Special Values: "MULTIPLE" and "OTHER"

The SDTMIG directs that, under certain circumstances, variables can be populated with the values "MULTIPLE" or "OTHER". Neither of these values is what might be called a "proper" value for the variable (i.e., a value that provides the the kind of information intended to be represented in the variable). Instead, these special values indicate that there are either multiple proper values or that the proper value collected was not in the list of values presented on the data collection form.

The "MULTIPLE" convention is introduced in SDTMIG in the section on Multiple Values for a Non-Result Qualifier Value (Section in v3.3, Section in earlier versions). All the examples in that section are for the AE domain, although the convention is certainly not limited to that domain. The examples show two values for location of an AE and, for a study with two study treatments, a relationship to study treatment for each of the two study treatments, and an action taken with study treatment for each of the study treatments. When a variable is populated with "MULTIPLE", the collected values are represented in supplemental qualifiers. The names of those supplemental qualifiers (QNAM values) may be differentiated by numbers when they are simply multiple answers to the same question, as in the AE location example. If the "MULTIPLE" values are answers to different questions, as in the cases of relationships to and actions taken with particular study treatments, the supplemental qualifiers will be named to indicate the different questions, as in the relationship to study treatment example, where the QNAM values include a fragment that is an abbreviation for the drug name.

The SDTMIG section "Specify" Values for Non-Result Qualifier Variables (Section in v3.3, Section in earlier versions) presents three options for representing values collected in an "Other, specify" field. One of those options is to populate the variable with "OTHER" and to present the collected value as a supplemental qualifier. An assumption in the Demographics domain (Assumption 6 in v3.1.2 through v3.3) specifies that this option is the one that should be used for data collected in an "Other, specify" for race.

These conventions for populating a variable with "MULTIPLE" or "OTHER" present a problem when validating data in variables with Controlled Terminology, since most Controlled Terminology includes only "proper" values. (There are, for example, a few QRS instrument response terminologies that do include the term "OTHER".) Validation software generally checks content of a variable against Controlled Terminology given for that variable in the Define-XML document, and may see values of "MULTIPLE" or "OTHER" as errors, since they are not values in the codelist.

CDISC is working out mechanisms for representing circumstances under which the value of "MULTIPLE" is allowed and circumstances under which the value "OTHER" is allowed. Some rules are easy to formulate, for example that neither value is allowed in a topic variable, identifier, timing variable, rule variable or result qualifier. Also, that "MULTIPLE" may not be used when values in a codelist are mutually exclusive (e.g., values of both "Y" and "N" are not allowed). However, complete rules for when each of the special values "MULTIPLE" and "OTHER" may or may not be used will need careful development.

What about SEND?

Datasets based on the SENDIG do not use either "MULTIPLE" or "OTHER" as special values.

SENDIG-based datasets do not use the MULTIPLE convention. The SENDIG, in the Section on Multiple Concepts Represented by Multiple Values (Section in both SENDIG v3.0 and v3.1), instead directs that the multiple values be represented in a semicolon-delimited list.

The use of "OTHER" is not needed for nonclinical studies, since those studies do not use case report forms, so "Other, specify" fields do not arise.