Skip to main content

Vocabulary Browser

The Vocabulary Browser is your primary tool for exploring the OMOP standardized vocabularies — the controlled terminologies that classify every clinical event recorded in an OMOP CDM database. Understanding and navigating these vocabularies is fundamental to OMOP-based research: every condition, drug, measurement, procedure, observation, and device in the CDM is represented by a concept with a unique integer identifier.

Parthenon's Vocabulary Browser combines traditional full-text search with AI-powered semantic search, letting you find concepts not just by keyword matching but by clinical meaning.

What is an OMOP Concept?

An OMOP concept is a single entry in the standardized vocabulary with the following attributes:

AttributeDescriptionExample
Concept IDUnique integer identifier201826
Concept NameHuman-readable labelType 2 diabetes mellitus
DomainClinical area the concept belongs toCondition
VocabularySource terminology systemSNOMED
Concept ClassGranularity level within the vocabularyClinical Finding
Standard ConceptWhether this is the OMOP standard for its domainStandard (S)
Concept CodeThe code in the source vocabulary44054006
Valid Start/EndValidity window for the concept2002-01-31 to 2099-12-31
Invalid ReasonIf deprecated, the reason (D=deleted, U=upgraded)NULL (valid)

The OMOP vocabulary contains over 7.2 million concepts drawn from dozens of vocabularies including SNOMED CT, RxNorm, RxNorm Extension, ICD-10-CM, ICD-10-PCS, LOINC, CPT4, HCPCS, ATC, MedDRA, UCUM, and many others.

Navigate to Vocabulary > Browse Concepts in the top navigation bar. The browser opens with an empty search field and filter panel. A source selector in the header indicates which data source's vocabulary is being searched — different sources may have different vocabulary versions loaded.

Searching Concepts

Type a search term in the search box and press Enter or click the search icon. The search uses full-text indexing on concept names, concept codes, and concept synonyms.

Search examples:

QueryFinds
diabetesAll concepts with "diabetes" in the name
201826The concept with ID 201826 directly
E11.9ICD-10-CM code E11.9 (Type 2 diabetes mellitus, without complications)
metforminDrug concepts containing "metformin"
HbA1cMeasurement concepts for hemoglobin A1c

The search supports partial matches and synonym lookups. For example, searching "T2DM" returns "Type 2 diabetes mellitus" because "T2DM" is registered as a concept synonym. Similarly, searching "heart attack" finds "Myocardial infarction" through synonym relationships.

Search by Concept ID

If you know the exact concept ID, type it directly in the search box. Parthenon recognizes numeric input and performs a direct ID lookup, which is faster than text search and always returns an exact match.

When an AI provider is configured (see Chapter 1: Introduction), the search box also supports semantic search. Toggle the AI search mode using the brain icon next to the search field.

In semantic mode, you can describe clinical concepts in natural language:

Natural Language QueryConcepts Found
"medications for lowering blood sugar"Metformin, Glipizide, Sitagliptin, Insulin, etc.
"lab tests for kidney function"Creatinine, eGFR, BUN, Cystatin C, etc.
"heart failure with reduced ejection fraction"Systolic heart failure, HFrEF, related SNOMED concepts
"drugs that cause QT prolongation"Amiodarone, Sotalol, Haloperidol, etc.

Semantic search uses the MedGemma model (or your configured AI provider) to understand the clinical intent behind your query and match it against concept descriptions, hierarchical relationships, and medical knowledge.

Semantic Search Availability

AI semantic search requires an active AI provider configured in Admin > AI Providers. If no provider is active, the semantic search toggle is disabled and only text search is available. The default Ollama/MedGemma configuration keeps all queries on-premises — no data leaves your network.

Filtering Results

The filter panel on the left side of the browser allows you to narrow search results by multiple criteria. Filters are applied additively (AND logic):

Domain Filter

Select one or more clinical domains to restrict results:

DomainContainsExample Concepts
ConditionDiagnoses, diseases, symptomsType 2 diabetes, Pneumonia, Chest pain
DrugMedications, ingredients, clinical drugsMetformin 500mg tablet, Aspirin
MeasurementLab tests, vital signs, assessmentsHbA1c, Blood pressure systolic, BMI
ProcedureSurgical and diagnostic proceduresColonoscopy, Coronary angiography
ObservationClinical observations, social historyTobacco use, Blood type
DeviceMedical devicesPacemaker, Insulin pump
VisitEncounter typesInpatient, Outpatient, ER visit
SpecimenBiological specimensBlood, Urine, Tissue biopsy
UnitUnits of measurementmg/dL, mmol/L, kg

