Understanding and Utilizing the RELREC Domain in SDTM
Introduction
The Study Data Tabulation Model (SDTM), adopted by the Clinical Data Interchange Standards Consortium (CDISC), is a standard model for organizing and formatting clinical trial data. This standardization promotes data consistency across trials and facilitates efficient data sharing and analysis. One critical component of the SDTM structure is the RELREC domain.
RELREC, short for Relationships Record, is a Special Purpose Relationship domain, which serves as a pivotal part of a study’s dataset in its ability to describe relationships between records for a subject or across different domains. Relationships can be explicitly stated, as noted in references or check boxes on the Case Report Form (CRF), or inexplicitly via the design of the CRF. For instance, the Hypoglycemia form’s collected data (CAH) could be mapped to various domains, such as CF, LB, etc.
This article provides a comprehensive exploration of the RELREC domain’s various aspects, focusing on its integral variables, and illustrates its utilization through some practical examples.
RELREC Variables
The RELREC domain incorporates seven key variables, namely, STUDYID, RDOMAIN, USUBJID, IDVAR, IDVARVAL, RELTYPE, and RELID. The meanings of these variables, as reflected in the SDTM guide, are as follows:
- STUDYID: This represents the Study Identifier. It’s the unique alphanumeric string that identifies the clinical trial study.
- RDOMAIN: This represents the two-letter domain code of the record in the relationship, like AE, CM, etc.
- USUBJID: This is the Unique Subject Identifier, which is a unique identifier for each subject within the study.
- IDVAR: This is the Unique Record Identifier, such as AESEQ. It points to the variable in the RDOMAIN that provides the unique identifier for the record.
- IDVARVAL: This represents the value of the corresponding IDVAR. It holds the unique record identifier value.
- RELTYPE: This denotes the Relationship Type. Its values should be either ONE or MANY. However, values are required only when identifying a relationship between datasets.
- RELID: This is the Relationship Identifier. It’s a unique alphanumeric string that defines a specific relationship.
The first six variables — STUDYID, RDOMAIN, USUBJID, IDVAR, IDVARVAL, and RELID — serve as key variables for the RELREC domain. They are used in creating a unique key for each record in the relationship domain.
Constructing a Relationship in RELREC
The process of constructing a relationship in RELREC entails the addition of a record to RELREC for each record intended to be related. A unique character identifier value for that relationship is then assigned. This mechanism allows for the versatile handling of data relationships, capturing the complex interconnectedness of various data elements within clinical research data.
Each record in the RELREC domain contains keys that identify a record or group of records and the relationship identifier stored in the RELID variable. The RELID value must be identical for all related records within each USUBJID.
The relationship is specified using the key variables as described earlier. For instance, to relate single records, a unique-record-identifier variable such as AESEQ in IDVAR can be employed. To relate groups of records, a grouping variable like LBGRPID in IDVAR can be utilized. Consequently, IDVARVAL will contain the value of the variable as indicated in IDVAR.
Practical Examples of Utilizing RELREC
Example 1: Relating Adverse Events and Concomitant Medications
Consider a clinical trial studying a novel drug, with subjects experiencing various adverse events (AEs) during the trial. In some instances, these AEs result in subjects taking concomitant medications. In such a scenario, the relationship between the AEs and the concomitant medications (CM) can be documented using the RELREC domain.
In this case, the RDOMAIN value will be AE for the AE records and CM for the concomitant medication records. The IDVAR would contain AESEQ for the AE records and CMSEQ for the CM records. IDVARVAL would hold the sequence number for the particular AE or CM. RELTYPE will specify whether there is a one-to-one (ONE) or one-to-many (MANY) relationship. The RELID value would be identical for all records related to a specific adverse event.
This way, a relationship can be established, illustrating a logical connection between an adverse event and a concomitant medication for a subject in a clinical trial.
Example 2: Relating Records Across Multiple Domains
Another scenario might involve relating records across several domains, each contributing to the overall understanding of a clinical event. For example, in a diabetes trial, there may be a need to link records from the Lab data (LB), the Concomitant Medications data (CM), and the Questionnaire data (QS) to track a hypoglycemic event.
In this situation, the RDOMAIN value will correspond to the domain of each related record (LB, CM, or QS). The IDVAR will hold the unique record identifier for each domain, such as LBSEQ, CMSEQ, or QSSEQ. The IDVARVAL will contain the unique sequence number for each domain record. Again, RELTYPE will indicate the type of relationship (ONE or MANY), and the same RELID will be assigned to all related records.
This establishes a complex relationship across multiple domains, allowing for a comprehensive view of a clinical event like a hypoglycemic event.
Conclusion
The RELREC domain plays a critical role in managing and interpreting complex clinical data in SDTM. Its ability to build relationships between single or multiple records and across different domains is invaluable in drawing meaningful insights from clinical trial data. A sound understanding of the RELREC domain’s function, its variables, and how to construct relationships within it is crucial for those dealing with clinical data management and analysis. The examples outlined in this article underscore the RELREC domain’s utility and its integral role in facilitating data-driven analyses in clinical trials.
