Create a variable specification

Describes a single variable to extract from the CDM. Variables reference a source table and column, and may include concept-level filtering and output formatting options. Variables are the atomic building blocks passed to omop_recipe via its variables argument (or grouped with omop_variable_block and passed via blocks).

Usage

omop_variable(
  name = NULL,
  table,
  column = NULL,
  concept_id = NULL,
  concept_name = NULL,
  type = c("auto", "numeric", "categorical", "date", "boolean", "integer", "character"),
  format = c("raw", "binary", "count", "first_value", "last_value", "mean", "min", "max",
    "time_since", "binned", "age", "sex_mf", "obs_duration", "drug_duration", "sum",
    "n_distinct", "sd", "cv", "slope", "abnormal_high", "abnormal_low", "gap_max",
    "gap_mean", "duration_sum", "prior_obs", "followup", "demo_missingness", "charlson",
    "chads2", "chadsvasc", "dcsi", "hfrs"),
  value_source = NULL,
  time_window = NULL,
  suffix_mode = c("index", "range", "label"),
  filters = list(),
  visit_filter = NULL,
  concept_col = NULL,
  expand = FALSE,
  reference_date = NULL,
  unit = NULL
)

Arguments

name: Character; output column name (auto-generated from concept_name, concept_id, or column if NULL).
table: Character; source OMOP CDM table (e.g. "condition_occurrence").
column: Character or NULL; source column to extract.
concept_id: Integer or NULL; concept ID filter (for concept columns).
concept_name: Character or NULL; human-readable concept name.
type: Character; variable type hint. One of "auto", "numeric", "categorical", "date", "boolean", "integer", "character".
format: Character; output format. One of "raw", "binary", "count", "first_value", "last_value", "mean", "min", "max", and "time_since", plus the documented derived and longitudinal summary formats. "time_since" requires a fixed reference_date; cohort-index recency remains episode-specific and is rejected. Calendar binning is configured with omop.date_handling, not as a variable format.
value_source: Character or NULL; column to extract value from (e.g. "value_as_number" for measurements).
time_window: Named list with start/end offsets relative to index date, or NULL for no window constraint.
suffix_mode: Character; how to name multi-column expansions ("index", "range", or "label").
filters: List of omop_filter objects to apply to this variable.
visit_filter: Named list list(concept_ids = ...) or NULL; restrict this variable's events to visits of those visit_concept_id values (via the visit_occurrence_id link).
concept_col: Character or NULL; override the concept column the concept_id/concept set scopes (default: the table's domain concept), e.g. "unit_concept_id" to extract a single unit for harmonization.
expand: Logical; if TRUE, expand the concept to include vocabulary descendants server-side (default FALSE).
reference_date: Character/Date or NULL; fixed ISO reference date required when format = "time_since".
unit: Character or NULL; "day" (default for time_since) or "month". Months are complete calendar months, not fixed 30-day intervals.

Value

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

Examples

if (FALSE) { # \dontrun{
recipe <- omop_recipe(
  variables = omop_variable(
    table = "condition_occurrence",
    concept_id = 201820,
    concept_name = "Type 2 diabetes",
    format = "binary"
  ),
  outputs = omop_output(type = "wide")
)
} # }

Usage

Arguments

Value

See also

Examples