ODM XML import validation in EDC software is best understood as a staged process rather than a single pass/fail event. The SAS Clinical Standards Toolkit documentation separates schema validation, import, validation of the SAS representation, and domain extraction into distinct steps [1][3]. Viedoc, by contrast, describes import as an upload-and-mapping workflow with explicit exclusions for medical coding, queries, review status, and clinical data history [2]. The practical question for a clinical data manager is therefore not whether an ODM file is valid in the abstract, but which layer of validation a given tool performs, what it silently ignores, and what remains unresolved after import. The following sections examine those layers using documented examples from SAS, Viedoc, OpenClinica, and Oracle, then propose a hypothetical validation scenario that makes the boundaries visible.
Schema validation is optional in some tools and consequential when skipped
The SAS Clinical Standards Toolkit supports CDISC ODM 1.3.0 and 1.3.1 and can import an ODM XML file into a SAS data set representation of either standard [1]. Its documented workflow begins with validating the ODM XML file against the ODM 1.3.1 XML schema using the odm_xmlvalidate macro [3]. A sample results data set in the 1.5 documentation shows schema validation issues with an invalid ODM XML file, indicating that the macro produces a record of problems rather than simply halting [3]. This output gives a data manager a reviewable artifact before any data is read into SAS.
The same documentation states that running odm_xmlvalidate is not required to import an ODM XML file [3]. The consequence is specific: importing an invalid file can produce an incomplete import because the odm_read macro ignores elements and attributes not defined by the XML schema [3]. This is a documented behavior, not a hypothetical risk. A team that skips schema validation may receive a SAS representation that appears successful while silently omitting non-schema content. The distinction matters because schema validation and import validation answer different questions: the first asks whether the XML conforms to the standard, the second asks whether the resulting SAS representation is internally consistent.
Post-import validation goes beyond XML structure
After import, the SAS toolkit validates the SAS representation of the ODM XML standard, and the documentation states that this goes further than XML schema validation [3]. One documented example is the assessment of foreign key relationships across data sets [3]. This is a relational integrity check that cannot be performed by an XML schema alone, because it depends on how ODM elements were distributed into SAS data sets during the read process. The final documented step is extraction of domain data from the ClinicalData or ReferenceData sections using the odm_extractdomaindata macro, with sample driver programs for one domain or all domains [3].
This sequence implies a division of labor: schema validation catches malformed or non-conforming XML, while representation validation catches problems introduced or revealed by the transformation into SAS structures. The documentation does not state that representation validation catches every possible data quality issue, and it does not describe checks for clinical plausibility or cross-study consistency. A data manager should treat the SAS representation validation as a structural and relational check, not as a substitute for downstream edit checks or medical review. The toolkit's documented scope is the ODM standard representation, not the clinical meaning of the imported values.
A checklist for planning an ODM import validation
Before importing, confirm which ODM version the file uses and which versions the target tool documents as supported. The SAS toolkit documents ODM 1.3.0 and 1.3.1 [1]; ODM v2.0 breaks backward compatibility and may serialize as JSON [5]. Run schema validation first and review the results data set, because skipping it can lead to an incomplete import when the read macro ignores non-schema elements [3]. After import, run representation validation and specifically review foreign key relationships across data sets [3]. Finally, document what the target excludes: Viedoc does not import medical coding, queries, review status, or clinical data history, and imported users require invitations before they can log in [2].
Vendor import workflows expose mapping decisions and explicit exclusions
Viedoc documents ODM import as a five-step workflow beginning with file upload in Viedoc Admin [2]. The system supports import in standard CDISC ODM format or using ODM files containing Viedoc extensions prefixed with v4:, such as files previously exported from Viedoc [2]. Study site mapping is based on site Name, with optional Code, TimeZone, v4:StudySiteSeqNo, and Address/Country fields; SiteRef is used later for assigning subjects to sites [2]. If v4:Address and Country are present they are used during import; otherwise the default is SE (Sweden) [2]. This default is a documented behavior that can affect data ownership and site attribution when source files lack address elements.
Viedoc also documents what the import does not include: medical coding, queries, review status, and clinical data history, with only snapshot supported [2]. Users are imported by full name and email address but are not active immediately; invitations must be sent using the imported email addresses before login access is available [2]. Functions are re-executed after import, which the documentation warns can cause issues when functions use current date or time, potentially changing existing event dates [2]. These are not validation failures in the schema sense, but they are import boundaries that a receiving study team must plan for.
Version boundaries and serialization choices complicate validation assumptions
CDISC describes ODM v2.0 as breaking backward compatibility with previous versions in several ways noted in the specification [5]. ODM v2.0 can be serialized as XML, JSON, or other formats, and an ODM XML schema is currently available [5]. This creates a practical validation boundary: a tool documented against ODM 1.3.0 or 1.3.1, such as the SAS Clinical Standards Toolkit versions cited here, is not documented as validating ODM v2.0 files [1][3]. A data manager receiving an ODM v2.0 file cannot assume that a 1.3.1 schema validator will produce meaningful results.
OpenClinica documents its ODM representation against CDISC ODM XML version 1.3 with OpenClinica extensions as implemented in OpenClinica 3.1 and later releases [6]. Its documentation states that an ODM file must contain only elements and attributes defined in the ODM standard schema or in a valid vendor extension schema, and must satisfy rules about element nesting and attribute value formats [6]. The namespace for version 1.3 is documented as cdisc.org/ns/odm/v1.3 [6]. These constraints are presented as syntactic constraints defined by the ODM standard, not as requirements unique to OpenClinica's implementation.
A hypothetical validation scenario for a migration between EDC systems
Consider a suggested workflow, not a tested procedure: a data manager receives an ODM 1.3.1 XML export from a source EDC and must import it into a target system. The first documented step would be schema validation against the ODM 1.3.1 schema, producing a results data set of any schema issues [3]. If the file contains vendor extensions from the source system, the target's documentation determines whether those extensions are accepted or ignored; Viedoc accepts its own v4: extensions, while OpenClinica permits valid vendor extension schemas [2][6].
After schema validation, the manager would run the import and then validate the resulting representation, checking foreign key relationships across data sets as documented in the SAS toolkit [3]. The final step would be domain extraction from ClinicalData or ReferenceData [3]. Throughout, the manager would track exclusions: if the target is Viedoc, medical coding, queries, review status, and clinical data history would not be imported [2]. If the target re-executes functions using current date or time, event dates may change [2]. This scenario illustrates that validation is a sequence of documented checks with known gaps, not a single guarantee of complete fidelity.
Sources & references
- Importing and Validating a CDISC ODM XML File :: SAS(R) Clinical Standards Toolkit 1.7: Getting Started, Second Edition · support.sas.com
- Viedoc Admin User Guide | Importing data from ODM file | Viedoc Learning · help.viedoc.net
- Importing and Validating a CDISC ODM XML File :: SAS(R) Clinical Standards Toolkit 1.5: Getting Started · support.sas.com
- Importing and Validating a CDISC ODM XML File :: SAS(R) Clinical Standards Toolkit 1.6: Getting Started · support.sas.com
- ODM v2.0 | CDISC · www.cdisc.org
- CDISC ODM XML - OpenClinica Reference Guide · docs.openclinica.com
- Digital Gateway User Guide · docs.oracle.com
- ODM | CDISC · www.cdisc.org