Data model considerations for clinical effectiveness researchers

Med Care. 2012 Jul;50 Suppl(0):S60-7. doi: 10.1097/MLR.0b013e318259bff4.


Introduction: Growing adoption of electronic health records and increased emphasis on the reuse and integration of clinical care and administration data require a robust informatics infrastructure to inform health care effectiveness in real-world settings. The Scalable Architecture for Federated Translational Inquiries Network (SAFTINet) was one of 3 projects receiving Agency for Healthcare Quality and Research funds to create a scalable, distributed network to support Comparative Effectiveness Research. SAFTINet's method of extracting and compiling data from disparate entities requires the use of a shared common data model. DATA MODELS: Focusing on the needs of CER investigators, in addition to other project considerations, we examined the suitability of several data models. Data modeling is the process of determining which data elements will be stored and how they will be stored, including their relationships and constraints. Addressing compromises between complexity and usability is critical to modeling decisions.

Case study: The SAFTINet project provides the case study for describing data model evaluation. A sample use case defines a cohort of asthma subjects that illustrates the need to identify patients by age, diagnoses, and medication use while excluding those with diagnoses that may often be misdiagnosed as asthma.

Discussion: The SAFTINet team explored several data models against a set of technical and investigator requirements to select a data model that best fit its needs and was conducive to expansion with new research requirements. Although SAFTINet ultimately chose the Observation Medical Outcomes Partnership common data model, other valid options exist and prioritization of requirements is dependent upon many factors.

Publication types

  • Research Support, N.I.H., Extramural
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Comparative Effectiveness Research / methods*
  • Data Mining*
  • Humans
  • Information Management / methods
  • Information Storage and Retrieval
  • Medical Informatics*
  • Medical Record Linkage
  • Models, Statistical*