Clinical Trial Analysis: Full End-to-End SDTM to ADaM Pipeline

CDISC-Compliant ADaM Dataset Construction and TLF Generation

Author

Ndoh Penn

Published

April 17, 2026

1. Setup and Environment

Introduction

This document presents a full end-to-end clinical trial analysis pipeline, covering:

  1. SDTM Data Loading — Demographics (DM), Adverse Events (AE), Exposure (EX), Disposition (DS), Vital Signs (VS)
  2. ADaM Dataset Construction — ADSL, ADAE, ADVS
  3. Quality Checks — Record counts, population flags, missing data summaries
  4. Tables, Listings, and Figures (TLFs) — Demographic summary, AE incidence table, Kaplan–Meier survival curve, vital signs over time

All derivations follow CDISC ADaM Implementation Guide conventions using the admiral package from the pharmaverse ecosystem.

2. Data Loading

In a production environment, SDTM .xpt files are loaded from a validated data transfer location. Here we use the admiral.test sample datasets to ensure full reproducibility.

Show Code 
# ---------------------------------------------------------------------------
# PRODUCTION: read from validated SDTM delivery path, e.g.:
#   sdtm_path <- "data/sdtm/"
#   dm <- read_xpt(file.path(sdtm_path, "dm.xpt"))
#   ae <- read_xpt(file.path(sdtm_path, "ae.xpt"))
# ---------------------------------------------------------------------------

data(admiral_dm)
data(admiral_ae)
data(admiral_ex)
data(admiral_ds)
data(admiral_vs)

# Rename to conventional SDTM domain names for clarity
dm <- admiral_dm
ae <- admiral_ae
ex <- admiral_ex
ds <- admiral_ds
vs <- admiral_vs

# Quick inventory check
domain_summary <- tibble::tibble(
  Domain    = c("DM", "AE", "EX", "DS", "VS"),
  Rows      = c(nrow(dm), nrow(ae), nrow(ex), nrow(ds), nrow(vs)),
  Subjects  = c(
    n_distinct(dm$USUBJID),
    n_distinct(ae$USUBJID),
    n_distinct(ex$USUBJID),
    n_distinct(ds$USUBJID),
    n_distinct(vs$USUBJID)
  ),
  Description = c(
    "Demographics",
    "Adverse Events",
    "Exposure",
    "Disposition",
    "Vital Signs"
  )
)

domain_summary |>
  gt() |>
  tab_header(title = "SDTM Domain Inventory") |>
  tab_style(
    style = cell_fill(color = "#2E75B6", alpha = 0.15),
    locations = cells_column_labels()
  ) |>
  fmt_number(columns = c(Rows, Subjects), decimals = 0)
SDTM Domain Inventory
Domain Rows Subjects Description
DM 306 306 Demographics
AE 1,191 225 Adverse Events
EX 591 254 Exposure
DS 850 306 Disposition
VS 29,643 254 Vital Signs

3. ADSL — Subject-Level Analysis Dataset

ADSL contains one row per subject and is the foundation for all subsequent ADaM datasets.

3.1 Treatment Variables

Show Code 
adsl <- dm |>
  # ---- Planned & Actual Treatment from first non-zero dose record ----
  admiral::derive_vars_merged(
    dataset_add = ex,
    by_vars    = exprs(USUBJID),
    order      = exprs(EXSEQ),
    mode       = "first",
    new_vars   = exprs(TRT01P = EXTRT, TRT01A = EXTRT),
    filter_add = EXDOSE > 0
  )

3.2 Safety Population Flag

Show Code 
adsl <- adsl |>
  # SAFFL = "Y" if subject received at least one dose
  admiral::derive_var_merged_exist_flag(
    dataset_add = ex,
    by_vars     = exprs(USUBJID),
    new_var     = SAFFL,
    condition   = EXDOSE > 0
  )

3.3 Intent-to-Treat Population Flag

Show Code 
adsl <- adsl |>
  # ITTFL = "Y" for all randomised subjects (ARM is not screen failure / not missing)
  dplyr::mutate(
    ITTFL = dplyr::if_else(
      !is.na(ARM) & ARM != "Screen Failure",
      "Y", "N",
      missing = "N"
    )
  )

