Notes on OMOP

OMOP is OHDSI's standardized database schema that allows for all the commonsensical benisons of standardization that you can think of (e.g. write analysis once, run anywhere). The key idea is that OMOP separates events from meaning: events live in the clinical tables (condition_occurence, drug_exposure) and meanings live in the concepts table.

Concept Standardization and IDs

OMOP Standardizes various vocabularies. So you can have a Concept ID in OMOP that maps to some Standard Concept ID. What is a Standard? In the case of conditions (the "Domain" in Athena), for example, it's SNOMED.

OMOP will map various vocabularies $V_n$ to a standard vocabulary $V_s$.

So when searching Athena for "Atrial fibrillation", you will see

• Concept ID 44821957
• Standard Concept ID 313217 because it maps to a standard that is SNOMED.

Querying

Generally speaking,

1. Start with person - who
2. Use observation_period to define when
3. Use condition_occurrence to define what happened
4. Use death to identify mortality outcomes
5. Use concept to make coded values interpretable

For cleanliness you may want to split things into concept info (from the concepts and concept_ancestor tables for example) and the evidence info (from condition_occurrence for example).

Resources

• Excellent overview of OMOP and Athena with some quick exercises.