Country / Region
EMEA
Tags
Clinical Practice, Implementation, Tooling
In April 2024, Hospital Clinic de Barcelona implemented a SNOMED CT-coded Health Problem List supported by terminology services and a natural language processing (NLP) tool to assist clinicians in real-time coding. The system's architecture comprises a Hybrid Coder (semantic and lexical), Reviewer, Manager, and a Terminology Service.
Initial results indicate that clinicians accepted 74.2% of the tool's suggested codes directly. An additional 23.3% required intervention from a professional medical coder, while only 2.5% of entries remained uncoded. The original catalogue included 7,364 SNOMED CT concepts, which has since expanded to 10,267 in response to clinical needs.
Key success factors included user-centered design tailored to various clinical environments (emergency, inpatient, outpatient), ensuring minimal disruption to clinical workflows. A governance model involving an Editorial Board of clinical experts helped resolve terminology issues. The use of a dedicated language reference set enabled appropriate granularity and assignment of preferred terms.
A major advantage of the implementation is the immediate availability of structured clinical data for both primary and secondary use, reinforcing user engagement by demonstrating tangible benefits early in the process.
Future developments include the integration of ORPHA codes to support the documentation of rare diseases. Nonetheless, achieving concept standardization remains a challenge, particularly in innovative care settings where clinical needs may surpass the specificity offered by existing terminologies.
Description
This work presents a novel experience in the implementation of multi-terminology services at a tertiary university hospital, used for the standardization of health problems from the very beginning of the care process.
Scope
SNOMED CT was selected as the reference terminology to record health problems due to its international recognition for coding from healthcare professionals' natural language expressions and its utility in linking with other classifications and local organizational catalogs. It was chosen over other vocabularies such as ICD because it is more comprehensive, clinically focused, and customizable, allowing domain-specific work, different levels of granularity, and integration with clinical guidelines and decision support systems through its hierarchical and transversal relationships.
How SNOMED CT will be used
The Clinician Workstation has a specific widget where the physician writes the Health Problem in natural language. Immediately the system offers a list of up to five candidate-SNOMED CT concepts coded by a Hybrid Coder (semantic and lexical). The descriptions viewed in the widget are the SNOMED CT Preferred terms. Clinician selects the desired one or chooses the option “none of them”. In the last case the textual expression remains uncoded until the Coder Expert reviews and solves it.
Why SNOMED CT will be used
Contact