3.4 Study Dates and Duration

Show Code 
adsl <- adsl |>
  # Convert ISO 8601 character dates to SAS-style numeric date variables
  admiral::derive_vars_dt(
    new_vars_prefix = "TRTS",
    dtc             = RFSTDTC   # Reference Start Date (first dose)
  ) |>
  admiral::derive_vars_dt(
    new_vars_prefix = "TRTE",
    dtc             = RFENDTC   # Reference End Date (last dose)
  ) |>
  # Duration on treatment (days)
  dplyr::mutate(
    TRTDURD = as.numeric(TRTEDT - TRTSDT) + 1L
  )

3.5 Baseline Age Group

Show Code 
adsl <- adsl |>
  dplyr::mutate(
    AGEGR1 = dplyr::case_when(
      AGE <  65             ~ "<65",
      AGE >= 65 & AGE < 75  ~ "65–74",
      AGE >= 75             ~ ">=75",
      TRUE                  ~ NA_character_
    ),
    AGEGR1 = factor(AGEGR1, levels = c("<65", "65–74", ">=75"))
  )

3.6 ADSL Quality Check

Show Code 
cat("========== ADSL Quality Check ==========\n")
========== ADSL Quality Check ==========
Show Code 
cat("Total subjects:         ", nrow(adsl), "\n")
Total subjects:          306
Show Code 
cat("Unique USUBJIDs:        ", n_distinct(adsl$USUBJID), "\n")
Unique USUBJIDs:         306
Show Code 
cat("Duplicates (should=0):  ", nrow(adsl) - n_distinct(adsl$USUBJID), "\n\n")
Duplicates (should=0):   0
Show Code 
cat("Population flags:\n")
Population flags:
Show Code 
adsl |>
  dplyr::count(SAFFL, ITTFL) |>
  dplyr::rename(`SAFFL` = SAFFL, `ITTFL` = ITTFL, `N` = n) |>
  knitr::kable() |>
  kableExtra::kable_styling(bootstrap_options = "striped", full_width = FALSE)
SAFFL ITTFL N
Y Y 168
NA N 52
NA Y 86

4. ADAE — Adverse Event Analysis Dataset

ADAE contains one row per adverse event per subject and is the primary dataset for safety analyses.

4.1 Merge Subject-Level Variables and Derive Analysis Dates

Show Code 
adae <- ae |>
  # Merge treatment and population flags from ADSL
  admiral::derive_vars_merged(
    dataset_add = dplyr::select(adsl, USUBJID, TRT01P, TRT01A, SAFFL, TRTSDT),
    by_vars     = exprs(USUBJID)
  ) |>
  # Restrict to safety population
  dplyr::filter(SAFFL == "Y") |>
  # AE start date (ISO 8601 → numeric)
  admiral::derive_vars_dt(
    new_vars_prefix = "AST",
    dtc             = AESTDTC
  ) |>
  # AE end date
  admiral::derive_vars_dt(
    new_vars_prefix = "AEN",
    dtc             = AEENDTC
  ) |>
  # Study day of AE onset relative to first dose
  dplyr::mutate(
    ASTDY = as.integer(ASTDT - TRTSDT) + 1L
  )

4.2 Derive Severity and Seriousness Flags

Show Code 
adae <- adae |>
  dplyr::mutate(
    # Serious AE flag
    AESERF = dplyr::if_else(AESER == "Y", "Y", "N", missing = "N"),
    # Grade 3+ severity flag (CTCAE-style: map AESEV text to numeric grade)
    AESEVN = dplyr::case_when(
      AESEV == "MILD"     ~ 1L,
      AESEV == "MODERATE" ~ 2L,
      AESEV == "SEVERE"   ~ 3L,
      TRUE                ~ NA_integer_
    ),
    CTC3FLG = dplyr::if_else(AESEVN >= 3L, "Y", "N", missing = "N"),
    # Treatment-emergent flag: AE onset on or after first dose date
    TRTEMFL = dplyr::if_else(
      !is.na(ASTDT) & !is.na(TRTSDT) & ASTDT >= TRTSDT,
      "Y", "N", missing = "N"
    )
  )

