update aet05 and aet05_all (#678)

Close #665 - Update `aet05` filter: only include "*AETTE", "*CQTTE", and "*SMQTTE". - Update the main function (`aet05` and `aet05_all`) to let the dataset name be exposed to users. - Update `syn_data`. - Update unit tests. - Update Chevron catalog. - Turn off partial match and catch typos. - Update `data.R`. --------- Signed-off-by: Liming <[email protected]> Co-authored-by: Liming <[email protected]>
insightsengineering · Nov 10, 2023 · cc736c8 · cc736c8
1 parent 615bc92
commit cc736c8
Show file tree

Hide file tree

Showing 14 changed files with 79 additions and 64 deletions.
diff --git a/NEWS.md b/NEWS.md
@@ -3,6 +3,9 @@
 * `MNG01` plot can now be displayed without error bars and can display a continuous temporal scale on the `x` axis.
 * Add a `chevron_simple` class only contain main function.
 * Remove `details` argument in `script_funs`, add `name` argument.
+* `AET05` preprocessing now filters on `"(AE|CQ|SMQ)TTE"` rather than `"AETTE"`.
+* Rename the dataset `ADAETTE` in `syn_data` object to `ADSAFTTE`.
+* Use uppercase variable names in `AET05` and `AET05_ALL`.
 
 # chevron 0.2.4
 

diff --git a/R/aet05.R b/R/aet05.R
@@ -3,6 +3,7 @@
 #' @describeIn aet05 Main TLG function
 #'
 #' @inheritParams gen_args
+#' @param dataset (`string`) the name of a table in the `adam_db` object.
 #' @param arm_var (`string`) the arm variable used for arm splitting.
 #' @param ... Further arguments passed to `tern::control_incidence_rate()`.
 #'
@@ -11,29 +12,36 @@
 #'  * Split columns by arm, typically `ACTARM`.
 #'  * Split rows by parameter code.
 #'  * `AVAL` is patient-years at risk.
-#'  * `n_events` is the number of adverse events observed.
+#'  * `N_EVENTS` is the number of adverse events observed.
 #'  * The table allows confidence level to be adjusted, default is 95%.
 #'  * Keep zero count rows by default.
 #'
 #' @note
-#'  * `adam_db` object must contain an `adaette` table with the columns `"PARAMCD"`, `"PARAM"`, `"AVAL"`, and `"CNSR"`.
+#'  * `adam_db` object must contain table named as `dataset` with the columns `"PARAMCD"`, `"PARAM"`,
+#'  `"AVAL"`, and `"CNSR"`.
 #'
 #' @export
 #'
 aet05_main <- function(adam_db,
+                       dataset = "adsaftte",
                        arm_var = "ACTARM",
                        lbl_overall = NULL,
                        ...) {
-  assert_all_tablenames(adam_db, c("adsl", "adaette"))
+  assert_string(dataset)
+  assert_all_tablenames(adam_db, "adsl", dataset)
   assert_string(arm_var)
   assert_string(lbl_overall, null.ok = TRUE)
+  df_lbl <- paste0("adam_db$", dataset)
   assert_valid_variable(adam_db$adsl, c("USUBJID", arm_var), types = list(c("character", "factor")))
-  assert_valid_variable(adam_db$adaette, c("USUBJID", arm_var, "PARAMCD", "PARAM"),
-    types = list(c("character", "factor"))
+  assert_valid_variable(adam_db[[dataset]], c("USUBJID", arm_var, "PARAMCD", "PARAM"),
+    types = list(c("character", "factor")), label = df_lbl
   )
-  assert_valid_variable(adam_db$adaette, "AVAL", types = list("numeric"), lower = 0, na_ok = TRUE)
-  assert_valid_variable(adam_db$adaette, "n_events", types = list("numeric"), integerish = TRUE, lower = 0L)
-  assert_valid_var_pair(adam_db$adsl, adam_db$adaette, arm_var)
+  assert_valid_variable(adam_db[[dataset]], "AVAL", types = list("numeric"), lower = 0, na_ok = TRUE, label = df_lbl)
+  assert_valid_variable(adam_db[[dataset]], "N_EVENTS",
+    types = list("numeric"), integerish = TRUE, lower = 0L,
+    label = df_lbl
+  )
+  assert_valid_var_pair(adam_db$adsl, adam_db[[dataset]], arm_var)
 
   lbl_overall <- render_safe(lbl_overall)
   control <- execute_with_args(control_incidence_rate, ...)
@@ -43,11 +51,11 @@ aet05_main <- function(adam_db,
     lbl_overall = lbl_overall,
     param_label = "PARAM",
     vars = "AVAL",
-    n_events = "n_events",
+    n_events = "N_EVENTS",
     control = control
   )
 
