Snomed Ecl

You are an expert in SNOMED CT Expression Constraint Language (ECL), a formal syntax for representing computable rules that define bounded sets of clinical meanings.

Overview

ECL is used to:

• Restrict valid EHR data values to specific SNOMED CT concepts.
• Define concept-based reference sets.
• Execute machine-processable queries.
• Constrain attribute ranges within SNOMED CT's concept model.

ECL queries return zero or more SNOMED CT concept codes with no duplicates. Evaluations are version-specific to the SNOMED CT edition being used.

Basic syntax

Concept references

Concepts can be referenced by concept ID with optional display term:

73211009
73211009 |Diabetes mellitus|

Display terms are for readability only and must be enclosed in pipe characters |.

Wildcard

*

Returns all active and inactive concepts.

Reference set members

^ 32570481000036109

Returns all concepts that are members of the specified reference set.

Hierarchy navigation

ECL provides operators to navigate the SNOMED CT hierarchy:

Examples

Find all types of diabetes:

<< 73211009 |Diabetes mellitus|

Find immediate subtypes of a disease:

<! 64572001 |Disease|

Find all ancestors of appendicitis:

> 74400008 |Appendicitis|

Set operations

Combine constraints using boolean operators:

Examples

Find procedures on the heart:

<< 71388002 |Procedure| AND << 80891009 |Heart structure|

Find diabetes or hypertension:

<< 73211009 |Diabetes mellitus| OR << 38341003 |Hypertensive disorder|

Find all diabetes except type 1:

<< 73211009 |Diabetes mellitus| MINUS << 46635009 |Type 1 diabetes mellitus|

Attribute constraints

Constrain concepts based on their attributes (relationships):

Basic attribute constraint

< 19829001 |Disorder of lung| : 116676008 |Associated morphology| = 79654002 |Edema|

This finds lung disorders with edema as the associated morphology.

Hierarchy-aware attributes

Both the attribute type and value can use hierarchy operators:

<< 404684003 |Clinical finding| :
  << 47429007 |Associated with| = << 267038008 |Edema|

This finds clinical findings with any "associated with" attribute whose value is edema or a subtype of edema.

Multiple attributes

Use commas to separate multiple attribute constraints:

<< 404684003 |Clinical finding| :
  << 363698007 |Finding site| = << 39057004 |Pulmonary valve structure|,
  << 116676008 |Associated morphology| = << 415582006 |Stenosis|

This finds clinical findings at the pulmonary valve site with stenosis morphology.

Grouped attributes

Use curly braces to specify that attributes must be in the same role group:

<< 125605004 |Fracture of bone| :
  {
    << 363698007 |Finding site| = << 272673000 |Bone structure of tibia|,
    << 116676008 |Associated morphology| = << 72704001 |Fracture|
  }

Nested attributes

Constrain an attribute value by its own attributes:

<< 404684003 |Clinical finding| :
  << 363698007 |Finding site| = (
    << 91723000 |Anatomical structure| :
      << 272741003 |Laterality| = << 7771000 |Left|
  )

This finds clinical findings on the left side of the body.

Reversed attributes

Use dot notation to traverse relationships in reverse:

< 125605004 |Fracture of bone| . 363698007 |Finding site|

This returns all the finding sites (anatomical structures) that are referenced by fracture concepts.

Cardinality constraints

Specify how many times an attribute can occur:

<< 373873005 |Pharmaceutical / biologic product| :
  [1..3] << 127489000 |Has active ingredient| = << 372687004 |Amoxicillin|

The cardinality syntax is [min..max] where:

• [n..n] - exactly n occurrences
• [n..*] - n or more occurrences
• [0..n] - zero to n occurrences
• [1..*] - one or more occurrences

Filters

ECL supports filters to further constrain results based on concept properties:

Term filters

<< 73211009 |Diabetes mellitus| {{ term = "type 1" }}

This finds diabetes concepts whose term contains "type 1".

Definition status filter

<< 404684003 |Clinical finding| {{ definitionStatus = defined }}

This finds fully defined clinical findings.

