Forums: Troubleshooting and Problems
Hi, in define.xml (2.0) datasets are not supposed to be listed in alphabetic order. The suggested order is given in section 3.4.2. Other Order Considerations for Elements of the Define-XML 2.0 specification:
Note that the SDTM-MSG indicates that the datasets should be displayed in the following Class order:
- TRIAL DESIGN
- SPECIAL PURPOSE - Subject-Level
- INTERVENTIONS
- EVENTS
- FINDINGS
- FINDINGS ABOUT – not yet added to the NCI/CDISC Controlled Terminology as of the time of writing this document
- RELATIONSHIP
For regulatory submissions of ADaM datasets, a standard order of display has not been established. However, the following Class order seems reasonable, as it is consistent with the ordering of Class values in ADaM 2.1:
- SUBJECT LEVEL ANALYSIS DATASET
- ADVERSE EVENTS ANALYSIS DATASET – not yet added to the NCI/CDISC Controlled Terminology as of the time of writing this document
- BASIC DATA STRUCTURE - datasets in alphabetical order by dataset name
- ADAM OTHER
Dear Dmitry,
Can you point me to the CDISC statement that for SDTM, the ItemGroupDefs (I presume that that is meant by John with "XML rendering") must not be in alphabetical order anymore within the sets of classes in the define.xml?
In section 3.4.2 of the define-XML 2.0 specification, I read: "The Define-XML specification does not require that ItemGroupDef elements related to the SDTM follow a particular order. However, for purposes of regulatory submissions, ItemGroupDef elements should follow the order recommended in the SDTM-MSG".
In the CDISC-SDTM-MSG (Metadata Submission Guide (2011), Section 3.2.1) I read : "All tabulation datasets, CDISC and sponsor-defined, must be included in the dataset-level metadata and they should be organized by their SDTM class. Datasets should be listed in alphabetical order by name attribute within each class in the define.xml file".
But maybe I missed something and there have been later CDISC statements about this order.
Hello Jozef,
You missed nothing, because CDISC has not written such a statement (yet). As you have perfectly shown the datasets should be first ordered in a certain way by class and then alphabetically. What confused John in his original message is why the order is not simply alphabetical by dataset name and is not consistent with a source spec file. Both Define-XML specification and SDTM-MSG perfectly answer this question.
Thanks Dmitry!
I must say honestly that I presonally don't like this "rule", as essentially it is about presentation to the reviewer, and presentation should be taken care of by the stylesheet (or any other tool used by the FDA/PMDA). Define-XML is primarily meant to be machine-consumable, but we also know that the FDA and PMDA do not use it in this way yet - they just look at what the browser shows them - that's it.
Hi Jozef,
Yes, you missed the second part of the statement you already highlighted:
"Datasets should be listed in alphabetical order by name attribute within each class in the define.xml file".
Regards,
Sergiy
My domains are listed in the *.xlsx spec file, but when rendered in XML, they appear in a semi-random order (Trial data sets at the top, SUPPs at the bottom,.everything else mixed up between the two.
They are listed in my specification in alpha order. What am I missing?
(I'm using Community version 2.1.2)