Mapping Structured EHR Ophthalmic Exam Data to Standard Vocabularies

Abstract

Purpose : Standardization of ophthalmology data in electronic health records (EHRs) by the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) is challenging due to differences in EHR implementations and potential coverage gaps in the OMOP CDM. In this study, we examined the concept coverage of two Cerner Millennium EHR implementations by the OMOP CDM.

Methods : Data elements for two ophthalmology modules, from the default Cerner Model Experience (Kansas City, MO) and a localized Cerner implementation at a large academic hospital, were classified into eight subject categories and mapped to the semantically closest standard concept in the OMOP CDM. Mappings were performed using the Athena web application, which compiles standard concepts listed in the OMOP CDM. Mappings were categorized as exact if the standard concept precisely represented the source data, wider if the mapping involved loss of information, narrower if there was additional information that was potentially inaccurate, or unmatched if no equivalent standard concept existed in the OMOP CDM.

Results : There were 409 data elements across six categories in the default Cerner module, and 947 data elements across eight categories in the local Cerner module (Figure 1a). Exact mappings existed for only 24.9% (102/409) and 18.3% (173/947) of data elements in each respective module (Figure 1b). No mappings were possible for 19.3% (79/409) of default Cerner and 26.4% (250/947) of local Cerner source data. More than half of the mappings in both Cerner modules suffered from information loss due to coverage gaps in the OMOP CDM that spanned all subject categories (Figure 2a, b). This was primarily due to insufficient granularity in OMOP standard concepts (91.7% and 93.0% of wider mappings in the default and local Cerner modules, respectively).

Conclusions : The OMOP CDM has considerable coverage gaps for all categories of the eye examination, which inhibits standardization of clinically relevant ophthalmology data. Advancing towards data standardization would therefore involve revising existing concepts or creating new standard concepts to increase clinical granularity in the OMOP CDM.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.

 

Figure 1. (a) Distribution of subject category of data elements (b) Distribution of source data mappings by match type for each Cerner module

 

Figure 2. Sankey diagrams of match type breakdown for (a) default Cerner module and (b) local Cerner module