Stream: united states
Topic: provider taxonomy vs specialty
Lee Surprenant (Mar 20 2020 at 15:25):
Got into a deep discussion around semantics of provider taxonomy vs specialty with our subject matter experts today. We have provider taxonomy codes that we're trying to map into the fields on a PractitionerRole resource.
I recommended we map those into the PractitionerRole.specialty element, but there was some concern over the term "specialty".
Apparently, "specialty" is only one of the 3 components that goes into the NUCC taxonomy code.
Can someone confirm whether PractitionerRole.specialty is the proper field for mapping the entire taxonomy code?
Lee Surprenant (Mar 20 2020 at 15:26):
The US Core PractitionerRole has an Extensible binding to https://www.hl7.org/fhir/us/core/ValueSet-us-core-provider-specialty.html ...is that the full set of NUCC codes or just a subset?
UPDATE: I was looking at the wrong one before (https://www.hl7.org/fhir/us/core/ValueSet-us-core-provider-role.html vs https://www.hl7.org/fhir/us/core/ValueSet-us-core-provider-specialty.html) ... now I see the specialty one contains all NUCC provider-taxonomy codes, so I think it answers my question
Lee Surprenant (Mar 20 2020 at 17:05):
is there an existing mapping from a specialty code (https://www.hl7.org/fhir/us/core/ValueSet-us-core-provider-specialty.html) to the corresponding "role" code (https://www.hl7.org/fhir/us/core/ValueSet-us-core-provider-role.html )...is it safe to just zero-out the specialty digits (chars 5-9) or we need a real mapping?
Robert McClure (Mar 20 2020 at 20:54):
@Lee Surprenant These Role and Speciality value sets are a bit of a mess and there is some work occurring for the Provider Directory project that is part of Da Vinci that is tackling this within that context. We're meeting next week and may have more clarity after that - noting that this is not US Core doing the work, it's is a project that is expected to be conformant, but may end up with reasons to request a variance. I really can't say anything more than that because it's still under discussion. Bottom line is I'd wait on mapping. I'd be very interested in what you currently have locally for these. Can you PM me, or comment here?
Lee Surprenant (Mar 20 2020 at 21:18):
Thanks Robert, thats really good to know. We're doing a mapping exercise now and it seems like the source system has a "Provider Taxonomy" field and a local-system-specific provider "type" code . So now we're trying to decide which one we use to populate the "PractitionerRole.code" value (which is required by US Core and has an Extensible binding to this subset of NUCC codes).
@Kim M DeMars is there any sample data we can share with Robert here (or privately in DM)?
Eric Haas (Mar 20 2020 at 21:53):
did some work on this for VHDIR and here is a discussion:
https://chat.fhir.org/#narrow/stream/179175-argonaut/topic/NUCC.20provider.20taxonomy
@Robert McClure I think the VHDIR work is being used by Da Vinci... hopefully they can reuse some of these hash tables too.
Kim M DeMars (Mar 23 2020 at 15:06):
Examples of "Provider Types (from Client specific data)03 Extended Care Facility
05 Home Health Agency
08 Clinic
09 Advance Practice Nurse
12 Special Services
14 Podiatrist
15 Chiropractor
17 Therapist
18 Optometrist
19 Optician
22 Local Health Department
24 Pharmacy
25 DME/Medical Supply Dealer
26 Transportation Provider
27 Dentist
28 Laboratory
29 Radiology
31 Physician
32 Nurse Midwife
Goal would be to be able to take the provider taxonomy and to ingest first the entire taxonomy and then to parse it into the 3 levels - level 1 provider type, level 2 provider classification and level 3 provider specialty. Then we have standardization and if needed can map client specific data elements to the standard.
Lee Surprenant (Mar 23 2020 at 18:22):
@Robert McClure is that what you were looking for? ^
To be clear, I think our decision point is whether to
A. try mapping these custom/local role codes (which may differ from 1 deployment to the next) to the NUCC ones listed at https://www.hl7.org/fhir/us/core/ValueSet-us-core-provider-role.html ; or
B. try mapping the full NUCC taxonomy codes into the higher-level codes listed at https://www.hl7.org/fhir/us/core/ValueSet-us-core-provider-role.html ; or
C. not use this valueset for PractitionerRole.code
I was hoping there would be some easy way to do option B
Lee Surprenant (Mar 23 2020 at 18:27):
@Eric Haas I see the snippet from your mapping from NUCC taxonomy code to NUCC role codes at https://chat.fhir.org/#narrow/stream/179175-argonaut/topic/NUCC.20provider.20taxonomy/near/157290544 ...are the full mappings published somewhere? If not, are you willing to share that full python dictionary?
