Differences

This shows you the differences between two versions of the page.

--- documentation:oncology:meeting_notes_2019_sep-03 [2019/09/05 17:22] – mgurley
+++ documentation:oncology:meeting_notes_2019_sep-03 [2019/09/05 17:23] (current) – mgurley
@@ Line 7: / Line 7: @@
     * Rimma Belenkaya asked that logic for the SQL ETL should skip any schema with an ambiguous site/histology combination in more than one NAACCR schema.  Not just ambiguous site/histology combinations.  The group agreed to this. See here: https://github.com/OHDSI/OncologyWG/issues/77.
     * We again reviewed the Symposium Vocabulary tasks: https://github.com/OHDSI/OncologyWG/issues?q=is%3Aopen+is%3Aissue+label%3ASymposium+label%3AVocabulary
-      * There was discussion on whether to move Drug Treatment NAACCR item codes into the new 'Regimen' domain.  For the following issue: https://github.com/OHDSI/OncologyWG/issues/69.  Mike Gurley raised that if we do this it leaves us without anything to put into the DRUG_EXPOSURE table.  So we should skip making any entry in DRUG_EXPOSURE for Drug Treatment NAACCR item codes.  Rimma argued if that if this is the case we should not move the DRUG Treatment NAACCR item codes into the Regimen domain. We should make them non-standard Drug domain entries and guide ETL developers to insert them as concept_id=0 but with the NAACCR concept in drug_source_concept_id.  And then create new or reuse high-level entries in the Regimen domain for the entry into Episode.  Rimma and Mike asked to create an entry in CONCEPT_RELATIONSHIP between the Drug Treatment NAACCR item codes and the high-level entries in the Regimen domain.  Christian Reich and Dymytry  argued that the CONCEPT_RELATIONSHIP entry is dirty and should not be done.  Insteady, they suggested ETL developers hardcode the mapping in their ETL logic.  Dymytry suggested re-using the Hemonc.org high-level 'Regimen Modality' concepts as the high-level concepts.
+      * There was discussion on whether to move Drug Treatment NAACCR item codes into the new 'Regimen' domain.  For the following issue: https://github.com/OHDSI/OncologyWG/issues/69.  Mike Gurley raised that if we do this it leaves us without anything to put into the DRUG_EXPOSURE table.  So we should skip making any entry in DRUG_EXPOSURE for Drug Treatment NAACCR item codes.  Rimma argued that if this is the case we should not move the DRUG Treatment NAACCR item codes into the Regimen domain. We should make them non-standard Drug domain entries and guide ETL developers to insert them as concept_id=0 but with the NAACCR concept in drug_source_concept_id.  And then create new or reuse high-level entries in the Regimen domain for the entry into Episode.  Rimma and Mike asked to create an entry in CONCEPT_RELATIONSHIP between the Drug Treatment NAACCR item codes and the high-level entries in the Regimen domain.  Christian Reich and Dymytry  argued that the CONCEPT_RELATIONSHIP entry is dirty and should not be done.  Insteady, they suggested ETL developers hardcode the mapping in their ETL logic.  Dymytry suggested re-using the Hemonc.org high-level 'Regimen Modality' concepts as the high-level concepts.
       * Dymytry reported that all of the symposium vocabulary tasks would be fixed and published by 9/5/2019.
       * Jermey Warner gave more detail and explanation about ambiguities in Hemonc.org pointed out by Christian.  See here: https://github.com/OHDSI/OncologyWG/issues/70#issuecomment-527222143