Skip to main content
v2026.11,986 entries · CC-BY 4.0

SNOMED CT Postcoordination, Compositional Grammar, and ECL Guide

How SNOMED CT Compositional Grammar (SCG) encodes postcoordinated clinical expressions and how Expression Constraint Language (ECL) queries FHIR value sets.

Written and maintained by CASRAI Editorial Board

Published

On this page

SNOMED CT (Systematized Nomenclature of Medicine — Clinical Terms) is not merely a flat code list or a simple tree like ICD-10-CM; it is a formal description-logic ontology built on an explicit Concept Model. Because clinical medicine involves combinatorial permutations of anatomical sites, morphologies, lateralities, severities, and infectious agents, SNOMED International provides two formal mechanisms for representing and querying clinical meaning: SNOMED CT Compositional Grammar (SCG) for encoding postcoordinated clinical expressions, and Expression Constraint Language (ECL) for querying intensional concept sets inside a FHIR terminology server.

What Is the Difference Between SNOMED CT Pre-Coordinated Concepts and Post-Coordinated Expressions?

Every clinical meaning in SNOMED CT can be represented using either pre-coordination or post-coordination:

  • Pre-coordination: Captures a complete clinical phrase within a single pre-existing SNOMED CT Identifier (SCTID)—for example, 80146002 |Appendectomy (procedure)| or 44054006 |Type 2 diabetes mellitus (disorder)|. Inside the SNOMED CT release files, a pre-coordinated concept’s defining relationships (`Finding site = Appendix structure`, `Method = Excision – action`) are already pre-authored in the ontology graph.
  • Post-coordination: Combines two or more existing SCTIDs at the point of clinical documentation or data capture using SNOMED CT Compositional Grammar (SCG) to express clinical detail that does not exist as a single pre-coordinated concept—avoiding combinatorial explosion in the core vocabulary while preserving full machine-computable semantics.

Syntax of SNOMED CT Compositional Grammar (SCG)

An SCG postcoordinated expression consists of one or more focus concepts followed by a colon (:) and one or more attribute-value refinements (separated by commas and grouped as needed inside curly-brace role groups { ... }):

=== Focus Concept : { Attribute1 = Value1, Attribute2 = Value2 } ===
71388002 |Procedure| :
  { 260686004 |Method| = 129304002 |Excision - action|,
    405813007 |Procedure site - Direct| = 66754008 |Appendix structure| }

Optional pipe-delimited human-readable term strings (|...|) can be included for readability during authoring or strips out to leave compact machine-normative SCTID strings (71388002:{260686004=129304002,405813007=66754008}). Each expression is prefixed by an equivalence status symbol—defaulting to === (equivalent to) when omitted, or explicitly marked as <<< (subtype of) when the postcoordinated expression is narrower than what its formal attributes capture alone.

How Does the SNOMED CT Concept Model Use Attribute-Value Relationships and Role Groups?

Authors cannot combine arbitrary concepts in a postcoordinated expression. Validity is governed by the SNOMED CT Machine Readable Concept Model (MRCM), which specifies for each domain (such as `<< 404684003 |Clinical finding|` or `<< 71388002 |Procedure|`) which attribute concepts are permitted and what range of value concepts each attribute may take:

SNOMED CT Attribute Concept Permitted Domain Permitted Value Range & Clinical Example
363698007 |Finding site| Clinical finding, Disorder << 442083009 |Anatomical or acquired body structure| (e.g., Left lower lobe of lung).
116676008 |Associated morphology| Clinical finding, Disorder << 49755003 |Morphologically abnormal structure| (e.g., 72704001 |Fracture|, 23583003 |Inflammation|).
272741003 |Laterality| Body structure, Clinical finding 7771000 |Left|, 24028007 |Right|, or 51440002 |Right and left (bilateral)|.
246112005 |Severity| Clinical finding << 272141005 |Severities| (e.g., 24484000 |Severe|, 255604002 |Mild|).
246075003 |Causative agent| Clinical finding, Disorder Organism, substance, or physical force (e.g., 840533007 |SARS-CoV-2|).