Eric Haas (Mar 23 2020 at 19:52):
lemme find them...
Eric Haas (Mar 23 2020 at 20:05):
https://github.com/HL7/VhDir/blob/master/notes_and_tools/example-generation/nucc_maps.py
Eric Haas (Mar 23 2020 at 20:06):
I can turn them into csv or other formats if you need, even a concept mapping.
Lee Surprenant (May 12 2020 at 16:11):
In prep for FHIR Connectathon 24, I noticed that the test data for the PDEX Plan-Net track is using the full taxonomy code instead of using the zeroed out values from the valueset.
Lee Surprenant (May 12 2020 at 16:13):
I'm thinking I'll open that as an issue at https://github.com/HL7-DaVinci/pdex-plan-net-sample-data but I found it pretty interesting given the discussion above
May Terry (May 12 2020 at 16:15):
cc: @Saul Kravitz re: PDex Plan-Net sample data.
Lee Surprenant (May 12 2020 at 16:22):
oh, actually, i guess the latest Provider Directory IG no longer includes an extensible binding to that subset of NUCC codes...so maybe its not an issue
Lee Surprenant (May 12 2020 at 16:23):
I was still looking at the snapshot at http://hl7.org/fhir/us/davinci-pdex-plan-net/2020Feb/StructureDefinition-plannet-PractitionerRole.html
Saul Kravitz (May 12 2020 at 16:43):
What do you mean by " zeroed out values from the valueset."?
Saul Kravitz (May 12 2020 at 16:44):
BTW: The Reference implementation still reflects the Ballotted IG, not the CI Build IG.
Lee Surprenant (May 12 2020 at 16:55):
I mean the difference between https://www.hl7.org/fhir/us/core/ValueSet-us-core-provider-role.html and https://www.hl7.org/fhir/us/core/ValueSet-us-core-provider-specialty.html
Lee Surprenant (May 12 2020 at 16:58):
in the balloted version of plan-net, PractitionerRole.code has an extensible binding to https://www.hl7.org/fhir/us/core/ValueSet-us-core-provider-role.html which is the subset of NUCC codes that includes type and classification but not specialty
Lee Surprenant (May 12 2020 at 16:58):
whereas the sample data includes full taxonomy codes (which include specialty)
Lee Surprenant (May 12 2020 at 17:00):
so in the sample data where it has:
"code": [
{
"coding": [
{
"system": "http://nucc.org/provider-taxonomy",
"code": "363LA2200X",
"display": "Adult Health"
}
],
"text": "Adult Health"
}
],
"specialty": [
{
"coding": [
{
"system": "http://nucc.org/provider-taxonomy",
"code": "363LA2200X",
"display": "Adult Health"
}
],
"text": "Adult Health"
}
],
I was expecting the code to be 363L00000X (Nurse Practitioner)
Lee Surprenant (May 12 2020 at 17:01):
anyway, I doubt its worth "fixing" since that latest version of the spec removes that binding
Robert McClure (May 13 2020 at 14:12):
@Lee Surprenant I'm working on some of this with @Saul Kravitz and I'm a bit confused by what you are saying. Can you explain what you mean by type, classification, specialty? NUCC PT does not have subsets designated as "type, classification, specialty."
Lee Surprenant (May 13 2020 at 16:21):
disclosure: i'm not a real subject-matter-expert on NUCC...I'm just translating what our business analyst @Kim M DeMars was telling me.
NUCC codes are hierarhical and composed of three components: type is the first 3 chars, classification is the next 1, then specialty is 5, then X. or something close to that
Lee Surprenant (May 13 2020 at 16:24):
so the code 363LA2200X would be the full taxonomy code...its fully specified and most specific and everyone agrees its a good fit for PractitionerRole.specialty
Lee Surprenant (May 13 2020 at 16:25):
the question is what to do for PractitionerRole.code
Lee Surprenant (May 13 2020 at 16:26):
in the balloted version, there was a valueset of NUCC codes which seems to roughly translate to "higher-level" NUCC codes which include the first 4 chars (type and classification), usually followed by 5 zeroes (no specialty)
Lee Surprenant (May 13 2020 at 16:31):
and in the generated sample data, the PractitionerRoles.code values are not members of that valueset
Last updated: Apr 12 2022 at 19:14 UTC