4.3 ADAE Quality Check

Show Code 
cat("========== ADAE Quality Check ==========\n")
========== ADAE Quality Check ==========
Show Code 
cat("Total AE records:       ", nrow(adae), "\n")
Total AE records:        890
Show Code 
cat("Subjects with AEs:      ", n_distinct(adae$USUBJID), "\n")
Subjects with AEs:       156
Show Code 
cat("Treatment-emergent AEs: ",
    sum(adae$TRTEMFL == "Y", na.rm = TRUE), "\n")
Treatment-emergent AEs:  839
Show Code 
cat("Serious AEs:            ",
    sum(adae$AESERF  == "Y", na.rm = TRUE), "\n\n")
Serious AEs:             3

5. ADVS — Vital Signs Analysis Dataset

ADVS is used for efficacy analyses involving continuous endpoint measurements (systolic BP, diastolic BP, heart rate, weight).

5.1 Build ADVS

Show Code 
# Parameters of interest
vs_params <- c("SYSBP", "DIABP", "PULSE", "WEIGHT")

advs <- vs |>
  # Merge subject-level variables
  admiral::derive_vars_merged(
    dataset_add = dplyr::select(adsl, USUBJID, TRT01P, TRT01A, SAFFL, TRTSDT),
    by_vars     = exprs(USUBJID)
  ) |>
  # Safety population only
  dplyr::filter(SAFFL == "Y", VSTESTCD %in% vs_params) |>
  # Analysis date
  admiral::derive_vars_dt(
    new_vars_prefix = "A",
    dtc             = VSDTC
  ) |>
  # Analysis value (numeric) and parameter label
  dplyr::mutate(
    AVAL    = VSSTRESN,
    PARAM   = VSTEST,
    PARAMCD = VSTESTCD,
    # Baseline: the last pre-dose measurement (VSBLFL == "Y" in SDTM, or derive below)
    ABLFL   = dplyr::if_else(VSBLFL == "Y", "Y", "N", missing = "N"),
    # Study day
    ADY     = as.integer(ADT - TRTSDT) + 1L
  )

# Merge baseline values back
# advs <- advs |>
#   admiral::derive_var_base(
#     by_vars    = exprs(USUBJID, PARAMCD),
#     source_var = AVAL,
#     new_var    = BASE
#   ) |>
#   # Change from baseline
#   dplyr::mutate(
#     CHG  = AVAL - BASE,
#     PCHG = round(100 * CHG / BASE, 2)
#   )

# # See which subject/parameter combos have duplicate baselines
# admiral::get_duplicates_dataset() |>
#   dplyr::select(USUBJID, PARAMCD, VSDTC, VSBLFL, ABLFL) |>
#   print(n = 30)


# ---- Re-derive ABLFL cleanly ----
  # Restrict candidates: non-missing AVAL, on or before first dose (ADY <= 1)
advs <- advs |>
  admiral::derive_var_extreme_flag(
    by_vars    = exprs(USUBJID, PARAMCD),
    order      = exprs(ADT, VSSEQ),   # last date, then highest sequence
    mode       = "last",              # last observation wins
    # filter     = ADY <= 1 & !is.na(AVAL),
    new_var    = ABLFL
  ) |>
  # Now derive BASE safely — guaranteed one baseline per USUBJID + PARAMCD
  admiral::derive_var_base(
    by_vars    = exprs(USUBJID, PARAMCD),
    source_var = AVAL,
    # new_var    = BASE
  ) |>
  dplyr::mutate(
    CHG  = AVAL - BASE,
    PCHG = dplyr::if_else(BASE != 0, round(100 * CHG / BASE, 2), NA_real_)
  )

advs |>
  dplyr::filter(ABLFL == "Y") |>
  dplyr::count(USUBJID, PARAMCD) |>
  dplyr::filter(n > 1)
Show Code 
cat("ADVS records:", nrow(advs),
    "| Subjects:", n_distinct(advs$USUBJID),
    "| Parameters:", n_distinct(advs$PARAMCD), "\n")
ADVS records: 16499 | Subjects: 168 | Parameters: 4

