SNOMED CT is intended to be used in healthcare:

• To provide effective and comprehensive coverage of terms

• As a terminological resource

• For implementation in electronic health applications

The purpose of SNOMED CT is to represent clinically relevant information reliably and reproducibly in electronic health applications, (most often electronic health records or EHRs) to support:

• Delivery of multidisciplinary, high-quality healthcare to individuals and populations

• Optimal retrieval, processing, and rendering of clinical information

• Effective use of clinical information consistently and reproducibly

• Use of clinical information for statistical and reporting purposes

The overall semantic interoperability of electronic health applications is achieved through the combined functioning of the information architecture of the application and the terminology that populates it. A basic principle of SNOMED CT is to create and maintain semantic interoperability of clinical information. Semantic interoperability is the capability of two or more systems to communicate and exchange information. Each system should be able to interpret the meaning of, and effectively use, received information. To achieve this goal, the meaning of the information must be agreed upon, consistent, and clearly expressed.

Context is an important part of representing clinically relevant information.

When entered in an EHR, concepts in the Procedure and Clinical finding hierarchies have the following default contexts.

• The procedure has actually occurred (versus, e.g. being planned or cancelled) or the finding is actually present (versus, e.g. being ruled out or considered)

• The procedure or finding refers to the patient of record (versus, e.g, a family member)

• The procedure or finding is occurring now or at a specified time (versus some time in the past)

When a concept is entered into an EHR, the information in the health record structure or its information model, can provide the context.

In addition to using the record structure to represent context, there may be a need to override the defaults and specify a particular context using the formal logic of the terminology. For that reason, SNOMED CT has developed a context model, i.e Situation with explicit context, to allow users and/or implementers to specify context using the terminology, without depending on a particular record structure. The Situation with explicit context hierarchy, and various attributes assigned to concepts in the hierarchy, accomplish this.

Designers and implementers of electronic health applications need guidance to identify which fields within their record structure will critically affect the meaning of concepts. They require open strategies to preserve meaning if concepts are retrieved or transferred and to allow detection of equivalence to constructs derived from alternative approaches.

(see also Situation with Explicit Context section)

Language

The terms required by users of a clinical terminology vary according to the local languages and dialects.

• When using a terminology, users must see terms in a language and dialect with which they are familiar. The terms must be clear and unambiguous independent of any hierarchical context or formal definition.

• The display of terms must not be confused by inclusion of terms in other languages or dialects.

• The terms used in different languages and dialects are not mutually exclusive. A term may be common to several languages or dialects of a language.

• When a code is presented without a specific reference to a term, an appropriate preferred term should be displayed. A term may be a preferred term in one dialect and a synonym in another.

Some terms differ only in spelling conventions (e.g. color vs. colour). The same spelling variants may recur in many different terms.

• It may be appropriate to recognize these cases and handle them differently from other term variants.

An individual instantiation of an application may only require access to a single language or dialect. It is inappropriate to install and maintain all language and dialect variants.

An application may need to support several languages with the ability to switch between languages and dialects in real-time to meet the needs of users.

Some specialties or disciplines prefer to use different terms to describe the same meaning. A particular specialist may use a precise term, while a generalist may use a different term to describe the same condition.

The following table lists factors affecting term use and examples of each.

Particular terms may be specific to an organization. They may not be included in the International Release of SNOMED CT. Organizations and users must be able to add terms or codes to SNOMED CT, without devaluing the main body of SNOMED CT.

It may be necessary to combine several subsets and extensions to meet the needs of a country, an organization, or a specialty. There must be consistent rules for combining subsets and extensions.

The requirements of a particular user may change according to the role they are performing. A single instance of an application may need to support different requirements of several users.

The headings in this section are the requirements identified in Desiderata for Controlled Medical Vocabularies in the Twenty-First Century by J.J. Cimino published in Methods of Information in Medicine 1998:37:394-403. Following each, is an explanation of the way in which SNOMED CT meets the requirement.

