c Christelle


Seeking some advice on the 3 columns in the title "Repeating", "Reference" in the datasets tab and "Mandatory" columns in the variables tab in the P21 spreadsheet to generate the define.xml for ADaM.

"Repeating": if set to "Yes", does it mean  the structure of the dataset should be one record per subject only, and so the key structure could be USUBJID only?

"Reference": if set to "Yes", does it means the dataset does not contain subject level data so the key structure would not reference a subject variable like "USUBJID"? Or is it possible to have an ADaM dataset that could be a reference dataset but still use a subject level variable in the key structure (for ex: specific exclusion for analysis at the subject level)? In the last scenario would it be acceptable. 

"Mandatory": If set to "Yes", does it mean the variable must be present because the ADaM IG has it as 'Expected' in the guidance and should not have missing value?



Forums: Define.xml

j Jozef
on February 2, 2021

Dear Christelle,

This is all well explained in the Define-XML Standard specification (https://www.cdisc.org/standards/data-exchange/define-xml). Please read it.
Repeating=Yes: It is expected that there can be more than 1 record per subject (i.e. per USUBJID). This usually means that besides USUBJID there are one or more additional keys for the dataset (provided by the KeySequence attribute in the ItemGroupDef/ItemRef).
Repeating=No. There may only be one record per subject (i.e. per USUBJID). This also means that there is only 1 key defined which is USUBJID.
Mandatory=Yes: There must be at least one record per subject for the given variable.
Mandatory=No: It can be that there are no records for the given variable for a subject.
As explained in the Define-XML specification (and also well explained at CDISC trainings), "Expected" translates to Mandatory=No in the define.xml. Mandatory=Yes is only set for "Required".
A sponsor can however always "promote" this. For example, if the ADaM or SDTM or SEND-IG states that a variable is "Expected", the sponsor is free to promote it to "Required", and than set the variable to Mandatory=Yes in the define.xml. The other way around (demoting) is of course not allowed.

With best regards,
Jozef Aerts
CDISC-Define-XML Team and Define-XML Trainer

a amparo
on June 22, 2021

Hi Jozef,

We also have the same question about this, according to your description "there are one or more additional keys for the dataset (provided by the KeySequence attribute in the ItemGroupDef/ItemRef)"  is the 'one key' except "STUDYID"? in example of DM dataset of the Define-XML Standard specification, it contains 2 keys (STUYID and USUBJID), please find attached snapshot.

Besides for TRIAL DESIGN domains, all of datasets should be Repeating=No, but some of their key variable is more than one (i.e. TA, key with STUDYID,ARMCD,TAETORD), this will also against the 'one key' rule.

Best regards,



j Jozef
on June 23, 2021

Yes, the first key usually is STUDYID. This is meant for the case that all metadata for different studies is stored into a (relational) database.
So, for DM, usually, one has KeySequence=1 for STUDYID, KeySequence=2 for USUBJID.
The choice of the keys is entirely your choice (but it must of course be correct to guarantee uniqueness), so you could also choose to only have 1 key for DM (which would then be USUBJID). But in the examples, we preferred to (also) have STUDYID as the first key, for the use case of setting up a cross-study database.

Hope this helps.


x Xinhua
on August 5, 2021

Hi, Jozef:

      In the case of VLM for expected/perm variable, what is the rule to define 'Yes'/'No' value for mandatory? for example, VSORRES where VSTESTCD eq 'WEIGHT' with ItemOID="IT.VS.VSORRES.WEIGHT".




j Jozef
on August 7, 2021

Think about it as a transpose, i.e. a "WEIGHT" column is created. If Mandatory="Yes", it is then expected that there is a value (at least one) for "WEIGHT" for each subject.
Does this help?

Want a demo?

Let’s Talk.

We're eager to share and ready to listen.

Cookie Policy

Pinnacle 21 uses cookies to make our site easier for you to use. By continuing to use this website, you agree to our use of cookies. For more info visit our Privacy Policy.