CDISC ODM

Purpose

(Source: CDISC web site http://www.cdisc.org/odm)

The CDISC ODM standard facilitates the archival and interchange of the metadata and data for clinical research. ODM is a vendor-neutral, platform-independent format for the interchange and archival of clinical study data. ODM includes the clinical data and its associated metadata, administrative data, reference data, and audit information. All of the information that needs to be shared during setup, operation, analysis, and submission, as well as for long-term retention as part of an archive, is included in ODM.

Release Dates

CDISC ODM, Version 1.3.0, December 15, 2006
CDISC ODM, Version 1.3.1, February 11, 2010

CDISC ODM 1.3.0 Reference Standard

The SAS Clinical Standards Toolkit 1.5 supports this CDISC ODM 1.3.0 functionality:

reading and representing in SAS a complete odm.xml file (specific limitations are noted below)
building an odm.xml file from a SAS representation of the ODM standard
schema-level validating of an odm.xml file
validating the structure and content of the SAS representation of an odm.xml file
identifying unsupported (unrecognized) ODM elements and attributes by using a sample tool
extracting one or more data sets from the ClinicalData or ReferenceData sections of the ODM XML file

The SAS Clinical Standards Toolkit 1.5 does not support this CDISC ODM 1.3.0 functionality:

reading or writing the DigitalSignatures section of the ODM
vendor or customer extensions of the ODM
processing is limited to a single ODM file (for example, the use of PriorFileOID to reference another file is ignored)
Full file metadata is expected in each file.
Effective support only for ODM FileType=Snapshot. The SAS Clinical Standards Toolkit 1.5 makes no attempt to process multiple transactions per data point; multiple transactions are saved in the SAS ODM representation for subsequent processing

The domain and column metadata that constitute the SAS representation of CDISC ODM 1.3.0 are derived from the global standards library in these formats:

as empty data sets (using the utility macro cst_createTablesForDataStandard)
as table metadata for 66 data sets (reference_tables in the standard metadata folder [see reference_tables (CDISC ODM 1.3.0)])
as column metadata for 315 columns in the 66 data sets (reference_columns in the standard metadata folder

As a general rule, the SAS representation of the CDISC ODM standard is patterned to match the XML element (data set) and attribute (column) structure of odm.xml. For example, consider this XML extract:

<ClinicalData StudyOID="P2006-101" MetadataVersionOID="101.01">
 <SubjectData SubjectKey="1000" TransactionType="Insert">
  <StudyEventData StudyEventOID="101.Screen">
   <FormData FormOID="101.DEMOG">
    <ItemGroupData ItemGroupOID="101.DM">
     <ItemDataString ItemOID="101.USUBJID">101-01-01</ItemDataString>
     <ItemDataString ItemOID="101.SEX">F</ItemDataString>
    </ItemGroupData>
   </FormData>
  </StudyEventData>
 </SubjectData>
</ClinicalData>

This table describes how the XML element and attribute information maps to the SAS representation.

Sample Mapping of odm.xml File to SAS Representation
XML Element or Attribute	SAS Data Set	SAS Column	SAS Column Value
<ClinicalData StudyOID="P2006-101" MetadataVersionOID="101.01">	ClinicalData	StudyOID MetaDataVersionOID	"P2006-101" "101.01"
<SubjectData SubjectKey="1000" TransactionType="Insert">	SubjectData	SubjectKey TransactionType	"1000" "Insert"
<StudyEventData StudyEventOID="101.Screen">	StudyEventData	StudyEventOID	"101.Screen"
<FormData FormOID="101.DEMOG">	FormData	FormOID	"101.DEMOG"
<ItemGroupData ItemGroupOID="101.DM">	ItemGroupData	ItemGroupOID	"101.DM"
<ItemDataString ItemOID="101.USUBJID">101-01-01</ItemDataString>	ItemData	ItemOID ItemDataType Value	"101.USUBJID" "ItemDataString" "101-01-01"
<ItemDataString ItemOID="101.SEX">F</ItemDataString>	ItemData	ItemOID ItemDataType Value	"101.SEX" "ItemDataString" "F"

This table lists the complete set of 66 tables that form the SAS Clinical Standards Toolkit 1.5 SAS representation of the CDISC ODM 1.3.0 standard.

reference_tables (CDISC ODM 1.3.0)
admindata	itemrangecheckvalues
annotation	itemrcformalexpression
annotationflag	itemrole
association	keyset
auditrecord	location
clinicaldata	locationversion
clitemdecodetranslatedtext	measurementunits
codelistitems	metadataversion
codelists	methoddefformalexpression
conditiondefformalexpression	methoddefs
conditiondefs	methoddeftranslatedtext
conditiondeftranslatedtext	mutranslatedtext
enumerateditems	odm
externalcodelists	presentation
formdata	protocoleventrefs
formdefarchlayouts	protocoltranslatedtext
formdefitemgrouprefs	rcerrortranslatedtext
formdefs	referencedata
formdeftranslatedtext	signature
imputationmethods	signaturedef
itemaliases	study
itemdata	studyeventdata
itemdefs	studyeventdefs
itemdeftranslatedtext	studyeventdeftranslatedtext
itemgroupaliases	studyeventformrefs
itemgroupdata	subjectdata
itemgroupdefitemrefs	user
itemgroupdefs	useraddress
itemgroupdeftranslatedtext	useraddressstreetname
itemmurefs	useremail
itemquestionexternal	userfax
itemquestiontranslatedtext	userlocationref
itemrangechecks	userphone

