Merge branch 'september-2024' into care_home_update

R/add_activity_after_death_flag.R

@@ -13,6 +13,8 @@ add_activity_after_death_flag <- function(
     year,
     deaths_data = read_file(get_all_slf_deaths_lookup_path()) %>%
       slfhelper::get_chi()) {
+  cli::cli_alert_info("Add activity after death flag function started at {Sys.time()}")
+
   # to skip warnings no visible binding for global variable ‘.’
   . <- NULL
 
@@ -85,7 +87,7 @@ add_activity_after_death_flag <- function(
   flag_data <- flag_data %>%
     dplyr::filter(.data$activity_after_death == 1) %>%
     # Remove temporary flag variables used to create activity after death flag and fill in missing death_date
-    dplyr::select(.data$year, .data$chi, .data$record_keydate1, .data$record_keydate2, .data$activity_after_death) %>%
+    dplyr::select(.data$year, .data$chi, .data$record_keydate1, .data$record_keydate2, .data$activity_after_death, .data$death_date_boxi) %>%
     dplyr::distinct()
 
   # Match activity after death flag back to episode file
@@ -94,7 +96,12 @@ add_activity_after_death_flag <- function(
       flag_data,
       by = c("year", "chi", "record_keydate1", "record_keydate2"),
       na_matches = "never"
-    )
+    ) %>%
+    dplyr::mutate(death_date = lubridate::as_date(ifelse(is.na(death_date) & !(is.na(death_date_boxi)),
+      death_date_boxi, death_date
+    ))) %>%
+    dplyr::select(-death_date_boxi)
+
 
 
   return(final_data)

R/add_nsu_cohort.R

@@ -13,6 +13,8 @@ add_nsu_cohort <- function(
     data,
     year,
     nsu_cohort = read_file(get_nsu_path(year)) %>% slfhelper::get_chi()) {
+  cli::cli_alert_info("Add NSU cohort function started at {Sys.time()}")
+
   year_param <- year
 
   if (!check_year_valid(year, "nsu")) {

R/add_ppa_flag.R

@@ -8,6 +8,8 @@
 #' @return A data frame to use as a lookup of PPAs
 #' @family episode_file
 add_ppa_flag <- function(data) {
+  cli::cli_alert_info("Add PPA flag function started at {Sys.time()}")
+
   check_variables_exist(
     data,
     variables = c(

R/correct_demographics.R

@@ -7,6 +7,8 @@
 #'
 #' @return episode files with updated date of birth and ages
 correct_demographics <- function(data, year) {
+  cli::cli_alert_info("Correct demographics function started at {Sys.time()}")
+
   # keep episodes with missing chi
   data_no_chi <- data %>%
     dplyr::filter(is_missing(.data$chi))

R/cost_uplift.R

@@ -5,6 +5,8 @@
 #' @return episode data with uplifted costs
 #' @family episode_file
 apply_cost_uplift <- function(data) {
+  cli::cli_alert_info("Apply cost uplift function started at {Sys.time()}")
+
   data <- data %>%
     # attach a uplift scale as the last column
     lookup_uplift() %>%

R/create_episode_file.R

@@ -199,6 +199,8 @@ create_episode_file <- function(
 #'
 #' @return `data` with only the `vars_to_keep` kept
 store_ep_file_vars <- function(data, year, vars_to_keep) {
+  cli::cli_alert_info("Store episode file variables function started at {Sys.time()}")
+
   tempfile_path <- get_file_path(
     directory = get_year_dir(year),
     file_name = stringr::str_glue("temp_ep_file_variable_store_{year}.parquet"),
@@ -236,6 +238,8 @@ store_ep_file_vars <- function(data, year, vars_to_keep) {
 #'
 #' @return The full SLF data.
 load_ep_file_vars <- function(data, year) {
+  cli::cli_alert_info("Load episode file variable function started at {Sys.time()}")
+
   tempfile_path <- get_file_path(
     directory = get_year_dir(year),
     file_name = stringr::str_glue("temp_ep_file_variable_store_{year}.parquet"),
@@ -263,6 +267,8 @@ load_ep_file_vars <- function(data, year) {
 #'
 #' @return A data frame with CIJ markers filled in for those missing.
 fill_missing_cij_markers <- function(data) {
+  cli::cli_alert_info("Fill missing cij markers function started at {Sys.time()}")
+
   fixable_data <- data %>%
     dplyr::filter(
       .data[["recid"]] %in% c("01B", "04B", "GLS", "02B", "DD") & !is.na(.data[["chi"]])
@@ -317,6 +323,8 @@ fill_missing_cij_markers <- function(data) {
 #'
 #' @return The data with CIJ variables corrected.
 correct_cij_vars <- function(data) {
+  cli::cli_alert_info("Correct cij variables function started at {Sys.time()}")
+
   check_variables_exist(
     data,
     c("chi", "recid", "cij_admtype", "cij_pattype_code")
@@ -358,6 +366,8 @@ correct_cij_vars <- function(data) {
 #'
 #' @return The data with cost including dna.
 create_cost_inc_dna <- function(data) {
+  cli::cli_alert_info("Create cost inc dna function started at {Sys.time()}")
+
   check_variables_exist(data, c("cost_total_net", "attendance_status"))
 