6. Tables, Listings & Figures (TLFs)

6.1 Table: Demographic and Baseline Characteristics

Show Code 
adsl |>
  dplyr::filter(SAFFL == "Y") |>
  dplyr::select(TRT01P, AGE, AGEGR1, SEX, RACE) |>
  dplyr::mutate(
    TRT01P = factor(TRT01P),
    SEX    = factor(SEX,  levels = c("M", "F"), labels = c("Male", "Female")),
    RACE   = stringr::str_to_title(RACE)
  ) |>
  gtsummary::tbl_summary(
    by         = TRT01P,
    label      = list(
      AGE    ~ "Age (years)",
      AGEGR1 ~ "Age Group",
      SEX    ~ "Sex",
      RACE   ~ "Race"
    ),
    statistic  = list(
      all_continuous()  ~ "{mean} ({sd})",
      all_categorical() ~ "{n} ({p}%)"
    ),
    digits     = all_continuous() ~ 1
  ) |>
  gtsummary::add_overall() |>
  gtsummary::add_p() |>
  gtsummary::bold_labels() |>
  gtsummary::modify_header(label ~ "**Characteristic**") |>
  gtsummary::modify_caption(
    "**Table 1. Demographic and Baseline Characteristics — Safety Population**"
  )
Table 1. Demographic and Baseline Characteristics — Safety Population
Characteristic Overall
N = 1681		 XANOMELINE
N = 1681		 p-value2
Age (years) 75.0 (8.1) 75.0 (8.1)
Age Group


    <65 19 (11%) 19 (11%)
    65–74 48 (29%) 48 (29%)
    >=75 101 (60%) 101 (60%)
Sex


    Male 78 (46%) 78 (46%)
    Female 90 (54%) 90 (54%)
Race


    American Indian Or Alaska Native 1 (0.6%) 1 (0.6%)
    Black Or African American 15 (8.9%) 15 (8.9%)
    White 152 (90%) 152 (90%)
1 Mean (SD); n (%)
2 NA

6.2 Table: Treatment-Emergent Adverse Events by System Organ Class

Show Code 
# Subjects per treatment arm (denominator for incidence %)
n_by_trt <- adsl |>
  dplyr::filter(SAFFL == "Y") |>
  dplyr::count(TRT01P, name = "N_ARM")

ae_soc <- adae |>
  dplyr::filter(TRTEMFL == "Y") |>
  dplyr::distinct(USUBJID, AEBODSYS, TRT01P) |>          # one record per subject per SOC
  dplyr::count(AEBODSYS, TRT01P, name = "n_subj") |>
  dplyr::left_join(n_by_trt, by = "TRT01P") |>
  dplyr::mutate(
    pct  = round(100 * n_subj / N_ARM, 1),
    cell = paste0(n_subj, " (", pct, "%)")
  ) |>
  dplyr::select(AEBODSYS, TRT01P, cell) |>
  tidyr::pivot_wider(names_from = TRT01P, values_from = cell, values_fill = "0 (0.0%)")