The highly structured nature of CDISC ODM data requires that any mapping to a relational format include a large number of data sets, with foreign key relationships to help preserve the intended non-relational object structure. In the SAS Clinical Standards Toolkit, foreign key relationships are enforced when validating the CDISC ODM data sets.

Field lengths in the CDISC ODM data sets are consistent by core data type. CDISC has not specified any limit to the length of most character fields. Arbitrary lengths have been chosen by data type. These lengths are listed in this table. In the table, standard data types are distilled into core data types. To be safe, larger lengths have been chosen to ensure that no data loss occurs in the SAS Clinical Standards Toolkit pre-installed data sets. Production tables might be compressed using SAS mechanisms to preserve disk space.

CDISC ODM Default Lengths by Data Type
Type Name	Length	Description
oid	64	A unique object identifier or a reference
text	2000	A character field that can accommodate a large number of characters
name	128	A descriptive identifier
value	512	An item of collected or reference data
path	512	An absolute or relative file system path or URL

The table metadata for the 66 data sets and the column metadata for the 315 columns in those data sets that comprise the SAS representation of the CDISC ODM 1.3.0 standard are in this folder:

global standards library directory/standards/cdisc-odm-1.3.0-1.5/metadata

Table metadata is in reference_tables.sas7bdat, and column metadata is in reference_columns.sas7bdat.

Only the ODM data set, which contains valid values for the FileOID, CreationDateTime, and FileType variables, is needed to create a minimal, but valid, CDISC ODM-compliant XML document. This is based on the CDISC ODM standard, which is flexible. All table and column names are case sensitive. They must be specified exactly as shown.

In the SAS implementation of the relational data model, the keys are extended to define a unique record in every SAS data set. For example, a unique record in the EnumeratedItems data set is defined by the variables FK_CODELISTS and CODEDVALUE. These SAS data set keys are in the table metadata in the SAS reference_tables data set.

Starting in ODM 1.3.0, there are two forms of the ItemData element, which is the element used by ODM for transmitting clinical data item values. These two forms are untyped and typed. Here is an example of a typed ItemData element:

Here is an example of an untyped ItemData element:

Both of these data values are stored in the Value variable in the ItemData SAS data set. In the case of typed data, the ItemDataType variable in the ItemData SAS data set has the data type (for example, Float). In the case of untyped data, the ItemDataType variable in the ItemData SAS data set is null.

Typed and untyped data transmission should not be mixed within a single ODM file. However, in the example provided by the SAS Clinical Standards Toolkit, both types are part of the same example for demonstration purposes.

In the SAS Clinical Standards Toolkit 1.5, the CDISC ODM standard supports reading and representing in SAS a complete odm.xml file, and building an odm.xml file. The SAS Clinical Standards Toolkit validates both the structure and content of the SAS representation of each odm.xml file and the structural integrity of that file. The SAS Clinical Standards Toolkit also supports the extraction of subject or reference data for a data set (such as an SDTM AE domain) from the odm.xml file.

To support all of this functionality, supplemental files include the following global standards library files:

A SAS format catalog (odmct.sas7bcat) in the formats folder provides valid values for selected columns in the 66 tables of the SAS representation.
The Messages data set in the messages folder provides error messaging for all Validation Master checks.
The Validation Master data set in the validation/control folder contains the superset of checks validating the structure and content of the 66 tables.
SAS code in the macros folder provides CDISC ODM-specific code that augments the code provided in the primary SAS Clinical Standards Toolkit autocall library (!sasroot/cstframework/sasmacro).

It is this set of files, in whole or in part, that defines the CDISC ODM 1.3.0 reference standard.

CDISC ODM 1.3.1 Reference Standard

The CDISC ODM 1.3.1 reference standard has the same functionality as CDISC ODM 1.3.0, with the following differences:

The SAS representation of CDISC ODM 1.3.1 includes 10 data sets in addition to those shown in reference_tables (CDISC ODM 1.3.0). The 10 additional data sets are listed in this table:

Additional CDISC ODM 1.3.1 Tables Not Included with CDISC ODM 1.3.0
codelistaliases	formaliases
codelistitemaliases	methodaliases
codelisttranslatedtext	mualiases
conditionaliases	protocolaliases
enumerateditemaliases	studyeventaliases

The table metadata for these 76 data sets can be found in the reference_tables data set in the standard metadata folder. Column metadata for the 352 columns in these 76 data sets can be found in the reference_columns data set in the standard metadata folder.

This set of files, in whole or in part, defines the CDISC ODM 1.3.1 reference standard.