Vocabulary Filter

Select specific source vocabularies:

  • SNOMED — Primary vocabulary for conditions, procedures, observations
  • RxNorm / RxNorm Extension — Drug ingredients, clinical drugs, branded drugs
  • ICD-10-CM — Diagnosis codes (US clinical modification)
  • ICD-10-PCS — Procedure codes (US procedure coding system)
  • LOINC — Laboratory and clinical observations
  • CPT4 / HCPCS — Procedure and service codes (US billing)
  • ATC — Anatomical Therapeutic Chemical drug classification
  • MedDRA — Medical Dictionary for Regulatory Activities
  • UCUM — Unified Code for Units of Measure

Standard Concept Filter

Filter by standard classification:

ValueMeaningWhen to Use
StandardThe OMOP standard concept for its domainDefault choice for cohort criteria — these are what the CDM stores
ClassificationA grouping concept used for hierarchyUseful for navigating the vocabulary tree
Non-StandardSource concept that maps to a standardUse when you need to find specific source codes (ICD-10, CPT)

Concept Class Filter

Filter by granularity within a vocabulary. Examples:

  • SNOMED: Clinical Finding, Procedure, Organism, Body Structure
  • RxNorm: Ingredient, Clinical Drug, Branded Drug, Clinical Drug Form
  • LOINC: Lab Test, Clinical Observation, Survey

Validity Filter

  • Valid only (default) — Excludes deprecated and invalid concepts
  • Include invalid — Shows all concepts including deprecated ones
Always Use Standard Concepts for Cohort Criteria

The OMOP CDM stores clinical data using standard concepts. When building cohort inclusion criteria, always select Standard concepts (SNOMED for conditions, RxNorm for drugs, LOINC for measurements). Non-standard concepts (ICD-10, CPT) appear in source fields but are not used for primary clinical data queries. The vocabulary browser highlights standard concepts with a blue "Standard" badge.

Concept Detail View

Click any concept in the search results to open the full concept detail view. This panel provides comprehensive information organized into several sections:

Metadata Header

Displays the concept ID, name, domain, vocabulary, class, standard flag, concept code, and validity dates at a glance.

Hierarchy — Ancestors

Shows broader concepts above the selected concept in the vocabulary hierarchy. Navigate upward to understand how a concept fits into the overall classification.

Example hierarchy for "Type 2 diabetes mellitus" (201826):

Endocrine disease
└── Disorder of glucose metabolism
└── Diabetes mellitus
└── Type 2 diabetes mellitus ← selected

Click any ancestor to navigate to that concept's detail view.

Hierarchy — Descendants

Shows narrower, more specific concepts below the selected concept. This is critically important for understanding what a concept's "Include descendants" flag will capture in a concept set.

Example descendants of "Type 2 diabetes mellitus" (201826):

  • Type 2 diabetes mellitus with ophthalmic complication
  • Type 2 diabetes mellitus with renal complication
  • Type 2 diabetes mellitus with neurological complication
  • Type 2 diabetes mellitus with peripheral circulatory complication
  • ... (potentially hundreds more)
Preview Descendants Before Building Concept Sets

Before adding a concept to a concept set with "Include descendants" checked, always review the descendants list in the detail view. You may find that descendants include unintended subtypes. For example, "Diabetes mellitus" includes both Type 1 and Type 2 — if you only want Type 2, start from concept 201826 instead of the parent concept.

Relationships

Displays all concept relationships defined in concept_relationship:

RelationshipDirectionExample
Is aUpward"Type 2 diabetes mellitus" Is a "Diabetes mellitus"
Maps toStandard mappingICD-10-CM E11.9 Maps to SNOMED 201826
Has ingredientDrug composition"Metformin 500mg tablet" Has ingredient "Metformin"
Has finding siteAnatomy link"Pneumonia" Has finding site "Lung structure"
SubsumesDownward"Diabetes mellitus" Subsumes "Type 2 diabetes mellitus"

Synonyms

Lists alternative names for the concept from concept_synonym. Synonyms are searchable — you can find concepts by any registered synonym.

Source Concepts

For a standard concept, this section shows all non-standard concepts that map to it. This is useful for understanding which source codes (ICD-10, CPT, etc.) resolve to this standard concept during ETL.