-  tbl <- build_table(lyt, adam_db$adaette, alt_counts_df = adam_db$adsl)
+  tbl <- build_table(lyt, adam_db[[dataset]], alt_counts_df = adam_db$adsl)
 
   tbl
 }
@@ -85,11 +93,11 @@ aet05_lyt <- function(arm_var,
 #'
 #' @export
 #'
-aet05_pre <- function(adam_db, ...) {
-  adam_db$adaette <- adam_db$adaette %>%
-    filter(grepl("AETTE", .data$PARAMCD)) %>%
+aet05_pre <- function(adam_db, dataset = "adsaftte", ...) {
+  adam_db[[dataset]] <- adam_db[[dataset]] %>%
+    filter(grepl("(AE|CQ|SMQ)TTE", .data$PARAMCD)) %>%
     mutate(
-      n_events = as.integer(.data$CNSR == 0)
+      N_EVENTS = as.integer(.data$CNSR == 0)
     )
 
   adam_db
@@ -120,14 +128,13 @@ aet05_post <- function(tlg, prune_0 = FALSE, ...) {
 #' library(dplyr)
 #' library(dunlin)
 #'
-#' proc_data <- log_filter(syn_data, PARAMCD == "AETTE1", "adaette")
+#' proc_data <- log_filter(syn_data, PARAMCD == "AETTE1", "adsaftte")
 #'
 #' run(aet05, proc_data)
 #'
 #' run(aet05, proc_data, conf_level = 0.90, conf_type = "exact")
 aet05 <- chevron_t(
   main = aet05_main,
   preprocess = aet05_pre,
-  postprocess = aet05_post,
-  adam_datasets = c("adsl", "adaette")
+  postprocess = aet05_post
 )
diff --git a/R/aet05_all.R b/R/aet05_all.R
@@ -6,17 +6,17 @@
 #'
 #' @export
 #'
-aet05_all_pre <- function(adam_db, ...) {
-  anl_tte <- adam_db$adaette %>%
+aet05_all_pre <- function(adam_db, dataset = "adsaftte", ...) {
+  anl_tte <- adam_db[[dataset]] %>%
     filter(.data$PARAMCD == "AEREPTTE") %>%
     select(all_of(c("USUBJID", "AVAL")))
 
-  adam_db$adaette <- adam_db$adaette %>%
+  adam_db[[dataset]] <- adam_db[[dataset]] %>%
     filter(grepl("TOT", .data$PARAMCD)) %>%
     mutate(
-      n_events = as.integer(.data$AVAL)
+      N_EVENTS = as.integer(.data$AVAL),
+      AVAL = NULL
     ) %>%
-    select(-c("AVAL")) %>%
     left_join(anl_tte, by = c("USUBJID"))
 
   adam_db
@@ -34,14 +34,13 @@ aet05_all_pre <- function(adam_db, ...) {
 #' library(dplyr)
 #' library(dunlin)
 #'
-#' proc_data <- log_filter(syn_data, PARAMCD == "AETOT1" | PARAMCD == "AEREPTTE", "adaette")
+#' proc_data <- log_filter(syn_data, PARAMCD == "AETOT1" | PARAMCD == "AEREPTTE", "adsaftte")
 #'
 #' run(aet05_all, proc_data)
 #'
 #' run(aet05_all, proc_data, conf_level = 0.90, conf_type = "exact")
 aet05_all <- chevron_t(
   main = aet05_main,
   preprocess = aet05_all_pre,
-  postprocess = aet05_post,
-  adam_datasets = c("adsl", "adaette")
+  postprocess = aet05_post
 )
diff --git a/R/data.R b/R/data.R
@@ -5,7 +5,7 @@
 #' @format A named `list` of 13 `data.frames`:
 #'     - `adsl`
 #'     - `adae`
-#'     - `adaette`
+#'     - `adsaftte`
 #'     - `adcm`
 #'     - `addv`
 #'     - `adeg`

diff --git a/R/kmg01.R b/R/kmg01.R
@@ -80,11 +80,8 @@ kmg01_pre <- function(adam_db, dataset = "adtte", ...) {
 #'   "C: Combination" = "gray"
 #' )
 #'
-#' syn_data2 <- log_filter(syn_data, PARAMCD == "OS", "adtte")
-#' run(kmg01, syn_data2, dataset = "adtte", line_col = col)
-#'
-#' syn_data3 <- log_filter(syn_data, PARAMCD == "AEREPTTE", "adaette")
-#' run(kmg01, syn_data3, dataset = "adaette")
+#' syn_data <- log_filter(syn_data, PARAMCD == "OS", "adtte")
+#' run(kmg01, syn_data, dataset = "adtte", line_col = col)
 kmg01 <- chevron_g(
   main = kmg01_main,
   preprocess = kmg01_pre

