FHIR Mapping for Concept Sources not mapped as "SAME-AS"

Implementing Terminology
1

The FHIR2 module only maps Concept sources with SAME-AS map type.

One Challenge we have with the IsantePlus-LIS integration is that most of the Test Concepts do not have the SAME-AS mapping to the LOINC source ie either have BROADER-THAN or NARROWER-THAN .

One work around i tried was to create an interceptor that modifies the Concept object and add the SAME-AS maptype to LOINC before the object is processed by the FHIR2 Concept translator , but that didnt seem very right solution as it ends up modifying concepts.

Can we assume that Concepts that are mapped as BROADER-THAN or NARROWER-THAN to the LOINC Source can also be accurately mapped to a LOINC Coding system “http://loinc.org” in FHIR ??

cc @akanter @ibacher @pmanko

2

Most labs as recorded in our dictionary do not have sufficient specificity to map SAME-AS to a given LOINC code. LIS should know which test they are doing and therefore it would be possible to have a CIEL concept which is SAME-AS the LOINC done by that particular machine/device or laboratory. But you can see how this creates a major problem for EHRs that need to support multiple labs. This is the heart of national (at least in the US) efforts under SHIELD and other programs around laboratory interoperability. I must admit that I have not QA’d the mapping types for the CIEL concepts to LOINC. I would say that if there is a single LOINC associated with the concept that it was probably the closest given the information in the term. I didn’t think that FHIR required SAME-AS only maps, however, simply a single code.

1 Like
3

Thanks @akanter , So we proposed Updating the concept translator to map Concept Sources based on these rules

  1. If there is a SAME-AS mapping, use that
  2. If there is no SAME-AS mapping and there is just one mapping for the concept source, use that one
  3. If there is no SAME-AS mapping and more than one mapping for a given concept source, we don’t output anything for that concept source
1 Like
4

This seems appropriate for LOINC maps which require a single LOINC code. For other FHIR resources where multiple codes are included (for example two NARROWER-THAN SNOMED CT codes) you would need a different approach since in those situations the concept requires post-coordination of the codes and that is a lot more complicated.

1 Like
5

@akanter Thats Right.

From the rules above , we ingore that Concept source and dont do any mapping

6

image
image1382Ă—216 24.3 KB
My point is that for multiple maps, if working OUTSIDE OF LAB, ignoring the coding is not sufficient. However, where you are trying to send lab orders/results, a single LOINC is always appropriate (there is not post-coordination for LOINC codes)

1 Like