How to make the OpenCDISC report clean

j jing

on August 17, 2017

Hi,

We try to make the report as clean as possible, but it seems we can't implement this. It seems there are some rules that conflict with each other, either way, there will be a warning in the report. For example, for the following 2 rules, if we add EPOCH, we will violate the 1st rule; if we remove EPOCH, we will violate the 2nd rule. Can you suggest what should we do to get rid of the warning?

SD1076/FDAC031

Model permissible variable added into standard domain

SDTM model variable may be added into standard domains according its domain general class, if there are no restrictions on their usage specified in IG.ALLWarningWarningXXX

SD1077/FDAC021

FDA Expected variable not foundVariables requested by FDA in policy documents should be included in the dataset. E.g., EPOCH and ELEMENT.ALLWarningXXX

Forums: SDTM

c Carlo

on August 18, 2017

You really shouldn't

In answer to the broad question, you really should not and CANNOT have a clean report, at least not under the current rules as they are implemented.

The current rule set is a mix of conformance, informational, and data quality checks, and understanding which kind a given rule is, and how to address it, is critical. Unfortunately, the information provided in the report is not always presented with enough clarity or context, and it is incumbent on the user to more broadly understand the issues.

In the instance you note above, while both rules are "Warnings" they are fundamentally different in terms of meaning and importance:

SD1076/FDAC031 is informational, and your only action should be to document in the cSDRG that yes, you did in fact use a variable allowed by the SDTM model.
SD1077/FDAC021 is identifying that you did not include a variable that FDA expects you to. If you opt not to include, for example, EPOCH, you would need to provide a rationale as to why you did not follow agency guidance, hopefully one with enough justification to be acceptable.

Obviously, and ideally, it would be clear that addressing SD1077/FDAC021 imposes a corrective action (add the variable), while SD1076/FDAC031 imposes a documentation activity.

The evolution of the ruleset this point has, unfortunately, led to a broad range of misunderstandings in terms of how companies interpret the findings and how they address them. With the publication of CDISC rules for SDTM, as well as the more thoughtful FDA ruleset, my hope is that future rules as implemented in Pinnacle21 (or any software) will be more clear in terms of meaning and resolution.

Regards,

Carlo