333 c02 link delayed discharges episodes

.github/actions/spelling/expect.txt

@@ -1,6 +1,7 @@
 Accom
 admloc
 admtype
+ADPE
 adtf
 arrivalmode
 arth
@@ -34,6 +35,7 @@ createslf
 dataframe
 datamart
 datazone
+datediff
 dateformat
 dateop
 datetime
@@ -82,6 +84,7 @@ hbtreatname
 hci
 HCP
 HHG
+hhg
 hjust
 hms
 homecare
@@ -172,6 +175,7 @@ smr
 SMRA
 smrtype
 SPARRA
+sparra
 spd
 SPSS
 spss
@@ -200,6 +204,7 @@ vline
 xintercept
 xlsx
 yearstay
+YYYYQX
 zihao
 zsav
 zstd

NAMESPACE

@@ -1,6 +1,7 @@
 # Generated by roxygen2: do not edit by hand
 
 export("%>%")
+export(add_dd)
 export(add_nsu_cohort)
 export(add_ppa_flag)
 export(check_year_format)
@@ -73,6 +74,7 @@ export(get_year_dir)
 export(is_date_in_fyyear)
 export(is_missing)
 export(la_code_lookup)
+export(last_date_month)
 export(latest_cost_year)
 export(latest_update)
 export(midpoint_fy)

