Snomed CT: the clinical terminology that improves the quality of patient care and the interoperability of global health data
Author: José Alberto Maldonado Segura. Responsable de Innovación y consultor en interoperabilidad semántica en Veratech for Health.
The use of SNOMED CT enables the improvement of the quality of patient care, the accurate identification of patient cohorts for clinical research, audits or reporting, and the interoperability of health data.
Clinical terminologies are one of the basic building blocks for the interoperability of health information. They provide standardized terms to refer to a clinical concept and thus enable consistent representation of information. SNOMED CT is a multilingual clinical terminology covering a wide range of medical domains managed by SNOMED International. It is considered the most extensive and essential terminology developed worldwide and currently contains more than 350,000 concepts. SNOMED CT is an ontology-based on concepts and semantic relationships between them where concepts represent a clinical idea (disease, procedure, organism, etc.). Each concept has an identifier and a set of descriptions. A relation represents an association between two concepts and defines the meaning of the concepts in a precise and actionable way.
Concepts in SNOMED CT are organized hierarchically by means of IS_A relationships. All concepts in SNOMED CT are subtypes of a root concept. "SNOMED CT Concept".There are 19 higher-level concepts under this root concept. They include clinical findings, procedures, organisms, pharmaceutical/biological products, etc. The rest of the concepts descend from these 19 top-level concepts. It is important to note that a concept can belong to several hierarchies, which makes SNOMED CT a terminology that supports poly-hierarchy. For example, bacterial pneumonia is both an infectious pneumonia and a bacterial lower respiratory tract infection. In addition to IS_A relations, it contains relations that define the meaning of the concept about other concepts, including essential features such as site of finding or morphology and complementary attributes such as severity or laterality.
Another highlight is its extensibility, which allows user organizations to add terms and relationships not found in the standard version to adapt the terminology to their specific needs.
SNOMED CT aims to standardize and codify the meanings used in health information globally. Está plenamente reconocido que su uso aporta múltiples beneficios. Podemos destacar que permite documentar y codificar la actividad clínica de manera precisa, detallada y específica lo que se traduce en una mejora de la calidad de la atención al paciente, ya que los profesionales de la salud pueden tomar decisiones más informadas. Su base ontología facilita un mecanismo de consulta que permite la identificación precisa de cohortes de pacientes relevantes para investigación clínica, auditorías o informes. Por último, facilita la interoperabilidad de los datos y por tanto su compartición entre personas, sistemas informáticos y organizaciones conservando su significado original.
To get the most out of SNOMED CT, a wide range of tools are needed to manage, explore and exploit terminology. These include terminology management systems to manage and maintain the terminology, ensuring its quality and integrity, viewers to search for concepts and relationships, coding support systems including natural language processing to detect and extract mentions of concepts in narrative texts, definition and management of mapping with other terminologies, definition, and management of subsets, query engines on the terminology itself, authoring environments to facilitate its extension or reasoners to exploit its ontological base. However, there is currently a lack of governance and infrastructure for the management of SNOMED CT, indicating the need for a more systematic and structured approach to ensure its proper management and successful implementation in clinical practice. Healthcare organizations need to establish a clear governance framework and provide adequate tools and resources to work with SNOMED CT effectively and efficiently.
Let us look at the use of terminology during data entry as an illustrative example. In this scenario, SNOMED CT is combined with the information model used to record the information, i.e. the user interface. Both technologies must be combined to faithfully capture the clinical meaning of the data, this is known as terminology linking. One type of terminology linkage is content linkage which allows us to define the set of admissible codes for an attribute of the information model, e.g. a subset of diseases. To define such links, it is necessary to have a query language capable of specifying terminological subsets without having to explicitly enumerate the concepts that form it, and an environment for editing, validating, and visualizing such subsets. See for example Figure 1, which shows a query defining the subset of non-food allergies consisting of 1921 concepts. Furthermore, it is necessary to establish where these subsets are stored, how they should evolve to adapt to new versions of the terminology, or how to make them public to facilitate their reuse.
Figure 1. The Subset of SNOMED CT concepts is defined using a query, in this case using the expression constraint language.
Veratech For Health has been actively working for more than 6 years on the development of tools to facilitate the exploitation and implementation of SNOMED CT in organizations, including query engines, viewers, and detection of mentions of concepts in narrative texts such as diseases or procedures. We are actively collaborating with the SNOMED CT National Reference Centre for Spain under the Ministry of Health of the Spanish Government in the implementation of an IT environment that facilitates the generation, publication, and consultation of the SNOMED CT national extension for Spain. This has involved the development of an environment for the classification and validation of the extension and a web browser for its consultation (https://snomedsns.es/).).
The former allows us to assign the correct hierarchies to concepts and to infer new relationships from the logical definition made by the author as well as to check the validity, e.g. the non-existence of logically equivalent concepts. For its part, the browser makes it possible to locate concepts using different search configurations such as "exact", "contains", "partial terms", etc. It is also possible to query the members of reference sets. The browser allows the search and downloading of subsets of concepts. For this purpose, it makes use of a proprietary implementation of the SNOMED CT Expression Constraint Language (ECL) that allows defining concept sets using constraints on the logical definition of concepts such as "lung diseases that are a type of edema", "food allergies" or "ophthalmological procedures", see Figure 1 for an example. As a result, it provides browsable concept trees with the ancestors and descendants of the found concept, the synonyms of the concept, its inferred relations, its OWL axioms, and the reference sets it belongs to, as well as a downloadable diagram of its defining relations.
There is still much work to be done regarding the adoption and implementation of SNOMED CT in organizations, however, it is recognized that this terminology will play a crucial role in the exploitation of health information.