CLINICAL CONTENT COVERAGE EVALUATION ON ICD-11: THE THIRD IN SERIES
Embargo until
Date
2019-08-26
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Johns Hopkins University
Abstract
Background and Objective: To better help store and render computer-based patient records in an accurate and efficient way, standardized and robust clinical classifications and taxonomies should be developed. To investigate whether major coding systems can solve the issue and help develop a reliable approach for evaluation, the authors conducted the study to look into the content coverage of major coding systems.
Methods: Clinical texts from four medical centers were retrieved and organized. A total of 14, 247 words were parse and 3061 clinical concepts were identified. The concepts were clustered into five different semantic groups. Each concept was then coded by ICD-11, SNOMED CT International, and Monarch Disease Ontology (MONDO). The score scaling is described here: 0 = no match, 1 = fair match. 2 = complete match.
Results: SNOMED CT International had an overall score of 1.55 and stood out in every category, except for Diagnoses, where ICD-11 demonstrated 1.86 in this category. ICD-11 had an overall score of 1.10 and showed improvements in every category, compared to ICD-9-CM and ICD-10-CM. The overall score MONDO is 0.43 but it scored 1.28 in Diagnoses.
Conclusions: No present classification systems captured all the concepts. SNOMED CT International has the highest overall score. ICD-11 performs better than SNOMED CT International in representing Diagnoses and performs better in every category than ICD-9-CM and ICD-10-CM. Considering the high score of ICD-11 in diagnoses and its fast internet-based developing environment, ICD-11 could be our next step towards improving healthcare information exchange (HIE).
Description
Keywords
Electronic Health Record, clinical classifications, taxonomy, evaluation,