Why Role Groups ({ ... }) Prevent Clinical Ambiguity

Suppose a complex surgical procedure removes an ovary using a salpingo-oophorectomy incision and also biopsies a lymph node. Without role groups, listing two `Method` attributes (`Excision`, `Biopsy`) and two `Procedure site` attributes (`Ovary`, `Lymph node`) in a flat list leaves it ambiguous whether the ovary was excised or biopsied. Enclosing each paired action and anatomical site inside its own curly-brace role group (`{ Method = Excision, Procedure site = Ovary }, { Method = Biopsy, Procedure site = Lymph node }`) binds each morphology or method to its exact anatomical target for description-logic classification.

How Does SNOMED CT Expression Constraint Language (ECL) Define Intensional ValueSets in FHIR?

Whereas Compositional Grammar (SCG) encodes a single clinical concept on a patient record (such as inside Condition.code or Observation.code), SNOMED CT Expression Constraint Language (ECL) is a query language used to select a set of concepts that satisfy hierarchical and attribute criteria:

  • Hierarchy Operators: < (strict descendants of), << (descendants or self of), > (ancestors of), >> (ancestors or self of), and ^ (members of a SNOMED CT Reference Set).
  • Attribute Refinements & Conjunctions: Uses : to filter concepts by their defining relationships, combined with boolean AND (or comma), OR, and MINUS (exclusion). For example, querying all infectious pneumonias caused by a bacterium:
    << 233604007 |Pneumonia| : 246075003 |Causative agent| = << 409822003 |Domain Bacteria|.
  • HL7 FHIR Implicit ValueSets: In HL7 FHIR R4, any ECL query can be turned into an implicit ValueSet canonical URL (`http://snomed.info/sct?fhir_vs=ecl/[url-encoded-ecl]`) or placed inside a ValueSet.compose.include.filter (`property = “constraint”`, `op = “=”`, `value = “[ecl]”`), allowing a terminology server’s `$expand` and `$validate-code` operations to evaluate description-logic subsumption dynamically.

Frequently Asked Questions

How are postcoordinated SNOMED CT expressions represented in an HL7 FHIR Coding element?

In HL7 FHIR, a postcoordinated SCG expression is placed directly inside Coding.code (using the normative SCTID-only syntax, with no pipe-delimited display terms) while Coding.system is set to http://snomed.info/sct. Because a postcoordinated string is not a single pre-assigned SCTID, validating it requires a FHIR terminology server equipped with an OWL description-logic classifier (such as Ontoserver or Snowstorm) that complies with the MRCM rules.

Can an OWL classifier recognize equivalence between a pre-coordinated concept and a post-coordinated expression?

Yes. Because SNOMED CT is authored in the OWL 2 EL description-logic profile, a classifier computes the normal form of both the pre-coordinated concept (`80146002 |Appendectomy|`) and the post-coordinated SCG expression (`71388002:{260686004=129304002,405813007=66754008}`) and proves via `$subsumes` that the two expressions are logically `equivalent`.

How do observational research data models like OMOP handle postcoordinated SNOMED CT codes?

Relational research schemas such as the OMOP Common Data Model expect a single integer `concept_id` in primary clinical event columns (`condition_concept_id`) rather than arbitrary runtime postcoordinated expressions. When source EHRs transmit postcoordinated SNOMED CT expressions, ETL pipelines either classify the expression to its closest pre-coordinated Standard Concept via subsumption or split qualifying modifiers into secondary qualifier/observation columns.

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · full access

Ask about SNOMED CT Postcoordination, Compositional Grammar, and ECL Guide

Full access: large monthly usage, up to 10 source passages behind each answer, and MCP and API access for Claude Code, Codex and other CLIs. Cancel any time.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →