1f)", total_n, total_n / N_total * 100), .before = 1 ) }) |> bind_rows() } # ── Build all rows matching FDA Table 6 ── ae_wide <- bind_rows( ae_flag_row(adae, adsl_saf, TRUE, "Any TEAE", arm_levels, N_total), ae_flag_row(adae, adsl_saf, AESER == "Y", "Any Serious AE (SAE)", arm_levels, N_total), ae_flag_row(adae, adsl_saf, AESDTH == "Y", "Any AE Leading to Death", arm_levels, N_total), ae_flag_row(adae, adsl_saf, AEREL %in% c("POSSIBLE", "PROBABLE"), "Any AE Related to Study Drug", arm_levels, N_total), severity_rows(adae, adsl_saf, arm_levels, N_total) ) |> mutate(across( all_of(c(arm_levels, "Total")), ~ replace_na(.x, "0 (0.0)") )) ``` ### cards ```{r} #| label: cards-code # Build one row per subject with dichotomous AE flags ae_subj <- adsl_saf |> select(USUBJID, TRT01A) |> left_join( adae |> group_by(USUBJID) |> summarise( any_teae = any(TRTEMFL == "Y", na.rm = TRUE), any_sae = any(AESER == "Y", na.rm = TRUE), any_death = any(AESDTH == "Y", na.rm = TRUE), any_related = any(AEREL %in% c("POSSIBLE", "PROBABLE"), na.rm = TRUE), max_sev = if (all(is.na(AESEV))) NA_character_ else c("MILD", "MODERATE", "SEVERE")[ max(match(AESEV, c("MILD", "MODERATE", "SEVERE")), na.rm = TRUE)], .groups = "drop" ), by = "USUBJID" ) |> mutate( across(c(any_teae, any_sae, any_death, any_related), ~ replace_na(.x, FALSE)), max_sev = factor(max_sev, levels = c("MILD", "MODERATE", "SEVERE")) ) ae_ard <- ard_stack( data = ae_subj, .by = "TRT01A", ard_dichotomous( variables = c(any_teae, any_sae, any_death, any_related), value = list( any_teae = TRUE, any_sae = TRUE, any_death = TRUE, any_related = TRUE ) ), ard_categorical(variables = "max_sev"), .overall = TRUE ) ae_wide_cards <- fr_wide_ard( ae_ard, statistic = list( dichotomous = "{n} ({p}%)", categorical = "{n} ({p}%)" ), decimals = c(p = 1), label = c( any_teae = "Any TEAE", any_sae = "Any Serious AE (SAE)", any_death = "Any AE Leading to Death", any_related = "Any AE Related to Study Drug", max_sev = "Maximum Severity" ) ) ``` ::: ## arframe Pipeline The rendered table below uses the **dplyr** data prep (`ae_wide`). The cards tab produces an equivalent `ae_wide_cards` — swap it in to use the cards path instead. ```{r} #| label: pipeline #| eval: false ae_wide |> fr_table() |> fr_titles( "Table 14.3.1", "Overview of Adverse Events", "Safety Population" ) |> fr_cols( category = fr_col(visible = FALSE), stat_label = fr_col("Event", width = 3.2), !!!setNames( lapply(arm_levels, function(a) fr_col(a, align = "decimal")), arm_levels ), Total = fr_col("Total", align = "decimal"), .n = n_vec ) |> fr_header(bold = TRUE, align = "center") |> fr_rows( group_by = list(cols = "category", label = "stat_label"), blank_after = "category", group_style = list(bold = TRUE) ) |> fr_styles( fr_row_style(rows = 1L, bold = TRUE) ) |> fr_footnotes( "TEAE = Treatment-Emergent Adverse Event (onset on or after first dose of study drug).", "Related = Investigator-assessed relationship of Possible or Probable.", "Maximum severity = worst severity across all TEAEs per subject.", "Subjects counted once per category. Percentages based on N per treatment group.", "CDISCPILOT01 Safety Population." ) ``` ## Rendered Table ```{r} #| label: table #| echo: false #| ref.label: pipeline ```