305 - Mapping the T1D Exchange Data Specification to the OMOP Common Data Model – Increasing Data Interoperability and Reusability
Monday, April 28, 2025
7:00am – 9:15am HST
Publication Number: 305.6003
Eric WIlliams, Ann & Robert H. Lurie Children's Hospital of Chicago, Houston, TX, United States; Brent Lockee, Children's Mercy Kansas City, Ka, MO, United States; Mark A. Clements, Children's Mercy Kansas City, Kansas City, MO, United States; Juan C. Espinoza, Ann & Robert H. Lurie Children's Hospital of Chicago, Chicago, IL, United States
Research Scientist Ann & Robert H. Lurie Children's Hospital of Chicago Houston, Texas, United States
Background: The T1D Exchange (T1DX) is a nonprofit organization dedicated to improving care for individuals with type 1 diabetes (T1D). In 2016, it established the T1DX Quality Improvement Collaborative (T1DX-QI) to accelerate QI interventions by leveraging a common data and intervention framework. The T1DX Data Specification (T1DX-DS) has evolved over time and now includes more than 120 distinct variables, categorized into seven groups: 1) Patients, 2) Providers, 3) Encounters, 4) Observations, 5) Conditions, 6) Medications, and 7) Diabetes. Common Data Models (CDM) are powerful tools that enable multi-organization collaborative research. The Observational Medical Outcomes Partnership (OMOP) CDM has become one of the most widely adopted CDM in healthcare research. A number of robust, open source tools already exist for OMOP, and many organizations already host their data in OMOP. Mapping to the T1DX-DS to OMOP would enable the T1DX-QI to leverage these tools and infrastructure to accelerate research. Objective: Develop a mapping between T1DX-DS and OMOP to facilitate data transformation. Design/Methods: The documentation for both T1DX-DS and OMOP was carefully reviewed, including data tables, variable descriptions, data types, data formats, and allowed values for each field. We then created mappings that highlighted correspondences and discrepancies between T1DX-DS and OMOP. Following this, we created 2 entity relationship diagrams (ERDs) to illustrate the relationships between OMOP and T1DX-DS Core Specification (Figure 1) and T1DX-DS Diabetes Specification (Figure 2). Results: T1DX-DS consists of six Core Specification tables and five Diabetes Specification tables; OMOP has 39 tables. The six Core Specification tables map to 10 OMOP Tables, while the five Diabetes Specification tables map to five OMOP tables (both also include the OMOP CONCEPT table). Many T1DX-DS tables combine variables from multiple OMOP tables. Several values in the Diseases Specification are not standard in EHRs and require custom collection; this results in unclear mapping to OMOP tables.
Conclusion(s): T1DX-DS and OMOP are compatible and mapping is feasible between them, though there is ambiguity on the best mapping for the unique diabetes specification variables; these may be generically mapped to the OBSERVATION table in OMOP, or it may be necessary to develop a novel table in OMOP. Further collaboration with experts on both will be necessary to resolve the ambiguity. Additional technical development and testing will be required before final documentation, widespread adoption, and training.
T1DX-DS Core Specification to OMOP CDM: Table Mapping Overview T1DX-DS Core to OMOP Table.jpegFigure 1 presents a table-only view of the mappings between T1DX-DS Core Specification and the OMOP CDM. In this figure, black tables represent T1DX-DS, while red tables denote OMOP. The tables positioned to the left of each T1DX-DS table are their corresponding OMOP tables.
T1DX-DS Diabetes Specification to OMOP: Table Mapping Overview T1DX-DS Diabetes to OMOP.jpegFigure 2 illustrates the detailed relationship between T1DX-DS Diabetes Data Specifications and OMOP tables. The colored lines are assigned to each T1DX-DS Diabetes Data Specification table. Please view the relationship diagram key in the figure for reference.