Stream: implementers
Topic: Profiling Identifier.use (was: Warning iso Error for miss...
David Simons (Mar 28 2022 at 12:35):
PS: Has there been guidance in the past on "usual" vs. "official", @Lloyd McKenzie ?
If I read the definitions below, in my interpretation there can realistically be only one 0..1 official Identifier on a Resource - the primary/main one in the given setting/ecosystem - while there can be multiple 0..* usual Identifiers. Because, if there's more than 1 official, which is the primary, main, most-trusted one?
http://hl7.org/fhir/identifier-use
Code Display Definition
usual Usual The identifier recommended for display and use in real-world interactions.
official Official **The identifier considered to be most trusted for the identification of this item**. Sometimes also known as "primary" and "main". The determination of "official" is subjective and implementation guides often provide additional guidelines for use.
René Spronk (Mar 28 2022 at 12:38):
As the qouted text states, it'd be up to an IG to define quite what 'official' means. There is no requirement that there be only one single official identifier, there could be multiple.
David Simons (Mar 28 2022 at 12:45):
René Spronk said:
As the qouted text states, it'd be up to an IG to define quite what 'official' means. There is no requirement that there be only one single official identifier, there could be multiple.
Right, the spec leaves a lot open - so curious what approach folks have been taking - there's not always an IG yet.
René Spronk (Mar 28 2022 at 12:56):
I've seen the following in quite a few places: 'official' identifiers are those that are reliable, backed up by semi-public government registries and government-issued id cards. These are the 'most trusted'. Hospital issued identifiers are 'usual', not 'official'.
David Simons (Mar 28 2022 at 13:04):
To me that is mixing the Identifier.system with the Identifier.use. The latter is about the use, not about the system, in my view.
A hospital system, could very well decide to use their MRN as the most trusted (official!) Identifier, used to aggregate data around a Patient, right? This is separate from a larger say nationwide ecosystem where a national identifier might be selected as official.
René Spronk (Mar 28 2022 at 13:07):
certainly.
David Simons (Mar 28 2022 at 13:09):
Dank René - I appreciate you thinking along!
René Spronk (Mar 28 2022 at 13:11):
Obviously in countries where there's no widely accepted government issued patient id (or: too many issuing government bodies) the use of 'use' will be more useful than in countries that a single well known patient Id.
If FHIR is used within the context of one single institution I'd expect that use would never be valued. It doesn't add anything above and beyond the ID system URI.
Lloyd McKenzie (Mar 28 2022 at 15:08):
Based on the definitions, I think that's backwards Rene. 'Official' is the one used by the system (i.e. the MRN), while 'usual' would be the ones commonly used/known - which would be the jurisdictional identifiers. What's 'official' to one system is typically not 'official' to another.
Jens Villadsen (Mar 28 2022 at 15:30):
Ehhh .... (just passing by here) - I don't agree. 'official' to me means something usually issued by state or regional authorites
Jens Villadsen (Mar 28 2022 at 15:31):
'usual' more describes a pattern of use/context of use. Eg. your passport ID is pretty official
Jens Villadsen (Mar 28 2022 at 15:31):
in theory that could also be usual
Jens Villadsen (Mar 28 2022 at 15:40):
In DK, our central reg. number would actually qualify for both
Brian Alper (Mar 28 2022 at 15:41):
To clarify the identifier.use codes I submitted FHIR-32308 https://jira.hl7.org/browse/FHIR-32308 with a proposed disposition of: Change the definition of "usual" to become "The identifier recommended for display and use in real-world interactions, used when such identifier is different from the "official" identifier."
Jens Villadsen (Mar 28 2022 at 15:42):
That wouldn't make sense in Denmark as it is the same
Brian Alper (Mar 28 2022 at 15:44):
That is why clarification is needed -- when the identifier.use is both 'official' and 'usual' which code should be used? The clarification would it make it more clear to use 'official' in that situation.
