ICD10CM upload / update

Prerequisites

• Basic knowledge of the OMOP representation of the ICD10CM vocabulary
• Schema DevV5 with copies of tables concept, concept_relationship and concept_synonym from ProdV5, fully indexed.
• Schema UMLS
• SNOMED must be loaded first
• Working directory, e.g. dev_icd10cm

Sequence of actions

1. Download the latest ICD-10-CM version from https://ftp.cdc.gov/pub/Health_Statistics/NCHS/Publications/ICD10CM/
2. Unzip the file ICD10CM_FYYYYY_code_descriptions.zip, extract icd10cm_order_YYYY.txt and rename to icd10cm.txt
3. Run create_source_tables.sql
4. Run in devv5 (with fresh vocabulary date and version, e.g. according "The 2018 ICD-10-CM codes are to be used from October 1, 2017 through September 30, 2018" so 20171001):

SELECT sources.load_input_tables('ICD10CM',TO_DATE('20170428','YYYYMMDD'),'ICD10CM FY2017 code descriptions');

5. Run FULL FastRecreate:

SELECT devv5.FastRecreateSchema(main_schema_name=>'devv5', include_concept_ancestor=>false, include_deprecated_rels=>true, include_synonyms=>true);

6. As described in the "manual_work" folder, upload concept_manual.csv and concept_relationship_manual.csv into eponymous tables, which exist by default in the dev schema after the FastRecreate. If you already have manual staging tables, obligatory create backups of them (e.g. concept_relationship_manual_backup_ddmmyy, concept_manual_backup_ddmmyy)
7. Run load_stage.sql for the first time to define problems in mapping. Note: Load_stage generates list of the relationships that need to be checked and modified by the medical coder and after uploading this data to server load_stage continues making proper relationships using this manually created table
8. Perform manual work described in manual_work folder
9. Run load_stage.sql for the second time to refresh ICD10CM
10. Run generic_update:

SELECT devv5.GenericUpdate();

11. Run manual_checks_after_generic.sql
12. If no problems, enjoy!