   # Create cost including DNAs and modify costs
@@ -382,6 +392,8 @@ create_cost_inc_dna <- function(data) {
 #'
 #' @return The data unchanged (the cohorts are written to disk)
 create_cohort_lookups <- function(data, year, update = latest_update()) {
+  cli::cli_alert_info("Create cohort lookups function started at {Sys.time()}")
+
   create_demographic_cohorts(
     data,
     year,
@@ -421,6 +433,8 @@ join_cohort_lookups <- function(
       col_select = c("anon_chi", "service_use_cohort")
     ) %>%
       slfhelper::get_chi()) {
+  cli::cli_alert_info("Join cohort lookups function started at {Sys.time()}")
+
   join_cohort_lookups <- data %>%
     dplyr::left_join(
       demographic_cohort,
@@ -447,6 +461,8 @@ join_sc_client <- function(data,
                            year,
                            sc_client = read_file(get_sc_client_lookup_path(year)) %>% slfhelper::get_chi(),
                            file_type = c("episode", "individual")) {
+  cli::cli_alert_info("Join social care client function started at {Sys.time()}")
+
   if (file_type == "episode") {
     # Match on client variables by chi
     data_file <- data %>%

R/create_individual_file.R

@@ -260,6 +260,8 @@ add_all_columns <- function(episode_file, year) {
 #' @param condition Condition to create new columns based on
 #' @family individual_file
 add_acute_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add acute columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition, episode = TRUE, cost = TRUE) %>%
@@ -271,6 +273,8 @@ add_acute_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_mat_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add maternity columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition, episode = TRUE, cost = TRUE) %>%
@@ -282,6 +286,8 @@ add_mat_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_mh_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add mental health columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition, episode = TRUE, cost = TRUE) %>%
@@ -293,6 +299,8 @@ add_mh_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_gls_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add geriatric long stay columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition, episode = TRUE, cost = TRUE) %>%
@@ -304,6 +312,8 @@ add_gls_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_op_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add outpatient columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file <- episode_file %>%
     add_standard_cols(prefix, condition)
@@ -327,6 +337,8 @@ add_op_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_ae_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add A&E columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition, cost = TRUE) %>%
@@ -338,6 +350,8 @@ add_ae_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_pis_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add prescribing columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition, cost = TRUE) %>%
@@ -349,6 +363,8 @@ add_pis_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_ooh_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add out of hours columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file <- episode_file %>%
     add_standard_cols(prefix, condition, cost = TRUE) %>%
@@ -384,6 +400,8 @@ add_ooh_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_dn_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add district nursing columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   if ("total_no_dn_contacts" %in% names(episode_file)) {
     episode_file %>%
@@ -407,6 +425,8 @@ add_dn_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_cmh_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add communicty mental health columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition) %>%
@@ -418,6 +438,8 @@ add_cmh_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_dd_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add delayed discharges columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   condition_delay <- substitute(condition & primary_delay_reason != "9")
   episode_file <- episode_file %>%
@@ -439,6 +461,8 @@ add_dd_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_nsu_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add non service users columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition) %>%
@@ -450,6 +474,8 @@ add_nsu_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_nrs_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add nrs columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition) %>%
@@ -461,6 +487,8 @@ add_nrs_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_hl1_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add homelessness columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition)
@@ -471,6 +499,8 @@ add_hl1_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_ch_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add care home columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition) %>%
@@ -501,6 +531,8 @@ add_ch_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_hc_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add home care columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
 