ae_soc |>
  dplyr::rename(`System Organ Class` = AEBODSYS) |>
  gt::gt() |>
  gt::tab_header(
    title    = "Table 2. Treatment-Emergent Adverse Events",
    subtitle = "Subjects with ≥1 TEAE by System Organ Class — Safety Population"
  ) |>
  gt::tab_style(
    style     = gt::cell_fill(color = "#2E75B6", alpha = 0.15),
    locations = gt::cells_column_labels()
  ) |>
  gt::tab_style(
    style     = gt::cell_text(weight = "bold"),
    locations = gt::cells_column_labels()
  ) |>
  gt::opt_row_striping()
Table 2. Treatment-Emergent Adverse Events
Subjects with ≥1 TEAE by System Organ Class — Safety Population
Body System or Organ Class XANOMELINE
CARDIAC DISORDERS 28 (16.7%)
CONGENITAL, FAMILIAL AND GENETIC DISORDERS 3 (1.8%)
EAR AND LABYRINTH DISORDERS 3 (1.8%)
EYE DISORDERS 2 (1.2%)
GASTROINTESTINAL DISORDERS 34 (20.2%)
GENERAL DISORDERS AND ADMINISTRATION SITE CONDITIONS 87 (51.8%)
IMMUNE SYSTEM DISORDERS 1 (0.6%)
INFECTIONS AND INFESTATIONS 22 (13.1%)
INJURY, POISONING AND PROCEDURAL COMPLICATIONS 10 (6%)
INVESTIGATIONS 12 (7.1%)
METABOLISM AND NUTRITION DISORDERS 3 (1.8%)
MUSCULOSKELETAL AND CONNECTIVE TISSUE DISORDERS 14 (8.3%)
NEOPLASMS BENIGN, MALIGNANT AND UNSPECIFIED (INCL CYSTS AND POLYPS) 3 (1.8%)
NERVOUS SYSTEM DISORDERS 45 (26.8%)
PSYCHIATRIC DISORDERS 18 (10.7%)
RENAL AND URINARY DISORDERS 6 (3.6%)
REPRODUCTIVE SYSTEM AND BREAST DISORDERS 1 (0.6%)
RESPIRATORY, THORACIC AND MEDIASTINAL DISORDERS 19 (11.3%)
SKIN AND SUBCUTANEOUS TISSUE DISORDERS 79 (47%)
SOCIAL CIRCUMSTANCES 1 (0.6%)
SURGICAL AND MEDICAL PROCEDURES 3 (1.8%)
VASCULAR DISORDERS 4 (2.4%)

6.3 Table: Serious Adverse Events Summary

Show Code 
adae |>
  dplyr::filter(TRTEMFL == "Y") |>
  dplyr::select(TRT01P, AESERF, CTC3FLG) |>
  dplyr::mutate(
    TRT01P = factor(TRT01P),
    AESERF  = factor(AESERF,  levels = c("Y", "N"), labels = c("Serious", "Non-serious")),
    CTC3FLG = factor(CTC3FLG, levels = c("Y", "N"), labels = c("Grade ≥3", "Grade <3"))
  ) |>
  gtsummary::tbl_summary(
    by    = TRT01P,
    label = list(AESERF ~ "Seriousness", CTC3FLG ~ "Severity Grade")
  ) |>
  gtsummary::add_overall() |>
  gtsummary::bold_labels() |>
  gtsummary::modify_caption("**Table 3. TEAE Seriousness and Severity — Safety Population**")
Table 3. TEAE Seriousness and Severity — Safety Population
Characteristic Overall
N = 8391		 XANOMELINE
N = 8391
Seriousness

    Serious 3 (0.4%) 3 (0.4%)
    Non-serious 836 (100%) 836 (100%)
Severity Grade

    Grade ≥3 35 (4.2%) 35 (4.2%)
    Grade <3 804 (96%) 804 (96%)
1 n (%)

6.4 Figure: Adverse Event Incidence by Treatment and SOC

Show Code 
ae_plot_data <- adae |>
  dplyr::filter(TRTEMFL == "Y") |>
  dplyr::distinct(USUBJID, AEBODSYS, TRT01P) |>
  dplyr::count(AEBODSYS, TRT01P) |>
  dplyr::left_join(n_by_trt, by = "TRT01P") |>
  dplyr::mutate(
    pct      = 100 * n / N_ARM,
    AEBODSYS = stringr::str_wrap(AEBODSYS, width = 30)
  )

ggplot2::ggplot(
    ae_plot_data,
    ggplot2::aes(x = pct, y = forcats::fct_reorder(AEBODSYS, pct), fill = TRT01P)
  ) +
  ggplot2::geom_col(position = "dodge", width = 0.65) +
  ggplot2::scale_fill_manual(
    values = c("#2E75B6", "#ED7D31", "#70AD47"),
    name   = "Treatment"
  ) +
  ggplot2::labs(
    x        = "Subjects with ≥1 TEAE (%)",
    y        = NULL,
    title    = "Treatment-Emergent Adverse Events by System Organ Class",
    subtitle = "Safety Population"
  ) +
  ggplot2::theme_minimal(base_size = 12) +
  ggplot2::theme(
    legend.position  = "bottom",
    panel.grid.major.y = ggplot2::element_blank(),
    plot.title       = ggplot2::element_text(face = "bold")
  )

