xDev Asia

What is Concept, Standard Concept vs Source Concept vs Classification Concept, concept_id vs source_value vs source_concept_id, Domain, Vocabulary, Concept Class, and how to look up on Athena.

Concept System — Source, Standard and Classification

Introduction

If OMOP CDM is a house, then Concept is the brick — everything is built from Concept. Understanding the Concept is the key to understanding CDM as a whole.

This article will explain: What is Concept? Is Standard vs Source Concept different? How to look up Concept on Athena? And the "holy trinity" concept_id / source_value / source_concept_id How does it work?

1. What is concept?

1.1. Definition

Concept = a unique medical "concept", represented by a number called concept_id.

Everything you encounter in medicine has a corresponding Concept:

Medical Entity	concept_id	concept_name	Vocabulary
Women	8532	Female	Gender
Type 2 diabetes	201826	Type 2 diabetes mellitus	SNOMED
Metformin 500mg tablets	1503297	metformin 500 MG Oral Tablet	RxNorm
HbA1c test	3004410	Hemoglobin A1c/Hemoglobin.total in Blood	LOINC
Percent unit	8554	percent	UCUM
Outpatient examination	9202	Outpatient Visit	Visit
Data from EHR	32817	EHR	Type Concept

1.2. CONCEPT table

Each Concept is saved as a row in the table CONCEPT:

SELECT *
FROM concept
WHERE concept_id = 201826;

Column	Value	Description
`concept_id`	201826	Unique ID
`concept_name`	Type 2 diabetes mellitus	Display name
`domain_id`	Conditions	Which domain does it belong to
`vocabulary_id`	SNOMED	Which vocabulary word
`concept_class_id`	Clinical Findings	Concept type
`standard_concept`	S	S = Standard
`concept_code`	44054006	Original code in vocabulary
`valid_start_date`	1970-01-01	Effective start date
`valid_end_date`	2099-12-31	Expiry date
`invalid_reason`	NULL	Invalid reason

2. Standard Concept vs Source Concept vs Classification

2.1. Three types of Concepts

  standard_concept column:
  ┌─────────────────────────────────────────────────────────────┐
  │                                                             │
  │  'S' = STANDARD CONCEPT                                    │
  │  → Dùng trong *_concept_id columns                          │
  │  → Là "đại diện chính thức" cho một khái niệm              │
  │  → VD: SNOMED 201826 "Type 2 diabetes mellitus"             │
  │                                                             │
  │  'C' = CLASSIFICATION CONCEPT                               │
  │  → Dùng để phân nhóm/phân cấp                              │
  │  → KHÔNG dùng trực tiếp trong clinical tables               │
  │  → VD: SNOMED parent concepts                               │
  │                                                             │
  │  NULL = NON-STANDARD (SOURCE) CONCEPT                       │
  │  → Mã từ vocabulary gốc                                     │
  │  → Lưu trong *_source_concept_id columns                    │
  │  → VD: ICD-10 E11 "Type 2 diabetes mellitus"               │
  │                                                             │
  └─────────────────────────────────────────────────────────────┘

2.2. Specific example: "Type 2 diabetes"

  ICD-10-CM 'E11'                         SNOMED 44054006
  ┌──────────────────────┐    Maps to     ┌──────────────────────────┐
  │ concept_id = 443238  │ ──────────→    │ concept_id = 201826      │
  │ standard_concept=NULL│                │ standard_concept = 'S'    │
  │ vocabulary=ICD10CM   │                │ vocabulary = SNOMED       │
  │ (Source Concept)     │                │ (Standard Concept)        │
  └──────────────────────┘                └──────────────────────────┘

  Trong bảng CONDITION_OCCURRENCE:
  ┌────────────────────────────────────────────────────────────────────┐
  │ condition_concept_id        = 201826    ← Standard (SNOMED)       │
  │ condition_source_value      = 'E11'     ← Text gốc từ HIS        │
  │ condition_source_concept_id = 443238    ← Source Concept (ICD-10) │
  └────────────────────────────────────────────────────────────────────┘

2.3. The holy trinity

Most clinical tables have 3 columns for each concept field:

Column	Purpose	Value
`*_concept_id`	Analysis — standard concept	Standard Concept ID (S)
`*_source_value`	Trace — original text value	Original text (eg: "E11", "Glucophage")
`*_source_concept_id`	Reverse Mapping — original concept	Non-standard Concept ID

  ┌── Dùng để phân tích (SELECT, GROUP BY, JOIN)
  │
  condition_concept_id = 201826  ← Standard SNOMED
                                                    ├── Truy nguyên nguồn gốc
  condition_source_value = 'E11'  ← Text gốc HIS   │
  condition_source_concept_id = 443238  ← ICD-10 ───┘