   episode_file <- episode_file %>%
@@ -545,6 +577,8 @@ add_hc_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_at_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add alarms telecare columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition) %>%
@@ -559,6 +593,8 @@ add_at_columns <- function(episode_file, prefix, condition) {
 #' @inheritParams add_acute_columns
 #' @family individual_file
 add_sds_columns <- function(episode_file, prefix, condition) {
+  cli::cli_alert_info("Add SDS columns function started at {Sys.time()}")
+
   condition <- substitute(condition)
   episode_file %>%
     add_standard_cols(prefix, condition) %>%
@@ -581,6 +617,8 @@ add_sds_columns <- function(episode_file, prefix, condition) {
 #' cij_pattype (lgl)
 #' @family individual_file
 add_ipdc_cols <- function(episode_file, prefix, condition, ipdc_d = TRUE, elective = TRUE) {
+  cli::cli_alert_info("Add ipdc columns function started at {Sys.time()}")
+
   condition_i <- substitute(eval(condition) & ipdc == "I")
   episode_file <- episode_file %>%
     dplyr::mutate(
@@ -625,6 +663,8 @@ add_ipdc_cols <- function(episode_file, prefix, condition, ipdc_d = TRUE, electi
 #' @param cost Whether to create prefix_cost col, e.g. "Acute_cost"
 #' @family individual_file
 add_standard_cols <- function(episode_file, prefix, condition, episode = FALSE, cost = FALSE) {
+  cli::cli_alert_info("Add standard columns function started at {Sys.time()}")
+
   if (episode) {
     episode_file <- dplyr::mutate(episode_file, "{prefix}_episodes" := dplyr::if_else(eval(condition), 1L, NA_integer_))
   }
@@ -703,6 +743,8 @@ recode_gender <- function(episode_file) {
 #' "dementia" and "dementia_date"
 #' @family individual_file
 condition_cols <- function() {
+  cli::cli_alert_info("Return condition columns function started at {Sys.time()}")
+
   conditions <- slfhelper::ltc_vars
   date_cols <- paste0(conditions, "_date")
   all_cols <- c(conditions, date_cols)
@@ -759,6 +801,8 @@ clean_individual_file <- function(individual_file, year) {
 #'
 #' @inheritParams clean_individual_file
 clean_up_gender <- function(individual_file) {
+  cli::cli_alert_info("Clean up gender column function started at {Sys.time()}")
+
   individual_file %>%
     dplyr::mutate(
       gender = dplyr::case_when(
@@ -785,6 +829,8 @@ join_slf_lookup_vars <- function(individual_file,
                                    col_select = c("gpprac", "cluster", "hbpraccode")
                                  ),
                                  hbrescode_var = "hb2018") {
+  cli::cli_alert_info("Join slf lookup variables function started at {Sys.time()}")
+
   individual_file <- individual_file %>%
     dplyr::left_join(
       slf_postcode_lookup,

R/fill_geographies.R

@@ -16,6 +16,8 @@ fill_geographies <- function(
       get_slf_gpprac_path(),
       col_select = c("gpprac", "cluster", "hbpraccode")
     )) {
+  cli::cli_alert_info("Fill geographies function started at {Sys.time()}")
+
   check_variables_exist(data, c(
     "chi",
     "postcode",

R/join_deaths_data.R

@@ -10,6 +10,8 @@ join_deaths_data <- function(
     data,
     year,
     slf_deaths_lookup = read_file(get_slf_deaths_lookup_path(year)) %>% slfhelper::get_chi()) {
+  cli::cli_alert_info("Join deaths data function started at {Sys.time()}")
+
   return(
     data %>%
       dplyr::left_join(

R/join_sparra_hhg.R

@@ -5,6 +5,8 @@
 #' @return The data including the SPARRA and HHG variables matched
 #' on to the episode file.
 join_sparra_hhg <- function(data, year) {
+  cli::cli_alert_info("Join SPARRA and HHG function started at {Sys.time()}")
+
   if (check_year_valid(year, "sparra")) {
     data <- dplyr::left_join(
       data,

R/link_delayed_discharge_eps.R

@@ -12,6 +12,8 @@ link_delayed_discharge_eps <- function(
     episode_file,
     year,
     dd_data = read_file(get_source_extract_path(year, "dd")) %>% slfhelper::get_chi()) {
+  cli::cli_alert_info("Link delayed discharge to episode file function started at {Sys.time()}")
+
   episode_file <- episode_file %>%
     dplyr::mutate(
       # remember to revoke the cij_end_date with dummy_cij_end

R/match_on_ltcs.R

@@ -12,6 +12,8 @@ match_on_ltcs <- function(
     data,
     year,
     ltc_data = read_file(get_ltcs_path(year)) %>% slfhelper::get_chi()) {
+  cli::cli_alert_info("Match on LTCs function started at {Sys.time()}")
+
   # Match on LTC lookup
   matched <- dplyr::left_join(
     data,