SNOMED CT content must be adequate both in scope and quality and must:

• Cover a wide variety of domains and different organizational needs, clinical disciplines, and medical specialties

• Meet the needs of an expanding scope, while retaining quality, with a structured systematic approach

Codes must have one meaning (nonvagueness) and no more than one meaning (nonambiguity

Key requirements that drive the design, development, and maintenance of SNOMED CT are as follows. They are related to:

1. Electronic health applications (most often electronic health records or EHRs)

   • Support for effective delivery of high quality healthcare to individuals and populations

2. The terminology

3. Implementation and migration

4. The intended user communities

   • International, multilingual applicability

   • Supporting particular localities

5. National and strategic priorities

These requirements are interrelated. The design objective is to enable all user communities to realize the potential benefits. However, the needs of different user communities may vary. To meet the overall objectives, the design must consider the entire range of needs. The approach must also be scalable in order to enable extension to new user communities.

A terminological resource is only one part of an electronic health application. Implementation of SNOMED CT should support applications in meeting user needs, rather than adding a burden to development.

The functions required to implement a terminology can be divided into those that are:

• Performed without reference to data stored in a particular application record structure.

• Involved in storing, retrieving, or processing application data.

Applications may make use of different aspects of SNOMED CT. Some may require SNOMED CT for a very limited range of uses for which there may be minimal value. These applications may not require all the functions for a full implementation or all the concepts and codes in SNOMED CT.

National and local extensions

SNOMED CT has an international and multilingual scope but can be localized to represent meanings and terms unique to particular organizations or localities. A National Extension includes content outside of the scope of the International Release, but necessary for national conformance and interoperability. Each member-state determines the application and interpretation of this scope and whether or not concepts should be added to their extension.

National Extension criteria include affirmative answers to the following:

• Is the concept outside of the scope of the International Release, but necessary for national conformance and interoperability?

• Is it useful throughout the national healthcare system?

• Does it need to be understandable throughout the national healthcare system?

• Does it need to be shared in a reproducible manner within the national healthcare system?

Extensions are created, structured, maintained, and distributed in accordance with SNOMED CT specifications and guidelines to ensure compatibility with the SNOMED CT International Release. Members may create, maintain, and distribute extensions to address specific national, regional, and language requirements. Affiliates may also create, maintain, and distribute extensions to meet the needs of particular software solutions and customers. Content that is within the scope of the International Release is restricted to the International Release and may not be modified or replaced by an extension, unless explicitly permitted by SNOMED International. Please see the Practical Guide to Extensions for more information.

SNOMED CT is not intended to cover all medical knowledge. Content that is strictly related to animals is out of the scope of the SNOMED CT international release. Non-human content may be included in a request for new content via the SNOMED International Content Request System (CRS) or may be identified in the International Release. Careful consideration is required to differentiate content that belongs in the International Release versus an extension. The basic principle is that content used in human medicine should be in the core. Content that is strictly non-human may be managed in an extension.

Examples of non-human content,

• Egg-related coelomitis (disorder)

• Dehorning (procedure)

• Bone structure of wing (body structure)

Types of content that should be in the core include the following:

• Diseases and findings. Anything that can occur in both humans and animals should be in the core.

• Material entities. A material entity is a concept found within the Substance, Physical object, Pharmaceutical/biologic product, Physical force, or Organism subhierarchies. Every substance that can cause adverse effects should be in the core (with the understanding that poisonings and adverse effects in humans may be caused by virtually any substance). Some material entities may be of interest only in a non-human or veterinary context. These entities may be added to, or left in, a veterinary extension.

• Organisms. Most organisms should be in the core:

The Veterinary Extension is publicly available to SNOMED International member countries and to Affiliate Licensees. To access to the Veterinary Extension, please see , or contact the VTSL via email at .

Classification-derived phrases are not accepted. Concepts with unclear, unspecified, or ambiguous meaning should not be used. Rejections are expected for requests with the following phrases:

• Not otherwise specified (NOS)

  • For example, Mental disorder, not otherwise specified

• Not elsewhere classified (NEC)

  • For example, Chronic hepatitis, not elsewhere classified

The determination of whether something is legal or illegal cannot be universally stated, as it is subject to jurisdictional variability and the specific laws, regulations, and interpretations that apply in different regions or legal systems. For this reason, the notion of legality should not be included in SNOMED CT International concepts.

Concepts referring to regulatory status or characterization (e.g., over-the-counter) are out of scope for the International Release . Meaning may vary by jurisdiction and may not be consistent internationally.

