Difference between rules SD1228 and CT2004

j Jozef

on August 21, 2019

Dear colleagues, dear SDTM experts,

I have trouble understanding the difference (and their implementation in P21) between rules SD1228 and rule CT2004.

Rule SD1228: "Variable should be populated with terms from its CDISC controlled terminology codelist, when its value level condition is met. New terms can be added as long as they are not duplicates, synonyms or subsets of existing standard terms."

Rule CT2004: "Variable must be populated with terms from its CDISC controlled terminology codelist, when its value level condition is met. New terms cannot be added into non-extensible codelists."

A few questions that pop up immediately:
a) is there a difference between "should" and "must"? If yes, where is this defined? HL7-documents e.g. start with such definitions. CDISC is just starting doing so.
b) Are these rules applied on the metadata (define.xml) level, i.e. is it checked in the define.xml whether the codelist is extensible, and then checked whether there are additional terms or not, and in the case there are (when allowed) whether these do lack the "def:ExtendedValue="Yes"" attribute pair?
c) or is the rule solely applied on the XPT files without looking into the define.xml?
d) Is, or how is it checked whether duplicates are present (I presume in the define.xml - that is feasible), whether one term is a synonym of another?

Many thanks in advance!!!

Jozef

Forums: SDTM

Sergiy

on August 28, 2019

P21

CT2004 is about CT, SD1228 is about define.xml

Hi Jozef,

a. Since beginning, original Pinnacle 21 approach was to use “must” for Errors and “should” for Warnings. See PhUSE guidance for developing validation rules. So, I believe that both CT2002 and SD1228 rules should use a “must” word. However, later many new stakeholders started being involved in rules development process and their interpretation of Error and Warning are not always the same including the same rule. Therefore, this original P21 good practice may be overlooked in some cases. As I said, formally SD1228 rule description is supposed to have a word “must” instead of “should’. We’ll discuss to fix it. Though, most people do not care.

b. SD1228 rule checks Codelist terms in define.xml file vs. actual study data. CT2004 rule compares study data vs. CDISC Controlled Terminology (CT).

c. CT2004 rule is about compliance with CDISC CT (doesn’t use define.xml). SD1228 is about consistency of study metadata with actual study data (uses define.xml).

d. Synonyms are not applicable to both rules. CT2004 rule is about non-extensible CDISC CT. Any new term is a formal error. Other CTxxxx checks related to extensible CDISC CT will benefit from implementing synonyms functionality. We are working on this now. However, it’s potentially different from your suggestion. For example, if Lab Test Code term is “SGOT”, then we know that it is a synonym of standard term “AST”. I would consider this case as an Error because it’s definitely incorrect implementation of CDISC CT. While Warning means a potential issue.

At this point we do not consider an implementation of your case due to potentially low business value for our users. It’s very rear case (I’ve never seen it so far), when the same information is presented using different synonyms in study data. For example, both “AST” and “SGOT” terms.

Kind Regards,

Sergiy