Skip to contents

One row per variable of the CDISC Clinical Data Acquisition Standards Harmonization (CDASH) model, at every version: 1.0, 1.1, 1.2, and 1.3. Built from CDISC Library CSV exports not redistributed with this package (see ig_sources); only variable metadata and the collection wording (question text, prompt) is carried, not the model's definitions, mapping instructions, or implementation notes. Retrieve it with get_cdash("CDASH").

Usage

cdash_model

Format

A data frame with columns:

standard

Always "CDASH".

version

Model version, e.g. "1.3".

class

Observation or special-purpose class, e.g. "Findings", "Identifiers".

domain

The domain a domain-specific variable belongs to (e.g. "AE"), or NA for class-level variables.

order

Position within the class and domain, as published.

variable

CDASH variable name, e.g. "--TERM".

label

Variable label. At most 40 characters except for 6 labels the model publishes longer (--TESTCD in 1.0 and 1.1; --ENDATF in all four versions).

domain_specific

TRUE where the model flags the variable as domain specific; NA where it does not say.

question_text

The CRF question text the model suggests.

prompt

The CRF field prompt the model suggests.

type

"Char" or "Num".

sdtm_target

The SDTM variable or variables the collected value maps to.

codelist_code

The CDISC CT codelist C-code the variable uses; NA when none.

Source

CDISC Library CSV exports, downloaded under CDISC's terms and not redistributed; see data-raw/README.md.

Details

A (standard, version, class, domain, variable) is unique. The model defines most variables once per class with a -- prefix (domain is NA) and some per domain.