A person's citizenship or legal residence is not an intrinsic characteristic of a person and is out of scope. SNOMED International recommends use of ISO country codes for recording residency.

Because of the jurisdictional and administrative aspects of medical insurance, this has been deemed out of scope for the SNOMED CT International release. It is up to individual member countries to determine if that type of content should be included in their extensions. Users should contact their country's extension administrator to determine if this type of content is acceptable.

Text that is protected by copyright will not be accepted for inclusion unless accompanied by a release from the copyright holder.

Knowledge representation in SNOMED CT involves modeling what we know about concepts to be necessarily true. Concepts are logically defined by their relationships to each other. Some knowledge provides valuable clues to the diagnostician, while not necessarily always present, i.e. it is uncertain or probabilistic knowledge. Attempts to capture probabilistic or uncertain knowledge are out of the scope of SNOMED CT.

22298006 | Myocardial infarction (disorder)|

Its terminological knowledge includes the following:

IS A: 
64572001 | Disease (disorder)|

Finding site: 
74281007 | Myocardium structure (body structure)|

Associated morphology: 
55641003 | Infarct (morphologic abnormality)|

These additional pieces of knowledge are variably present and therefore represent uncertain or probabilistic knowledge about myocardial infarction:

• Crushing substernal chest pain

• Diaphoresis

• Arrhythmia

• ST-segment elevation on EKG

• Elevated cardiac enzymes

Another example:

Its terminological knowledge includes the following:

These additional pieces of knowledge are variably present and therefore represent uncertain or probabilistic knowledge about appendicitis:

• Central abdominal pain that migrates to the right lower quadrant

• Rebound tenderness over McBurneys point

• Anorexia

• Nausea

The anticipated benefits of SNOMED CT are derived from use of information to support effective delivery of high quality healthcare to individuals and populations.

Individuals

Aide-memoire for clinicians

Clinically relevant information in an electronic health record acts as an aide-memoire for the clinician, enabling recall of previous interactions.

Structured data entry

Structured data entry enhances the value of an electronic health record in various ways. It may:

• Simplify recording of frequently collected data

• Ensure that information is collected in a reliable and reproducible way

• Help clinicians to think logically about a patient's condition

Clinical applications may combine several data entry methods. Some of the most commonly used methods are as follows:

• Searching a coded terminology for matching terms using words or phrases

• Navigating a hierarchical structure to refine or generalize meanings

• Using templates or protocols to record structured information; may be based on answers to questions or values entered on a data entry form

• Parsing of natural language to identify and retrospectively code and structure data

Data entry may require selection from a list. Such lists must be manageable in size and appropriate to the needs of the user.

• A multilingual, multidisciplinary terminology requires mechanisms that limit and/or prioritize access to terms and codes in ways that are appropriate to:

  • Languages and dialects

  • Countries, organizations, disciplines, specialties, and users

  • Contexts within a record or protocol

When a code is entered in a record it may require structured entry of additional qualifying information.

• Qualifying information may be coded.

  • For example,

    • The code named removal of kidney may require a statement of laterality.

• Qualifying information may be numeric.

To meet all the needs for coded structured data entry in a health record, a terminology must have an adequate scope.

• The main body of SNOMED CT covers the required scope.

  • It may be difficult to meet the needs of some organizations, specialties, and users; they may need specific terms or codes to meet their own operational requirements. Therefore, SNOMED CT is structured to allow for additions to meet specific needs.

A clinical terminology requires frequent changes including new codes, terms, and relationships between codes. Changes may be required due to new:

• Health risks

• Health and disease process information

• Drugs, investigations, therapies, and procedures

The presentation of clinical information may:

• Highlight key information and indicate links between items, thus helping clinicians understand patients' conditions.

• Be determined entirely by record structure without regard to the terminological resource (e.g.,may be in chronological order, by author, or by the type of recorded event).

• Be enhanced based on its semantic content (e.g., grouping procedures, investigation results, or observations relevant to a particular disease process).

