Forums: Validation Rule Suggestions
Hi Karl-Stephan,
Thank you for this interesting suggestion!
It could be added it as a new Notice rule like "Custom analysis variable". Some people will appreciate it, other ones will hate it as extra validation issues to explain.
However, I see many implementation challenges. For example, what about standard SDTM variables and custom SDTM variables from SUPP-- datasets? They should not be reported by this new validation rule.
Kind Regards,
Sergiy
I think the implementation challenges for marking NSVs (Non-Standard Variables in the official CDISC language) are minor. This can easily be accomplished by a checkbox in the graphical user interface. When the checkbox is not checked, such tests are not executed. So one could then generate a report for internal usage with checks for NSVs and "custom analysis variables" and another one for submission without these checks performed.
My own SDTM-ETL mapping software has many such options.
Also remark that define.xml 2.1 will have a mechanism for marking NSVs, this although checking whether a variable is an NSV can also be done by a simple RESTful web service query to CDISC's "CDISC Library" system (formerly known as SHARE 2.0 API), Company-internal RESTful web services can also easily be used to implement company-internal validation or quality rules. We have done so for different sponsors.
ADaMiG (v1.0 and 1.1) is strictly in favour of restricting the allowed scenarios for creating new variables. While the various fragments - prefixed radicals (e.g. 'WA' used for references to wash-out) and suffixes *FL, *DT etc - suggest this should probably not even be a warning, it would be great of p21 could identify all variables not in ADSL and not named in one of ADaMiG, OCCDS or TTE.