Example: Standard concept "Type 2 diabetes mellitus" (201826) is the target of mappings from:

  • ICD-10-CM E11.9 (Type 2 diabetes mellitus, without complications)
  • ICD-10-CM E11 (Type 2 diabetes mellitus)
  • ICD-9-CM 250.00 (Diabetes mellitus without mention of complication)
  • Read V2 C109. (Non-insulin dependent diabetes mellitus)

Concept Comparison Tool

Parthenon provides a side-by-side concept comparison tool at Vocabulary > Compare Concepts (or navigable via the URL path /vocabulary/compare). This tool is useful when you need to decide between similar concepts for a concept set.

Using the Comparison Tool

  1. Navigate to Vocabulary > Compare Concepts.
  2. Search and select two or more concepts to compare.
  3. The comparison view displays the concepts in columns with rows for each attribute:
    • Metadata (ID, name, domain, vocabulary, class, standard status)
    • Ancestor count and hierarchy depth
    • Descendant count
    • Relationship count
    • Synonym count
    • Source concept mapping count

This helps you make informed decisions about which concept to use. For example, comparing "Diabetes mellitus type 2" across SNOMED and ICD-10 shows which has more descendants, more source mappings, and which is the standard concept.

Standard vs. Source Concepts

Understanding the distinction between standard and source (non-standard) concepts is essential for effective use of the OMOP CDM:

Standard Concepts

  • One vocabulary is designated as "standard" for each domain (e.g., SNOMED for conditions, RxNorm for drugs, LOINC for measurements)
  • Clinical data in the CDM is stored using standard concept IDs in the primary *_concept_id columns
  • Always use standard concepts when building cohort criteria — they are what the CDM queries against

Source Concepts

  • Source concepts preserve the original terminology from the source data (ICD-10, CPT, local codes)
  • They are stored in *_source_concept_id columns and *_source_value text fields
  • Source concepts have "Maps to" relationships pointing to their corresponding standard concepts
  • Use source concepts only when you need exact source code matching (rare in most research)

Classification Concepts

  • These are grouping nodes in vocabulary hierarchies (e.g., ATC drug classes)
  • They are not directly used in CDM data columns
  • They are useful for navigating the hierarchy and understanding concept relationships

Vocabulary Versions

The vocabulary version loaded in each data source is shown in the browser header when a source is selected. Vocabulary releases from Athena are dated (e.g., v5.0 2026-02-25) and periodically include:

  • New concepts added to source vocabularies
  • Deprecated concepts marked as invalid with invalid_reason = 'D' (deleted) or 'U' (upgraded/replaced)
  • Reclassified concepts moved between domains or classes
  • New relationships linking concepts across vocabularies
  • Updated synonyms in multiple languages
Review Concept Sets After Vocabulary Upgrades

When your administrator uploads a new vocabulary version, existing concept sets should be reviewed. A concept that was standard may have been deprecated, or new descendants may have been added that change the scope of a concept set with "Include descendants" enabled. Use the concept set Preview feature to verify resolved concept IDs after any vocabulary update.

Checking Vocabulary Version

To see which vocabulary version a source is using:

  1. Open Data Sources and click on the source
  2. The vocabulary version is listed under the vocabulary daimon metadata
  3. Alternatively, in the Vocabulary Browser, the version is shown in the header bar next to the source name

Practical Workflows

Finding the Right Concept for a Condition

  1. Search for the condition name (e.g., "myocardial infarction")
  2. Filter by Domain: Condition and Standard Concept: Standard
  3. Review the top results — look for the broadest concept that captures your intent
  4. Open the concept detail view and review descendants to understand scope
  5. Add to a concept set with appropriate flags (see Chapter 4: Concept Sets)

Finding Drug Concepts at the Ingredient Level

  1. Search for the drug name (e.g., "metformin")
  2. Filter by Domain: Drug and Concept Class: Ingredient
  3. The ingredient concept captures all formulations, strengths, and brands
  4. Use "Include descendants" in your concept set to automatically include all clinical drug forms

Mapping Source Codes to Standard Concepts

  1. Search for the source code directly (e.g., "E11.9" for ICD-10-CM)
  2. Open the concept detail view
  3. Look at the Relationships section for "Maps to" relationships
  4. The target of the "Maps to" relationship is the standard concept you should use

Next Steps

Now that you understand how to navigate vocabularies and find concepts, proceed to Chapter 4: Concept Sets to learn how to assemble concepts into reusable lists that power cohort definitions and analyses.