The Study Data Tabulation Model, or SDTM, is a CDISC metadata model for the uniform organization of data from clinical trials. It maps the observations collected or imported for a study into standardized domains and variables without altering their original meaning. Thus, SDTM establishes the tabular baseline for regulatory data review.
This standardized organization facilitates the consistent review of recurring data patterns across different studies.
Role in the Clinical Trial Data Flow
SDTM sits between data collection at the investigational site and regulatory submission. Data from eCRFs, laboratories, electronic patient diaries, or external vendors are first cleaned and then mapped to the specified SDTM domains. For example, the adverse events domain consolidates event data, while demographic, exposure, and laboratory information each receive their own distinct structures.
The standard describes the data as tabulations, not as final statistical analyses. Its domains represent collected or received information in a format that enables consistent review and reporting. For FDA regulatory submissions, the Study Data Technical Conformance Guide classifies SDTM under clinical tabulation datasets; nonclinical toxicology tabulations are handled separately via SEND.
Domains, Observation Classes, and Metadata
The conceptual modeling begins with the question of what type an observation is. For this purpose, SDTM uses the observation classes Interventions, Events, Findings, and Findings About Events or Interventions. This classification dictates the appropriate domains, variable roles, and relationships between datasets. Consequently, a laboratory test is modeled differently than an administered study medication or a concomitant disease.
The model alone is not sufficient for operational implementation. The associated Implementation Guide specifies which domains and variables are to be used for specific content. Additionally, Define-XML explains the metadata of a specific submission, such as variable definitions, permissible values, and origin. This allows a reviewer to understand what a variable means rather than having to rely on project-specific nomenclature.
SDTM domains follow a common vocabulary so that similar observations are not renamed from study to study. In addition to the content, the modeling considers the relationship to the subject, time points, and other observations. Precise domain selection prevents clinically distinct concepts from being mixed in the same table, which would leave the purpose of individual rows unclear to the regulatory authority.
For submitted data, a distinction is often made between specialized and general domains. This distinction is not merely about file naming: it influences which variables are expected and how the link between data collection, standardization, and subsequent review is established. CDISC also links domains to controlled terminology to ensure uniform interpretation of permissible values.
Distinction from ADaM
SDTM maps the observations available for the study into a submission-ready tabular structure. The Analysis Data Model (ADaM) builds on this tabular data and creates datasets for defined statistical analyses. Analysis populations, derived visit windows, or imputed values therefore belong in ADaM, not in the SDTM tabulation.
This distinction is crucial for traceability: SDTM preserves the nature of the standardized source data, while ADaM explicitly defines the analytical purpose of a variable. The overarching term CDISC standards encompasses both levels but does not replace the respective modeling decisions. SDTM should also not be equated with an eCRF; form-based data capture follows upstream requirements and can be designed according to CDASH.
Relevance for clinical trials
In SDTM projects, data management, medical coding, external data vendors, and biostatistics must define early on which raw data will be mapped to which domain and how origin, units, and controlled terminology will be represented. Late changes to eCRF fields, laboratory files, or vendor specifications otherwise result in mapping adjustments that cascade all the way to Define-XML and the Reviewer’s Guides. Particularly critical for regulatory review are comprehensible relationships between datasets, consistent identifiers, and the separation of collected data from analysis derivations.
Full-service CROs like Mediconomics support SDTM implementations with data management plans, SDTM mapping specifications, reconciliation of laboratory and external data sources, programming of tabulation datasets, and metadata review prior to submission. This collaboration integrates CRF annotation, coding of events and concomitant medications, data cleaning, and the creation of study-specific data review documents.
Frequently Asked Questions (FAQ)
Does SDTM only contain data from the eCRF?
No. SDTM can also structure data from central laboratories, imaging, wearables, or other qualified sources, provided they are incorporated into the study and their meaning is preserved in the dataset.
May SDTM contain derived or imputed values?
SDTM is used to present standardized tabulation data. The FDA guidance separates this level from ADaM and specifies that SDTM and SEND datasets should not contain imputed data.
Is SDTM a requirement for every European study?
The specific requirement depends on the submission pathway and the target regulatory authority. SDTM is a CDISC standard for structured study data; for FDA submissions, the acceptance of supported standards is governed by the Data Standards Catalog and the Technical Conformance Guide.
Regulatory References
- CDISC Study Data Tabulation Model (SDTM) – defines the metadata model and the classification of standardized study domains.
- CDISC SDTM Implementation Guide – specifies the implementation of domains and variables in tabulation datasets.
- FDA Study Data Technical Conformance Guide – describes SDTM as the format for clinical tabulation datasets in FDA submissions.
- FDA Data Standards Catalog – lists the data standards supported by the FDA and their scope of application.