Figure 1. Incidence of TEAEs by System Organ Class and Treatment Group

6.5 Figure: Vital Signs — Mean Change from Baseline Over Time

Show Code 
# Bin study days into nominal weeks
advs_sysbp <- advs |>
  dplyr::filter(PARAMCD == "SYSBP", !is.na(CHG), !is.na(TRT01P)) |>
  dplyr::mutate(
    WEEK = dplyr::case_when(
      ADY <= 0   ~ "Baseline",
      ADY <= 14  ~ "Week 2",
      ADY <= 28  ~ "Week 4",
      ADY <= 56  ~ "Week 8",
      ADY <= 84  ~ "Week 12",
      TRUE       ~ "Week 12+"
    ),
    WEEK = factor(WEEK, levels = c("Baseline", "Week 2", "Week 4", "Week 8", "Week 12", "Week 12+"))
  ) |>
  dplyr::group_by(TRT01P, WEEK) |>
  dplyr::summarise(
    mean_chg = mean(CHG,  na.rm = TRUE),
    se_chg   = sd(CHG,   na.rm = TRUE) / sqrt(dplyr::n()),
    n        = dplyr::n(),
    .groups  = "drop"
  )

ggplot2::ggplot(
    advs_sysbp,
    ggplot2::aes(x = WEEK, y = mean_chg, colour = TRT01P, group = TRT01P)
  ) +
  ggplot2::geom_hline(yintercept = 0, linetype = "dashed", colour = "grey60") +
  ggplot2::geom_line(linewidth = 1) +
  ggplot2::geom_point(size = 3) +
  ggplot2::geom_errorbar(
    ggplot2::aes(ymin = mean_chg - se_chg, ymax = mean_chg + se_chg),
    width = 0.2
  ) +
  ggplot2::scale_colour_manual(
    values = c("#2E75B6", "#ED7D31", "#70AD47"),
    name   = "Treatment"
  ) +
  ggplot2::labs(
    x        = "Study Visit",
    y        = "Mean Change from Baseline (mmHg)",
    title    = "Systolic Blood Pressure: Mean Change from Baseline",
    subtitle = "Safety Population — Mean ± SE"
  ) +
  ggplot2::theme_minimal(base_size = 12) +
  ggplot2::theme(
    legend.position = "bottom",
    plot.title      = ggplot2::element_text(face = "bold"),
    axis.text.x     = ggplot2::element_text(angle = 30, hjust = 1)
  )

Figure 2. Mean Change from Baseline in Systolic Blood Pressure Over Study Weeks

6.6 Figure: Kaplan–Meier Time to First TEAE

Show Code 
# Build subject-level time-to-first-TEAE dataset
first_ae <- adae |>
  dplyr::filter(TRTEMFL == "Y", !is.na(ASTDY)) |>
  dplyr::group_by(USUBJID) |>
  dplyr::slice_min(ASTDY, n = 1, with_ties = FALSE) |>
  dplyr::ungroup() |>
  dplyr::select(USUBJID, TIME = ASTDY, EVENT = TRTEMFL)

