BYOC_ID: 45357 Refactor NRS

NAMESPACE

@@ -139,7 +139,6 @@ export(process_sc_all_alarms_telecare)
 export(process_sc_all_care_home)
 export(process_sc_all_home_care)
 export(process_sc_all_sds)
-export(process_slf_deaths_lookup)
 export(process_tests_acute)
 export(process_tests_ae)
 export(process_tests_alarms_telecare)

R/create_episode_file.R

@@ -168,7 +168,7 @@ create_episode_file <- function(
     ) %>%
     join_deaths_data(
       year,
-      slf_deaths_lookup
+      BYOC_MODE = BYOC_MODE
     ) %>%
     write_temp_data(year, file_name = "ep_temp5", write_temp_to_disk) %>%
     add_activity_after_death_flag(year,

R/get_slf_lookup_paths.R

@@ -82,20 +82,31 @@ get_slf_deaths_lookup_path <- function(year, ...) {
 #'
 #' @param ... additional arguments passed to [get_file_path()]
 #' @param update the update month (defaults to use [latest_update()])
+#' @param BYOC_MODE BYOC_MODE, Boolean type
 #'
 #' @export
 #' @family slf lookup file path
 #' @seealso [get_file_path()] for the generic function.
-get_combined_slf_deaths_lookup_path <- function(update = latest_update(), ...) {
+get_combined_slf_deaths_lookup_path <- function(update = latest_update(),
+                                                BYOC_MODE = FALSE,
+                                                ...) {
   # Note this name is very similar to the existing slf_deaths_lookup_path which returns the path for
   # the refined_death with deceased flag for each financial year.
   # This function will return the combined financial
   # years lookup i.e. all years put together.
-  combined_slf_deaths_lookup_path <- get_file_path(
-    directory = fs::path(get_slf_dir(), "Deaths"),
-    file_name = stringr::str_glue("anon-combined_slf_deaths_lookup_{update}.parquet"),
-    ...
-  )
+  if (isTRUE(BYOC_MODE)) {
+    combined_slf_deaths_lookup_path <- file.path(
+      denodo_output_path(),
+      "anon-combined_slf_deaths_lookup.parquet"
+    )
+  } else {
+    combined_slf_deaths_lookup_path <- get_file_path(
+      directory = fs::path(get_slf_dir(), "Deaths"),
+      file_name = stringr::str_glue("anon-combined_slf_deaths_lookup_{update}.parquet"),
+      ...
+    )
+  }
+
   return(combined_slf_deaths_lookup_path)
 }
 
@@ -112,7 +123,7 @@ get_combined_slf_deaths_lookup_path <- function(update = latest_update(), ...) {
 #' @export
 #' @family slf lookup file path
 #' @seealso [get_file_path()] for the generic function.
-get_slf_chi_deaths_path <- function(update = latest_update(), ...) {
+get_slf_chi_deaths_path <- function(update = latest_update(), BYOC_MODE, ...) {
   slf_chi_deaths_path <- get_file_path(
     directory = fs::path(get_slf_dir(), "Deaths"),
     file_name = stringr::str_glue("anon-chi_deaths_{update}.parquet"),

R/join_deaths_data.R

@@ -2,15 +2,23 @@
 #'
 #' @param data Episode file data
 #' @param year financial year, e.g. '1920'
-#' @param slf_deaths_lookup The SLF deaths lookup.
+#' @param BYOC_MODE BYOC mode
 #'
 #' @return The data including the deaths lookup matched
 #'         on to the episode file.
 join_deaths_data <- function(
   data,
   year,
-  slf_deaths_lookup = read_file(get_slf_deaths_lookup_path(year))
+  BYOC_MODE = FALSE
 ) {
+  slf_deaths_lookup <- read_file(get_combined_slf_deaths_lookup_path(BYOC_MODE = BYOC_MODE)) %>%
+    # Filter the chi death dates to the FY as the lookup is by FY
+    dplyr::filter(fy == year) %>%
+    # use the BOXI NRS death date by default, but if it's missing, use the chi death date.
+    dplyr::mutate(
+      deceased = TRUE
+    )
+
   data <- data %>%
     dplyr::left_join(
       slf_deaths_lookup %>%

