Handle Unknown/Other Data in ValueSets

[torstees]

Requester information

• Name: Eric Torstenson

• Affiliations: VUMC/AnVIL

Request Details

• Purpose: Other/Unknown/Declined are not part of standard code systems, including Race, Ethnicity and likely others, but are important indicators that may have meaning beyond simply being missing.

Race/Ethnicity=Other is definitely meaningful, especially for older datasets where the check boxes may not have been particularly complete.

Declined...

Unknown may or may not be informative. If it isn't informative, then it may not have any place in the FHIR representation but it may be present in the datasets from which the data is being pulled.

• Who it benefits: For Race, this is a standard issue that will be encountered by any self-reported race where the options are limited. Unknown and Declined are likely common as well, depending on the nature of the source data.

• Use case: While ingesting demographic data, users may encounter values that don't actually represent tangible race or ethnicities

[torstees]

Meen suggested using a subset of FHIR NullFlavor value sets: https://www.hl7.org/fhir/v3/NullFlavor/cs.html

This covers Unknown and Other. Declined could be encoded as "ASKU/Asked but Unknown"

This will require the codes decided upon to be added to the ValueSet used for the Race/Ethnicity extension

[RobertJCarroll]

I don't think we want to change the US Core extension: https://hl7.org/fhir/us/core/StructureDefinition-us-core-race.html

From reading about the ValueSet, https://hl7.org/fhir/us/core/ValueSet-omb-race-category.html , it looks like there are some mixed messages regarding the inclusion of other codes... not sure how to reconcile this.

My initial reaction is that perhaps it just should be left empty if there is no answer that maps to one of the 5 omb codes. If it is truly unknown or no response, that's fine. But if it's actually "Other" that does mean it does not have an OMB Code.