Merge branch 'master' into episode/write_dataset

R/fill_geographies.R

@@ -120,7 +120,7 @@ fill_postcode_geogs <- function(data) {
       hscp = dplyr::coalesce(.data$hscp2018, .data$hscp),
       lca = dplyr::coalesce(.data$lca, .data$lca_old)
     ) %>%
-    dplyr::select(-"hb2018", -"hscp2018", -"lca_old")
+    dplyr::select(!c("hb2018", "hscp2018", "lca_old", "most_recent_postcode"))
 
   return(filled_postcodes)
 }
@@ -148,7 +148,7 @@ fill_gpprac_geographies <- function(data) {
     dplyr::mutate(
       hbpraccode = dplyr::coalesce(.data$hbpraccode, .data$hbpraccode_old)
     ) %>%
-    dplyr::select(-"hbpraccode_old")
+    dplyr::select(!c("hbpraccode_old", "most_recent_gpprac"))
 
   return(filled_gpprac)
 }

R/link_delayed_discharge_eps.R

@@ -1,16 +1,18 @@
 #' Link  Delayed Discharge to WIP episode file
 #'
-#' @param data The input data frame
+#' @param episode_file The episode file
 #' @param year The year being processed
+#' @param dd_data The processed DD extract
 #'
 #' @return A data frame with the delayed discharge cohort added and linked
 #' using the `cij_marker`
 #'
 #' @family episode file
-link_delayed_discharge_eps <- function(data, year) {
-  year_param <- year
-
-  data <- data %>%
+link_delayed_discharge_eps <- function(
+    episode_file,
+    year,
+    dd_data = read_file(get_source_extract_path(year, "DD"))) {
+  episode_file <- episode_file %>%
     dplyr::mutate(
       # remember to revoke the cij_end_date with dummy_cij_end
       cij_start_date_lower = .data$cij_start_date - lubridate::days(1L),
@@ -34,8 +36,7 @@ link_delayed_discharge_eps <- function(data, year) {
 
   ## handling DD ----
   # no flag for last reported
-  dd_data <-
-    read_file(get_source_extract_path(year_param, "DD")) %>%
+  dd_data <- dd_data %>%
     dplyr::mutate(
       # remember to revoke the keydate2 and amended_dates with dummy_keydate2
       is_dummy_keydate2 = is.na(.data$record_keydate2),
@@ -53,11 +54,12 @@ link_delayed_discharge_eps <- function(data, year) {
     x$dummy_keydate2 <= y$dummy_cij_end
   )
 
-  data <- dd_data %>%
-    dplyr::inner_join(data,
-      by = by_dd,
-      suffix = c("_dd", "")
-    ) %>%
+  linked_data <- dplyr::inner_join(
+    x = dd_data,
+    y = episode_file,
+    by = by_dd,
+    suffix = c("_dd", "")
+  ) %>%
     # Remove existing beddays as we're re-calculating them for this cohort
     dplyr::select(-dplyr::ends_with("beddays")) %>%
     dplyr::arrange(
@@ -78,7 +80,7 @@ link_delayed_discharge_eps <- function(data, year) {
     ) %>%
     # determine DD quality
     dplyr::mutate(
-      dd_type = dplyr::if_else(
+      dd_quality = dplyr::if_else(
         is.na(.data$cij_marker),
         "no-cij",
         dplyr::case_when(
@@ -199,8 +201,8 @@ link_delayed_discharge_eps <- function(data, year) {
           .default = "-"
         )
       ),
-      dd_type = factor(
-        .data$dd_type,
+      dd_quality = factor(
+        .data$dd_quality,
         levels = c(
           "1",
           "1P",
@@ -224,16 +226,16 @@ link_delayed_discharge_eps <- function(data, year) {
 
       # For "1APE", assign 1APE cij_end_date to record_keydate2_dd
       record_keydate2_dd = dplyr::if_else(
-        .data$dd_type == "1APE" | .data$dd_type == "3ADPE",
+        .data$dd_quality == "1APE" | .data$dd_quality == "3ADPE",
         max(.data$record_keydate1_dd, .data$cij_end_date),
         .data$record_keydate2_dd
       ),
       datediff_end = abs(.data$cij_end_date - .data$record_keydate2_dd),
       datediff_start = .data$cij_start_date - .data$record_keydate1_dd
     ) %>%
-    dplyr::filter(.data$dd_type != "-") %>%
+    dplyr::filter(.data$dd_quality != "-") %>%
     dplyr::mutate(smrtype = dplyr::case_match(
-      as.character(.data$dd_type),
+      as.character(.data$dd_quality),
       c(
         "1",
         "1P",
@@ -262,7 +264,7 @@ link_delayed_discharge_eps <- function(data, year) {
       .data$record_keydate1_dd,
       .data$record_keydate2_dd,
       .data$dummy_id,
-      .data$dd_type,
+      .data$dd_quality,
       .data$datediff_end,
       dplyr::desc(.data$datediff_start)
     ) %>%
@@ -289,7 +291,9 @@ link_delayed_discharge_eps <- function(data, year) {
       .data$record_keydate2_dd,
       count_last = TRUE
     ) %>%
-    dplyr::mutate(yearstay = rowSums(dplyr::pick(dplyr::ends_with("_beddays")))) %>%
+    dplyr::mutate(
+      yearstay = rowSums(dplyr::pick(dplyr::ends_with("_beddays")))
+    ) %>%
     # tidy up and rename columns to match the format of episode files
     dplyr::select(
       "year" = "year_dd",
@@ -303,9 +307,8 @@ link_delayed_discharge_eps <- function(data, year) {
       "age",
       "gpprac",
       "postcode" = "postcode_dd",
-      "lca" = "dd_responsible_lca",
+      "dd_responsible_lca",
       "hbtreatcode" = "hbtreatcode_dd",
-      "original_admission_date",
       "delay_end_reason",
       "primary_delay_reason",
       "secondary_delay_reason",
@@ -320,13 +323,13 @@ link_delayed_discharge_eps <- function(data, year) {
       "cij_delay",
       "location",
       "spec" = "spec_dd",
-      "dd_type",
+      "dd_quality",
       dplyr::ends_with("_beddays"),
       "yearstay"
     ) %>%
-    # combine DD with episode data
+    # Combine DD with episode data
     dplyr::bind_rows(
-      data %>%
+      episode_file %>%
         dplyr::select(
           -c(
             "cij_start_date_lower",
@@ -340,5 +343,5 @@ link_delayed_discharge_eps <- function(data, year) {
         )
     )
 
