gms | German Medical Science

Background: The adoption of the laboratory coding system LOINC, Logical Observation Identifier Names and Codes, is growing worldwide. In Germany, interest in using LOINC has largely increased recently due to the Medical Informatics Initiative (MII) [1].

LOINC nowadays includes almost 100,000 different codes, a large set that easily becomes unmanageable for search and mapping. Thus, mapping local codes to LOINC based on smaller, locally harmonized subsets of LOINC is more efficient and can effectively reduce costs [2]. Therefore, Regenstrief Institute defined a “Universal Lab Order” subset (n = 1522) that shall account for 95 % of testing in US laboratories [3] by applying the Pareto principle as in [4].

Similarly, the MII is working on a set of 300 common laboratory tests based on previously analyzed test frequencies at five locations. Each test allows multiple LOINC codes for representation; adding up to currently 935 codes that shall cover 80% of lab orders [5].

When defining specific subsets their suitability needs to be guaranteed to ensure valid usage; thus, evaluation is essential. Therefore, a visual presentation using LOINC's internal part hierarchies can provide valuable insights into the subsets' properties and their coverage regarding different areas, particularly when aggregated on various hierarchical levels.

Methods: In previous efforts the unreleased LOINC part hierarchies were used to create a comprehensive OWL-ontology and – based thereupon – a visualization of hierarchical structures was developed, dynamically generated by an embedded reasoner for a user-defined set of LOINC codes [6]. The authors extended the proposed tooling to allow an integrated comparison of two LOINC subsets with possible overlap. The subset codes are internally added as Individuals of the corresponding LOINC code classes in the ontology. So, each code or hierarchical class can be evaluated for any Individuals belonging to it and their referenced subset(s). A color-coded representation of subset membership is implemented throughout the application. The aforementioned LOINC subsets were used exemplarily to test the tool's capacities for subset comparison and evaluation.

Results: A visual presentation of LOINC subsets' similarities and differences is achieved, both for individual codes and – most importantly – for aggregating hierarchy superclasses of various LOINC axis or part categories.

For the application example these characteristics were revealed: The Regenstrief subset proved to be far more varied while the MII subset shows comparably low variety in utilized System parts, even on higher aggregation levels (1 vs. 5 and 5 vs. 20 different classes for the two highest levels). Furthermore, it often specifies multiple subclass codes of one class referenced by the Regenstrief set (e.g. seven slightly different tests for two less granular bilirubin in urine measurements). An expected fundamental disparity in usage of Property categories, due to regional preference of molar- (substance) over mass-based reporting [7], could only be observed partially.

Conclusion: The presented tool facilitates ontology-based, visual comparisons of LOINC subset properties, thereby providing unique insights for evaluation. Due to the stated importance of subset definition, the tool's usage in various national and local projects can be considered, including Austrian, Belgian, Dutch and IHE subsets [4], [8], [9], [10].

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.