Open BSnoeijerCD opened 1 month ago
@BSnoeijerCD I can only think of the following three rules:
@BSnoeijerCD I can only think of the following three rules:
- Study identifiers must be unique - there must be no more than one study identifier for each combination of text, identifierScope.identifierScheme, and identifierScope.identifier. I am assuming that there can only be one legal address for any organization (that assigns study identifiers).
- Only one identifier is expected within each identifier scope - no more than one reference identifier is expected for each combination of identifierScope.identifierScheme, and identifierScope.identifier. Note if this is changed from a warning to an error, then the previous rule is unnecessary.
- Reference identifiers are expected to be unique - no more than one reference identifier is expected for each combination of text, type.code, (type.codeSystem, type.codeSystemVersion, type.decode?), identifierScope.identifierScheme, and identifierScope.identifier. (However, I'm guessing it would be OK for a study to be included in more than one clinical development plan defined by a sponsor, so there's no reference identifier rule corresponding with the second study identifier rule above). Should this be an error instead of a warning?
@BSnoeijerCD in reply to your comments:
There was a typo in my original definition - I'd intended this to be:
Only one study identifier is expected within each identifier scope - no more than one reference identifier is expected for each combination of identifierScope.identifierScheme, and identifierScope.identifier.
So this would correspond with your comments for 1 - "you cannot have more than 1 identifier for 1 organization" - and I agree it should only be a warning. I wouldn't object to adding the notes
attribute to StudyIdentifier
. If we think it might be useful for ReferenceIdentifier
too, then it should be added to Identifier
to be inherited by both StudyIdentifier
and ReferenceIdentifer
. However, even if we add the notes
attribute, I don't think it should be referenced by this rule.
@ASL-rmarshall
@ASL-rmarshall : Could you review the following? Apart from the above, because we changed the naming of variable studyIdentifierScope to scope, we need to update CHK0012 and CHK0102 to make them only applicable to V3.0 and create corresponding ones with different variable names for v4.0. For these checks I set v4.0 to N and I created corresponding checks CHK0193 and CHK0193 for version 4.
@BSnoeijerCD
Based on comments above, it looks like we need 3 new checks:
Are these OK? If so, I have a couple of additional questions:
If we're reviewing/updating CHK0012 and CHK0102 then I think we should implement the change agreed in "Q12" on the Questions sheet - i.e., recombine these checks into a single check that says "Every study version must have one, and only one, sponsor identifier" (or "Of the study identifiers defined for a study version, exactly one must have an identifier scope that references a clinical study sponsor organization."). Because of the renaming of the scope
attribute, I agree that this single check specification would then need to be implemented as one rule for V3 and one for V4.
If you agree, should we delete CHK0012 and CHK0102 and create a completely new (recombined) check for V3, or should we reuse and modify either CHK0012 or CHK0102 and delete the other? (We'd need a new check for V4 regardless - either CHK0193 or CHK0194 ... or reuse either CHK0012 or CHK0102 for this?).
Also, the cardinality for studyIdentifiers
was 0.. in V3 but has been changed to 1.. for V4. Does this mean that we need a V3 rule for "Every study version must have at least one study identifier"?
Consider a check for there should only be one sponsor identifier. It is useful to be able to pick out "THE" sponsor identifier. This should probably be an error.
The three checks above should probably be warnings rather than errors
@dih-cdisc :
@ASL-rmarshall: I would suggest not to change any V3 checks unless there is a strong reason to do so. We want to remain as stable as possible. But making it more simple and thus combining CHK0012 and CHK0102 if that is possible with the CORE engine is better indeed. I think we are almost there. Let's go through it together in the spreadsheet to make it complete.
@BSnoeijerCD As discussed on 2024-0813:
We agreed that only the first 2 checks are needed (we won't check for more than one reference identifier per type - a study referenced by more than one clinical development plan or PIP will not be reported as a warning). There will be separate V3/V4 rules for each of these checks.
We agreed not to check for the same identifier used as both study identifier and reference identifier.
The question about how to identify duplicate type is now redundant as we're not checking by type any more.
We agreed to recombine the checks. CHK0012 and CHK0193 have been updated to specify that there must be exactly one sponsor identifier. CHK0102 and CHK0194 have been removed.
CHK0012 and CHK0193 will be changed to check for "Every study design must have exactly one sponsor identifier" instead of "Of the study identifiers defined for a study design, exactly one must be a sponsor identifier".
@ASL-rmarshall :
Pediatric investigational plan number needs to be addressed. It is more a reference than an identifier. See EMA template https://www.ema.europa.eu/en/documents/template-form/paediatric-investigation-plan-pip-key-elements-guidance_en.pdf See feature ticket #420