Table 4 - Patient Disposition, Pooled Analyses
Closed this issue · 3 comments
Table Layout & Design
Please refer to page 14-15 of the pdf here for more details: FDA-2022-N-1961-0002_attachment_1.pdf
Description
Patient Disposition, Pooled Analyses
Analysis Datasets
Any dataset - df
ADSL - alt_counts_df (if required)
When using data from the scda package, please use datasets with archive_name set to "rcd_2022_10_13".
Analysis Filtering
Column Variable
ARM (ignore "Risk Difference" column for now. This will be added in at a later date once functionality is available.)
Additional Notes
N/A
See other table functions as reference
#' FDA Table 4: Patient Disposition, Pooled Analyses
#'
#' @details
#' * df must contain SAFFL, USUBJID, and the variables specified by arm_var
#' and trtedtm_var.
#' * If specified, alt_counts_df must contain SAFFL, USUBJID, and the variable specified by arm_var.
#' * Flag variables (i.e. XXXFL) are expected to have two levels: "Y" (true) and "N" (false). Missing values in
#' flag variables are treated as "N".
#' * Columns are split by arm. Overall population column is excluded by default (see lbl_overall argument).
#' * Numbers in table "Patients Treated" section are the absolute numbers of patients and fraction of N.
#' * All-zero rows are not removed by default (see prune_0 argument).
#' * Records with missing treatment start and/or end datetime are excluded from all calculations.
#'
#' @inheritParams argument_convention
#' @examples
#' adsl <- scda::synthetic_cdisc_dataset("rcd_2022_10_13", "adsl")
#'
#' tbl <- make_table_04(df = adsl)
#' tbl
#'
#' @export
make_table_04 <- function(df,
alt_counts_df = NULL,
show_colcounts = TRUE,
arm_var = "ARM",
lbl_overall = NULL,
prune_0 = FALSE,
annotations = NULL) {
checkmate::assert_subset(c("SAFFL", "USUBJID", "ARM",
"AESER", "ITTFL", "SAFFL", "PPROTFL",
"EOTSTT", "DCTREAS", "EOSSTT", "DCSREAS"), names(df))
checkmate::assert_flag_variables(df, "SAFFL")
df <- df %>%
filter(SAFFL == "Y") %>%
df_explicit_na() %>%
mutate(
RAN = with_label(AESER != '', "Patients randomized"), #should be the flag "RANDFL"?
ITT = with_label(ITTFL == 'Y', "ITT/mITT population"),
SAF = with_label(SAFFL == 'Y', "Safety population"),
# PPP = with_label(PPROTFL == 'Y', "Per-protocol population"), #could not find flag representing per protocal population?
DISCSD = with_label(EOTSTT == 'DISCONTINUED', "Discontinued study drug"),
# DISCSD_AE = with_label(EOTSTT == 'DISCONTINUED' & DCTREAS == 'ADVERSE EVENT', 'Adverse event'), #could not find DCTREAS?
# DISCSD_LOE = with_label(EOTSTT == 'DISCONTINUED' & DCTREAS == 'LACK OF EFFICACY', 'Lack of efficacy'),
# DISCSD_PD = with_label(EOTSTT == 'DISCONTINUED' & DCTREAS == 'PROTOCOL DEVIATION', 'Protocol deviation'),
# DISCSD_DT = with_label(EOTSTT == 'DISCONTINUED' & DCTREAS == 'DEATH', 'Death'),
# DISCSD_WBS = with_label(EOTSTT == 'DISCONTINUED' & DCTREAS == 'WITHDRAWAL BY SUBJECT', 'Withdrawal by subject'),
# DISCSD_OTH = with_label(EOTSTT == 'DISCONTINUED' & DCTREAS == 'OTHER', 'Other'),
DISCS = with_label(EOSSTT == 'DISCONTINUED', 'Discontinued study'),
DISCS_DT = with_label(EOSSTT == 'DISCONTINUED' & DCSREAS == 'DEATH', 'Death'),
DISCS_LFU = with_label(EOSSTT == 'DISCONTINUED' & DCSREAS == 'LOST TO FOLLOW-UP', 'Lost to follow-up'),
DISCS_WBS = with_label(EOSSTT == 'DISCONTINUED' & DCSREAS == 'WITHDRAWAL BY SUBJECT', 'Withdrawal by subject'),
DISCS_PHD = with_label(EOSSTT == 'DISCONTINUED' & DCSREAS == 'PHYSICIAN DECISION', 'Physician decision'),
DISCS_PD = with_label(EOSSTT == 'DISCONTINUED' & DCSREAS == 'PROTOCOL VIOLATION', 'Protocol deviation'),
DISCS_OTH = with_label(EOSSTT == 'DISCONTINUED' & DCSREAS == 'OTHER', 'Other')
)
alt_counts_df <- alt_counts_df_preproc(alt_counts_df, arm_var)
lyt <- basic_table_annot(show_colcounts, annotations) %>%
split_cols_by_arm(arm_var, lbl_overall) %>%
count_patients_with_flags(
var = "USUBJID",
flag_variables = var_labels(df[, "RAN"]),
table_names = "ran"
) %>%
count_patients_with_flags(
var = "USUBJID",
flag_variables = var_labels(df[, c("ITT", "SAF", "PPP")]),
.indent_mods = 1L,
table_names = "ran_fl"
) %>%
count_patients_with_flags(
var = "USUBJID",
flag_variables = var_labels(df[, "DISCSD"]),
table_names = "discsd"
) %>%
count_patients_with_flags(
var = "USUBJID",
flag_variables = var_labels(df[, c('DISCSD_AE', 'DISCSD_LOE', 'DISCSD_PD', 'DISCSD_DT', 'DISCSD_WBS', 'DISCSD_OTH')]),
.indent_mods = 1L,
table_names = "discsd_fl"
) %>%
count_patients_with_flags(
var = "USUBJID",
flag_variables = var_labels(df[, "DISCS"]),
table_names = "discs"
) %>%
count_patients_with_flags(
var = "USUBJID",
flag_variables = var_labels(df[, c('DISCS_DT', 'DISCS_LFU', 'DISCS_WBS', 'DISCS_PHD', 'DISCS_PD', 'DISCS_OTH')]),
.indent_mods = 1L,
table_names = "discs_fl")
tbl <- build_table(lyt, df = df, alt_counts_df = alt_counts_df)
if (prune_0) tbl <- prune_table(tbl)
tbl
}