All datasets will adhere to the record structure defined for the domain, regardless of whether multiple values are collected for a variable. Guidance in this section describes conventions for maintaining the record structure defined for a domain or dataset when multiple values are collected. When concepts are related to either Nonclinical or Product Impact on Individual Health use cases but not both, this is denoted in the Implementation column.  


NumRecord and Variable PopulationImplementation
1Multiple values for an Intervention or Event topic variable

Product Impact on Individual Health only:

If multiple values are reported for an Intervention or Event general-observation class topic variable (e.g., --TRT in Intervention,  --TERM in Events), then the applicant will split the values into multiple records or otherwise resolve the multiplicity per the data management standard operating procedures (e.g., query the data for clarification).

  • The Disposition (DS) dataset is an exception to the general rule of splitting multiple topic values into separate records. For DS, 1 record for each disposition or protocol milestone. For cases of multiple reasons for discontinuation, refer to the domain specifications for DS.
2Multiple values for a Findings result variable

If multiple result values (represented in --ORRES) are collected for a test, then multiple records will be represented for the test (represented in --TESTCD).

3

Multiple values for a variable with role:

  • Grouping qualifier
  • Synonym qualifier
  • Record qualifier
  • Variable qualifier

Product Impact on Individual Health only:

For multiple values (e.g., due to a "Check all that apply" instruction on a CRF):

  • The variable will be populated with "MULTIPLE".
  • The SUPP-- dataset will represent individual responses.
    • The values for QNAM will be sequential variable names, formed by appending a 1-digit integer, beginning with 1, to the original domain variable name. In cases where the standard domain variable name is already 8 characters in length, applicants will replace the last character with a digit when creating values for QNAM.
    • The value for QLABEL should be the original domain variable label for all QNAM values.
4Multiple values for a parameter

If multiple values (--VAL) are reported for a parameter in a Trial Design dataset (e.g., TS), multiple records should be submitted for the --PARMCD, unless the CDISC notes for the parameter indicate otherwise (e.g., TXPARMCD=LIGHT) 

5

Single concept represented by multiple values

Nonclinical only:

  • A single value may represent a combination of multiple contributing items, delimited by slashes ("/"). In such cases, the combination is the collected value, the value analyzed, and so on—not the individual contributing items composing the value.
  • When controlled terminology applies to the variable, the combination of concepts may already exist as a controlled terminology term. If this is not the case, then a new term will be constructed by combining existing controlled terms, when possible. 
  • For example, DEGENERATION/REGENERATION is a microscopic finding that is a single concept represented by multiple values (which is different from having 2 separately collected findings of degeneration and regeneration, with 2 separate records respectively).
6

Multiple concepts represented by multiple values

Nonclinical only:

  • A variable value may represent multiple individual values. In such cases, each contributing value is a separately collected or analyzed item.
  • When controlled terminology applies to the variable, controlled terms for each value will be represented delimited by semicolons.
  • For example, multiple distinct values may be collected for directionality (--DIR), may be collected for a result. Values may be represented using controlled terminology as "VENTRAL;SURFACE".

  • No labels