Create a filter specification

Filters restrict the population or events included in the extraction. There are two executable levels: "population" (person-level inclusion criteria) and "row" (event-level restrictions). Post-extraction transformations belong in the output specification rather than in a filter; the retired "output" filter level is rejected. Filters are passed to omop_recipe via its filters argument and can be nested into groups with omop_filter_group.

Usage

omop_filter(
  type = c("sex", "age_range", "age_group", "cohort", "has_concept", "not_has_concept",
    "concept_count", "prior_observation", "followup", "visit_count", "has_measurement",
    "missing_measurement", "date_range", "concept_set", "value_bin", "value_concept",
    "custom"),
  level = c("population", "row"),
  params = list(),
  label = NULL
)

omop_filter_sex(value)

omop_filter_age(min = 0, max = 150, year = NULL, reference_date = NULL)

omop_filter_age_group(groups, year = NULL, reference_date = NULL)

omop_filter_cohort(cohort_definition_id)

omop_filter_has_concept(
  concept_id,
  table,
  concept_name = NULL,
  window = NULL,
  min_count = 1L,
  reference_date = NULL
)

omop_filter_date_range(start = NULL, end = NULL, date_column = NULL)

omop_filter_value(
  column = "value_as_number",
  threshold,
  direction = c("above", "below"),
  safe_bins = NULL
)

omop_filter_value_concept(
  concept_ids,
  column = "value_as_concept_id",
  concept_name = NULL
)

omop_filter_not_has_concept(
  concept_id,
  table,
  concept_name = NULL,
  window = NULL,
  reference_date = NULL
)

omop_filter_concept_count(
  concept_id,
  table,
  min_count = 2L,
  concept_name = NULL,
  window = NULL,
  reference_date = NULL
)

omop_filter_prior_observation(min_days = 365L, reference_date = NULL)

omop_filter_followup(min_days = 30L, reference_date = NULL)

omop_filter_visit_count(
  min_count = 1L,
  visit_concept_id = NULL,
  window = NULL,
  reference_date = NULL
)

omop_filter_has_measurement(
  concept_id,
  min_value = NULL,
  max_value = NULL,
  safe_bins = NULL,
  window = NULL,
  reference_date = NULL
)

omop_filter_missing_measurement(
  concept_id,
  window = NULL,
  reference_date = NULL
)

Arguments

type: Character; executable filter type. Population filters are "sex", "age_range", "age_group", "cohort", "has_concept", "not_has_concept", "concept_count", "prior_observation", "followup", "visit_count", "has_measurement", and "missing_measurement". Row filters are "date_range", "concept_set", "value_bin", "value_concept", and the fail-closed typed "custom" predicate.
level: Character; "population" or "row". When omitted, the unique executable level for type is selected. Output-level filters are not part of the executable Recipe contract.
params: Named list; filter-specific parameters (varies by type).
label: Character or NULL; human-readable description (auto-generated from type and params if NULL).
value: Character; sex value. Accepts "F", "f", "female", "Female", "FEMALE", "M", "m", "male", "Male", "MALE" — normalized internally to "F" or "M".
min: Numeric; minimum age (inclusive)
max: Numeric; maximum age (inclusive)
year: Integer or NULL; explicit calendar-year anchor (shorthand for July 1 of that year). A cohort index supplies the anchor when omitted.
reference_date: Date/string or NULL; fixed anchor for window when the population has no cohort index. An index is used when this is omitted.
groups: Character vector; age group labels (e.g. c("18-24", "25-34"))
cohort_definition_id: Integer; existing OMOP cohort_definition_id to require for membership.
concept_id: Integer scalar or vector; measurement concept ID(s) to check absence of (a vector requires all of them to be absent)
table: Character; which OMOP table to check
concept_name: Character or NULL; human-readable name
window: Named list with start/end index-relative day offsets, or NULL; restricts absence to that window (e.g. "no HbA1c in the prior year")
min_count: Integer; minimum number of visits
start: Character; inclusive start date in ISO YYYY-MM-DD form.
end: Character; inclusive end date in ISO YYYY-MM-DD form and not before start. The server applies the authoritative minimum disclosure-safe width configured by the data controller.
date_column: Character or NULL; explicit OMOP date column. When NULL, recipe_to_plan infers the standard date column from the output's OMOP table.
column: Character; the value-concept column (default "value_as_concept_id")
threshold: Numeric; threshold value
direction: Character; "above" or "below"
safe_bins: Server-issued result for the same measurement concept from ds.omop.safe.cutpoints(). Required whenever a numeric range is supplied; client-authored thresholds are not executable.
concept_ids: Integer scalar or vector; the value concept(s) to keep (a record matches if its value concept is any of them)
min_days: Integer; minimum days of followup
visit_concept_id: Integer scalar or vector, or NULL; visit type filter (a vector counts visits of any of the given types)
min_value: Numeric or NULL; minimum value
max_value: Numeric or NULL; maximum value

Value

An omop_filter object (a named list with class "omop_filter").

Details

Convenience constructors are provided for common filter types: omop_filter_sex, omop_filter_age, omop_filter_age_group, omop_filter_has_concept, omop_filter_date_range, omop_filter_value.

Examples

if (FALSE) { # \dontrun{
f <- omop_filter(type = "sex", level = "population",
                 params = list(value = "F"))
} # }

Usage

Arguments

Value

Details

See also

Examples