Facilitating pre-operative assessment guidelines representation using SNOMED CT

To investigate whether SNOMED CT covers the terms used in pre-operative assessment guidelines, and if necessary, how the measured content coverage can be improved.

Pre-operative assessment guidelines were retrieved from the websites of (inter)national anesthesia-related societies. The recommendations in the guidelines were rewritten to "IF condition THEN action" statements to facilitate data extraction. Terms were extracted from the IF-THEN statements and mapped to SNOMED CT. Content coverage was measured by using three scores: no match, partial match and complete match. Non-covered concepts were evaluated against the SNOMED CT editorial documentation.

From 6 guidelines, 133 terms were extracted, of which 71% (n=94) completely matched with SNOMED CT concepts. Disregarding the vague concepts in the included guidelines SNOMED CT's content coverage was 89%. Of the 39 non-completely covered concepts, 69% violated at least one of SNOMED CT's editorial principles or rules. These concepts were categorized based on four categories: non-reproducibility, classification-derived phrases, numeric ranges, and procedures categorized by complexity.

Guidelines include vague terms that cannot be well supported by terminological systems thereby hampering guideline-based decision support systems. This vagueness reduces the content coverage of SNOMED CT in representing concepts used in the pre-operative assessment guidelines. Formalization of the guidelines using SNOMED CT is feasible but to optimize this, first the vagueness of some guideline concepts should be resolved and a few currently missing but relevant concepts should be added to SNOMED CT.