Skip to content

Create a population node

Source: R/recipe.R
omop_population.Rd

A recipe defines one or more populations and every output targets one (via omop_output(..., population_id=)). Each recipe starts with an implicit "base" population representing all persons; additional populations are passed to omop_recipe via its populations argument.

Usage

omop_population(
  id = "base",
  label = "Base Population",
  parent_id = NULL,
  filters = list(),
  cohort_definition_id = NULL,
  episode_policy = NULL,
  union = NULL,
  intersect = NULL,
  setdiff = NULL,
  index_event = NULL
)

Arguments

id

Character; population ID (must be unique within the recipe).

label

Character; human-readable label.

parent_id

Character or NULL; parent population ID (NULL for root). Informational provenance only; set-op membership is the executable dependency.

filters

List of omop_filter or omop_filter_group objects (criteria populations only).

cohort_definition_id

Integer or NULL; base cohort definition ID (if the population is defined by a pre-existing cohort).

episode_policy

Character or NULL; explicit semantics for index-dependent filters when the index cohort can contain multiple episodes per person. One of "any_episode", "all_episodes", "first_episode", or "last_episode". Without a policy the server rejects index-dependent filtering of recurrent cohorts.

union, intersect, setdiff

Character vector of two or more population IDs to derive this population from by the named set operation on the person key. Exactly one may be supplied, and only for a set-op population (mutually exclusive with filters / cohort_definition_id).

index_event

An omop_index_event or NULL. When present, population filters are evaluated for each retained event episode and the event's start/end dates are preserved.

Value

An omop_population object. A set-op population carries a $setop = list(op, members) field; a criteria population carries $filters (and optionally $cohort_definition_id and $episode_policy).

Details

A population is one of two kinds, which are mutually exclusive:

  • criteria-defined — a person-level inclusion tree given in filters (any mix of omop_filter / omop_filter_group at the "population" level, e.g. sex + has_concept + has_measurement). It compiles to the same cohort filter tree the server builds the base cohort from.

  • set-op derived — built from other populations by a set operation on the person key. Supply exactly one of union, intersect, or setdiff as a character vector of two or more population IDs; the server folds the named members with the matching algebra (ds.omop.cohort.combine's .cohortCombine). setdiff keeps persons in the first member and not in the rest.

See also

omop_recipe, omop_filter, ds.omop.cohort.combine

Examples

if (FALSE) { # \dontrun{
# Criteria population.
females <- omop_population(id = "females", label = "Female patients",
                           filters = list(omop_filter_sex("F")))

# Set-op population: persons in EITHER of two criteria subgroups.
either <- omop_population(id = "either", label = "diabetic or hypertensive",
                          union = c("diabetic", "hypertensive"))
} # }