Rule FDAC017/FDA018 in XQuery

Description: SDTM Required variable not found / NULL value in variable marked as Required

Rules FDAC017 and FDAC018 can/must be taken together when using Dataset-XML as the transport format. The reason is that Dataset-XML nicely (as it should) separates data from metadata. SAS-XPT however (a format from the mainframe era) contains some metadata (variable name, label and datatype - but only text or numeric) and all other metadata must be in the define.xml. However "define.xml is leading"!

So, when implementing this rule, we will read the define.xml file, and look for these variables that are marked as "required", i.e. for which the value of the "Mandatory" attribute is "Yes". Then we will look for records in the Dataset-XML for which such a data point is missing.

This of course requires that the "Mandatory" field in the define.xml is correctly implemented, i.e. that for SDTM required variables, the value of the "Mandatory" attribute is "Yes".
The second approach is that the rule implementation looks up whether the variable is a required variable from the SDTM-IG (version dependent). Unfortunately however, the SDTM-IGs are still published as PDF, which is not machine-readable. Essentially, SHARE should deliver this information by a web service. This feature has not been implemented yet in SHARE, so we developed a RESTful web service "Get whether an SDTM variable is required, expected or permissible", for which you can find the specification here. We will demonstrate such a "web service lookup" in one of the other SDTM rules.

Explanation

• As in the previous example, the first lines are comments and the necessary namespace declarations
• Lines 10-12 provide the path to the location of the define.xml file. In this case, the define.xml is located in the native XML database in the collection "/db/fda_submissions/cdisc01"
• In line 14, we combine the base path and define.xml into a single document using the doc() function
• In line 17, we start iterating over all dataset definitions, i.e. over all "ItemGroupDef" elements
• In line 19, we read whether the attribute is "required" (Mandatory="Yes") and store that into the $mandatory attribute
• The location of the dataset is detected in lines 21-22
• In line 25, we start iterating over all the records in each of the datasets

Remark that in Dataset-XML, a record is defined by an "ItemGroupData" element. The record number is provided by the "data:ItemGroupDataSeq" attribute.

The second part of the XQuery comes as:

• The record number is taken in line 26
• Within the record, an iteration is made over all required variables (line 28)
• The variable name is taken from the define.xml. This is necessary, as as well define.xml as Dataset-XML use OIDs (ODM identifiers) for the variables.
• Line 31 then tests whether there is a data point for the required variable. If not, the "return" statement is executed
• Line 32 then returns the error messages, adding metadata such as the record number, the variable name, the dataset name, the last update of the rule, ...

Courtesy of XML4Pharma - last update: May 2016