Build End of Life / Mortality Cohort — build_end_of_life

Constructs a standardized End of Life / Mortality cohort in DuckDB SQL. Supports four mortality ascertainment paradigms:

"in_hospital": Death during index stay (expired/died in hospital).
"post_discharge": Death within (t_discharge + gap_days, t_discharge + mortality_window_days].
"fixed_window": SARD-style EOL prediction within (t_index + gap_days, t_index + mortality_window_days].
"composite_readmit_or_death": Clinical composite of acute readmission OR all-cause death post-discharge.

Resolves dual OMOP death sources:

death table: death_date linked to person_id.
visit_occurrence table: discharged_to_concept_id IN (4216643, 4155309) (Expired / Hospice).

This is a thin wrapper over the shared study-cohort engine (see define_study_cohort(), which generalises it); its legacy conventions are pinned here and covered by golden-file tests.

Usage

build_end_of_life_cohort(
  con,
  cohort_id = 1,
  outcome_cohort_id = 2,
  cohort_name = "End of Life Cohort",
  cohort_description = NULL,
  mortality_type = "post_discharge",
  target_visit_concept_ids = 9201,
  outcome_visit_concept_ids = 9201,
  mortality_window_days = 30,
  gap_days = 0,
  min_age = 18,
  washin_days = 365,
  require_verified_followup = TRUE,
  index_selection_rule = "random",
  random_state = 42,
  table_name = NULL,
  schema = "main",
  overwrite = TRUE
)

build_mortality_cohort(
  con,
  cohort_id = 1,
  outcome_cohort_id = 2,
  cohort_name = "End of Life Cohort",
  cohort_description = NULL,
  mortality_type = "post_discharge",
  target_visit_concept_ids = 9201,
  outcome_visit_concept_ids = 9201,
  mortality_window_days = 30,
  gap_days = 0,
  min_age = 18,
  washin_days = 365,
  require_verified_followup = TRUE,
  index_selection_rule = "random",
  random_state = 42,
  table_name = NULL,
  schema = "main",
  overwrite = TRUE
)

Arguments

con: Active DuckDB DBI connection.
cohort_id: Target cohort definition ID in OMOP cohort table (default 1).
outcome_cohort_id: Optional outcome cohort definition ID (default 2). Set NULL to skip outcome cohort table population.
cohort_name: Display name for cohort_definition table.
cohort_description: Detailed description for cohort_definition table.
mortality_type: Mode of mortality ascertainment: "in_hospital", "post_discharge", "fixed_window", or "composite_readmit_or_death".
target_visit_concept_ids: Visit concept IDs defining qualifying index encounters (default 9201: Inpatient). Set NULL for any visit.
outcome_visit_concept_ids: Visit concept IDs defining readmissions for composite mode (default 9201).
mortality_window_days: Number of days for mortality outcome window (default 30).
gap_days: Days between index date / discharge and start of outcome window (default 0).
min_age: Minimum patient age at index encounter (default 18).
washin_days: Baseline continuous observation lookback requirement before index stay (default 365).
require_verified_followup: Logical. If TRUE, excludes lost-to-follow-up patients without confirmed survival follow-up (default TRUE).
index_selection_rule: "random" (reproducible seed), "first", or "last".
random_state: Seed for reproducible random sampling (default 42).
table_name: Optional custom table name in DuckDB to materialize cohort records with metadata.
schema: CDM schema containing tables (default "main").
overwrite: Logical. If TRUE, cleans existing cohort and cohort_definition entries for cohort_id and outcome_cohort_id (default TRUE).

Value

A data.frame containing qualifying cohort records with metadata.