R/add_dd.R

@@ -0,0 +1,314 @@
+#' Add Delay Discharge to working file
+#'
+#' @param data The input data frame
+#' @param year The year being processed
+#'
+#' @return A data frame linking delay discharge cohorts
+#' @export
+#'
+#' @family episode file
+add_dd <- function(data, year) {
+  year_param <- year
+
+  data <- data %>%
+    dplyr::mutate(
+      # remember to revoke the cij_end_date with dummy_cij_end
+      cij_start_date_lower = .data$cij_start_date - lubridate::days(1),
+      cij_end_date_upper = cij_end_date + lubridate::days(1),
+      cij_end_month = last_date_month(cij_end_date),
+      is_dummy_cij_start = is.na(cij_start_date) & !is.na(cij_end_date),
+      dummy_cij_start = dplyr::if_else(
+        is_dummy_cij_start,
+        lubridate::as_date("1900-01-01"),
+        cij_start_date_lower
+      ),
+      is_dummy_cij_end = !is.na(cij_start_date) & is.na(cij_end_date),
+      dummy_cij_end = dplyr::if_else(
+        is_dummy_cij_end,
+        lubridate::today(),
+        cij_end_month
+      )
+    )
+
+  ## handling DD ----
+  # no flag for last reported
+  dd_data <-
+    read_file(get_source_extract_path(year_param, "DD")) %>%
+    dplyr::rename(
+      record_keydate1 = keydate1_dateformat,
+      record_keydate2 = keydate2_dateformat
+    ) %>%
+    dplyr::mutate(
+      # remember to revoke the keydate2 and amended_dates with dummy_keydate2
+      is_dummy_keydate2 = is.na(record_keydate2),
+      dummy_keydate2 = dplyr::if_else(is_dummy_keydate2,
+        lubridate::today(),
+        record_keydate2
+      ),
+      dummy_id = dplyr::row_number()
+    )
+
+  by_dd <- dplyr::join_by(
+    chi,
+    x$record_keydate1 >= y$dummy_cij_start,
+    x$dummy_keydate2 <= y$dummy_cij_end
+  )
+  data <- dd_data %>%
+    dplyr::inner_join(data,
+      by = by_dd,
+      suffix = c("_dd", "")
+    ) %>%
+    dplyr::arrange(cij_start_date, cij_end_date, cij_marker, postcode) %>%
+    # remove duplicate rows, but still got some duplicate mis-matches
+    dplyr::distinct(
+      chi,
+      cij_start_date,
+      cij_end_date,
+      cij_marker,
+      record_keydate1_dd,
+      record_keydate2_dd,
+      .keep_all = TRUE
+    ) %>%
+    # determine DD quality
+    dplyr::mutate(
+      dd_type = dplyr::if_else(
+        is.na(cij_marker),
+        "no-cij",
+        dplyr::case_when(
+          # "1"	"Accurate Match - (1)"
+          # "1P"	"Accurate Match (allowing +-1 day) - (1P)"
+          # "1A"	"Accurate Match (has an assumed  end date) - (1A)"
+          # "1AP"	"Accurate Match (allowing +-1 day and has an assumed end date) - (1AP)"
+          # "2"	"Starts in CIJ - (2)"
+          # "2D"	"Starts in CIJ (ends one day after) - (2D)"
+          # "2DP"	"Starts in CIJ (allowing +-1 day and ends one day after) - (2DP)"
+          # "2A"	"Starts in CIJ (Accurate Match after correcting assumed end date) - (2A)"
+          # "2AP"	"Starts in CIJ (Accurate Match (allowing +-1 day) after correcting assumed end date) - (2AP)"
+          # "3"	"Ends in CIJ - (3)"
+          # "3D"	"Ends in CIJ (starts one day before) - (3D)"
+          # "3DP"	"Ends in CIJ (allowing +-1 day and starts one day before) - (3DP)"
+          # "4"	"Matches unended MH record - (4)"
+          # "4P" "Matches unended MH record (allowing -1 day) - (4P)"
+          # "-" "No Match (We don't keep these)".
+
+          # If we use record_keydate2_dd,
+          # we implicitly mean is_dummy_keydate2 needs to be FALSE.
+          # Given that in DD files,
+          # we only keep the records with missing keydate2 for 04B, mental health,
+          # and drop the records with missing keydate2 for other recid,
+          # it should be ok to only use dummy_keydate2 for "4"(s).
+
+          # "1"	"Accurate Match - (1)"
+          record_keydate1_dd >= cij_start_date &
+            record_keydate2_dd <= cij_end_date &
+            !amended_dates ~ "1",
+
+          # "1P"	"Accurate Match (allowing +-1 day) - (1P)"
+          record_keydate1_dd >= cij_start_date_lower &
+            record_keydate2_dd <= cij_end_date_upper &
+            !amended_dates ~ "1P",
+
+          # "1A"	"Accurate Match (has an assumed end date) - (1A)"
+          record_keydate1_dd >= cij_start_date &
+            record_keydate2_dd <= cij_end_date &
+            amended_dates ~ "1A",
+
+          # "1AP"	"Accurate Match (allowing +-1 day and has an assumed end date) - (1AP)"
+          record_keydate1_dd >= cij_start_date_lower &
+            record_keydate2_dd <= cij_end_date_upper &
+            amended_dates ~ "1AP",
+
+          # "1APE"	the CIJ ends during the month but the delay has an end date of the end of the month
+          record_keydate1_dd >= cij_start_date_lower &
+            record_keydate2_dd == cij_end_month &
+            amended_dates ~ "1APE",
+
+          # "2"	"Starts in CIJ - (2)"
+          record_keydate1_dd >= cij_start_date &
+            record_keydate1_dd <= cij_end_date &
+            record_keydate2_dd > cij_end_date &
+            !amended_dates ~ "2",
+
+          # "2D"	"Starts in CIJ (ends one day after) - (2D)"
+          record_keydate1_dd >= cij_start_date &
+            record_keydate1_dd <= cij_end_date &
+            record_keydate2_dd > cij_end_date_upper &
+            !amended_dates ~ "2D",
+
+          # "2DP"	"Starts in CIJ (allowing +-1 day and ends one day after) - (2DP)"
+          record_keydate1_dd >= cij_start_date_lower &
+            record_keydate1_dd <= cij_end_date_upper &
+            record_keydate2_dd > cij_end_date_upper &
