LOINC vs SNOMED CT: Key Differences Explained

In the healthcare informatics standards ecosystem, LOINC and SNOMED CT are two of the most widely used and referenced terminologies. However, their purposes are fundamentally different and complementary. Understanding when and how to use each one is essential for any clinical informatics professional or healthcare system integrator working with lab data.

Origins and governance

LOINC: the language of observations

LOINC (Logical Observation Identifiers Names and Codes) was created in 1994 by the Regenstrief Institute at Indiana University. Its goal was to solve a concrete problem: each laboratory used its own internal codes to identify tests, making automated comparison and exchange of results between institutions impossible.

LOINC is maintained as a free, open standard, managed by the Regenstrief Institute with semiannual updates. In 2026, the LOINC database contains over 100,000 terms covering laboratory observations, vital signs, clinical questionnaires, documents, and imaging results.

Each LOINC code is defined by six axes: component (analyte), property (measurement type), timing, system (specimen), scale, and method. For example, code 2345-7 specifically identifies glucose, measured as mass concentration, at a point in time, in serum/plasma, with a quantitative scale, and no specified method.

SNOMED CT: the clinical ontology

SNOMED CT (Systematized Nomenclature of Medicine — Clinical Terms) is a much broader clinical ontology, maintained by SNOMED International. Its origins trace back to the College of American Pathologists in the 1960s, but the modern version (SNOMED CT) was consolidated in 2002 with the merger of SNOMED RT and the NHS's Clinical Terms Version 3.

SNOMED CT is a concept-based terminology with over 350,000 active concepts organized in hierarchies covering virtually every domain of medicine: clinical findings, procedures, organisms, substances, body structures, devices, and more. Unlike LOINC, SNOMED CT requires a national license (free for SNOMED International member countries, paid for others).

Fundamental differences

Scope and purpose

The most important difference between LOINC and SNOMED CT is their scope. LOINC is designed specifically to identify observations: what was measured, how it was measured, and in what sample. SNOMED CT is designed to represent complete clinical meaning: diagnoses, procedures, findings, causes, and relationships between concepts.

In practical terms: LOINC answers the question "what test is this?", while SNOMED CT answers "what does this result mean clinically?". A high hemoglobin result has a LOINC code that identifies the test (718-7) and could have a SNOMED CT code identifying the clinical finding ("polycythemia" or "elevated hemoglobin").

Internal structure

LOINC uses a flat, axis-based structure. Each code is a combination of six attributes that uniquely identify the observation. There is no formal hierarchy between codes: the serum glucose code is not a "child" of the generic glucose code in a formal ontological sense.

SNOMED CT, in contrast, is a complete ontology with hierarchical relationships (is-a), defining attributes, and description logic. Each concept has a formal definition that enables automated reasoning. The concept "increased blood glucose" is a subtype of "blood glucose finding", which in turn is a subtype of "clinical chemistry finding".

Laboratory granularity

For laboratory data specifically, LOINC offers superior granularity. LOINC distinguishes between glucose measured in serum vs. whole blood, between a point-in-time measurement and a 24-hour urine measurement, between a quantitative and a qualitative result. SNOMED CT can represent these concepts, but with less detail for methodological and specimen variations.

This greater granularity of LOINC is precisely what makes it indispensable for lab data exchange: when a system receives a LOINC code, it knows exactly what observation it represents, without ambiguity.

How they complement each other

The FHIR model for joint use

The FHIR R4 standard is designed to use LOINC and SNOMED CT together. In an Observation resource, the code field uses LOINC to identify what was observed, while the interpretation or valueCodeableConcept field can use SNOMED CT to represent the clinical significance of the result.

