4 Radnor Corporate Center
Suite 350
Radnor, PA 19087
+1 (888) 507-2270
Dataset Level - STRUCTURE text
Forums: Define.xml
LexThere is no length…
There is no length restriction for the structure attribute in the Define-XML document. Systems that implement these Define-XML attributes in tables (SAS, Oracle, ...) may have restrictions.
There is no standard for the 'structure' text. Examples given in the SDTM-IG are just like that: examples. Your values may be different for some datasets depending on your study.
You may also want to look at the PhUSE Define-XML Version 2.0 Completion Guidelines:
https://www.phusewiki.org/docs/WorkingGroups/New%20Template%20Deliverables/Optimizing%20the%20Use%20of%20Data%20Standards/Define-XML%20Version%202.0%20Completion%20Guidelines%20.pdf
Best,
Lex
Dataset Level - STRUCTURE text
As Lex already wrote, this is just free text, without any length limitation.
My own way of creating it (automatically) is by taking the keys (KeySequence attribute) from the define.xml in the reverse order (from high to low), with exception of STUDYID and then generate a sentence of the structure: "One record per {VARA} per {VARB} per {VARC} ..."
E.g. when my keys are (in the following order): STUDYID, USUBJID, VISIT, LBTPT, LBTESTCD => One record per LBTESTCD per LBTPT per VISIT per USUBJID.
Anyone can of course choose his/her own way, manually or automatically.
What I have seen however, is that people tend to just copy the sentence from the SDTM-IG. That is of course often not correct...
Hi all
At the dataset level, there is the variable called STRUCTURE, does anyone know if there is a max length for this field ? Also is it permitted to change this text, I know the KEYS can be adjust per the SDTMIG but there is no guidance on the STRUCTURE.
Best
Colm