3. Domain — Which table does Concept belong to?

3.1. Main Domains

Domain	Destination table	Example
Conditions	CONDITION_OCCURRENCE	Diabetes, pneumonia
Drugs	DRUG_EXPOSURE	Metformin, Amoxicillin
Procedure	PROCEDURE_OCCURRENCE	Endoscopy, surgery
Measurement	MEASUREMENT	HbA1c, blood pressure, BMI
Observation	OBSERVATION	Smoking, family history
Device	DEVICE_EXPOSURE	Stents, pacemakers
Specimen	SPECIMEN	Blood samples, tissue samples
Visit	VISIT_OCCURRENCE	Outpatient examination, hospitalization
Gender	PERSON	Male, Female
Race	PERSON	Asian, White
Type Concept	All tables	EHR, Claim, Lab
Unit	MEASUREMENT	mg/dL, %, mmHg
Route	DRUG_EXPOSURE	Oral, IV, Topical

3.2. Why is Domain important?

The domain decides which table the record is in. Here are the core ETL rules:

  Dữ liệu nguồn: "ICD-10: Z87.891 — History of nicotine dependence"
  
  Bước 1: Tra cứu ICD-10 Z87.891 trên Athena
  Bước 2: Tìm Standard Concept → maps to SNOMED concept
  Bước 3: Standard Concept thuộc domain "Observation"
  Bước 4: Lưu vào bảng OBSERVATION (không phải CONDITION!)
  
  ⚠️ Dù ICD-10 thường gắn với Condition domain,
  nhưng "History of" map sang Observation domain

4. Vocabulary — Concept Origin

4.1. Important Vocabularies

  ┌──────────────────────────────────────────────────────────────────┐
  │  VOCABULARY CHÍNH TRONG OMOP CDM                                 │
  │                                                                   │
  │  ┌─────────────┐   Conditions (diagnoses, symptoms)              │
  │  │  SNOMED CT  │   → Standard vocabulary cho Condition domain     │
  │  └─────────────┘                                                  │
  │                                                                   │
  │  ┌─────────────┐   Drugs (medications, vaccines)                  │
  │  │  RxNorm     │   → Standard vocabulary cho Drug domain          │
  │  └─────────────┘                                                  │
  │                                                                   │
  │  ┌─────────────┐   Measurements (lab tests, vitals)              │
  │  │   LOINC     │   → Standard vocabulary cho Measurement domain  │
  │  └─────────────┘                                                  │
  │                                                                   │
  │  ┌─────────────┐   Procedures                                     │
  │  │  SNOMED CT  │   → Standard vocabulary cho Procedure domain     │
  │  │  CPT4       │   → US-specific procedures                       │
  │  └─────────────┘                                                  │
  │                                                                   │
  │  SOURCE VOCABULARIES (Non-standard, cần mapping):                 │
  │  ICD-10-CM/VN, ICD-9-CM, NDC, HCPCS, ATC, Read, MedDRA         │
  └──────────────────────────────────────────────────────────────────┘

4.2. Standard vs Non-standard Vocabulary

Type	Example	standard_concept	Used in column
Standard	SNOMED CT, RxNorm, LOINC	'S'	`*_concept_id`
Non-standard	ICD-10, NDC, ATC, MedDRA	NULL	`*_source_concept_id`
Classification	SNOMED hierarchy nodes	'C'	Used in hierarchical queries

4.3. Mapping process

  HIS: "E11" (ICD-10)
       │
       │  Tra bảng CONCEPT_RELATIONSHIP
       │  relationship_id = 'Maps to'
       ↓
  Source Concept: 443238 (ICD-10 E11)
       │
       │  Maps to
       ↓
  Standard Concept: 201826 (SNOMED Type 2 diabetes mellitus)

SQL lookup mapping:

-- Tìm Standard Concept từ ICD-10 code 'E11'
SELECT
    c1.concept_id   AS source_concept_id,
    c1.concept_name AS source_name,
    c1.vocabulary_id AS source_vocab,
    c2.concept_id   AS standard_concept_id,
    c2.concept_name AS standard_name,
    c2.vocabulary_id AS standard_vocab
FROM concept c1
JOIN concept_relationship cr
    ON c1.concept_id = cr.concept_id_1
    AND cr.relationship_id = 'Maps to'
JOIN concept c2
    ON cr.concept_id_2 = c2.concept_id
    AND c2.standard_concept = 'S'
WHERE c1.concept_code = 'E11'
  AND c1.vocabulary_id = 'ICD10CM';

5. Concept Class — Detailed classification

Each Concept belongs to a Concept Class that indicates which level of vocabulary it belongs to:

Domain	Concept Class	Example
Conditions	Clinical Findings	Type 2 diabetes mellitus
Drugs	Ingredient	Metformin
Drugs	Clinical Drug	Metformin 500 MG Oral Tablet
Drugs	Branded Drug	Glucophage 500 MG Oral Tablet
Measurement	Lab Test	Hemoglobin A1c
Measurement	Clinical Observation	Bodyweight
Procedure	Procedure	Coronary artery bypass grafting
Observation	Context-dependent	History of clinical findings

