Code of practice

Tim Benson

SNOMED CT is, along with HL7, one of the core standards that under-pins the NHS Connecting for Health programme. SNOMED CT is the dictionary of clinical terms used in clinical communications and electronic healthcare records including the National Care Records Service.

Special computer-based dictionaries such as SNOMED CT are required because traditional medical terminology is ambiguous, with many synonyms (two or more ways of saying the same thing) and homonyms (the same term meaning different things depending on context). Computers work by exact matching digital strings – they cannot cope with the ambiguity of natural clinical language.

The term “coding” is frequently used to describe the task of classifying diagnoses and procedures, but this is misleading. Codes and classifications do different jobs. Codes are unambiguous identifiers of concepts – for example CDG identifies Charles De Gaulle airport in Paris. Classifications put instances into categories according to pre-defined criteria. Codes identify, while classifications group for analysis.

SNOMED CT is both a coding scheme, identifying concepts and terms, and a multi-dimensional classification, enabling concepts to be related to each other, grouped and analysed according to different criteria.

Building blocks

SNOMED CT uses numeric codes to identify every instance of its three core building blocks: concepts, descriptions and relationships. Each concept represents a single specific meaning; each description associates a single term with a concept (any concept may have any number of descriptions or names); and each relationship represents a logical relationship between two concepts.

SNOMED CT is large, containing over 300,000 concepts, 1 million descriptions and 1.5 million relationships. In comparison, ICD-10 has 10,760 classes (excluding Chapter XX, external causes) and comes in three large volumes.

SNOMED codes are linked within the dictionary to define relationships between concepts. For example, the concept “fracture of the tibia” has an explicit relationship with the concept “tibia” to define the site of fracture. Relationships are also used to build expressions within patient records. For example, the concept of fracture of shaft of tibia can be qualified by laterality (laterality = right) and by fracture type (fracture type = spiral). This is known as post-coordination.

SNOMED CT is only used in computer systems – it cannot be used manually. First because it is so large, but more importantly because it works in a different way than earlier coding schemes such as ICD or the Read Codes. In these schemes the relationships between concepts is specified within the code itself. For example, in ICD-10, the code for fracture of shaft of tibia is S82.2, which is a specialisation of S82 (fracture of lower leg including ankle).

In SNOMED CT this taxonomy (hierarchy) is maintained as separate relationships, which rely on computer software to work. SNOMED CT’s relationship mechanism is more complex than a code-dependant hierarchy, but is enormously more powerful and flexible. It allows any concept to be classified or qualified in any number of ways.

History of SNOMED CT

Where has SNOMED CT come from? In 1999 the NHS and the College of American Pathologists (CAP) agreed to merge SNOMED with the NHS Clinical Terms (the Read Codes) to produce a single joint clinical terminology – SNOMED CT (Clinical Terminology). The merger was completed in 2002 with the first release of SNOMED CT.

SNOMED has a long history. Back in 1971, Roger Côté, a Canadian Pathologist, published the Standard Nomenclature of Pathology (SNOP) which describes pathology findings using four axes: topography (anatomic site affected), morphology (structural changes associated with disease), aetiology (the cause of disease) and function (physiologic alterations associated with disease). This was adopted by the College of American Pathologists (CAP).

In 1975 Côté and colleagues extended SNOP by adding additional dimensions covering diseases, procedures, organisms and so on to give it a broader scope with the name Systematised Nomenclature of Human and Veterinary Medicine (SNOMED).

While SNOMED has its origins in North American pathology laboratories, the Read Codes were developed during the mid-1980s as a comprehensive coding and classification scheme for anything that a British GP might want to record in an electronic patient record. Their simple hierarchical structure was based around the organisation of the International Classification of Diseases, the OPCS Classification of Procedures and the British National Formulary for medicines.

SNOMED CT is a true merger. Every Read Code ever released is present in SNOMED CT so migration from Read Codes to SNOMED CT will not result in loss of information.

SNOMED CT provides an extensible foundation for expressing clinical data in both local systems and in the National Care Record System. However, SNOMED CT is much larger and more complex than any previous coding scheme. The relationship structure is completely different. One question remains is how effectively and enthusiastically will Local Service Providers and clinical application suppliers take full advantage of its power.

Tim Benson
Abies

Links

Abies (SNOMED CT Training)
SNOMED website
NHS Connecting for Health (SNOMED CT site)