R/process_extract_nrs_deaths.R

@@ -6,13 +6,26 @@
 #' @param data The extract to process
 #' @param year The year to process, in FY format.
 #' @param write_to_disk (optional) Should the data be written to disk default is
+#' @param BYOC_MODE BYOC_MODE
+#' @param run_id run_id for BYOC
+#' @param run_date_time run_date_time for BYOC
 #' `TRUE` i.e. write the data to disk.
 #'
 #' @return the final data as a [tibble][tibble::tibble-package].
 #' @export
 #' @family process extracts
-process_extract_nrs_deaths <- function(data, year, write_to_disk = TRUE) {
-  log_slf_event(stage = "process", status = "start", type = "deaths", year = year)
+process_extract_nrs_deaths <- function(data,
+                                       year,
+                                       write_to_disk = TRUE,
+                                       BYOC_MODE = FALSE,
+                                       run_id = NA,
+                                       run_date_time = NA) {
+  log_slf_event(
+    stage = "process",
+    status = "start",
+    type = "deaths",
+    year = year
+  )
 
   stopifnot(length(year) == 1L)
 
@@ -25,16 +38,32 @@ process_extract_nrs_deaths <- function(data, year, write_to_disk = TRUE) {
       year = year,
       gpprac = convert_eng_gpprac_to_dummy(.data$gpprac),
       smrtype = add_smrtype(.data$recid)
+    ) %>%
+    dplyr::mutate(
+      run_id = run_id,
+      run_date_time = run_date_time
     )
 
   if (write_to_disk) {
     deaths_clean %>%
-      write_file(get_source_extract_path(year, "deaths", check_mode = "write"),
+      write_file(
+        get_source_extract_path(
+          year,
+          "deaths",
+          check_mode = "write",
+          BYOC_MODE = BYOC_MODE
+        ),
+        BYOC_MODE = BYOC_MODE,
         group_id = 3356 # sourcedev owner
       )
   }
 
-  log_slf_event(stage = "process", status = "complete", type = "deaths", year = year)
+  log_slf_event(
+    stage = "process",
+    status = "complete",
+    type = "deaths",
+    year = year
+  )
 
   return(deaths_clean)
 }

R/process_refined_death.R

@@ -14,19 +14,33 @@
 #' @family process extracts
 process_refined_death <- function(
   it_chi_deaths = read_file(get_slf_chi_deaths_path()),
-  write_to_disk = TRUE
+  write_to_disk = TRUE,
+  BYOC_MODE = FALSE,
+  run_id = NA,
+  run_date_time = NA
 ) {
-  log_slf_event(stage = "process", status = "start", type = "refined_death", year = "all")
+  log_slf_event(
+    stage = "process",
+    status = "start",
+    type = "refined_death",
+    year = "all"
+  )
 
   years_list <- years_to_run()
 
   nrs_all_years <- lapply(years_list, (\(year) {
     read_extract_nrs_deaths(
       year,
-      get_boxi_extract_path(year, type = "deaths")
+      denodo_connect = get_denodo_connect(BYOC_MODE = BYOC_MODE),
+      get_boxi_extract_path(year, type = "deaths", BYOC_MODE = BYOC_MODE),
+      BYOC_MODE = BYOC_MODE
     ) %>%
-      process_extract_nrs_deaths(year,
-        write_to_disk = write_to_disk
+      process_extract_nrs_deaths(
+        year,
+        write_to_disk = write_to_disk,
+        BYOC_MODE = BYOC_MODE,
+        run_id = run_id,
+        run_date_time = run_date_time
       )
   })) %>%
     data.table::rbindlist()
@@ -59,16 +73,23 @@ process_refined_death <- function(
     dplyr::arrange(.data$death_date) %>%
     dplyr::distinct(.data$anon_chi, .keep_all = TRUE) %>%
     dplyr::ungroup()
+  # run_id and run_date_time should have been added in process_extract_nrs_deaths()
 
   if (write_to_disk) {
     write_file(
       refined_death,
-      get_combined_slf_deaths_lookup_path(create = TRUE),
+      get_combined_slf_deaths_lookup_path(create = TRUE, BYOC_MODE = BYOC_MODE),
+      BYOC_MODE = BYOC_MODE,
       group_id = 3206 # hscdiip owner
     )
   }
 