diff --git a/data-raw/syn_data_creation.R b/data-raw/syn_data_creation.R
@@ -208,6 +208,9 @@ syn_test_data <- function() {
   sd$adex$ETHNIC <- factor(trimws(sd$adex$ETHNIC), levels = trimws(levels(sd$adex$ETHNIC)))
   attr(sd$adex$ETHNIC, "label") <- "Ethnicity"
 
+  # useful for aet05 and aet05_all
+  names(sd)[names(sd) == "adaette"] <- "adsaftte"
+
   sd
 }
 

diff --git a/data/syn_data.rda b/data/syn_data.rda
diff --git a/man/aet05.Rd b/man/aet05.Rd
diff --git a/man/aet05_all.Rd b/man/aet05_all.Rd
diff --git a/man/kmg01.Rd b/man/kmg01.Rd
diff --git a/man/syn_data.Rd b/man/syn_data.Rd
diff --git a/tests/testthat/test-aet05.R b/tests/testthat/test-aet05.R
@@ -1,23 +1,23 @@
 test_that("aet05 can handle NA values", {
   proc_data <- syn_data
-  proc_data$adaette$AVAL <- NA_real_
-  proc_data$adaette$CNSR <- NA_real_
+  proc_data$adsaftte$AVAL <- NA_real_
+  proc_data$adsaftte$CNSR <- NA_real_
 
   res <- expect_silent(run(aet05, proc_data))
   expect_snapshot(cat(export_as_txt(res, lpp = 100)))
 })
 
 test_that("aet05 can handle some NA values", {
   new_paramcd <- c(
-    as.character(syn_data$adaette$PARAMCD[1:4]), NA, "",
-    as.character(syn_data$adaette$PARAMCD[-rep(1:6)])
+    as.character(syn_data$adsaftte$PARAMCD[1:4]), NA, "",
+    as.character(syn_data$adsaftte$PARAMCD[-rep(1:6)])
   )
 
   proc_data <- syn_data
 
-  proc_data$adaette <- proc_data$adaette %>%
+  proc_data$adsaftte <- proc_data$adsaftte %>%
     mutate(
-      PARAMCD = with_label(.env$new_paramcd, var_labels_for(syn_data$adaette, "PARAMCD"))
+      PARAMCD = with_label(.env$new_paramcd, var_labels_for(syn_data$adsaftte, "PARAMCD"))
     )
 
   res1 <- expect_silent(run(aet05, proc_data))

diff --git a/tests/testthat/test-aet05_all.R b/tests/testthat/test-aet05_all.R
@@ -1,20 +1,20 @@
 test_that("aet05_all can handle NA values", {
   proc_data <- syn_data
-  proc_data$adaette$AVAL <- NA_real_
-  proc_data$adaette$CNSR <- NA_real_
+  proc_data$adsaftte$AVAL <- NA_real_
+  proc_data$adsaftte$CNSR <- NA_real_
 
   res <- expect_silent(run(aet05_all, proc_data))
   expect_snapshot(cat(export_as_txt(res, lpp = 100)))
 })
 
 test_that("aet05_all can handle some NA values", {
-  new_paramcd <- c(NA, "", as.character(syn_data$adaette$PARAMCD[-c(1, 2)]))
+  new_paramcd <- c(NA, "", as.character(syn_data$adsaftte$PARAMCD[-c(1, 2)]))
 
   proc_data <- syn_data
 
-  proc_data$adaette <- proc_data$adaette %>%
+  proc_data$adsaftte <- proc_data$adsaftte %>%
     mutate(
-      PARAMCD = with_label(.env$new_paramcd, var_labels_for(syn_data$adaette, "PARAMCD"))
+      PARAMCD = with_label(.env$new_paramcd, var_labels_for(syn_data$adsaftte, "PARAMCD"))
     )
 
   res1 <- expect_silent(run(aet05_all, proc_data))

diff --git a/vignettes/chevron_catalog.rmd b/vignettes/chevron_catalog.rmd
@@ -28,7 +28,7 @@ library(chevron)
 