Jens Villadsen (Mar 28 2022 at 15:53):
arhh okay ... just got into context. Yes I agree
Lloyd McKenzie (Mar 28 2022 at 17:11):
You have to be driven by the definitions of the codes, not the typical meaning of the word.
"official": The identifier considered to be most trusted for the identification of this item. Sometimes also known as "primary" and "main". The determination of "official" is subjective and implementation guides often provide additional guidelines for use.
"usual": The identifier recommended for display and use in real-world interactions.
And that (in Canada at least), would make "official" the MRN, and "usual" the provincial health number.
René Spronk (Mar 28 2022 at 17:23):
What's the point? Such usage of use sound pretty useless to me (pun intended). I'm with Jens, 'official implies usual'. An official identifier that's not meant for display/actual use (usual) - can't think of any.
Lloyd McKenzie (Mar 28 2022 at 17:26):
I'm not in love with 'official' vs. 'use' for identifiers. They make much more sense for names and addresses. But 'official' definitely means "most trusted within that system", and that's typically the system-assigned id, not a centrally assigned id.
Jens Villadsen (Mar 28 2022 at 17:32):
Disagree - big time
Jens Villadsen (Mar 28 2022 at 17:34):
Scandinavian societies are based upon trust (for better and for worse) - with a huge trust on central services and authorities
Jens Villadsen (Mar 28 2022 at 17:35):
Meaning that Id's issued by central services has more credit than those issued locally
Jens Villadsen (Mar 28 2022 at 17:36):
That may correlate to the fact that the countries are small in numbers
Jens Villadsen (Mar 28 2022 at 17:38):
Citizens that is
Jens Villadsen (Mar 28 2022 at 17:41):
It would actually be illegal to have a patient record without an Identifier from the central registry (with few exceptions)
John Moehrke (Mar 28 2022 at 17:49):
there might not be as much dissagreement as it might appear, although I do think there is some. Lloyd is indicating official is ... "within that system"... if the Patient resource is being managed within the EHR, then it is possible that they might choose to have the EHR assigned MRN be the official.. where as if the Patient is managed within a state patient registry, then the state identity would be the official... I can understand this perspective, but I do find it odd.
John Moehrke (Mar 28 2022 at 17:50):
that said, I tend to not bother with the .use; when I want the national ID, I look for the .system value of the national ID issuer.
David Simons (Mar 29 2022 at 07:23):
I am also with Lloyd - we must not interpret "official" per se as "governmental"- with the given definitions of identifier-use.
If it is governmental _issued_, with all its trust levels, the Identifiersystem will show that. Same with other institutional Identifiers.
Similarly, the MRN will have a system showing it was issued locally by a health system.
Only once making the choice to _use_ such an Identifier in a health-system integration/ecosystem as the _main_/_primary_ Identifier for a data subject (e.g. Patient, Device, ...), that makes it use='official', at least that is how I interpret the definitions.
NB: use, not issue!
Some deployment choose to use a governmental-issued Identifier (and corresponding Identifier.system) as official, other deployments may choose to use another MRN-based Identifier as official.
Agreed that that same system+value can also be usual - and such the multiplicity of 0..1 Identifier.use may lead to repetition of the same Identifier with multiple uses. In that sense, I also agree that official does not excluse usual use of that same Identifier system+value.
Also, and that was part of my original question - I expect there to only be 1 Identifier to be listed as official, on a given FHIR Resource, in a given deployment/use of the data. Otherwise, what's the added value of Identifier.use attribute, if I cannot filter on it - to find the main/primary (singular!) Identifier - without also filtering on Identifier.system ?
Jens Villadsen (Mar 29 2022 at 08:11):
a note on that: I don't see a problem in having multiple official Identifiers on a given FHIR resource. It probably happens seldomly.
David Simons (Mar 29 2022 at 08:19):
Thanks all - getting your viewpoints is appreciated!
An alternative apporach I can take is a constraint like use{official, usual}.exists implies system.exists, or just system.exists on Patient.Identifiers, and query/filter by system, primarily.
Last updated: Apr 12 2022 at 19:14 UTC