-  log_slf_event(stage = "process", status = "complete", type = "refined_death", year = "all")
+  log_slf_event(
+    stage = "process",
+    status = "complete",
+    type = "refined_death",
+    year = "all"
+  )
 
   return(refined_death)
 }

R/read_extract_nrs_deaths.R

@@ -5,68 +5,68 @@
 #' @export
 read_extract_nrs_deaths <- function(
   year,
-  file_path = get_boxi_extract_path(year = year, type = "deaths")
+  denodo_connect = get_denodo_connect(BYOC_MODE = BYOC_MODE),
+  file_path = get_boxi_extract_path(year, type = "deaths", BYOC_MODE = BYOC_MODE),
+  BYOC_MODE
 ) {
-  log_slf_event(stage = "read", status = "start", type = "deaths", year = year)
+  year <- check_year_format(year, "fyyear")
+  c_year <- convert_fyyear_to_year(year)
 
-  extract_nrs_deaths <- read_file(file_path,
-    col_types = readr::cols_only(
-      "Death Location Code" = readr::col_character(),
-      "Geo Council Area Code" = readr::col_character(),
-      "Geo Postcode [C]" = readr::col_character(),
-      "Geo HSCP of Residence Code - current" = readr::col_character(),
-      "NHS Board of Occurrence Code - current" = readr::col_character(),
-      "NHS Board of Residence Code - current" = readr::col_character(),
-      "Pat Date Of Birth [C]" = readr::col_date(format = "%Y/%m/%d %T"),
-      "Date of Death(99)" = readr::col_date(format = "%Y/%m/%d %T"),
-      "Pat Gender Code" = readr::col_double(),
-      "anon_chi" = readr::col_character(),
-      "Place Death Occurred Code" = readr::col_character(),
-      "Post Mortem Code" = readr::col_character(),
-      "Prim Cause of Death Code (6 char)" = readr::col_character(),
-      "Sec Cause of Death 0 Code (6 char)" = readr::col_character(),
-      "Sec Cause of Death 1 Code (6 char)" = readr::col_character(),
-      "Sec Cause of Death 2 Code (6 char)" = readr::col_character(),
-      "Sec Cause of Death 3 Code (6 char)" = readr::col_character(),
-      "Sec Cause of Death 4 Code (6 char)" = readr::col_character(),
-      "Sec Cause of Death 5 Code (6 char)" = readr::col_character(),
-      "Sec Cause of Death 6 Code (6 char)" = readr::col_character(),
-      "Sec Cause of Death 7 Code (6 char)" = readr::col_character(),
-      "Sec Cause of Death 8 Code (6 char)" = readr::col_character(),
-      "Sec Cause of Death 9 Code (6 char)" = readr::col_character(),
-      "Unique Record Identifier" = readr::col_character(),
-      "GP practice code(99)" = readr::col_character()
-    )
+  log_slf_event(
+    stage = "read",
+    status = "start",
+    type = "deaths",
+    year = year
+  )
+
+  on.exit(try(DBI::dbDisconnect(denodo_connect), silent = TRUE), add = TRUE)
+
+  extract_nrs_deaths <- dplyr::tbl(
+    denodo_connect,
+    # TODO: check table name
+    dbplyr::in_schema("sdl", "sdl_nrs_deaths_episodes")
   ) %>%
-    dplyr::rename(
-      death_location_code = "Death Location Code",
-      lca = "Geo Council Area Code",
-      postcode = "Geo Postcode [C]",
-      hscp = "Geo HSCP of Residence Code - current",
-      death_board_occurrence = "NHS Board of Occurrence Code - current",
-      hbrescode = "NHS Board of Residence Code - current",
