Skip to main content
Note: TRUST is supporting use cases across the local healthcare sector.

Click here to view our use cases.

TRUST Platform

Using OMOP Concepts

Find the right standard concept for each of your local variables

TRUST Members' Portal (opens in new tab)
Last updated on 20 October 2026

The Singapore harmonised OMOP vocabulary is a curated reference for mapping source clinical terms to OMOP concepts. You can explore it in OMOP Concepts Search. This guide explains how to find and choose the right OMOP concept for each of your variables.

Cleaning your data beforehand and loading the mapped data into OMOP tables are covered in Deploying OMOP.

Where to look, in order

  1. Singapore harmonised vocabulary

    OMOP Concepts Search

  2. ATHENA

    then Usagi for long lists

  3. Feedback to NDCT

    NDCT common mailbox

Before you start

  • You need to be a TRUST Member. Find out how to register.

  • Have a list of your cohort's variables at hand, with a short description and a few example values for each.

  • Log in to the TRUST Members' Portal and open Metadata Catalogue › OMOP Concepts Search.

1. Search OMOP Concepts Search

Start by searching for the source term or code. Use keywords, source codes, abbreviations or alternative clinical terms. If another cohort has already mapped a variable like yours, reuse that concept: it keeps mappings consistent and saves you the work. One search box looks across concept names, descriptions, local variable names and concept IDs at the same time. Some ways to search:

Start with the clinical term

Use plain words, such as "hypertension" rather than "HTN".

Try your own variable name

Another Singapore cohort may already have mapped a variable like yours.

Narrow with filters

Use the Domain and Vocabulary filters when a search returns too many results.

Already have an ID?

Type the concept ID to go straight to that record, then check the name matches.

Hundreds of variables? Your scripts can query the same index through the OMOP Concept Search API.

<internal note - replace> Screenshot of search portal

2. Identify the OMOP domain

Work out which OMOP domain the information belongs to. The domain decides how, and in which table, the data is held in the OMOP Common Data Model. A diagnosis is a Condition; a blood pressure reading is a Measurement.

Examples of OMOP domains

3. Refine and validate the concept

Choose the vocabulary

Filter the results by the vocabulary that suits your domain.

Vocabulary

Use it for

SNOMED

Clinical conditions, most observations, most procedures and some measurements

LOINC

Laboratory tests and measurements

RxNorm / RxNorm Extension

Medications and drug products

UCUM

Units of measurement

Validate the concept

Review each candidate concept against your variable with this checklist.

  1. Concept name

    Does it describe what your variable records?

  2. Clinical meaning

    Does it mean the same as your variable, in context?

  3. Concept class

    Is it the right kind of concept, such as a disorder or a lab test?

  4. Specificity

    Is it as specific as your variable, no broader and no narrower?

  5. Standard

    Is it a standard concept?

  6. Validity

    Is it still valid in the current release?

Reading a concept record

Reading a concept record

  1. Concept ID - The number you record in your mapping. Names can be reworded between releases, so always use the ID

  2. Domain - Decides which OMOP table your data goes into. Check it fits what your variable records: a diagnosis is a Condition, a blood pressure reading is a Measurement.

  3. Vocabulary - The standard the concept comes from, such as SNOMED for clinical findings.

  4. Synonyms - Local variables from Singapore cohorts that are already mapped to this concept. If one looks like your variable, that's a good sign you've found the right concept.

Recording your choice

Record each mapping in the same format the National Data Curation Team uses, so your mappings can be shared and reused. Always record the concept ID, as names can be reworded between releases. Here's an example for a cohort variable called htn_dx.

a. Search "hypertension" and open Essential hypertension.

b. Check the domain is Condition, which fits a diagnosis variable, and that similar variables from other cohorts appear under Synonyms.

c. Grade how close the match is, then add a row to your mapping table.

Match grades

Equal

The concept matches your variable exactly.

Equivalent

The concept mostly matches your variable.

Wider / Narrower

The concept is broader or narrower than your variable.

Example of a concept ID search

Variables with categorical answers, such as country of birth, also get a child table that maps each answer value. See the detailed Using OMOP Concepts guide in the Members' Portal for both formats.

Note the vocabulary version

Concepts can change between releases. Note the version you mapped against (currently v1.1) so you can recheck when a new one comes out.

Get a second pair of eyes

Ask a colleague, ideally a clinician, to check mappings for key variables before you use them.

Check what the variable really records. A self-reported questionnaire answer and a doctor's diagnosis may need different concepts even when the wording is the same. If you're unsure, you can consult the Data Standardisation Technical Working Group (part of the CoP).

More questions, such as when to use a concept set instead of a single concept? See the OMOP FAQs.

4. Search ATHENA, or use Usagi

First try a broader or alternative term, clear the Domain filter, and search your variable name to see how other cohorts handled something similar. If OMOP Concepts Search still has no suitable mapping:

  1. Search ATHENA (opens in new tab), the OHDSI vocabulary browser, for a valid, standard OMOP concept. Apply the same domain, vocabulary and validation checks.

  2. If your source data already uses a coded vocabulary, such as ICD-10, look the code up in ATHENA and follow its Maps to relationship to the standard concept.

  3. For a long list of terms, use Usagi (opens in new tab), an OHDSI tool that suggests concepts in bulk for you to review.

Setting up Usagi is covered in the detailed guide in the Members' Portal.

5. Give feedback on mappings

If a mapping in the Singapore harmonised OMOP vocabulary looks incorrect or questionable, or one is missing, tell NDCT through the NDCT common mailbox. Feedback feeds into future releases of the vocabulary.

Where you can, include:

  • the source term or code

  • the existing mapping, if there is one

  • your proposed alternative

  • why you think it should change

Email the NDCT team

6. Get guidance from NDCT

For complex mapping questions, contact the National Data Curation Team (NDCT) through the same mailbox. Subject to the team's scope and availability, NDCT can advise on terminology mapping and related topics, including ETL design, data standardisation and data quality assessment.