link cup to acute

Public-Health-Scotland · May 23, 2024 · a967d5c · a967d5c
1 parent 3869b09
commit a967d5c
Show file tree

Hide file tree

Showing 4 changed files with 60 additions and 4 deletions.
diff --git a/R/get_boxi_extract_path.R b/R/get_boxi_extract_path.R
@@ -16,6 +16,7 @@ get_boxi_extract_path <- function(
       "ae",
       "ae_cup",
       "acute",
+      "acute_cup",
       "cmh",
       "deaths",
       "dn",
@@ -44,6 +45,7 @@ get_boxi_extract_path <- function(
     "ae" ~ "A&E-episode-level-extract",
     "ae_cup" ~ "A&E-UCD-CUP-extract",
     "acute" ~ "Acute-episode-level-extract",
+    "acute_cup" ~ "Actue-cup-extract",
     "cmh" ~ "Community-MH-contact-level-extract",
     "dn" ~ "District-Nursing-contact-level-extract",
     "gp_ooh-c" ~ "GP-OoH-consultations-extract",

diff --git a/R/process_extract_acute.R b/R/process_extract_acute.R
@@ -12,7 +12,11 @@
 #' @return the final data as a [tibble][tibble::tibble-package].
 #' @export
 #' @family process extracts
-process_extract_acute <- function(data, year, write_to_disk = TRUE) {
+process_extract_acute <- function(data,
+                                  year,
+                                  acute_cup_path = get_boxi_extract_path(year, "acute_cup"),
+                                  write_to_disk = TRUE) {
+
   # Only run for a single year
   stopifnot(length(year) == 1L)
 
@@ -58,8 +62,43 @@ process_extract_acute <- function(data, year, write_to_disk = TRUE) {
     ) %>%
     # Add oldtadm as a factor with labels
     dplyr::mutate(oldtadm = factor(.data$oldtadm,
-      levels = 0L:8L
-    ))
+                                   levels = 0L:8L
+    )) %>%
+    dplyr::mutate(
+      unique_row_num = dplyr::row_number()
+    )
+
+  acute_cup <- read_file(
+    path = acute_cup_path,
+    col_type = readr::cols(
+      "Acute Admission Date" = readr::col_date(format = "%Y/%m/%d %T"),
+      "Acute Discharge Date" = readr::col_date(format = "%Y/%m/%d %T"),
+      "Acute Admission Type Code" = readr::col_character(),
+      "Acute Discharge Type Code" = readr::col_character(),
+      "Case Reference Number [C]" = readr::col_character(),
+      "UPI Number [C]" = readr::col_character(),
+      "CUP Marker" = readr::col_integer(),
+      "CUP Pathway Name" = readr::col_character()
+    )
+  ) %>% dplyr::select(
+    chi = "UPI Number [C]",
+    case_reference_number = "Case Reference Number [C]",
+    record_keydate1 = "Acute Admission Date",
+    record_keydate2 = "Acute Discharge Date",
+    tadm = "Acute Admission Type Code",
+    disch = "Acute Discharge Type Code",
+    cup_marker = "CUP Marker",
+    cup_pathway = "CUP Pathway Name"
+  ) %>% dplyr::distinct()
+
+  acute_clean <- acute_clean %>%
+    dplyr::left_join(acute_cup,
+                     by = c("record_keydate1",
+                            "record_keydate2",
+                            "case_reference_number",
+                            "chi",
+                            "tadm",
+                            "disch"))
 
   acute_processed <- acute_clean %>%
     dplyr::select(
@@ -106,7 +145,9 @@ process_extract_acute <- function(data, year, write_to_disk = TRUE) {
       "cost_total_net",
       tidyselect::ends_with("_beddays"),
       tidyselect::ends_with("_cost"),
-      "uri"
+      "uri",
+      "cup_marker",
+      "cup_pathway"
     ) %>%
     dplyr::arrange(.data$chi, .data$record_keydate1)
 

diff --git a/R/read_extract_acute.R b/R/read_extract_acute.R
@@ -149,6 +149,7 @@ read_extract_acute <- function(year, file_path = get_boxi_extract_path(year = ye
       ipdc = "Inpatient Day Case Identifier Code",
       cij_ipdc = "CIJ Inpatient Day Case Identifier Code (01)",
       lineno = "Line Number (01)",
+      case_reference_number = "Case Reference Number [C]",
       GLS_record = "GLS Record"
     ) %>%
     # replace NA in cost_total_net by 0

diff --git a/_targets.R b/_targets.R
@@ -281,6 +281,16 @@ list(
       get_boxi_extract_path(year = year, type = "gp_ooh-c"),
       format = "file"
     ),
+    tar_target(
+      acute_cup_path,
+      get_boxi_extract_path(year, type = "acute_cup"),
+      format = "file"
+    ),
+    tar_target(
+      gp_ooh_cup_path,
+      get_boxi_extract_path(year, type = "gp_ooh_cup"),
+      format = "file"
+    ),
     tar_qs(
       ooh_data,
       read_extract_gp_ooh(
@@ -294,6 +304,7 @@ list(
     tar_target(source_acute_extract, process_extract_acute(
       acute_data,
       year,
+      acute_cup_path,
       write_to_disk = write_to_disk
     )),
     tar_target(
@@ -419,6 +430,7 @@ list(
     tar_target(source_ooh_extract, process_extract_gp_ooh(
       year,
       ooh_data,
+      gp_ooh_cup_path,
       write_to_disk = write_to_disk
     )),
     tar_target(