Interfaces between recorded clinical information and appropriate decision support tools and reference works may assist the clinician in selecting diagnostic tests, making diagnoses, and choosing treatment. Decision support requires selective retrieval and processing of information in an individual health record to determine whether the patient has particular characteristics relevant to the decision support protocol. The algorithms for establishing the presence of characteristics should include relationships between coded meanings and other aspects of record structure. Performance is also important, as decision support algorithms are typically run in real-time during data recording. Decision support algorithms may:

• Depend on numeric or other values (and their units) associated with particular observations

• Include the context in which information is recorded, e.g., the date of recording and any stated relationships between individual items of information

• Include information such as age, sex, clinical conditions, findings, surgical procedures, medication, and social/environmental factors, such as occupation

• Use codes or identifiers from other terminologies, classifications, or proprietary schemes. Mapping tables are required to allow applications that use a terminology to interface with these resources

Effective delivery of high quality healthcare to individuals requires communication between those involved in providing care. This requires communication within and across teams or organizations.

The primary objective of many clinical communications is to convey information from human to human. Communications with this purpose should include human-readable text. Relying on text from coded data is not recommended. Coded data is therefore not relevant to the requirement for human-to-human communication.

A receiving application may process clinical communications. This information may need to be retrieved and processed to meet terminology requirements. To meet terminology requirements, messages and other means of electronic communication must permit the communication of SNOMED CT identifiers and associated structures.

Communication specifications, such as those produced by HL7 and CENTC251, define the structures to meet requirements. The coded information is used in two distinct situations:

• Coded elements that must be filled with codes enumerated in the specifications. The codes enumerated in the specifications generally communicate, mission critical features of the message. Some of the enumerated codes and the codes in a clinical terminology may have overlapping meanings.

• Coded elements that are populated with clinical codes from appropriate coding schemes. The open coded elements may require the full expressiveness of a terminology. Some of the open coded elements may be restricted to codes that express particular types of meaning.

  • For example,

There are two situations in which communication of coded information may be of value for human-to-human communication. They are where:

• The storage capacity or communication bandwidth is restricted. Receiving applications must contain (or have real-time access to) a table listing the text description associated with each code.

• The translation between the languages of the sender and the recipient is needed. A coded representation of a meaning may allow the appropriate description in the recipient's language.

Recording a particular code may trigger a communication. And, receipt of a code, may trigger specific processing in the receiving application.

• For example,

  • Recording a decision to prescribe a medicine might trigger an electronic prescription sent to the pharmacy. Receipt of such a prescription might trigger dispensing and stocking activities.

The relationship of a trigger is an additional characteristic of a code that may be context dependent.

Patients may wish to view, and comprehend, their own records. For SNOMED CT to meet this requirement, the inclusion of patient-friendly terms should be considered. However, this requirement should not take precedence over accurate professional terminology.

Patients may also be allowed to contribute to their own records, i.e., be users of SNOMED CT.

• For example,

  • Patients with diabetes may monitor and record their blood glucose levels.

The provision of effective high-quality care to populations requires an understanding of the state of health and healthcare needs of that population. Information recorded about individual patients must be available for analysis to determine trends.

• It must be possible to analyze data recorded with SNOMED CT.

Population trends are usually monitored at a higher level, using codes that are more general than those used in individual patient records. This may be accomplished through one or both of the following methods:

• Using hierarchical relationships and/or equivalences defined within SNOMED CT.

• Mapping SNOMED CT codes to codes in appropriate classifications.

Appropriate analysis of information requires reliable and reproducible queries.

• The scope of SNOMED CT must cover the types of information relevant to analysis.

• Analysis may require data about multiple clinical characteristics. Queries must account for both the terminology and the record structure.

The requirements for analysis of quality of service are similar to those for analysis of health needs. The main difference is that the scope of the analysis must be extended to cover consultations, referrals, procedures, medications, and other interventions.

The requirements for research are also similar to those for analysis of health needs, however, there is a need to allow for:

• Recording interventions in ways that do not compromise blind and double blind trials.

• Adding SNOMED CT content for experimental observations or treatments, which may never require permanent addition to the terminology.

The management and funding of healthcare delivery often depends on recording and reporting of particular information, e.g. bundled or packaged care. Automating this process offers a way of reducing bureaucratic overhead, i.e. mapping clinical information recorded with SNOMED CT to appropriate forms.

Some information required for management and funding purposes is specifically related to claims for particular events or services.

• For example,

  • Funding general practitioners in the NHS is dependent on meeting immunization administration and cervical cytology screening targets.