-      dob = "Pat Date Of Birth [C]",
-      record_keydate1 = "Date of Death(99)",
-      gender = "Pat Gender Code",
-      anon_chi = "anon_chi",
-      place_death_occurred = "Place Death Occurred Code",
-      post_mortem = "Post Mortem Code",
-      deathdiag1 = "Prim Cause of Death Code (6 char)",
-      deathdiag2 = "Sec Cause of Death 0 Code (6 char)",
-      deathdiag3 = "Sec Cause of Death 1 Code (6 char)",
-      deathdiag4 = "Sec Cause of Death 2 Code (6 char)",
-      deathdiag5 = "Sec Cause of Death 3 Code (6 char)",
-      deathdiag6 = "Sec Cause of Death 4 Code (6 char)",
-      deathdiag7 = "Sec Cause of Death 5 Code (6 char)",
-      deathdiag8 = "Sec Cause of Death 6 Code (6 char)",
-      deathdiag9 = "Sec Cause of Death 7 Code (6 char)",
-      deathdiag10 = "Sec Cause of Death 8 Code (6 char)",
-      deathdiag11 = "Sec Cause of Death 9 Code (6 char)",
-      uri = "Unique Record Identifier",
-      gpprac = "GP practice code(99)"
-    )
+    dplyr::select(
+      death_location_code = "death_location_code",
+      lca = "geo_council_area_code",
+      postcode = "geo_postcode",
+      hscp = "geo_hscp_of_residence_code_curr",
+      death_board_occurrence = "nhs_board_of_occurrence_code_curr",
+      hbrescode = "nhs_board_of_residence_code_curr",
+      dob = "patient_dob",
+      record_keydate1 = "patient_dod",
+      gender = "pat_gender_code",
+      chi = "patient_chi",
+      place_death_occurred = "place_death_occurred_code",
+      post_mortem = "post_mortem_code",
+      deathdiag1 = "primary_cause_of_death_code",
+      deathdiag2 = "secondary_cause_of_death_0_code",
+      deathdiag3 = "secondary_cause_of_death_1_code",
+      deathdiag4 = "secondary_cause_of_death_2_code",
+      deathdiag5 = "secondary_cause_of_death_3_code",
+      deathdiag6 = "secondary_cause_of_death_4_code",
+      deathdiag7 = "secondary_cause_of_death_5_code",
+      deathdiag8 = "secondary_cause_of_death_6_code",
+      deathdiag9 = "secondary_cause_of_death_7_code",
+      deathdiag10 = "secondary_cause_of_death_8_code",
+      deathdiag11 = "secondary_cause_of_death_9_code",
+      uri = "unique_record_identifier",
+      gpprac = "gp_practice_code",
+      date_of_death_financial_year = "date_of_death_financial_year"
+    ) %>%
+    dplyr::filter(
+      date_of_death_financial_year == c_year
+    ) %>%
+    dplyr::select(-"date_of_death_financial_year") %>%
+    dplyr::collect() %>%
+    slfhelper::get_anon_chi()
 
-  log_slf_event(stage = "read", status = "complete", type = "deaths", year = year)
+  log_slf_event(
+    stage = "read",
+    status = "complete",
+    type = "deaths",
+    year = year
+  )
 
   return(extract_nrs_deaths)
 }

_targets.R

@@ -286,7 +286,10 @@ list(
     refined_death_data,
     process_refined_death(
       it_chi_deaths = it_chi_deaths_data,
-      write_to_disk = write_to_disk
+      write_to_disk = write_to_disk,
+      BYOC_MODE = BYOC_MODE,
+      run_id = run_id,
+      run_date_time = run_date_time
     )
   ),
   ### Social Care - 'All' data -----------------------------------------------