How to Create New Specifications

If data cannot be represented in a domain dataset defined in this guide, then a custom domain may be used. A custom domain may only be created when data are different in nature and are not in scope for domains described in this guide. Custom domains will be used to represent logically related observations based on the scientific subject matter of the data and will not be created based on:

• The timing of collected observations (e.g., all vital signs measurements will be represented in domain Vital Signs (VS) irrespective of when measurements occurred).
• How collected data are used (e.g., all vital signs measurements will be represented in domain VS and a custom "safety" domain will not be created for measurements used to assess safety).
• Data collection methodology (e.g., all vital signs measurements will be represented in domain VS irrespective of whether measurements were recorded using separate CRFs).

Prior to creating a custom domain, confirm that none of the existing published domains will fit the need. Once confirmed, drafting a specification upfront, using conventions in Section 2.8.1, How to Read Domain Specifications, is recommended to ensure expectations for the custom domain are clear. Custom domains and corresponding specifications must be created based on the three general observation classes, Interventions, Events, and Findings, described in the SDTM. In most cases, the choice of observation class appropriate to a specific collection of data can be easily determined according to descriptions of these classes in the SDTM. The majority of data, which typically consists of measurements or responses to questions, usually at specific visits or time points, will fit the Findings general observation class. 

The overall process for creating a custom domain is as follows:

1. Establish a common topic or topics for the data. The common topic or topics will reflect a collection of logically related observations based on the scientific subject matter of the data.
   1. If more than one topic is identified, then more than one domain may be needed.
      1. In such cases, consider whether topics are hierarchical in nature where data for one topic must be observed before data for a second topic can be observed. If a hierarchical relationship between topics exists, then paired domains will be created (e.g., Pharmacokinetics Concentrations (PC) and Pharmakinetics Parameters (PP) is an established domain pair). Relationships between records in paired domains may then be represented in the Related Records (RELREC) dataset as appropriate.
2. Categorize data within the domain using Grouping Qualifier variables (e.g., --CAT, --SCAT) and identify other Qualifiers applicable to the data (e.g., --METHOD, --SPEC) as appropriate.
3. Look for a domain within this guide to serve as a prototype. If no domain seems appropriate, choose the general observation class in the SDTM (Interventions, Events, or Findings) that best fits the data given the topic of the observations.
4. Select variables for the domain from the SDTM. Selection of variables must align with SDTM Usage Restrictions. As illustrated in the following figure, the general approach for selecting variables for a custom domain is to:
   1. Include applicable Identifier variables. Identifier variables STUDYID, USUBJID, DOMAIN, and --SEQ are required in all domains based on the general observation classes. Additional Identifiers may be added as needed.
   2. Include the Topic variable from the SDTM general observation class (e.g., --TESTCD for Findings). 
   3. Include the relevant Qualifier variables from the identified SDTM general observation class.
   4. Include the applicable SDTM Timing variables. In general, the domain must have at least one timing variable.
   5. Determine the two-character domain code.
      
      1. To eliminate the risk of using a name that CDISC later determines to have a different meaning, domain codes beginning with the letters X, Y, and Z have been reserved for the creation of custom domains. Any letter or number may be used in the second position. The use of codes beginning with X, Y, or Z is optional, and not required for custom domains.
5. In some cases, applicants may need to define new custom domains and may be concerned that CDISC domain codes defined in the future will conflict with those they choose to use.  
   1. Apply the domain code to the appropriate variables in the domain by replacing all variable prefixes (shown in the SDTM as “--“) with the domain code.
   2. Set the order of variables consistent with the order defined in the SDTM for the general observation class.
   3. Adjust the labels of the variables only as appropriate to convey their meaning in the context of the data in the newly created domain. Use title case for all labels.

   4. Ensure that appropriate standard variables are being properly applied by comparing their use in the custom domain to their use in related TIG domains.

   5. Place any non-SDTM variables in a Supplemental Qualifier dataset.