Model-Driven and Scalable Authoring

Drug extension content is not typically authored manually on a concept-by-concept basis. Instead, it is derived from existing national drug dictionaries using a structured and scalable process.

This approach is comparable to the development of terminologies such as LOINC, where content is generated based on combinations of well-defined attributes.

Process

Drug extension development involves several common challenges, including:

• Variations in ingredient representation across local dictionaries

• Limited granularity in source data (e.g. missing strength or form details)

• Rounding rules for strength normalization

• Non-normalized strength expressions, including numerator/denominator inconsistencies

Addressing these challenges requires careful normalization and alignment with the SNOMED CT concept model.

Introduction

National drug extensions enable the representation of real-world medicinal products within the SNOMED CT drug model, complementing the international semantic backbone with country-specific content such as authorized products and packaged medicines.

The development of a drug extension requires adherence to a set of principles that ensure:

• alignment with the SNOMED CT concept model

• consistency across international and national content

• scalability through semi-automated processes

• interoperability with local drug dictionaries and regulatory sources

This chapter outlines the key principles for both overall extension strategy and authoring processes.

Drug extensions are built on top of the SNOMED CT International Edition and must align with its concept model and hierarchy. The international model provides the core semantic structure through concepts such as:

• substances

• medicinal products (MP)

• medicinal product forms (MPF)

• clinical drugs (CD)

In some cases, national extensions may need to create concepts that logically belong to the international model, such as:

• new substances

• MPs, MPFs, or CDs not yet present

These concepts may subsequently be promoted to the International Edition following review by SNOMED International.

The creation of international-level concepts by National Release Centers (NRCs) introduces coordination challenges, particularly where multiple countries may require the same concepts.

To reduce duplication and competition:

• SNOMED International is taking a more proactive role in creating substance concepts, for example by aligning with regulatory sources such as the European Medicines Agency (EMA)

• The creation of Medicinal Products (MPs) is also planned centrally

• Requests for MPFs and CDs can be submitted by NRCs and incorporated into the International Edition

This approach supports consistency, reduces duplication of effort, and ensures a shared foundation across extensions.

A clear distinction should be maintained between:

• International content: abstract, reusable concepts supporting interoperability

• National extension content: real-world products, packages, and regulatory-specific information

National extensions should:

• reuse international concepts wherever possible

• only create new concepts when necessary

• ensure that extension concepts are logically aligned with the international model

SNOMED CT Extension Format – Drug Content in RF2

Overview

SNOMED CT drug content, whether part of the International Edition or a national drug extension, is distributed using Release Format 2 (RF2). RF2 provides a structured, tabular representation of terminology components, enabling consistent storage, exchange, and processing of SNOMED CT data.

In RF2, a medicinal product is not represented as a single record, but as a set of coordinated entries across multiple files:

• the Concept file identifies the drug concept

• the Description file provides human-readable names

• the Relationship file defines the meaning of the drug

• Reference sets (refsets) provide additional context such as language preferences and mappings

Each row in these files represents a single component, and together they form a computable definition of a medicinal product.

Consider the following real-world product:

Brand X amoxicillin 500 mg oral capsule (real clinical drug)

In RF2, this product is reconstructed from multiple components across files.

The Concept file contains one row per concept. For a drug, this row represents the existence of the medicinal product.

In drug context:

• This row represents the Real Clinical Drug (RCD) concept

• The concept itself has no meaning without descriptions and relationships

• It acts as the anchor identifier for all related information

The Description file provides the human-readable terms associated with the concept.

In drug context:

• The Fully Specified Name (FSN) provides a precise and unambiguous description of the medicinal product

• Synonyms provide alternative terms used in prescribing, dispensing, and user interfaces

The acceptability of synonyms (e.g. which term is preferred for display) is not defined in the Description file itself. Instead, it is specified in a Language Reference Set, which indicates:

• which synonym is the preferred term

• which synonyms are acceptable alternatives

The Relationship file represent the defining properties (the definition) of the drug using structured relationships.

Each row represents a single relationship between concepts.

In drug context:

These relationships define the clinical meaning of the drug:

• what substance it contains

• how it is administered

• how strong it is

In drug context:

These relationships distinguish this concept as a Real Clinical Drug (RCD):

• same clinical definition as other amoxicillin 500 mg capsules

• but a specific branded and authorized product

Now consider:

BrandX amoxicillin 500 mg oral capsule box of 24 (real clinical drug package)

This concept includes additional packaging relationships:

In drug context:

• These relationships describe how the product is packaged and supplied

• Combined with product identity, this forms a Real Clinical Drug Package (RPCD)

In addition to SNOMED CT content, extensions often include mapping artifacts that link SNOMED CT concepts to local drug dictionary identifiers.

These mappings support:

• interoperability between systems

• migration from legacy terminologies

• integration with national regulatory and reimbursement systems

Reference sets provide additional, non-defining information.

In drug context:

• Links the SNOMED CT concept to a local drug dictionary or regulatory identifier

• Supports integration with pharmacy, billing, and regulatory systems