+            !amended_dates ~ "2DP",
+
+          # "2A"	"Starts in CIJ (Accurate Match after correcting assumed end date) - (2A)"
+          record_keydate1_dd >= cij_start_date &
+            record_keydate1_dd <= cij_end_date &
+            record_keydate2_dd > cij_end_date &
+            amended_dates ~ "2A",
+
+          # "2AP"	"Starts in CIJ (Accurate Match (allowing +-1 day) after correcting assumed end date) - (2AP)"
+          record_keydate1_dd >= cij_start_date_lower &
+            record_keydate1_dd <= cij_end_date_upper &
+            record_keydate2_dd > cij_end_date_upper &
+            # record_keydate2_dd == cij_end_month &
+            amended_dates ~ "2AP",
+
+          # "3"	"Ends in CIJ - (3)"
+          record_keydate1_dd <= cij_start_date &
+            record_keydate2_dd >= cij_start_date &
+            record_keydate2_dd <= cij_end_date &
+            !amended_dates ~ "3",
+
+          # "3D"	"Ends in CIJ (starts one day before) - (3D)"
+          record_keydate1_dd <= cij_start_date_lower &
+            record_keydate2_dd >= cij_start_date &
+            record_keydate2_dd <= cij_end_date &
+            !amended_dates ~ "3D",
+
+          # "3DP"	"Ends in CIJ (allowing +-1 day and starts one day before) - (3DP)"
+          record_keydate1_dd <= cij_start_date_lower &
+            record_keydate2_dd >= cij_start_date_lower &
+            record_keydate2_dd <= cij_end_date_upper &
+            !amended_dates ~ "3DP",
+
+          # "3ADPE"
+          record_keydate1_dd <= cij_start_date_lower &
+            record_keydate2_dd >= cij_start_date_lower &
+            record_keydate2_dd <= cij_end_month &
+            amended_dates ~ "3ADPE",
+
+          # "4"	"Matches unended MH record - (4)"
+          recid == "04B" &
+            record_keydate1_dd >= cij_start_date &
+            is_dummy_cij_end ~ "4",
+
+          # "4P"	"Matches unended MH record (allowing -1 day) - (4P)"
+          recid == "04B" &
+            record_keydate1_dd >= cij_start_date_lower &
+            is_dummy_cij_end ~ "4P",
+
+          # "-" "No Match (We don't keep these)"
+          .default = "-"
+        )
+      ),
+      dd_type = factor(
+        dd_type,
+        levels = c(
+          "1",
+          "1P",
+          "1A",
+          "1AP",
+          "2",
+          "2D",
+          "2DP",
+          "2A",
+          "2AP",
+          "3",
+          "3D",
+          "3DP",
+          "1APE",
+          "3ADPE",
+          "4",
+          "4P",
+          "-"
+        )
+      ),
+
+      # For "1APE", assign 1APE cij_end_date to record_keydate2_dd
+      record_keydate2_dd = dplyr::if_else(
+        dd_type == "1APE" | dd_type == "3ADPE",
+        cij_end_date,
+        record_keydate2_dd,
+      ),
+      datediff_end = abs(cij_end_date - record_keydate2_dd),
+      datediff_start = cij_start_date - record_keydate1_dd
+    ) %>%
+    dplyr::filter(dd_type != "-") %>%
+    dplyr::mutate(smrtype_dd = dplyr::case_when(
+      dd_type %in% c(
+        "1",
+        "1P",
+        "1A",
+        "1AP",
+        "1APE",
+        "2",
+        "2D",
+        "2DP",
+        "2A",
+        "2AP",
+        "3",
+        "3D",
+        "3DP",
+        "3ADPE",
+        "4",
+        "4P"
+      ) ~ "DD-CIJ",
+      dd_type %in% c("no-cij") ~ "DD-No CIJ"
+    )) %>%
+    # remove duplicated rows when many to many inner join
+    # keep the records that closest to the cij record
+    dplyr::arrange(
+      chi,
+      original_admission_date,
+      record_keydate1_dd,
+      record_keydate2_dd,
+      dummy_id,
+      dd_type,
+      datediff_end, -datediff_start
+    ) %>%
+    dplyr::distinct(postcode,
+      record_keydate1_dd,
+      record_keydate2_dd,
+      .keep_all = TRUE
+    ) %>%
+    # tidy up and rename columns to match the format of episode files
+    dplyr::select(
+      "year" = "year_dd",
+      "recid" = "recid_dd",
+      "record_keydate1" = "record_keydate1_dd",
+      "record_keydate2" = "record_keydate2_dd",
+      "smrtype" = "smrtype_dd",
+      "chi",
+      "gender",
+      "dob",
+      "age",
+      "gpprac",
+      "postcode" = "postcode_dd",
+      "lca" = "dd_responsible_lca",
+      "hbtreatcode" = "hbtreatcode_dd",
+      "original_admission_date",
+      "amended_dates",
+      "delay_end_reason",
+      "primary_delay_reason",
+      "secondary_delay_reason",
+      "cij_marker",
+      "cij_start_date",
+      "cij_end_date",
+      "cij_pattype_code",
+      "cij_ipdc",
+      "cij_admtype",
+      "cij_adm_spec",
+      "cij_dis_spec",
+      "location",
+      "spec" = "spec_dd",
+      "dd_type"
+    ) %>%
+    # combine DD with episode data
+    dplyr::bind_rows( # restore cij_end_date
+      data %>%
+        dplyr::select(
+          -c(
+            "cij_start_date_lower",
+            "cij_end_date_upper",
+            "cij_end_month",
+            "is_dummy_cij_start",
+            "dummy_cij_start",
+            "is_dummy_cij_end",
+            "dummy_cij_end"
+          )
+        )
+    )
+
+  return(data)
+}

R/last_date_month.R

@@ -0,0 +1,16 @@
+#' Return the end date of the month of the given date
+#'
+#' @description Return the end date of the month of the given date
+#'
+#' @param x a date with a date format
+#'
+#' @return a vector of dates of the end date of the FY year
+#' @export
+#'
+#' @examples
+#' last_date_month(lubridate::as_date("2020-02-05"))
+#'
+#' @family date functions
+last_date_month <- function(x) {
+  return(lubridate::ceiling_date(x, "month") - lubridate::days(1))
+}