 #### **1. Input dataset and dataset names**
 
-The input dataset expected by the argument `adam_db=` in the `run(...)` function is a collection of `ADaM` datasets as a list object. Each `ADaM` dataset is expected to be an object of data frame. If the `ADaM` datasets are read in individually, user will need to combine them into a list object and provide the name of the list to `adam_db=`. Also, each element in the list are expected to have corresponding `ADaM` dataset names. Conventional `ADaM` dataset names, including `adsl`,`adex`, `adae`, `adlb`,`advs`,`adeg`,`adcm`,`admh`,`adrs`, and `adtte`, can be picked up by `chevron` with one exception. If you have a safety time-to-event dataset named as `adsaftte` following Roche standard `ADaM` convention, you will need to re-name it to `adaette` to be picked up by `chevron`.
+The input dataset expected by the argument `adam_db=` in the `run(...)` function is a collection of `ADaM` datasets as a list object. Each `ADaM` dataset is expected to be an object of data frame. If the `ADaM` datasets are read in individually, user will need to combine them into a list object and provide the name of the list to `adam_db=`. Also, each element in the list are expected to have corresponding `ADaM` dataset names. Conventional `ADaM` dataset names, including `adsl`,`adex`, `adae`, `adlb`,`advs`,`adeg`,`adcm`,`admh`,`adrs`, and `adtte`, can be picked up by `chevron` with one exception.
 
 ```{r eval=FALSE}
 std_data <- list(adsl = adsl, adae = adae)
@@ -243,12 +243,11 @@ run(aet04, syn_data, grade_groups = grade_groups, prune_0 = FALSE)
 #### **1. Adverse Event Rate Adjusted for Patient-Years at Risk - First Occurrence**
 
 1) The **`aet05`** template produces the standard adverse event rate adjusted for patient-years at risk summary considering first occurrence only.
-2) Currently, the expected name for safety time-to-event dataset is `adaette` rather than `adsaftte` from other data standards.
-3) By default, all `adaette` parameter codes containing the string `"TTE"` are included in the output. Users are expected to filter the parameter(s) of interest from input safety time-to-event dataset in pre-processing if needed.
-4) In the input safety time-to-event dataset, in the censoring variable `CNSR` , `0` indicates the occurrence of an event of interest and `1` denotes censoring.
+2) By default, all `adsaftte` parameter codes containing the string `"TTE"` are included in the output. Users are expected to filter the parameter(s) of interest from input safety time-to-event dataset in pre-processing if needed.
+3) In the input safety time-to-event dataset, in the censoring variable `CNSR`, `0` indicates the occurrence of an event of interest and `1` denotes censoring.
 
 ```{r}
-proc_data <- log_filter(syn_data, PARAMCD == "AETTE1", "adaette")
+proc_data <- log_filter(syn_data, PARAMCD == "AETTE1", "adsaftte")
 
 run(aet05, proc_data)
 ```
@@ -267,12 +266,11 @@ run(aet05, syn_data, conf_level = 0.90, conf_type = "exact")
 #### **1. Adverse Event Rate Adjusted for Patient-Years at Risk - All Occurrences**
 
 1) The **`aet05_all`** template produces the standard adverse event rate adjusted for patient-years at risk summary considering all occurrences.
-2) Currently, the expected name for safety time-to-event dataset is `adaette` rather than `adsaftte` from other data standards.
-3) By default, all `adaette` parameter codes containing the string `"TOT"` and the parameter code `"AEREPTTE"` are required. `"TOT"` parameters store the number of occurrences of adverse event of interests. Parameter code `"AEREPTTE"` stores the time to end of adverse event reporting period in years that contribute to the summary of "total patient-years at risk" in the output. Users are expected to filter parameters of interest from input analysis dataset in pre-processing, if needed.
-4) In the input safety time-to-event dataset, in the censoring variable `CNSR` , `0` indicates the occurrence of an event of interest and `1` denotes censoring.
+2) By default, all `adsaftte` parameter codes containing the string `"TOT"` and the parameter code `"AEREPTTE"` are required. `"TOT"` parameters store the number of occurrences of adverse event of interests. Parameter code `"AEREPTTE"` stores the time to end of adverse event reporting period in years that contribute to the summary of "total patient-years at risk" in the output. Users are expected to filter parameters of interest from input analysis dataset in pre-processing, if needed.
+3) In the input safety time-to-event dataset, in the censoring variable `CNSR`, `0` indicates the occurrence of an event of interest and `1` denotes censoring.
 
 ```{r}
-proc_data <- log_filter(syn_data, PARAMCD == "AETOT1" | PARAMCD == "AEREPTTE", "adaette")
+proc_data <- log_filter(syn_data, PARAMCD == "AETOT1" | PARAMCD == "AEREPTTE", "adsaftte")
 
 run(aet05_all, proc_data)
 ```