Background: Patient data from general practices is already used for many types of epidemiological research and increasingly, primary care systems to facilitate randomized clinical trials. The EU funded project TRANSFoRm aims to create a Learning Healthcare System at a European level that is able to support all types of research using primary care data, to recruit patients and follow patients in clinical studies and to improve diagnosis and therapy. The implementation of such a Learning Healthcare System needs an information model for clinical research (CRIM), as an informational backbone to integrate aspects of primary care with clinical trials and database searches. Methods: Workflow descriptions and corresponding data objects of two clinical use cases (Gastro-Oesophageal Reflux Disease and Type 2 Diabetes) were described in UML activity diagrams. The components of activity diagrams were mapped to information objects of PCROM (Primary Care Research Object Model) and BRIDG (Biomedical Research Integrated Domain Group) and evaluated. The class diagram of PCROM was adapted to comply with workflow descriptions. Results: The suitability of PCROM, a primary care information model already used for clinical trials, to act as an information model for TRANSFoRm was evaluated and resulted in its extension with 14 new information object types, two extensions of existing objects and the introduction of two new high-ranking concepts (CARE area and ENTRY area). No PCROM component was redundant. Our result illustrates that in primary care based research an important but underestimated portion of research activity takes place in the area of care (e.g. patient consultation, screening, recruitment and response to adverse events). The newly introduced CARE area for care-related research activities accounts for this shift and includes Episode of Care and Encounter as two new basic elements. In the ENTRY area different aspects of data collection were combined, including data semantics for observations, assessment activities, intervention activities and patient reporting to enable case report form (CRF) based data collection combined with decision support. Conclusions: Research with primary care data needs an extended information model that covers research activities at the care site which are characteristic for primary care based research and the requirements of the complicated data collection processes.
Bibliographical noteFunding Information:
We thank Anna Nixon Andreasson and Lars Agreus for their input in the workflow description of the GORD use case, Hilde Bastiaens, Paul van Royen, and Peter Leysen for their input in the workflow description of the Diabetes use case and Vasa Curcin and Mark McGilchrist for reviewing the manuscript. Special thanks to Jean-Francois Ethier for insights into CDIM. TRANSFoRm is partially funded by the European Commission - DG INFSO (FP7 247787).