Obtaining a multi-organization OMOP CDM repository from two heterogeneous EHR ecosystems: a flexible methodology based on Detailed Clinical Models

Abstract Background Standardized repositories of real-world data provide a mechanism for semantic convergence of data from different heterogeneous organizations for secondary use. However, it is common for these repositories to be populated from locally designed information systems, which generates inefficient processes that are not reusable in other organizations or projects. Objective Design and application of a methodology based on the Detailed Clinical Models (DCM) paradigm for allowing the flexible and harmonized implementation of a real world-data (RWD) repository from two technically and organizationally heterogeneous EHR ecosystems. Material and methods First, the DCM paradigm was used for the design of common information objects. Second, a set of clinical archetypes were implemented according to ISO 13606 standard. Third, an OMOP CDM muti-organization repository was implemented for COVID-19 research. Finally, the quality of the data obtained with the aforementioned process was evaluated. Results The main result was the proposal of a methodology for obtaining harmonized EHR-derived datasets using clinical archetypes as a convergence mechanism between local organization-dependent EHR designs. In addition, the application of this method also generated a set of reusable implementation results: (1) the catalog of clinical archetypes, (2) the definition of the transformation process from the archetypes to the OMOP CDM model, and (3) the EHR-derived dataset obtained. Conclusions The flexibility of the methodology made possible the adoption by two digitally mature tertiary hospitals, without altering the platforms already in place. Likewise, the method is agnostic to organizations, to persistence and exchange standards to be obtained, and to application health conditions. Therefore, it can be concluded that the implemented methodology constitutes an innovative and transferable solution to obtain RWD datasets in an efficient, flexible and reusable way.