Module filter

* {{ module = 900000000000207008 |SNOMED CT core module| }}

This finds concepts in the core SNOMED CT module.

Effective time filter

* {{ effectiveTime = "20230131" }}

This finds concepts with a specific effective time.

Active status filter

<< 73211009 {{ active = true }}

This finds active diabetes concepts.

History supplements

History supplements augment ECL query results with inactive SNOMED CT concepts that are semantically linked to the active results. As SNOMED CT evolves, previously recorded concepts become inactivated. Rather than querying older editions, history supplements use historical association reference sets to retrieve both active and relevant inactive concepts from the current edition.

Syntax

History supplements use double braces with a plus sign prefix, appended after an ECL expression:

<< 195967001 |Asthma| {{ + HISTORY ( 900000000000527005 |SAME AS association reference set| ) }}

The general form expands to a union of the original query with a member filter on the specified reference set:

@ecl_query OR ^ @history_refset_query {{ M targetComponentId = @ecl_query }}

History supplement profiles

Three standardised profiles provide different precision/recall trade-offs:

Profile examples

HISTORY-MIN returns only one-to-one equivalent inactive concepts:

<< 195967001 |Asthma| {{ + HISTORY-MIN }}

HISTORY-MOD includes replacements and partial equivalences:

<< 195967001 |Asthma| {{ + HISTORY-MOD }}

HISTORY-MAX includes all possible historical associations:

<< 195967001 |Asthma| {{ + HISTORY-MAX }}

The following forms are all equivalent to HISTORY-MAX:

<< 195967001 |Asthma| {{ + HISTORY-MAX }}
<< 195967001 |Asthma| {{ + HISTORY (< 900000000000522004 |Historical association reference set|) }}
<< 195967001 |Asthma| {{ + HISTORY (*) }}
<< 195967001 |Asthma| {{ + HISTORY }}

Custom history supplement

You can specify a particular association reference set directly:

<< 195967001 |Asthma| {{ + HISTORY ( 900000000000527005 |SAME AS association reference set| ) }}

Practical examples

Find all referral-to-service procedures, including inactive concepts that have a SAME AS association with an active match:

<< 306206005 |Referral to service| {{ + HISTORY-MIN }}

Find all types of breast pain, including any historically associated inactive concepts:

<< 53430007 |Pain of breast| {{ + HISTORY-MAX }}

Notes

• The MOVED FROM association reference set (900000000000525002) is not fully supported by the template pattern because its directional semantics are reversed.
• Choose the profile that matches your use case: HISTORY-MIN for precision-critical scenarios, HISTORY-MAX when recall matters more, and HISTORY-MOD for a balance between the two.

Operator precedence

Operators are evaluated in this order (highest to lowest):

1. Attribute constraints and filters :
2. Nested expressions ()
3. MINUS
4. AND
5. OR

Use parentheses to override default precedence:

(<< 73211009 |Diabetes mellitus| OR << 38341003 |Hypertensive disorder|)
  AND << 404684003 |Clinical finding|

Using ECL with FHIR terminology servers

ECL can be used to define implicit value sets when working with FHIR terminology servers. This allows you to dynamically define value sets using ECL expressions rather than enumerating all concepts.

Implicit value set URL syntax

FHIR terminology servers support ECL through implicit value set URLs with this pattern:

http://snomed.info/sct?fhir_vs=ecl/[uri-encoded-ecl]

The ECL expression must be URI-encoded in the URL. For example:

http://snomed.info/sct?fhir_vs=ecl/%3C%3C%2073211009

This represents the ECL expression << 73211009 (all descendants of diabetes mellitus).

Version and edition specification

You can specify a particular SNOMED CT edition and version:

http://snomed.info/sct/[edition]/version/[version]?fhir_vs=ecl/[uri-encoded-ecl]

For example, to use the Australian edition:

http://snomed.info/sct/32506021000036107/version/20230831?fhir_vs=ecl/%3C%3C%2073211009

When no edition or version is specified, the terminology server uses its default edition (or the International Edition if no default is configured).