Pinnacle Community 2.1

Define 1.0 Configuration

Hello Cathy,

The Define 1.0 and 2.0 configurations from previous versions have been combined into 1 Define.xml configuration. The difference between the "Define" and "Define (PMDA)" configurations is discussed by Brad here http://www.opencdisc.org/forum/general-question-about-data-xxx-fda-vs-data-xxx-pmda-vs-data-xxx#comment-1960.

define.xml validation in Pinnacle 2.1 community

As one of the developers of the Define-XML standard (both v.1.0 and 2.0) I must say that I am not happy at all with this "unification", this as the rules in define.xml 1.0 and 2.0 are rather different.
When I validate a define.xml v.1.0 file that I consider to be 100% valid (it was published by the CDISC define.xml Team some 10 years ago) I get a lot of errors DD0068 (Invalid use of Length), and an error DD0074 (Variable/Value Level Origin Type mismatch), as well as a few errors DD0055 (invalid Class value). I consider these as false positives as:
- In define.xml 1.0, "Length" was required for DataType="text", "integer" and "float". It was however NOT forbidden for other datatypes.
- In define.xml 1.0, there were no clear rules about "Origin", the specification states:
"Values such as 'CRF Page numbers', 'derived', or variable references". When I use a variable reference, I get the DD074 error. I.e. in define.xml 1.0, the value of "Origin" was essentially free text, without any specific rules.
- In define.xml 1.0, there was no enumeration for "def:Class". The specification states:
"General class of the data domain (e.g., "Events", "Interventions", "Findings", "Special Purpose", "Analysis", "Selection", "Support", and "Legacy"). If I put in "Study Design" (which I consider completely legal), error DD0055 is thrown.

So I suggest that Pinnacle21 community returns to the earlier design that the user can choose between define.xml 1.0 and 2.0, with different rules: it is surely not correct to apply 2.0 rules to a 1.0 file.

Define 1.0 v 2.0

Define 2.0 (and the soon to be 2.1 version) has many advantages for clearly communicating study metadata.    Because define 1.0 is a 10+ year standard (Released for Implementation Feb 10, 2005), shouldn't there be an industry move towards the more modern standard ?   I feel like this promotes the evolution of the standard which in general allows for better metadata.  

For things like "Study Design" as a class value, it seems like CDISC is clear in what the list of valid classes should be in the IG.  Why not use Trial Design instead ?   Same for other datatypes, origin and use of length.  if you have valid use cases for your terms, can you propose them to CDISC for the Define 2.1 version ?

Define 1.0 vs 2.0

Mike,

I very much agree with you that Define-XML v2.0 (and soon to be v2.1) has many advantages for clearly communicating study metadata. We all hope that the industry will move to this version very soon.

Having said that, I also agree with Jozef, that we can not retrospectively apply the stricter v2.0 rules to a 10+ year old standard.The Define-XML v2.0 specification does not specify rules for Define-XML v1.0.

The FDA and PMDA still accept Define.xml v1.0. I hope they would send a strong message that they want to receive Define-XML v2.0

Pinnacle 21 Community validates both define 1.0 and 2.0

This should help clear things up and make everyone happy.   We don't support generating a define.xml 1.0.  

But if you provide your own Define.xml 1.0 for validation, it will run the 1.0 rules.   We simplied the drop down to only list Define and Define (PMDA), but the logic will read the version in your Define and decide which rules to apply.

Pinnacle 21 Community validates both define 1.0 and 2.0

Mike,

Thanks for clarifying that Define-XML v2.0 rules will not be used for Define-XML v1.0 validation.

Lex

This does not seem to be the

This does not seem to be the case.  Define 2.0 has different origins than did Define 1.0, and the origins in 2.0 do not fit well with SEND files.  Even though we have a Define 1.0 file, we are getting errors related to define 2.0. 

This does not seem to be the

This does not seem to be the case.  Define 2.0 has different origins than did Define 1.0, and the origins in 2.0 do not fit well with SEND files.  Even though we have a Define 1.0 file, we are getting errors related to define 2.0.