Observational Health Data Sciences and Informatics

The Standardized Vocabularies are constructed with a few principles in mind. Not every principle can be executed to perfection, but it represents a general motivation and direction of the ongoing improvement and development process:

• Comprehensive Coverage of all clinical entities in a domain. The Standardized Vocabularies intend to capture every event that is relevant to the patient's clinical experience (e.g. Conditions, Procedures, Exposures to Drug, etc.) and some of the administrative artifacts of the healthcare system (e.g. Visits, Care Sites, etc.). This means that in many cases a single Vocabulary is not sufficient to cover a Domain. For example, RxNorm, which is the source for Ingredients and drug products in the Drug Domain, does not cover products not marketed in the USA, and therefore Concepts from other Vocabularies (ATC, Multilex) have to be used to complement RxNorm in covering the Domain.
• Avoidance of Duplication: Only one Standard Concept per clinical entity. This is easier if the Concepts of a Domain are very precisely defined (e.g. the Drug domain). For Domains where the Concepts have nuanced meanings (e.g. Condition, Procedure) this is a more challenging task.
• Meaningful and high-quality hierarchy involving all concepts of a domain. This entails the solution of two separate problems:
  • Each Concepts should have one or more classifications (bottom up).
  • Each Classification should contain all the relevant Concepts (top down).

The top-down problem has the additional difficulty in that Classifications are not always generally agreed on. For example, whether diabetic complications are part of the Condition Diabetes or are separate is differently answered by the respective SNOMED and ICD-9 hierarchies.

• Life cycle keeping data representation up to date but supporting the processing of deprecated Concepts. This means that for each deprecated Concept a replacement Concept has to be assigned capturing the meaning of the clinical entity.
• Precise unique Domain assignment for each Standard Concept, and combination of Domains for “dirtier” non-standard Source Concepts.