Clinical terminology mapping to improve semantic interoperability in Australian General Practice

Julie Gordon, Graeme Miller, Carmen Wong, Julia Leahy, Helena Britt, Richard Madden

Background: Clinical terminologies allow clinicians to record standardised clinical data within electronic health records (EHRs). The University of Sydney develops the ICPC-2 PLUS terminology, used in EHRs for general practice, Indigenous health care and community health. Clinical data transfer between health sectors often requires maps between terminologies. Vendors using ICPC-2 PLUS requested a map from ICPC-2 PLUS to the Australian version of SNOMED CT (called SNOMED CT-AU) to promote interoperability.

Aims: To create a one-to-one map of best fit from ICPC-2 PLUS symptoms and diagnosis terms to SNOMED CT-AU and assess quality of mapping results.

Methods: Two mappers independently used the CSIRO’s Snapper tool to generate automated maps, which they reviewed and assessed for quality, creating manual maps where required. Review by two map leads (one GP and one senior health information manager) resolved discrepancies.

Results: Using ICPC-2 PLUS April 2020 and SNOMED CT-AU 31 March 2020, 5,629 of 5,880 ICPC-2 PLUS terms were mapped to SNOMED CT-AU (95.7%). Only 77.7% represented acceptable matches. Remaining maps were rated as ‘best fit’ (meaning lost during mapping ) (12.7%) or precoordination existed affecting mapping (5.3%). No match was found for remaining 4.3%.

Conclusions: Substantial variance existed in the quality of mapping results. Use of maps with lower accuracy ratings create risk which must be carefully assessed during implementation. Acceptable risk levels depend on map use—clinical use requires higher accuracy, while lower match levels might be tolerated for statistical purposes.