Coverage of rare disease names in standard terminologies and implications for patients, providers, and research

Area of Science:

• Medical Informatics
• Health Data Standards
• Rare Diseases Research

Background:

• Research on rare diseases is challenging due to small patient populations.
• Electronic Health Records (EHR) offer a potential data source if rare disease patients can be accurately identified.
• Standardized medical terminologies are essential for data interoperability and patient cohort identification.

Purpose of the Study:

• To evaluate the coverage and specificity of different medical terminologies for identifying rare diseases in EHR data.
• To compare the mapping capabilities of UMLS, ICD-9-CM, ICD-10-CM, and SNOMED CT for a large set of rare diseases.
• To determine which terminology best facilitates the identification of rare disease patients for research.

Main Methods:

• Assessed the coverage of 6,519 rare disease names using the Unified Medical Language System (UMLS).
• Mapped diseases to ICD-9-CM, ICD-10-CM, and SNOMED CT, analyzing direct matches and broader matches via SNOMED CT to ICD mappings.
• Evaluated the specificity by counting unique disease-to-code mappings for each terminology.

Main Results:

• SNOMED CT demonstrated the highest coverage, matching 44% of rare diseases directly and providing broader matches for an additional 25% to ICD-10-CM.
• SNOMED CT also showed the highest specificity, with 85% of its codes mapping to a single disease.
• ICD-9-CM and ICD-10-CM had lower direct coverage (11% and 21% respectively) and specificity (62% and 73% respectively).

Conclusions:

• SNOMED CT provides superior coverage and specificity for identifying rare diseases in EHR data compared to ICD-9-CM and ICD-10-CM.
• The use of SNOMED CT can significantly enhance the ability to identify patient cohorts for rare disease research.
• Improved patient identification through robust terminologies like SNOMED CT can advance rare disease research and evidence-based care.