Drug Concept Class — Particularly important

RxNorm organizes drugs by level:

  INGREDIENT (hoạt chất)
  └── Metformin (concept_id = 1503297)
       │
       ├── CLINICAL DRUG FORM
       │   └── Metformin Oral Tablet
       │        │
       │        ├── CLINICAL DRUG (hoạt chất + liều + dạng)
       │        │   └── Metformin 500 MG Oral Tablet
       │        │        │
       │        │        └── BRANDED DRUG (tên thương mại)
       │        │            └── Glucophage 500 MG Oral Tablet
       │        │
       │        └── CLINICAL DRUG
       │            └── Metformin 1000 MG Oral Tablet
       │
       └── CLINICAL DRUG FORM
           └── Metformin Extended Release Oral Tablet

6. Look up Concept on Athena

6.1. What is Athena?

Athena (https://athena.ohdsi.org/) is a free web tool to look up Standardized Vocabularies.

6.2. Search instructions

Step 1: Access https://athena.ohdsi.org/ (need to create free account)

Step 2: Enter a keyword, for example "diabetes type 2"

Step 3: Filter results:

Standard Concept: select "Standard" to see only standard concepts
Domain: select "Condition" if searching for a disease
Vocabulary: select "SNOMED" or "ICD10CM"

Step 4: See concept details:

Concept ID, name, class, domain, vocabulary
Relationships tab: see "Maps to", "Is a", "Has component"
Hierarchy tab: see father/child concept

6.3. Popular lookup example

  Tìm bệnh tiểu đường type 2:
  → Search: "type 2 diabetes"
  → Filter: Domain=Condition, Standard=Standard
  → Kết quả: concept_id=201826, SNOMED "Type 2 diabetes mellitus"

  Tìm thuốc Metformin:
  → Search: "metformin"
  → Filter: Domain=Drug, Concept Class=Ingredient
  → Kết quả: concept_id=1503297, RxNorm "metformin"

  Tìm xét nghiệm HbA1c:
  → Search: "hemoglobin a1c"
  → Filter: Domain=Measurement, Standard=Standard
  → Kết quả: concept_id=3004410, LOINC "Hemoglobin A1c/Hemoglobin.total"

7. concept_id = 0 — When unable to map

When source data cannot be mapped to Standard Concept:

-- Một mã thuốc nội bộ bệnh viện không có trong RxNorm
drug_concept_id        = 0              -- Không map được!
drug_source_value      = 'THUOC_BV_001' -- Vẫn giữ mã gốc
drug_source_concept_id = 0              -- Cũng không có source concept

This is completely valid. OMOP CDM allows it concept_id = 0, but the goal is to minimize the number of records with concept_id = 0 by:

Use SOURCE_TO_CONCEPT_MAP for custom mappings
Use Usagi tool for semi-automatic mapping
Ask the community to add new concepts to the vocabulary

8. Practice: Read a CDM record

For the following record in CONDITION_OCCURRENCE:

condition_occurrence_id    = 50001
person_id                  = 12345
condition_concept_id       = 201826
condition_start_date       = '2024-06-10'
condition_start_datetime   = '2024-06-10 09:30:00'
condition_end_date         = NULL
condition_end_datetime     = NULL
condition_type_concept_id  = 32817
condition_status_concept_id = 32902
provider_id                = 5001
visit_occurrence_id        = V001
condition_source_value     = 'E11'
condition_source_concept_id = 443238
condition_status_source_value = 'admitting'

Decoding:

School	Value	Meaning
condition_concept_id = 201826	SNOMED "Type 2 diabetes mellitus"	Standard diagnosis
condition_type_concept_id = 32817	"EHR"	Data from EMR system
condition_status_concept_id = 32902	"Primary diagnosis"	Primary Diagnosis
condition_source_value = 'E11'	ICD-10-CM original code	Original code from HIS
condition_source_concept_id = 443238	ICD-10-CM concept for "E11"	Original concept
provider_id = 5001	Diagnosis Doctor	Look up table PROVIDER
visit_occurrence_id = V001	Outpatient visits	Look up the table VISIT_OCCURRENCE

Summary

In this article, you have learned:

Concept = medical concept, represented by unique concept_id
3 types of Concept: Standard (S), Classification (C), Source (NULL)
Triple columns: *_concept_id / *_source_value / *_source_concept_id
Domain decides which table the record is in
Vocabulary: SNOMED (Condition), RxNorm (Drug), LOINC (Measurement)
Concept Class: detailed classification (Ingredient vs Clinical Drug...)
Athena: free concept lookup tool
concept_id = 0: valid when map is not possible

Next post: We'll start by exploring the first table — PERSON — where patient demographic information is stored.

Lesson 3: Understanding the Concept — The heart of OMOP CDM