-  return(data)
+  return(linked_data)
 }

R/process_extract_delayed_discharges.R

@@ -12,10 +12,10 @@
 #' @return the final data as a [tibble][tibble::tibble-package].
 #' @export
 #' @family process extracts
-
-process_extract_delayed_discharges <- function(data,
-                                               year,
-                                               write_to_disk = TRUE) {
+process_extract_delayed_discharges <- function(
+    data,
+    year,
+    write_to_disk = TRUE) {
   # Only run for a single year
   stopifnot(length(year) == 1L)
 
@@ -48,7 +48,8 @@ process_extract_delayed_discharges <- function(data,
       record_keydate1 = .data[["rdd"]],
       record_keydate2 = .data[["delay_end_date"]]
     ) %>%
-    # Use end of the month date for records with no end date (but we think have ended)
+    # Use end of the month date for records with no end date
+    # (but we think have ended)
     # Create a flag for these records
     dplyr::mutate(
       month_end = lubridate::ceiling_date(.data[["monthflag"]], "month") - 1L,
@@ -83,7 +84,8 @@ process_extract_delayed_discharges <- function(data,
       # Flag records with no end date
       not_mh_spec = is.na(.data$record_keydate2) & !(.data$spec %in% mh_spec)
     ) %>%
-    # Keep only records which have an end date (except Mental Health) and fall within our dates.
+    # Keep only records which have an end date (except Mental Health) and fall
+    # within our dates.
     dplyr::filter(.data$dates_in_fyyear, !.data$not_mh_spec)
 
   dd_final <- dd_clean %>%
@@ -106,9 +108,10 @@ process_extract_delayed_discharges <- function(data,
     )
 
   if (write_to_disk) {
-    dd_final %>%
-      # Save as rds file
-      write_file(get_source_extract_path(year, "DD", check_mode = "write"))
+    write_file(
+      dd_final,
+      get_source_extract_path(year, "DD", check_mode = "write")
+    )
   }
 
   return(dd_final)

R/process_sc_all_home_care.R

@@ -147,9 +147,6 @@ process_sc_all_home_care <- function(
         )
     )
 
-
-  # Outfile ---------------------------------------
-
   merge_data <- pivoted_hours %>%
     # group the data to be merged
     dplyr::group_by(
@@ -169,16 +166,19 @@ process_sc_all_home_care <- function(
       # Store the period for the latest submitted record
       sc_latest_submission = dplyr::last(.data$period),
       # Sum the (quarterly) hours
-      dplyr::across(tidyselect::starts_with("hc_hours_"), sum),
-      dplyr::across(tidyselect::starts_with("hc_cost_"), sum),
+      dplyr::across(
+        c(dplyr::starts_with("hc_hours_"), -"hc_hours_derived"),
+        sum
+      ),
+      dplyr::across(dplyr::starts_with("hc_cost_"), sum),
       # Shouldn't matter as these are all the same
       dplyr::across(c("gender", "dob", "postcode"), dplyr::first)
     ) %>%
     dplyr::ungroup()
 
 
   # Create Source variables---------------------------------------
-  final_data <- merge_data %>%
+  all_hc_processed <- merge_data %>%
     # rename
     dplyr::rename(
       record_keydate1 = "hc_service_start_date",
@@ -198,15 +198,16 @@ process_sc_all_home_care <- function(
     # person_id
     create_person_id(type = "SC") %>%
     # compute lca variable from sending_location
-    dplyr::mutate(sc_send_lca = convert_sending_location_to_lca(.data$sending_location))
-
-  # Save outfile---------------------------------------------------
+    dplyr::mutate(
+      sc_send_lca = convert_sending_location_to_lca(.data$sending_location)
+    )
 
   if (write_to_disk) {
-    # Save .rds file
-    final_data %>%
-      write_file(get_sc_hc_episodes_path(check_mode = "write"))
+    write_file(
+      all_hc_processed,
+      get_sc_hc_episodes_path(check_mode = "write")
+    )
   }
 
-  return(final_data)
+  return(all_hc_processed)
 }

R/run_episode_file.R

@@ -100,6 +100,7 @@ run_episode_file <- function(processed_data_list, year, write_to_disk = TRUE) {
     join_cohort_lookups(year) %>%
     join_sparra_hhg(year) %>%
     fill_geographies() %>%
+    join_deaths_data(year) %>%
     load_ep_file_vars(year)
 
   if (write_to_disk) {