{
  "resourceType": "Observation",
  "code": {
    "coding": [{
      "system": "http://loinc.org",
      "code": "2345-7",
      "display": "Glucose [Mass/volume] in Serum or Plasma"
    }]
  },
  "valueQuantity": {
    "value": 250,
    "unit": "mg/dL",
    "system": "http://unitsofmeasure.org"
  },
  "interpretation": [{
    "coding": [{
      "system": "http://snomed.info/sct",
      "code": "75540009",
      "display": "High"
    }]
  }]
}

In this example, LOINC identifies the test (serum glucose) and SNOMED CT classifies the interpretation (high value). Both standards working together provide a complete and unambiguous representation of the lab result.

Cross-mapping

LOINC and SNOMED CT maintain official mappings between their terminologies. SNOMED International and the Regenstrief Institute jointly publish an equivalence table that relates concepts from both systems. These mappings allow systems to translate between terminologies according to the receiver's needs.

However, mappings are not always one-to-one. A LOINC code may map to multiple SNOMED CT concepts depending on context, and vice versa. That is why best practices recommend including both codings in FHIR resources whenever possible, rather than relying on mapping at reception time.

When to use each standard

Use LOINC when...

• Identifying a lab test: LOINC is the de facto standard for coding what was observed in a lab result. The EHDS and most European regulations specify LOINC for lab data.
• Exchanging results between labs: LOINC's granularity ensures the receiver correctly interprets the observation.
• Mapping data from unstructured reports: Data extraction pipelines like MedExtract use LOINC as the target code for test names extracted from PDF or scanned reports.
• Building a FHIR DiagnosticReport: The Observation code field should use LOINC according to standard implementation guides.

Use SNOMED CT when...

• Representing clinical findings: A diagnosis, condition, or finding in the clinical record is best coded with SNOMED CT.
• Needing hierarchical reasoning: SNOMED CT's ontology enables queries like "all blood chemistry findings" that automatically include subtypes.
• Coding procedures: Laboratory procedures (the act of performing the test, not the result) are represented in SNOMED CT.
• Representing clinical interpretation: The clinical significance of an abnormal result is best expressed with SNOMED CT.

Use both when...

• Building complete FHIR resources: Best practice is to include LOINC in code and SNOMED CT in interpretation, valueCodeableConcept, or bodySite.
• Needing maximum interoperability: Some receiving systems expect LOINC, others expect SNOMED CT. Including both maximizes compatibility.
• Meeting European regulations: The EHDS references both standards for different aspects of clinical data.

Implications for lab data extraction

When extracting data from lab reports via OCR or document processing, mapping to LOINC is the primary step: each extracted test name is associated with its corresponding LOINC code. This is the process MedExtract implements with its proprietary matching cascade, ranging from exact matching to semantic matching with clinical embeddings.

Mapping to SNOMED CT is typically a secondary step added during FHIR resource generation. Once the result has its LOINC code and numeric value, the system can automatically generate the SNOMED CT interpretation based on the reference range and observed value.

Other relevant standards

ICD-10

ICD-10 (International Classification of Diseases) is used primarily for coding diagnoses and encounter reasons, not for lab data directly. However, lab results can be associated with ICD-10 codes as clinical justification for the test.

UCUM

UCUM (Unified Code for Units of Measure) is the standard for coding units of measurement in clinical data. While LOINC identifies what was measured and SNOMED CT interprets the meaning, UCUM ensures that the units of the numeric value are universally understood.

HL7

HL7 is not a terminology but a messaging standard that uses both LOINC and SNOMED CT. HL7 v2 messages (OBX segments) and FHIR R4 resources are the containers that carry the codes from both terminologies.

Conclusion

LOINC and SNOMED CT do not compete with each other: they complement each other. LOINC is the right choice for identifying lab tests and clinical observations with granular precision. SNOMED CT is the right choice for representing clinical meaning, findings, and relationships between medical concepts. Modern healthcare systems and clinical data pipelines like MedExtract must handle both standards, using each in its optimal context and maintaining the necessary mappings to ensure complete interoperability.

For organizations beginning their journey toward lab data interoperability, the priority should be LOINC mapping for all lab results, followed by SNOMED CT adoption for the clinical interpretation layer. This phased approach allows pragmatic progress without the complexity of implementing both standards simultaneously from day one.