The scope of SNOMED CT must be adequate to meet these needs, or must be capable of extension to meet these needs, without presenting irrelevant terms or coded meanings to those not requiring them.

Organizations, such as WHO and some government bodies, require specific data related to healthcare statistics. Organizations should be able to use clinical information recorded with SNOMED CT. When this is not possible, the clinical information should at least support their manual generation. Using structured data entry allows for direct mapping to statutory national and international classifications such as ICD, CPT, OPC, etc.

Population-based preventive care should be offered to specific groups, based on sex, age, medical history, and other factors. Health information applications based on information recorded with SNOMED CT can be used to identify patients, so they can be offered appropriate care.

SNOMED CT includes 19 domains arranged in a polyhierarchical structure. Each hierarchy is an ordered organization of concepts linked together through IS A relationships. Each concept may have one or more parents.

The hierarchical arrangement is helpful for locating concepts, grouping similar concepts, and conveying meaning. For example, if we see the concept cell under the concept anatomic entity we will understand the intended meaning as different than if it appeared under the concepts room or power source(Desiderata for Controlled Medical Vocabularies in the Twenty-First Century by J.J. Cimino published in Methods of Information in Medicine 1998:37:394-403).

Concepts are linked to their more general parent concept codes directly above them in a hierarchy. Concepts with more general meanings are usually presented as being at the top of the hierarchy and then at each level down the hierarchy, the meanings become increasingly more specific or specialized.

The domains contain all of the components (clinical, administrative, database structure, as well as other components that express how the domains relate to each other) necessary to create SNOMED CT concepts and maintain the database structure.

The following table lists the domains, definitions, and examples. *Those without a concept model are marked with an asterisk.

The scale, or level of detail, in a terminology is called granularity. Concepts and meanings range from very general, or coarse, to very specific, or fine. SNOMED CT has multiple granularities, which is an important component of terminologies that are multipurpose. The broader meanings are useful for aggregation (e.g. Clinical finding, Procedure, etc.), but are not intended for recording individual patient data.

The progressive levels of refinement are used to meet clinical data requirements. There are, however, limits to the degree of precoordination of certain types of complex statements.

In general, concepts in SNOMED CT should name things that exist in the real world. The concepts are usually names or short noun phrases, not complete sentences or paragraphs.

SNOMED CT is intended to be used with electronic health applications that can support full clinical statements, along with their attributions, dates, times, and statement interrelationships. It may be challenging to balance SNOMED CT content with the needs of those using electronic health applications. For example, some older applications may require concepts outside of the scope of SNOMED CT. SNOMED CT tries to maximize its usefulness and at the same time minimize precoordination.

What is SNOMED CT?

SNOMED CT is a high-quality, comprehensive, international, logic-based reference terminology that is used to present clinically relevant information. It began with the union of NHS Clinical Terms Version 3 and SNOMED RT; this provided the initial scope which has since been updated to reflect contemporary clinical practice and changes in medical technology.

Content development is provided by expert clinicians driven by the requirements of user communities. This includes core content for use internationally and content relevant to national extensions for local implementation.

Its logic-based definitions represent terminological knowledge, or what is always true about the meaning of concepts. It consists of codes, that correspond to concepts, arranged in a polyhierarchical manner, as well as relationships between the concepts, which further define the meaning.

Description logic (DL) is the formal foundation of meaning in SNOMED CT. The way that concepts have been modeled in SNOMED CT permits them to be represented using description logic. A DL reasoner is used to classify SNOMED CT. The DL reasoner also helps test expressions for subsumption and equivalence.

Why use SNOMED CT?

It supports semantic interoperability and multi-purpose use within electronic health applications (primarily electronic health records or EHRs) and has many advantages over other terminologies. They include:

• Consistent, and formal expansion of, content through centralized authoring and maintenance (International Release)

• Flexibility to meet most terminological requirements based on national, regional, language, application, or customer (Extensions)

• Clear, singular meaning of concepts

• Reliable, consistent, and reproducible clinical documentation

• Enhanced high-quality healthcare delivery to individuals and populations

A summary of the SNOMED CT requirements is as follows. Additional information may be found throughout this guide, as well as in other documents on the SNOMED International website.

Terminology Structure Requirements