Term sets: A transparent and reproducible representation of clinical code sets

Area of Science:

• Health Informatics
• Biomedical Research Methodology

Background:

• Clinical code sets are crucial for research with electronic healthcare data.
• Current methods for creating code sets limit research reproducibility and adherence to FAIR principles.

Purpose of the Study:

• To introduce 'term sets' as a FAIR-aligned alternative to traditional code sets.
• To demonstrate the utility and benefits of term sets for research transparency and reusability.

Main Methods:

• Described a new code set representation using natural language inclusion and exclusion terms (term sets).
• Formally proved the equivalence between code sets and term sets.
• Demonstrated utility by converting published code sets into term sets.

Main Results:

• Thirty-one code sets were converted into term sets.
• Term sets were, on average, 74% the size of their original code sets.
• Identified deficiencies in original code sets when four term sets were larger.

Conclusions:

• Term sets concisely represent any code set, improving research examination and reuse.
• Term sets enhance research reproducibility, resist terminology changes, and are less error-prone.
• Open-source software is available to support researchers using term sets.