Hi, I am currently mapping and loading ECG data into our staging OMOP implementation at Tufts Medical Center. I have mapping for both electrocardiograph interpretation statements and measurements for the top 3 electrocardiograph companies (Philips' 0A/0B/0C, GE's 12SL, and the GLASGOW criteria) as well as global (ex Heart rate, PR Interval, QTc, etc) and lead level measurements (ex V2 ST J).
The approach I've taken is to define a new ECG vocabulary/concepts for each of manufacturer criteria and measurements, then concept_relationship entries to link those ECG concepts where possible to SNOMED and LOINC concepts. The ETL then preferentially utilize the SNOMED/LOINC concepts where available to load observations and measurements, otherwise uses the ECG vocabulary concepts.
Confirmed ECG statements that are not matched to SNOMED concepts presently go into the note table and typed accordingly so it can be easily identified as ECG notes vs EHR or other clinician notes.
Using the above - each confirmed ECG generates about 500 measurements (490) and observations (~5-10) into our OMOP database.
Happy to share our mapping.