Subject Disposition

# ------------------------------------------------------------------------------
# f-disposition — Subject Disposition (flow)
# ------------------------------------------------------------------------------
# Target: Figure 10-1 of the CDISCPILOT01 clinical study report (27 June 2006),
# page 24: subjects screened, of whom the screen failures and those randomised
# into the treatment phase, of whom those who completed Week 24 and those who
# completed the study through Week 26. Five counts, drawn as a flow.
#
# The screened population is the study's SDTM DM domain — 306 subjects, 52 of
# whom never reached ADSL — which is why this is the one display that reads
# `dm` and why prepare_data() keeps DM whole (#63, D0032). Every other count is
# ADSL's: randomised subjects, COMP24FL, and the complement of DISCONFL that the
# populations table already prints as "Complete Study".
# ------------------------------------------------------------------------------
id: f-disposition
title: "Subject Disposition"
regulatory_id: DSF01
type: figure
dataset: dm
analysis_set: all
denominator: adsl
group:
  - TRT01P
total: false
analyses:
  - name: flow
    method: figure
    custom: ard_disposition_flow
    label: "Subjects screened, randomised, and completing the study"

id: f-disposition
title: "Subject Disposition"
study: CDISCPILOT01
population_label: "All Subjects"
cutoff: "2014-07-01"
source: "Source: dm (the study's SDTM demographics domain) and adsl (CDISCPILOT01 ADaM package, phuse-org/phuse-scripts). Data cut-off: 2014-07-01."
footnotes:
  - "Subjects screened are every subject in the study's SDTM DM domain; screen failures are those DM labels as such. Randomised subjects are every subject in ADSL; completed Week 24 is COMP24FL, and completed the study through Week 26 is the complement of DISCONFL — the definition Table 14-1.01 states for Complete Study."
  - "Drawn as the reference report's Figure 10-1: 306 screened, of whom 52 screen failures and 254 randomised; of those, 118 completed Week 24 and 110 completed the study."
columns:
  order: ["All Subjects"]
format:
  count: "{n}"
rows:
  - { analysis: flow, variable: DSFLOW, level: screened, pattern: count, label: "Subjects screened" }
  - { analysis: flow, variable: DSFLOW, level: screen_failures, pattern: count, label: "Screen failures", indent: 1 }
  - { analysis: flow, variable: DSFLOW, level: randomised, pattern: count, label: "Randomised, entered treatment phase", indent: 1 }
  - { analysis: flow, variable: DSFLOW, level: completed_wk24, pattern: count, label: "Completed Week 24", indent: 2 }
  - { analysis: flow, variable: DSFLOW, level: completed_study, pattern: count, label: "Completed study through Week 26", indent: 3 }
figure:
  kind: flow
  analysis: flow
  width: 760
  height: 330
  boxes:
    - { level: screened, label: "Subjects Screened" }
    - { level: randomised, label: "Randomized, entered Treatment Phase", from: screened }
    - { level: screen_failures, label: "Screen failures", from: screened, side: right }
    - { level: completed_wk24, label: "Completed Week 24", from: randomised }
    - { level: completed_study, label: "Completed Study through Week 26", from: completed_wk24 }
variants:
  post_text: {}
  in_text:
    title: "Subject Disposition"

# Subject disposition counts for the flow figure (Figure 10-1 of the reference
# report). Five numbers, each a count of subjects: screened (every subject in
# DM), screen failures (DM's own arm label), randomised into the treatment phase
# (every subject in ADSL), completed Week 24 (COMP24FL) and completed the study
# through Week 26 (the complement of DISCONFL, the definition the populations
# table states for "Complete Study").
#
# Contract: (data, spec, denominator) -> a {cards}-shaped ARD. `data` is DM,
# whole; `denominator` is the prepared ADSL.

ard_disposition_flow <- function(data, spec, denominator) {
  blank_na <- function(x) {
    x <- as.character(x)
    x[is.na(x)] <- ""
    x
  }
  screened <- length(unique(data$USUBJID))
  screen_failures <- length(unique(data$USUBJID[blank_na(data$ARM) == "Screen Failure"]))
  randomised <- length(unique(denominator$USUBJID))
  completed_wk24 <- sum(blank_na(denominator$COMP24FL) == "Y")
  completed_study <- sum(blank_na(denominator$DISCONFL) != "Y")
  stages <- c(
    screened = screened, screen_failures = screen_failures, randomised = randomised,
    completed_wk24 = completed_wk24, completed_study = completed_study
  )
  data.frame(
    group1 = "population",
    group1_level = "All Subjects",
    group2 = NA_character_,
    group2_level = NA_character_,
    variable = "DSFLOW",
    variable_level = names(stages),
    context = "subject_count",
    stat_name = "n",
    stat_label = "n",
    stat = I(as.list(unname(stages))),
    warning = NA_character_,
    error = NA_character_,
    stringsAsFactors = FALSE
  )
}