Define.xml is "arguably the most important part of the electronic dataset submission for regulatory review," according to The FDA’s Technical Conformance Guide. It helps reviewers gain familiarity with study data, its origins and derivations, as well as sponsor-specific implementation of CDISC standards. In addition to the define.xml, the Technical Conformance Guide also asks that sponsors submit a stylesheet. A stylesheet transforms the define.xml into a human readable structure. Most sponsors choose to use the stylesheet provided by PhUSE.  In November 2018, the PhUSE Define-XML 2.0 Style Sheet Recommendations Working group released an updated stylesheet. In a recent webinar, we compared the old stylesheet (published in 2015) to the new one and have highlighted our findings in the slides below. We also took some audience questions and have published them in the Q & A section below the slides.

Define Questions:

Q. What is difference between origin and source?
The def:Origin element was introduced in Define-XML 2.0.0. It is intended to define the Origin metadata for variables in SDTM or SEND datasets or Source metadata for ADaM datasets. Source on the other hand is def:Origin/Description which must be provided to indicate the source when def:Origin@Type=”Predecessor”. Conventions described in IG documents for the specific standard should be followed.

Q. How are hyperlinks created in the define?
Hyperlinks are generated when there are def:leaf elements. See section 5.3.15 of the Define-XML Specification v2.0 for more information.

Q. What is a "formal expression"?
A formal expression provides machine readable code to compute or impute the value of an ItemDef. See section 5.1.13.1 of the define-xml specification for more information. In P21E, the formal expression is created when Expression Context and Expression Code are populated on the methods tab.

Q. How should you handle multiple origins for the same variable (i.e. CRF, Assigned)?
When there is more than one origin for a variable, value-level metadata should be created for that variable.

Q. Should the length column show the assigned length of variable or max length of the values in the variable?
From the Define-XML Specification v2.0, Section 5.3.11 ItemDef Element, page 75, the business rule for the length is: "Length should be defined as the maximum expected variable length."

Q. Should we have labels for all VLM?
Yes, labels should be provided for VLM.

Q. Does FDA accept define.xml v1.0 or define.xml v2.0?
According to the FDA Data Standards Catalog, support for define 1.0 ended on March 15th 2018. Use version 2.0.

Q. How do we decide which variable will go into value level metadata?
Please see P21 blog on the Do's and Don'ts of define.xml for more information on when to created VLM

Q. Are there any best practices for creating codeslists?
Please see P21 blog on the Do's and Don'ts of define.xml for more information on when to created VLM

Q. How can we define the order value in codelists?
Individual terms within codelists can be ordered using the OrderNumber attribute on the EnumeratedItem and CodeListItem elements.

Stylesheet Questions:

Q. Does the new stylesheet work with define 2.1?
The latests stylesheet is for define 2.0. There is a different stylesheet for define 2.1 available on the CDISC website. 

Q. Can I start using the updated Style sheet for submissions?
There is no formal requirement (at this time) regarding which version of the stylesheet to submit. P21 recommends using the latest.

Q. Has the P21 define Excel spec spreadsheet structure changed at all to accommodate the new stylesheet?
No – the P21 define Excel spec is independent of the new stylesheet.

Q. Should we mention what stylesheet is used in reviewer's guide?
The PhUSE template does not provide a row in the study data standards and inventory section for this information; however, you can always add it but there is no formal requirement at this time.

Q. Does this stylesheet work for SEND, SDTM, and ADaM defines?
Yes – the new stylesheet is compatible with all three standards

Q. Is there a document explaining the new changes that we can refer to?
Yes – The full list of changes are listed in the change log which can be found at the top of the stylesheet file