OHDSI Home | Forums | Wiki | Github

ICD10PCS standardization - LOINC usage


(Eduard Korchmar) #1

During ongoing procedure consolidation process, 7th level ICD10PCS concepts are going to be integrated in SNOMED procedure subhierarchy, essentially becoming extension of existing SNOMED model. These relations in early version are already written in standardized vocabularies and we plan on releasing a much cleaner version very soon – most likely this week.

However, ICD10PCS includes a chapter for “Measurement and Monitoring” – all codes that start with character ‘4’. Given that SNOMED has poor hierarchy for measurement procedures and SNOMED’s attribute model doesn’t accommodate aspects of measurements well, we consider standardizing them to LOINC vocabulary instead. LOINC has own hierarchy built from LOINC part codes (LP) and could serve as a great target.

My question is: would it be still convenient for vocabulary users if part of the ICD10PCS vocabulary becomes mapped to (or becomes descendants of) LOINC concepts? This will also mean that these concepts’ domain_id will be “Measurement” instead of “Procedure” – this could also possibly break things?

For context:
4A033R1 Measurement of Arterial Saturation, Peripheral, Percutaneous Approach (ICD10PCS)
currently ‘Is a’ 104847001 Oxygen saturation measurement (SNOMED). We could manually extend definitions to 113080007 Oxygen saturation measurement, arterial (SNOMED), since it will be not automatically picked up (any reference to arteries/arterial blood is absent from attributes).

At the same time, 2708-6 Oxygen saturation in Arterial blood (LOINC) has corresponding relation to LOINC classification/atom concepts – which are not currently present in CDM, but could be added.

All in all, relations to LOINC concepts would be easier to implement since it is better defined by attributes.

(Christian Reich) #2

Convenient probably isn’t the right criterion, even though we should be pragmatic about our choices. But first, it should be right.

Measurement of arterial saturation is clearly a Measurement and should be mapped to LOINC. The problem is the percutaneous approach. That is an attribute of a Procedure, not a Measurement. But I agree with you, the Measurement component is more important than the Procedure component. So, LOINC mapping should be all good.

(Vojtech Huser) #3

7th level ICD10PCS concepts are going to be integrated in SNOMED procedure subhierarchy, essentially becoming extension of existing SNOMED model

what vocabulary_id will be used for this extension?

(Eduard Korchmar) #4

For nearest time, all relations (hierarchical and attribute-value) are built directly between concepts from ICD10PCS and SNOMED.

2722336 Division of Cerebral Hemisphere, Open Approach (ICD10PCS) “Is a” 4062872 Open division of brain tissue (SNOMED)
2722336 Division of Cerebral Hemisphere, Open Approach (ICD10PCS) “Has dir proc site” 4157335 Cerebral hemisphere structure (SNOMED)

Eventually we will have a dedicated vocabulary_id for SNOMED extension concepts which will serve as mapping targets for ICD10PCS concepts to separate source-given vs. OMOP-built relations.

(Christian Reich) #5


BTW: We have to bring this approach to the CDM WG and get the community to form an opinion. Don’t think it will be a problem, but this should be an open community.