search

OHDSI / OncologyWG

Oncology Working Group Repository
https://ohdsi.github.io/OncologyWG
Apache License 2.0
56 stars 24 forks source link

For source metastases records that are conditions: map to condition, measurement, or both? #556

Open rtmill opened 1 year ago

rtmill commented 1 year ago

To help with context, I did more of a dive into current metastasis mappings. Whatever decisions we end up making, I believe it's clear the current state is not consistent.

Initial thoughts: Even if we synchronize these vocabularies into a singular hierarchical form, we're still going to need an additional qualifier on other types of concepts that are indicative of the existence of metastasis if we are going to facilitate comprehensive phenotyping. To that end, one solution would be to also create relationships to those concepts:

e.g. All concepts under: https://athena.ohdsi.org/search-terms/terms/1635142

Anyway...

With the caveat of defining the "source metastasis condition concepts" as :

select concept_id, concept_name, domain_id, vocabulary_id, concept_code, standard_concept
     FROM concept
     WHERE Concept_id in (
        SELECT concept_id_2
        FROM concept_relationship cr
        WHERE concept_id_1 = 44498002 -- ICDO Neoplasm, metastatic
        AND relationship_id = 'Histology of ICDO'
        )
     union 
        select concept_id, concept_name, domain_id, vocabulary_id, concept_code, standard_concept
        from concept
        where vocabulary_id in ('ICD10CM', 'ICD9CM')
        and lower(concept_name) like 'secondary%'
        and lower(concept_name) like '%neoplasm%'
    union 
        select concept_id, concept_name, domain_id, vocabulary_id, concept_code, standard_concept
        from concept
        where concept_id in (
            select descendant_concept_id
            from concept_ancestor ca 
            where ancestor_concept_id = 432851 -- SNOMED: secondary malignant neoplastic disease
        )

<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:x="urn:schemas-microsoft-com:office:excel" xmlns="http://www.w3.org/TR/REC-html40">

1) Counts source concepts, and the corresponding domain mapping, by vocabulary Takeaway: There is significant inconsistency as to which domain(s) these are mapped to.

vocabulary_id source_concept_count mapped_to_condition mapped_to_measurement
ICD10CM 98 46 52
ICD9CM 43 31 12
ICDO3 229 5 224
Nebraska Lexicon 7 6 0
SNOMED 847 847 0

2) Counts source concepts, and the corresponding vocabulary mapping Takeaway: There is significant inconsistency as to which vocabularies these are mapped to.

vocabulary_id concept_count map_to_snomed map_to_cancermodifier map_to_icdo3
ICD10CM 59 46 52 0
ICD9CM 31 31 12 0
ICDO3 229 0 224 5
Nebraska Lexicon 7 0 0 0
SNOMED 847 847 0 0

Concept level mapping counts attached: mets_source_mapping_compare.csv