1777 Sentry Parkway West
Building 17, Suite 405
Blue Bell, PA 19422
+1 (888) 507-2270
Extent of Value Level Metadata
Extent of Value Level Metadata
To what extend you provide "ValueLevel" metadata in your define.xml is your choice. These are there to make the life easier for the reviewer in order to understand the submission.
As already said in another post (https://www.pinnacle21.com/forum/define-comment-requirements), every reviewer is different.
What I usually try to do (but that is my choice, other people may do it differently) is to "group" ValueLevel metadata.
When it is only for the "Origin", you can e.g. group by LBCAT when each distinct value of LBCAT represents a single page or section in the CRF.
When it however is about units (e.g. in VS), you can either provide ValueLevel metadata based on the test (i.e. VSTESTCD value), but even there, you may want to group.
For example SYSBP and DIABP have the same unit, so you can have a single "Where-clause" for both of them.
Remark also that ValueLevel metadata can be used to do "transpose" on the dataset, e.g. to make each different test a column in the dataset.
Will the reviewer do this (and has the reviewer the capability to do so)? Maybe you can look in your crystal ball ...
Extent of Value Level Metadata
I understands the grouping of records (ex. using LBNAM to group records and assign an Origin), but is there a way to enter VLM for a variable without considering —TESTCD at all? All of the examples I’m able to find reference a —TESTCD in the WHERE clause. What I’m wanting to do is something like creating a VLM for SUBJID, with a WHERE clause of LBNAM = ‘External’ so I can identify a separate Origin. However, when looking at the Define.XML, the VLM records require a Label, which based on my examples is always equal to an LBTEST value, but I don’t want to reference a specific test. Is there something else I should input as the label value for my SUBJID example above?
Extent of Value Level Metadata
I would say that "whereclauses" are indeed for 80-90% based on -TESTCD. But they don't have to be.
I often use "whereclauses" based on --CAT and even sometimes on --SCAT.
Use of LBNAM is surely a good idea when some of the lab data come e.g. from a separate lab data transfer, and some from the CRF.
For the case of "external" lab data (i.e. that are not from a CRF), you can then indeed uses VLM for USUBJID based on LBNAM, there is nothing against that.
As said, major goal of VLM is to allow the reviewer to understand the submitted information.
When creating the VLM for a domain (LB for example), to what extent do we need to report multiple origins for variables? To specify, say SUBJID, LBORRES, LBORRESU, LBMETHOD, LBDTC are all collected on the CRF and in an external for multiple LBTESTCDs. Do we need to create a row in the VLM for each variable, in relation to each LBTESTCD for each Origin, or is there a simpler way to report the multiple Origins for each variable without having to create a row in the VLM for each variable for each LBTESTCD from each source?