# All safety-pop subjects; those without an AE are censored at TRTDURD
km_data <- adsl |>
  dplyr::filter(SAFFL == "Y", !is.na(TRT01P)) |>
  dplyr::select(USUBJID, TRT01P, TRTDURD) |>
  dplyr::left_join(first_ae, by = "USUBJID") |>
  dplyr::mutate(
    TIME  = dplyr::coalesce(TIME, TRTDURD, 1L),
    EVENT = dplyr::if_else(EVENT == "Y", 1L, 0L, missing = 0L),
    TIME  = pmax(TIME, 1L)
  )

km_fit <- survival::survfit(
  survival::Surv(TIME, EVENT) ~ TRT01P,
  data = km_data
)

survminer::ggsurvplot(
  km_fit,
  data         = km_data,
  risk.table   = TRUE,
  pval         = TRUE,
  conf.int     = TRUE,
  xlab         = "Days Since First Dose",
  ylab         = "Probability of Remaining TEAE-Free",
  title        = "Time to First TEAE — Safety Population",
  legend.title = "Treatment",
  palette      = c("#2E75B6", "#ED7D31", "#70AD47"),
  ggtheme      = ggplot2::theme_minimal(base_size = 12)
)

Figure 3. Kaplan–Meier Estimate of Time to First Treatment-Emergent Adverse Event

7. Final Dataset Export

Show Code 
# In a validated production environment, export as SAS Transport (.xpt) format
# using haven::write_xpt() for submission to regulatory agencies (FDA, EMA).
#
# Example:
#   haven::write_xpt(adsl, path = "data/adam/adsl.xpt", version = 5, name = "ADSL")
#   haven::write_xpt(adae, path = "data/adam/adae.xpt", version = 5, name = "ADAE")
#   haven::write_xpt(advs, path = "data/adam/advs.xpt", version = 5, name = "ADVS")

cat("=====================================================\n")
=====================================================
Show Code 
cat("  Final ADaM Dataset Summary\n")
  Final ADaM Dataset Summary
Show Code 
cat("=====================================================\n")
=====================================================
Show Code 
tibble::tibble(
  Dataset     = c("ADSL", "ADAE", "ADVS"),
  Rows        = c(nrow(adsl), nrow(adae), nrow(advs)),
  Variables   = c(ncol(adsl), ncol(adae), ncol(advs)),
  Subjects    = c(
    n_distinct(adsl$USUBJID),
    n_distinct(adae$USUBJID),
    n_distinct(advs$USUBJID)
  ),
  `Pop Flag`  = c("SAFFL / ITTFL", "SAFFL", "SAFFL"),
  `Key Vars`  = c(
    "TRT01P/A, TRTSDT, TRTEDT, TRTDURD, AGEGR1",
    "TRTEMFL, AESERF, CTC3FLG, ASTDT, ASTDY",
    "AVAL, BASE, CHG, PCHG, PARAMCD, ADY"
  )
) |>
  gt::gt() |>
  gt::tab_header(title = "ADaM Datasets Ready for TLF Generation") |>
  gt::tab_style(
    style     = gt::cell_fill(color = "#2E75B6", alpha = 0.15),
    locations = gt::cells_column_labels()
  ) |>
  gt::tab_style(
    style     = gt::cell_text(weight = "bold"),
    locations = gt::cells_column_labels()
  )
ADaM Datasets Ready for TLF Generation
Dataset Rows Variables Subjects Pop Flag Key Vars
ADSL 306 33 306 SAFFL / ITTFL TRT01P/A, TRTSDT, TRTEDT, TRTDURD, AGEGR1
ADAE 890 47 156 SAFFL TRTEMFL, AESERF, CTC3FLG, ASTDT, ASTDY
ADVS 16499 37 168 SAFFL AVAL, BASE, CHG, PCHG, PARAMCD, ADY

Conclusion

This report has demonstrated a complete, CDISC-compliant end-to-end pipeline using the pharmaverse admiral package:

Phase Datasets Key Deliverables
Data Loading DM, AE, EX, DS, VS Domain inventory table
ADaM Construction ADSL, ADAE, ADVS Treatment vars, population flags, dates, baseline
Quality Checks All Record counts, duplicates, missing data
TLFs ADSL + ADAE + ADVS Demographics table, AE incidence, KM curve, vital signs
Export ADSL, ADAE, ADVS Ready for write_xpt() submission packages

All datasets are audit-ready and reproducible. Next steps would include ADLB (Laboratory), ADTTE (Time-to-Event), statistical model outputs, and integration into an RTF/PDF regulatory submission package.