PhUSE Working Group RGvD - Meeting Minutes 2015-07-21
From PHUSE Wiki
- Sandy (Astra Zeneca)
- Steve (Cytel)
- Janet (Merck)
- Cathy (Novartis)
- Ellen Asam (Merck)
- Continue define.xml review
- Sandy e-mailed the revised schedule prior to the meeting
- Using the V2.0.0 example file: define2-0-0-example-sdtm.xml (or .html):
- We revisited the Table of Contents. Sandy queried the group whether they sorted the dataset by the variables in the Keys column. Although the datasets were not part of the charter of this group, if we were to make a recommended best practice, we could put it as a special section in our document.
- The Structure column seemed to have a format that did not follow the same order as the variables in the Keys column. If we were to recommend a best practice on changing the format to match the order followed by the Keys column please email a recommendation on how to phrase the best practice.
- Split datasets were next discussed
- The example file showed how the QC domain was split: each split dataset had it’s own line however most of the metadata was repeated.
- We continued with the Controlled Terminology section
- The group was queried whether their company included all possible terms or just the values which were present in the data. Most companies include all values if there isa pick-list on the CRF If there is any free text collected or not a finite set of values, then only those that appeared in the data are displayed.
- There was some further discussion about origins.
- STRESC/STRESN – should origins be Derived or Assigned, should it vary from domain to domain? Derived is used in the example define doc.
- TESTCD vs. TEST: which one links to the CRF and which gets Assigned? Typically the TEST description comes from the CRF and the test code is an assigned value.
- Next we reviewed the Value List section
- No new challenges were identified here that did not also apply to variable-level tables.
- Then we reviewed the Comments section
- This is a new section in V2 and appeared to repeat text from the Derivation/Comment column in the Variables and Value-level sections.
- It’s purpose was not apparent to the group. Sandy will check with Mike Molter to see what the logic was for including it.
- Last we discussed Controlled Terms/Format column and Codelists
- New in v2.0, text values which do not utilize codes are listed in a single column style codelist and lists which include codes have two columns, with code on left and decode on right. Within the controlled terms columns (variable and value level), sometimes the individual values are listed and sometimes there is just a link to the codelist present. It is difficult to tell what criteria is used for each style. Some further research is needed and we will discuss at the next meeting.
Next meeting: 2015-08-04
- Sandy to check with Mike Molter regarding Comments section in define.xml v2.0.
- All team members to research how codelists are displayed within the Controlled Terms/Format columns and consider best practices for populating this column.